1 /* SPDX-License-Identifier: GPL-2.0 */ 2 #ifndef _LINUX_FS_H 3 #define _LINUX_FS_H 4 5 #include <linux/linkage.h> 6 #include <linux/wait_bit.h> 7 #include <linux/kdev_t.h> 8 #include <linux/dcache.h> 9 #include <linux/path.h> 10 #include <linux/stat.h> 11 #include <linux/cache.h> 12 #include <linux/list.h> 13 #include <linux/list_lru.h> 14 #include <linux/llist.h> 15 #include <linux/radix-tree.h> 16 #include <linux/rbtree.h> 17 #include <linux/init.h> 18 #include <linux/pid.h> 19 #include <linux/bug.h> 20 #include <linux/mutex.h> 21 #include <linux/rwsem.h> 22 #include <linux/mm_types.h> 23 #include <linux/capability.h> 24 #include <linux/semaphore.h> 25 #include <linux/fcntl.h> 26 #include <linux/fiemap.h> 27 #include <linux/rculist_bl.h> 28 #include <linux/atomic.h> 29 #include <linux/shrinker.h> 30 #include <linux/migrate_mode.h> 31 #include <linux/uidgid.h> 32 #include <linux/lockdep.h> 33 #include <linux/percpu-rwsem.h> 34 #include <linux/workqueue.h> 35 #include <linux/delayed_call.h> 36 #include <linux/uuid.h> 37 #include <linux/errseq.h> 38 39 #include <asm/byteorder.h> 40 #include <uapi/linux/fs.h> 41 42 struct backing_dev_info; 43 struct bdi_writeback; 44 struct bio; 45 struct export_operations; 46 struct hd_geometry; 47 struct iovec; 48 struct kiocb; 49 struct kobject; 50 struct pipe_inode_info; 51 struct poll_table_struct; 52 struct kstatfs; 53 struct vm_area_struct; 54 struct vfsmount; 55 struct cred; 56 struct swap_info_struct; 57 struct seq_file; 58 struct workqueue_struct; 59 struct iov_iter; 60 struct fscrypt_info; 61 struct fscrypt_operations; 62 63 extern void __init inode_init(void); 64 extern void __init inode_init_early(void); 65 extern void __init files_init(void); 66 extern void __init files_maxfiles_init(void); 67 68 extern struct files_stat_struct files_stat; 69 extern unsigned long get_max_files(void); 70 extern unsigned int sysctl_nr_open; 71 extern struct inodes_stat_t inodes_stat; 72 extern int leases_enable, lease_break_time; 73 extern int sysctl_protected_symlinks; 74 extern int sysctl_protected_hardlinks; 75 76 typedef __kernel_rwf_t rwf_t; 77 78 struct buffer_head; 79 typedef int (get_block_t)(struct inode *inode, sector_t iblock, 80 struct buffer_head *bh_result, int create); 81 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset, 82 ssize_t bytes, void *private); 83 84 #define MAY_EXEC 0x00000001 85 #define MAY_WRITE 0x00000002 86 #define MAY_READ 0x00000004 87 #define MAY_APPEND 0x00000008 88 #define MAY_ACCESS 0x00000010 89 #define MAY_OPEN 0x00000020 90 #define MAY_CHDIR 0x00000040 91 /* called from RCU mode, don't block */ 92 #define MAY_NOT_BLOCK 0x00000080 93 94 /* 95 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond 96 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open() 97 */ 98 99 /* file is open for reading */ 100 #define FMODE_READ ((__force fmode_t)0x1) 101 /* file is open for writing */ 102 #define FMODE_WRITE ((__force fmode_t)0x2) 103 /* file is seekable */ 104 #define FMODE_LSEEK ((__force fmode_t)0x4) 105 /* file can be accessed using pread */ 106 #define FMODE_PREAD ((__force fmode_t)0x8) 107 /* file can be accessed using pwrite */ 108 #define FMODE_PWRITE ((__force fmode_t)0x10) 109 /* File is opened for execution with sys_execve / sys_uselib */ 110 #define FMODE_EXEC ((__force fmode_t)0x20) 111 /* File is opened with O_NDELAY (only set for block devices) */ 112 #define FMODE_NDELAY ((__force fmode_t)0x40) 113 /* File is opened with O_EXCL (only set for block devices) */ 114 #define FMODE_EXCL ((__force fmode_t)0x80) 115 /* File is opened using open(.., 3, ..) and is writeable only for ioctls 116 (specialy hack for floppy.c) */ 117 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100) 118 /* 32bit hashes as llseek() offset (for directories) */ 119 #define FMODE_32BITHASH ((__force fmode_t)0x200) 120 /* 64bit hashes as llseek() offset (for directories) */ 121 #define FMODE_64BITHASH ((__force fmode_t)0x400) 122 123 /* 124 * Don't update ctime and mtime. 125 * 126 * Currently a special hack for the XFS open_by_handle ioctl, but we'll 127 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon. 128 */ 129 #define FMODE_NOCMTIME ((__force fmode_t)0x800) 130 131 /* Expect random access pattern */ 132 #define FMODE_RANDOM ((__force fmode_t)0x1000) 133 134 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */ 135 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000) 136 137 /* File is opened with O_PATH; almost nothing can be done with it */ 138 #define FMODE_PATH ((__force fmode_t)0x4000) 139 140 /* File needs atomic accesses to f_pos */ 141 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000) 142 /* Write access to underlying fs */ 143 #define FMODE_WRITER ((__force fmode_t)0x10000) 144 /* Has read method(s) */ 145 #define FMODE_CAN_READ ((__force fmode_t)0x20000) 146 /* Has write method(s) */ 147 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000) 148 149 /* File was opened by fanotify and shouldn't generate fanotify events */ 150 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000) 151 152 /* File is capable of returning -EAGAIN if I/O will block */ 153 #define FMODE_NOWAIT ((__force fmode_t)0x8000000) 154 155 /* 156 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector 157 * that indicates that they should check the contents of the iovec are 158 * valid, but not check the memory that the iovec elements 159 * points too. 160 */ 161 #define CHECK_IOVEC_ONLY -1 162 163 /* 164 * Attribute flags. These should be or-ed together to figure out what 165 * has been changed! 166 */ 167 #define ATTR_MODE (1 << 0) 168 #define ATTR_UID (1 << 1) 169 #define ATTR_GID (1 << 2) 170 #define ATTR_SIZE (1 << 3) 171 #define ATTR_ATIME (1 << 4) 172 #define ATTR_MTIME (1 << 5) 173 #define ATTR_CTIME (1 << 6) 174 #define ATTR_ATIME_SET (1 << 7) 175 #define ATTR_MTIME_SET (1 << 8) 176 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */ 177 #define ATTR_ATTR_FLAG (1 << 10) 178 #define ATTR_KILL_SUID (1 << 11) 179 #define ATTR_KILL_SGID (1 << 12) 180 #define ATTR_FILE (1 << 13) 181 #define ATTR_KILL_PRIV (1 << 14) 182 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */ 183 #define ATTR_TIMES_SET (1 << 16) 184 #define ATTR_TOUCH (1 << 17) 185 186 /* 187 * Whiteout is represented by a char device. The following constants define the 188 * mode and device number to use. 189 */ 190 #define WHITEOUT_MODE 0 191 #define WHITEOUT_DEV 0 192 193 /* 194 * This is the Inode Attributes structure, used for notify_change(). It 195 * uses the above definitions as flags, to know which values have changed. 196 * Also, in this manner, a Filesystem can look at only the values it cares 197 * about. Basically, these are the attributes that the VFS layer can 198 * request to change from the FS layer. 199 * 200 * Derek Atkins <warlord@MIT.EDU> 94-10-20 201 */ 202 struct iattr { 203 unsigned int ia_valid; 204 umode_t ia_mode; 205 kuid_t ia_uid; 206 kgid_t ia_gid; 207 loff_t ia_size; 208 struct timespec ia_atime; 209 struct timespec ia_mtime; 210 struct timespec ia_ctime; 211 212 /* 213 * Not an attribute, but an auxiliary info for filesystems wanting to 214 * implement an ftruncate() like method. NOTE: filesystem should 215 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL). 216 */ 217 struct file *ia_file; 218 }; 219 220 /* 221 * Includes for diskquotas. 222 */ 223 #include <linux/quota.h> 224 225 /* 226 * Maximum number of layers of fs stack. Needs to be limited to 227 * prevent kernel stack overflow 228 */ 229 #define FILESYSTEM_MAX_STACK_DEPTH 2 230 231 /** 232 * enum positive_aop_returns - aop return codes with specific semantics 233 * 234 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has 235 * completed, that the page is still locked, and 236 * should be considered active. The VM uses this hint 237 * to return the page to the active list -- it won't 238 * be a candidate for writeback again in the near 239 * future. Other callers must be careful to unlock 240 * the page if they get this return. Returned by 241 * writepage(); 242 * 243 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has 244 * unlocked it and the page might have been truncated. 245 * The caller should back up to acquiring a new page and 246 * trying again. The aop will be taking reasonable 247 * precautions not to livelock. If the caller held a page 248 * reference, it should drop it before retrying. Returned 249 * by readpage(). 250 * 251 * address_space_operation functions return these large constants to indicate 252 * special semantics to the caller. These are much larger than the bytes in a 253 * page to allow for functions that return the number of bytes operated on in a 254 * given page. 255 */ 256 257 enum positive_aop_returns { 258 AOP_WRITEPAGE_ACTIVATE = 0x80000, 259 AOP_TRUNCATED_PAGE = 0x80001, 260 }; 261 262 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */ 263 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct 264 * helper code (eg buffer layer) 265 * to clear GFP_FS from alloc */ 266 267 /* 268 * oh the beauties of C type declarations. 269 */ 270 struct page; 271 struct address_space; 272 struct writeback_control; 273 274 /* 275 * Write life time hint values. 276 */ 277 enum rw_hint { 278 WRITE_LIFE_NOT_SET = 0, 279 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE, 280 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT, 281 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM, 282 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG, 283 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME, 284 }; 285 286 #define IOCB_EVENTFD (1 << 0) 287 #define IOCB_APPEND (1 << 1) 288 #define IOCB_DIRECT (1 << 2) 289 #define IOCB_HIPRI (1 << 3) 290 #define IOCB_DSYNC (1 << 4) 291 #define IOCB_SYNC (1 << 5) 292 #define IOCB_WRITE (1 << 6) 293 #define IOCB_NOWAIT (1 << 7) 294 295 struct kiocb { 296 struct file *ki_filp; 297 loff_t ki_pos; 298 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2); 299 void *private; 300 int ki_flags; 301 enum rw_hint ki_hint; 302 } __randomize_layout; 303 304 static inline bool is_sync_kiocb(struct kiocb *kiocb) 305 { 306 return kiocb->ki_complete == NULL; 307 } 308 309 /* 310 * "descriptor" for what we're up to with a read. 311 * This allows us to use the same read code yet 312 * have multiple different users of the data that 313 * we read from a file. 314 * 315 * The simplest case just copies the data to user 316 * mode. 317 */ 318 typedef struct { 319 size_t written; 320 size_t count; 321 union { 322 char __user *buf; 323 void *data; 324 } arg; 325 int error; 326 } read_descriptor_t; 327 328 typedef int (*read_actor_t)(read_descriptor_t *, struct page *, 329 unsigned long, unsigned long); 330 331 struct address_space_operations { 332 int (*writepage)(struct page *page, struct writeback_control *wbc); 333 int (*readpage)(struct file *, struct page *); 334 335 /* Write back some dirty pages from this mapping. */ 336 int (*writepages)(struct address_space *, struct writeback_control *); 337 338 /* Set a page dirty. Return true if this dirtied it */ 339 int (*set_page_dirty)(struct page *page); 340 341 int (*readpages)(struct file *filp, struct address_space *mapping, 342 struct list_head *pages, unsigned nr_pages); 343 344 int (*write_begin)(struct file *, struct address_space *mapping, 345 loff_t pos, unsigned len, unsigned flags, 346 struct page **pagep, void **fsdata); 347 int (*write_end)(struct file *, struct address_space *mapping, 348 loff_t pos, unsigned len, unsigned copied, 349 struct page *page, void *fsdata); 350 351 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */ 352 sector_t (*bmap)(struct address_space *, sector_t); 353 void (*invalidatepage) (struct page *, unsigned int, unsigned int); 354 int (*releasepage) (struct page *, gfp_t); 355 void (*freepage)(struct page *); 356 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter); 357 /* 358 * migrate the contents of a page to the specified target. If 359 * migrate_mode is MIGRATE_ASYNC, it must not block. 360 */ 361 int (*migratepage) (struct address_space *, 362 struct page *, struct page *, enum migrate_mode); 363 bool (*isolate_page)(struct page *, isolate_mode_t); 364 void (*putback_page)(struct page *); 365 int (*launder_page) (struct page *); 366 int (*is_partially_uptodate) (struct page *, unsigned long, 367 unsigned long); 368 void (*is_dirty_writeback) (struct page *, bool *, bool *); 369 int (*error_remove_page)(struct address_space *, struct page *); 370 371 /* swapfile support */ 372 int (*swap_activate)(struct swap_info_struct *sis, struct file *file, 373 sector_t *span); 374 void (*swap_deactivate)(struct file *file); 375 }; 376 377 extern const struct address_space_operations empty_aops; 378 379 /* 380 * pagecache_write_begin/pagecache_write_end must be used by general code 381 * to write into the pagecache. 382 */ 383 int pagecache_write_begin(struct file *, struct address_space *mapping, 384 loff_t pos, unsigned len, unsigned flags, 385 struct page **pagep, void **fsdata); 386 387 int pagecache_write_end(struct file *, struct address_space *mapping, 388 loff_t pos, unsigned len, unsigned copied, 389 struct page *page, void *fsdata); 390 391 struct address_space { 392 struct inode *host; /* owner: inode, block_device */ 393 struct radix_tree_root page_tree; /* radix tree of all pages */ 394 spinlock_t tree_lock; /* and lock protecting it */ 395 atomic_t i_mmap_writable;/* count VM_SHARED mappings */ 396 struct rb_root_cached i_mmap; /* tree of private and shared mappings */ 397 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */ 398 /* Protected by tree_lock together with the radix tree */ 399 unsigned long nrpages; /* number of total pages */ 400 /* number of shadow or DAX exceptional entries */ 401 unsigned long nrexceptional; 402 pgoff_t writeback_index;/* writeback starts here */ 403 const struct address_space_operations *a_ops; /* methods */ 404 unsigned long flags; /* error bits */ 405 spinlock_t private_lock; /* for use by the address_space */ 406 gfp_t gfp_mask; /* implicit gfp mask for allocations */ 407 struct list_head private_list; /* for use by the address_space */ 408 void *private_data; /* ditto */ 409 errseq_t wb_err; 410 } __attribute__((aligned(sizeof(long)))) __randomize_layout; 411 /* 412 * On most architectures that alignment is already the case; but 413 * must be enforced here for CRIS, to let the least significant bit 414 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON. 415 */ 416 struct request_queue; 417 418 struct block_device { 419 dev_t bd_dev; /* not a kdev_t - it's a search key */ 420 int bd_openers; 421 struct inode * bd_inode; /* will die */ 422 struct super_block * bd_super; 423 struct mutex bd_mutex; /* open/close mutex */ 424 void * bd_claiming; 425 void * bd_holder; 426 int bd_holders; 427 bool bd_write_holder; 428 #ifdef CONFIG_SYSFS 429 struct list_head bd_holder_disks; 430 #endif 431 struct block_device * bd_contains; 432 unsigned bd_block_size; 433 u8 bd_partno; 434 struct hd_struct * bd_part; 435 /* number of times partitions within this device have been opened. */ 436 unsigned bd_part_count; 437 int bd_invalidated; 438 struct gendisk * bd_disk; 439 struct request_queue * bd_queue; 440 struct backing_dev_info *bd_bdi; 441 struct list_head bd_list; 442 /* 443 * Private data. You must have bd_claim'ed the block_device 444 * to use this. NOTE: bd_claim allows an owner to claim 445 * the same device multiple times, the owner must take special 446 * care to not mess up bd_private for that case. 447 */ 448 unsigned long bd_private; 449 450 /* The counter of freeze processes */ 451 int bd_fsfreeze_count; 452 /* Mutex for freeze */ 453 struct mutex bd_fsfreeze_mutex; 454 } __randomize_layout; 455 456 /* 457 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache 458 * radix trees 459 */ 460 #define PAGECACHE_TAG_DIRTY 0 461 #define PAGECACHE_TAG_WRITEBACK 1 462 #define PAGECACHE_TAG_TOWRITE 2 463 464 int mapping_tagged(struct address_space *mapping, int tag); 465 466 static inline void i_mmap_lock_write(struct address_space *mapping) 467 { 468 down_write(&mapping->i_mmap_rwsem); 469 } 470 471 static inline void i_mmap_unlock_write(struct address_space *mapping) 472 { 473 up_write(&mapping->i_mmap_rwsem); 474 } 475 476 static inline void i_mmap_lock_read(struct address_space *mapping) 477 { 478 down_read(&mapping->i_mmap_rwsem); 479 } 480 481 static inline void i_mmap_unlock_read(struct address_space *mapping) 482 { 483 up_read(&mapping->i_mmap_rwsem); 484 } 485 486 /* 487 * Might pages of this file be mapped into userspace? 488 */ 489 static inline int mapping_mapped(struct address_space *mapping) 490 { 491 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root); 492 } 493 494 /* 495 * Might pages of this file have been modified in userspace? 496 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff 497 * marks vma as VM_SHARED if it is shared, and the file was opened for 498 * writing i.e. vma may be mprotected writable even if now readonly. 499 * 500 * If i_mmap_writable is negative, no new writable mappings are allowed. You 501 * can only deny writable mappings, if none exists right now. 502 */ 503 static inline int mapping_writably_mapped(struct address_space *mapping) 504 { 505 return atomic_read(&mapping->i_mmap_writable) > 0; 506 } 507 508 static inline int mapping_map_writable(struct address_space *mapping) 509 { 510 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ? 511 0 : -EPERM; 512 } 513 514 static inline void mapping_unmap_writable(struct address_space *mapping) 515 { 516 atomic_dec(&mapping->i_mmap_writable); 517 } 518 519 static inline int mapping_deny_writable(struct address_space *mapping) 520 { 521 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ? 522 0 : -EBUSY; 523 } 524 525 static inline void mapping_allow_writable(struct address_space *mapping) 526 { 527 atomic_inc(&mapping->i_mmap_writable); 528 } 529 530 /* 531 * Use sequence counter to get consistent i_size on 32-bit processors. 532 */ 533 #if BITS_PER_LONG==32 && defined(CONFIG_SMP) 534 #include <linux/seqlock.h> 535 #define __NEED_I_SIZE_ORDERED 536 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount) 537 #else 538 #define i_size_ordered_init(inode) do { } while (0) 539 #endif 540 541 struct posix_acl; 542 #define ACL_NOT_CACHED ((void *)(-1)) 543 #define ACL_DONT_CACHE ((void *)(-3)) 544 545 static inline struct posix_acl * 546 uncached_acl_sentinel(struct task_struct *task) 547 { 548 return (void *)task + 1; 549 } 550 551 static inline bool 552 is_uncached_acl(struct posix_acl *acl) 553 { 554 return (long)acl & 1; 555 } 556 557 #define IOP_FASTPERM 0x0001 558 #define IOP_LOOKUP 0x0002 559 #define IOP_NOFOLLOW 0x0004 560 #define IOP_XATTR 0x0008 561 #define IOP_DEFAULT_READLINK 0x0010 562 563 struct fsnotify_mark_connector; 564 565 /* 566 * Keep mostly read-only and often accessed (especially for 567 * the RCU path lookup and 'stat' data) fields at the beginning 568 * of the 'struct inode' 569 */ 570 struct inode { 571 umode_t i_mode; 572 unsigned short i_opflags; 573 kuid_t i_uid; 574 kgid_t i_gid; 575 unsigned int i_flags; 576 577 #ifdef CONFIG_FS_POSIX_ACL 578 struct posix_acl *i_acl; 579 struct posix_acl *i_default_acl; 580 #endif 581 582 const struct inode_operations *i_op; 583 struct super_block *i_sb; 584 struct address_space *i_mapping; 585 586 #ifdef CONFIG_SECURITY 587 void *i_security; 588 #endif 589 590 /* Stat data, not accessed from path walking */ 591 unsigned long i_ino; 592 /* 593 * Filesystems may only read i_nlink directly. They shall use the 594 * following functions for modification: 595 * 596 * (set|clear|inc|drop)_nlink 597 * inode_(inc|dec)_link_count 598 */ 599 union { 600 const unsigned int i_nlink; 601 unsigned int __i_nlink; 602 }; 603 dev_t i_rdev; 604 loff_t i_size; 605 struct timespec i_atime; 606 struct timespec i_mtime; 607 struct timespec i_ctime; 608 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */ 609 unsigned short i_bytes; 610 unsigned int i_blkbits; 611 enum rw_hint i_write_hint; 612 blkcnt_t i_blocks; 613 614 #ifdef __NEED_I_SIZE_ORDERED 615 seqcount_t i_size_seqcount; 616 #endif 617 618 /* Misc */ 619 unsigned long i_state; 620 struct rw_semaphore i_rwsem; 621 622 unsigned long dirtied_when; /* jiffies of first dirtying */ 623 unsigned long dirtied_time_when; 624 625 struct hlist_node i_hash; 626 struct list_head i_io_list; /* backing dev IO list */ 627 #ifdef CONFIG_CGROUP_WRITEBACK 628 struct bdi_writeback *i_wb; /* the associated cgroup wb */ 629 630 /* foreign inode detection, see wbc_detach_inode() */ 631 int i_wb_frn_winner; 632 u16 i_wb_frn_avg_time; 633 u16 i_wb_frn_history; 634 #endif 635 struct list_head i_lru; /* inode LRU list */ 636 struct list_head i_sb_list; 637 struct list_head i_wb_list; /* backing dev writeback list */ 638 union { 639 struct hlist_head i_dentry; 640 struct rcu_head i_rcu; 641 }; 642 atomic64_t i_version; 643 atomic_t i_count; 644 atomic_t i_dio_count; 645 atomic_t i_writecount; 646 #ifdef CONFIG_IMA 647 atomic_t i_readcount; /* struct files open RO */ 648 #endif 649 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */ 650 struct file_lock_context *i_flctx; 651 struct address_space i_data; 652 struct list_head i_devices; 653 union { 654 struct pipe_inode_info *i_pipe; 655 struct block_device *i_bdev; 656 struct cdev *i_cdev; 657 char *i_link; 658 unsigned i_dir_seq; 659 }; 660 661 __u32 i_generation; 662 663 #ifdef CONFIG_FSNOTIFY 664 __u32 i_fsnotify_mask; /* all events this inode cares about */ 665 struct fsnotify_mark_connector __rcu *i_fsnotify_marks; 666 #endif 667 668 #if IS_ENABLED(CONFIG_FS_ENCRYPTION) 669 struct fscrypt_info *i_crypt_info; 670 #endif 671 672 void *i_private; /* fs or device private pointer */ 673 } __randomize_layout; 674 675 static inline unsigned int i_blocksize(const struct inode *node) 676 { 677 return (1 << node->i_blkbits); 678 } 679 680 static inline int inode_unhashed(struct inode *inode) 681 { 682 return hlist_unhashed(&inode->i_hash); 683 } 684 685 /* 686 * inode->i_mutex nesting subclasses for the lock validator: 687 * 688 * 0: the object of the current VFS operation 689 * 1: parent 690 * 2: child/target 691 * 3: xattr 692 * 4: second non-directory 693 * 5: second parent (when locking independent directories in rename) 694 * 695 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two 696 * non-directories at once. 697 * 698 * The locking order between these classes is 699 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory 700 */ 701 enum inode_i_mutex_lock_class 702 { 703 I_MUTEX_NORMAL, 704 I_MUTEX_PARENT, 705 I_MUTEX_CHILD, 706 I_MUTEX_XATTR, 707 I_MUTEX_NONDIR2, 708 I_MUTEX_PARENT2, 709 }; 710 711 static inline void inode_lock(struct inode *inode) 712 { 713 down_write(&inode->i_rwsem); 714 } 715 716 static inline void inode_unlock(struct inode *inode) 717 { 718 up_write(&inode->i_rwsem); 719 } 720 721 static inline void inode_lock_shared(struct inode *inode) 722 { 723 down_read(&inode->i_rwsem); 724 } 725 726 static inline void inode_unlock_shared(struct inode *inode) 727 { 728 up_read(&inode->i_rwsem); 729 } 730 731 static inline int inode_trylock(struct inode *inode) 732 { 733 return down_write_trylock(&inode->i_rwsem); 734 } 735 736 static inline int inode_trylock_shared(struct inode *inode) 737 { 738 return down_read_trylock(&inode->i_rwsem); 739 } 740 741 static inline int inode_is_locked(struct inode *inode) 742 { 743 return rwsem_is_locked(&inode->i_rwsem); 744 } 745 746 static inline void inode_lock_nested(struct inode *inode, unsigned subclass) 747 { 748 down_write_nested(&inode->i_rwsem, subclass); 749 } 750 751 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass) 752 { 753 down_read_nested(&inode->i_rwsem, subclass); 754 } 755 756 void lock_two_nondirectories(struct inode *, struct inode*); 757 void unlock_two_nondirectories(struct inode *, struct inode*); 758 759 /* 760 * NOTE: in a 32bit arch with a preemptable kernel and 761 * an UP compile the i_size_read/write must be atomic 762 * with respect to the local cpu (unlike with preempt disabled), 763 * but they don't need to be atomic with respect to other cpus like in 764 * true SMP (so they need either to either locally disable irq around 765 * the read or for example on x86 they can be still implemented as a 766 * cmpxchg8b without the need of the lock prefix). For SMP compiles 767 * and 64bit archs it makes no difference if preempt is enabled or not. 768 */ 769 static inline loff_t i_size_read(const struct inode *inode) 770 { 771 #if BITS_PER_LONG==32 && defined(CONFIG_SMP) 772 loff_t i_size; 773 unsigned int seq; 774 775 do { 776 seq = read_seqcount_begin(&inode->i_size_seqcount); 777 i_size = inode->i_size; 778 } while (read_seqcount_retry(&inode->i_size_seqcount, seq)); 779 return i_size; 780 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT) 781 loff_t i_size; 782 783 preempt_disable(); 784 i_size = inode->i_size; 785 preempt_enable(); 786 return i_size; 787 #else 788 return inode->i_size; 789 #endif 790 } 791 792 /* 793 * NOTE: unlike i_size_read(), i_size_write() does need locking around it 794 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount 795 * can be lost, resulting in subsequent i_size_read() calls spinning forever. 796 */ 797 static inline void i_size_write(struct inode *inode, loff_t i_size) 798 { 799 #if BITS_PER_LONG==32 && defined(CONFIG_SMP) 800 preempt_disable(); 801 write_seqcount_begin(&inode->i_size_seqcount); 802 inode->i_size = i_size; 803 write_seqcount_end(&inode->i_size_seqcount); 804 preempt_enable(); 805 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT) 806 preempt_disable(); 807 inode->i_size = i_size; 808 preempt_enable(); 809 #else 810 inode->i_size = i_size; 811 #endif 812 } 813 814 static inline unsigned iminor(const struct inode *inode) 815 { 816 return MINOR(inode->i_rdev); 817 } 818 819 static inline unsigned imajor(const struct inode *inode) 820 { 821 return MAJOR(inode->i_rdev); 822 } 823 824 extern struct block_device *I_BDEV(struct inode *inode); 825 826 struct fown_struct { 827 rwlock_t lock; /* protects pid, uid, euid fields */ 828 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */ 829 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */ 830 kuid_t uid, euid; /* uid/euid of process setting the owner */ 831 int signum; /* posix.1b rt signal to be delivered on IO */ 832 }; 833 834 /* 835 * Track a single file's readahead state 836 */ 837 struct file_ra_state { 838 pgoff_t start; /* where readahead started */ 839 unsigned int size; /* # of readahead pages */ 840 unsigned int async_size; /* do asynchronous readahead when 841 there are only # of pages ahead */ 842 843 unsigned int ra_pages; /* Maximum readahead window */ 844 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */ 845 loff_t prev_pos; /* Cache last read() position */ 846 }; 847 848 /* 849 * Check if @index falls in the readahead windows. 850 */ 851 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index) 852 { 853 return (index >= ra->start && 854 index < ra->start + ra->size); 855 } 856 857 struct file { 858 union { 859 struct llist_node fu_llist; 860 struct rcu_head fu_rcuhead; 861 } f_u; 862 struct path f_path; 863 struct inode *f_inode; /* cached value */ 864 const struct file_operations *f_op; 865 866 /* 867 * Protects f_ep_links, f_flags. 868 * Must not be taken from IRQ context. 869 */ 870 spinlock_t f_lock; 871 enum rw_hint f_write_hint; 872 atomic_long_t f_count; 873 unsigned int f_flags; 874 fmode_t f_mode; 875 struct mutex f_pos_lock; 876 loff_t f_pos; 877 struct fown_struct f_owner; 878 const struct cred *f_cred; 879 struct file_ra_state f_ra; 880 881 u64 f_version; 882 #ifdef CONFIG_SECURITY 883 void *f_security; 884 #endif 885 /* needed for tty driver, and maybe others */ 886 void *private_data; 887 888 #ifdef CONFIG_EPOLL 889 /* Used by fs/eventpoll.c to link all the hooks to this file */ 890 struct list_head f_ep_links; 891 struct list_head f_tfile_llink; 892 #endif /* #ifdef CONFIG_EPOLL */ 893 struct address_space *f_mapping; 894 errseq_t f_wb_err; 895 } __randomize_layout 896 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */ 897 898 struct file_handle { 899 __u32 handle_bytes; 900 int handle_type; 901 /* file identifier */ 902 unsigned char f_handle[0]; 903 }; 904 905 static inline struct file *get_file(struct file *f) 906 { 907 atomic_long_inc(&f->f_count); 908 return f; 909 } 910 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count) 911 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1) 912 #define file_count(x) atomic_long_read(&(x)->f_count) 913 914 #define MAX_NON_LFS ((1UL<<31) - 1) 915 916 /* Page cache limit. The filesystems should put that into their s_maxbytes 917 limits, otherwise bad things can happen in VM. */ 918 #if BITS_PER_LONG==32 919 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT) 920 #elif BITS_PER_LONG==64 921 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX) 922 #endif 923 924 #define FL_POSIX 1 925 #define FL_FLOCK 2 926 #define FL_DELEG 4 /* NFSv4 delegation */ 927 #define FL_ACCESS 8 /* not trying to lock, just looking */ 928 #define FL_EXISTS 16 /* when unlocking, test for existence */ 929 #define FL_LEASE 32 /* lease held on this file */ 930 #define FL_CLOSE 64 /* unlock on close */ 931 #define FL_SLEEP 128 /* A blocking lock */ 932 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */ 933 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */ 934 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */ 935 #define FL_LAYOUT 2048 /* outstanding pNFS layout */ 936 937 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE) 938 939 /* 940 * Special return value from posix_lock_file() and vfs_lock_file() for 941 * asynchronous locking. 942 */ 943 #define FILE_LOCK_DEFERRED 1 944 945 /* legacy typedef, should eventually be removed */ 946 typedef void *fl_owner_t; 947 948 struct file_lock; 949 950 struct file_lock_operations { 951 void (*fl_copy_lock)(struct file_lock *, struct file_lock *); 952 void (*fl_release_private)(struct file_lock *); 953 }; 954 955 struct lock_manager_operations { 956 int (*lm_compare_owner)(struct file_lock *, struct file_lock *); 957 unsigned long (*lm_owner_key)(struct file_lock *); 958 fl_owner_t (*lm_get_owner)(fl_owner_t); 959 void (*lm_put_owner)(fl_owner_t); 960 void (*lm_notify)(struct file_lock *); /* unblock callback */ 961 int (*lm_grant)(struct file_lock *, int); 962 bool (*lm_break)(struct file_lock *); 963 int (*lm_change)(struct file_lock *, int, struct list_head *); 964 void (*lm_setup)(struct file_lock *, void **); 965 }; 966 967 struct lock_manager { 968 struct list_head list; 969 /* 970 * NFSv4 and up also want opens blocked during the grace period; 971 * NLM doesn't care: 972 */ 973 bool block_opens; 974 }; 975 976 struct net; 977 void locks_start_grace(struct net *, struct lock_manager *); 978 void locks_end_grace(struct lock_manager *); 979 bool locks_in_grace(struct net *); 980 bool opens_in_grace(struct net *); 981 982 /* that will die - we need it for nfs_lock_info */ 983 #include <linux/nfs_fs_i.h> 984 985 /* 986 * struct file_lock represents a generic "file lock". It's used to represent 987 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to 988 * note that the same struct is used to represent both a request for a lock and 989 * the lock itself, but the same object is never used for both. 990 * 991 * FIXME: should we create a separate "struct lock_request" to help distinguish 992 * these two uses? 993 * 994 * The varous i_flctx lists are ordered by: 995 * 996 * 1) lock owner 997 * 2) lock range start 998 * 3) lock range end 999 * 1000 * Obviously, the last two criteria only matter for POSIX locks. 1001 */ 1002 struct file_lock { 1003 struct file_lock *fl_next; /* singly linked list for this inode */ 1004 struct list_head fl_list; /* link into file_lock_context */ 1005 struct hlist_node fl_link; /* node in global lists */ 1006 struct list_head fl_block; /* circular list of blocked processes */ 1007 fl_owner_t fl_owner; 1008 unsigned int fl_flags; 1009 unsigned char fl_type; 1010 unsigned int fl_pid; 1011 int fl_link_cpu; /* what cpu's list is this on? */ 1012 wait_queue_head_t fl_wait; 1013 struct file *fl_file; 1014 loff_t fl_start; 1015 loff_t fl_end; 1016 1017 struct fasync_struct * fl_fasync; /* for lease break notifications */ 1018 /* for lease breaks: */ 1019 unsigned long fl_break_time; 1020 unsigned long fl_downgrade_time; 1021 1022 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */ 1023 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */ 1024 union { 1025 struct nfs_lock_info nfs_fl; 1026 struct nfs4_lock_info nfs4_fl; 1027 struct { 1028 struct list_head link; /* link in AFS vnode's pending_locks list */ 1029 int state; /* state of grant or error if -ve */ 1030 } afs; 1031 } fl_u; 1032 } __randomize_layout; 1033 1034 struct file_lock_context { 1035 spinlock_t flc_lock; 1036 struct list_head flc_flock; 1037 struct list_head flc_posix; 1038 struct list_head flc_lease; 1039 }; 1040 1041 /* The following constant reflects the upper bound of the file/locking space */ 1042 #ifndef OFFSET_MAX 1043 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1))) 1044 #define OFFSET_MAX INT_LIMIT(loff_t) 1045 #define OFFT_OFFSET_MAX INT_LIMIT(off_t) 1046 #endif 1047 1048 extern void send_sigio(struct fown_struct *fown, int fd, int band); 1049 1050 /* 1051 * Return the inode to use for locking 1052 * 1053 * For overlayfs this should be the overlay inode, not the real inode returned 1054 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are 1055 * equal. 1056 */ 1057 static inline struct inode *locks_inode(const struct file *f) 1058 { 1059 return f->f_path.dentry->d_inode; 1060 } 1061 1062 #ifdef CONFIG_FILE_LOCKING 1063 extern int fcntl_getlk(struct file *, unsigned int, struct flock *); 1064 extern int fcntl_setlk(unsigned int, struct file *, unsigned int, 1065 struct flock *); 1066 1067 #if BITS_PER_LONG == 32 1068 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *); 1069 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int, 1070 struct flock64 *); 1071 #endif 1072 1073 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg); 1074 extern int fcntl_getlease(struct file *filp); 1075 1076 /* fs/locks.c */ 1077 void locks_free_lock_context(struct inode *inode); 1078 void locks_free_lock(struct file_lock *fl); 1079 extern void locks_init_lock(struct file_lock *); 1080 extern struct file_lock * locks_alloc_lock(void); 1081 extern void locks_copy_lock(struct file_lock *, struct file_lock *); 1082 extern void locks_copy_conflock(struct file_lock *, struct file_lock *); 1083 extern void locks_remove_posix(struct file *, fl_owner_t); 1084 extern void locks_remove_file(struct file *); 1085 extern void locks_release_private(struct file_lock *); 1086 extern void posix_test_lock(struct file *, struct file_lock *); 1087 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *); 1088 extern int posix_unblock_lock(struct file_lock *); 1089 extern int vfs_test_lock(struct file *, struct file_lock *); 1090 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *); 1091 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl); 1092 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl); 1093 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type); 1094 extern void lease_get_mtime(struct inode *, struct timespec *time); 1095 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv); 1096 extern int vfs_setlease(struct file *, long, struct file_lock **, void **); 1097 extern int lease_modify(struct file_lock *, int, struct list_head *); 1098 struct files_struct; 1099 extern void show_fd_locks(struct seq_file *f, 1100 struct file *filp, struct files_struct *files); 1101 #else /* !CONFIG_FILE_LOCKING */ 1102 static inline int fcntl_getlk(struct file *file, unsigned int cmd, 1103 struct flock __user *user) 1104 { 1105 return -EINVAL; 1106 } 1107 1108 static inline int fcntl_setlk(unsigned int fd, struct file *file, 1109 unsigned int cmd, struct flock __user *user) 1110 { 1111 return -EACCES; 1112 } 1113 1114 #if BITS_PER_LONG == 32 1115 static inline int fcntl_getlk64(struct file *file, unsigned int cmd, 1116 struct flock64 __user *user) 1117 { 1118 return -EINVAL; 1119 } 1120 1121 static inline int fcntl_setlk64(unsigned int fd, struct file *file, 1122 unsigned int cmd, struct flock64 __user *user) 1123 { 1124 return -EACCES; 1125 } 1126 #endif 1127 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg) 1128 { 1129 return -EINVAL; 1130 } 1131 1132 static inline int fcntl_getlease(struct file *filp) 1133 { 1134 return F_UNLCK; 1135 } 1136 1137 static inline void 1138 locks_free_lock_context(struct inode *inode) 1139 { 1140 } 1141 1142 static inline void locks_init_lock(struct file_lock *fl) 1143 { 1144 return; 1145 } 1146 1147 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl) 1148 { 1149 return; 1150 } 1151 1152 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl) 1153 { 1154 return; 1155 } 1156 1157 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner) 1158 { 1159 return; 1160 } 1161 1162 static inline void locks_remove_file(struct file *filp) 1163 { 1164 return; 1165 } 1166 1167 static inline void posix_test_lock(struct file *filp, struct file_lock *fl) 1168 { 1169 return; 1170 } 1171 1172 static inline int posix_lock_file(struct file *filp, struct file_lock *fl, 1173 struct file_lock *conflock) 1174 { 1175 return -ENOLCK; 1176 } 1177 1178 static inline int posix_unblock_lock(struct file_lock *waiter) 1179 { 1180 return -ENOENT; 1181 } 1182 1183 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl) 1184 { 1185 return 0; 1186 } 1187 1188 static inline int vfs_lock_file(struct file *filp, unsigned int cmd, 1189 struct file_lock *fl, struct file_lock *conf) 1190 { 1191 return -ENOLCK; 1192 } 1193 1194 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl) 1195 { 1196 return 0; 1197 } 1198 1199 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl) 1200 { 1201 return -ENOLCK; 1202 } 1203 1204 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type) 1205 { 1206 return 0; 1207 } 1208 1209 static inline void lease_get_mtime(struct inode *inode, struct timespec *time) 1210 { 1211 return; 1212 } 1213 1214 static inline int generic_setlease(struct file *filp, long arg, 1215 struct file_lock **flp, void **priv) 1216 { 1217 return -EINVAL; 1218 } 1219 1220 static inline int vfs_setlease(struct file *filp, long arg, 1221 struct file_lock **lease, void **priv) 1222 { 1223 return -EINVAL; 1224 } 1225 1226 static inline int lease_modify(struct file_lock *fl, int arg, 1227 struct list_head *dispose) 1228 { 1229 return -EINVAL; 1230 } 1231 1232 struct files_struct; 1233 static inline void show_fd_locks(struct seq_file *f, 1234 struct file *filp, struct files_struct *files) {} 1235 #endif /* !CONFIG_FILE_LOCKING */ 1236 1237 static inline struct inode *file_inode(const struct file *f) 1238 { 1239 return f->f_inode; 1240 } 1241 1242 static inline struct dentry *file_dentry(const struct file *file) 1243 { 1244 return d_real(file->f_path.dentry, file_inode(file), 0, 0); 1245 } 1246 1247 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl) 1248 { 1249 return locks_lock_inode_wait(locks_inode(filp), fl); 1250 } 1251 1252 struct fasync_struct { 1253 spinlock_t fa_lock; 1254 int magic; 1255 int fa_fd; 1256 struct fasync_struct *fa_next; /* singly linked list */ 1257 struct file *fa_file; 1258 struct rcu_head fa_rcu; 1259 }; 1260 1261 #define FASYNC_MAGIC 0x4601 1262 1263 /* SMP safe fasync helpers: */ 1264 extern int fasync_helper(int, struct file *, int, struct fasync_struct **); 1265 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *); 1266 extern int fasync_remove_entry(struct file *, struct fasync_struct **); 1267 extern struct fasync_struct *fasync_alloc(void); 1268 extern void fasync_free(struct fasync_struct *); 1269 1270 /* can be called from interrupts */ 1271 extern void kill_fasync(struct fasync_struct **, int, int); 1272 1273 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force); 1274 extern int f_setown(struct file *filp, unsigned long arg, int force); 1275 extern void f_delown(struct file *filp); 1276 extern pid_t f_getown(struct file *filp); 1277 extern int send_sigurg(struct fown_struct *fown); 1278 1279 /* 1280 * sb->s_flags. Note that these mirror the equivalent MS_* flags where 1281 * represented in both. 1282 */ 1283 #define SB_RDONLY 1 /* Mount read-only */ 1284 #define SB_NOSUID 2 /* Ignore suid and sgid bits */ 1285 #define SB_NODEV 4 /* Disallow access to device special files */ 1286 #define SB_NOEXEC 8 /* Disallow program execution */ 1287 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */ 1288 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */ 1289 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */ 1290 #define SB_NOATIME 1024 /* Do not update access times. */ 1291 #define SB_NODIRATIME 2048 /* Do not update directory access times */ 1292 #define SB_SILENT 32768 1293 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */ 1294 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */ 1295 #define SB_I_VERSION (1<<23) /* Update inode I_version field */ 1296 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */ 1297 1298 /* These sb flags are internal to the kernel */ 1299 #define SB_SUBMOUNT (1<<26) 1300 #define SB_NOREMOTELOCK (1<<27) 1301 #define SB_NOSEC (1<<28) 1302 #define SB_BORN (1<<29) 1303 #define SB_ACTIVE (1<<30) 1304 #define SB_NOUSER (1<<31) 1305 1306 /* 1307 * Umount options 1308 */ 1309 1310 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */ 1311 #define MNT_DETACH 0x00000002 /* Just detach from the tree */ 1312 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */ 1313 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */ 1314 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */ 1315 1316 /* sb->s_iflags */ 1317 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */ 1318 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */ 1319 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */ 1320 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */ 1321 1322 /* sb->s_iflags to limit user namespace mounts */ 1323 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */ 1324 1325 /* Possible states of 'frozen' field */ 1326 enum { 1327 SB_UNFROZEN = 0, /* FS is unfrozen */ 1328 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */ 1329 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */ 1330 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop 1331 * internal threads if needed) */ 1332 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */ 1333 }; 1334 1335 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1) 1336 1337 struct sb_writers { 1338 int frozen; /* Is sb frozen? */ 1339 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */ 1340 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS]; 1341 }; 1342 1343 struct super_block { 1344 struct list_head s_list; /* Keep this first */ 1345 dev_t s_dev; /* search index; _not_ kdev_t */ 1346 unsigned char s_blocksize_bits; 1347 unsigned long s_blocksize; 1348 loff_t s_maxbytes; /* Max file size */ 1349 struct file_system_type *s_type; 1350 const struct super_operations *s_op; 1351 const struct dquot_operations *dq_op; 1352 const struct quotactl_ops *s_qcop; 1353 const struct export_operations *s_export_op; 1354 unsigned long s_flags; 1355 unsigned long s_iflags; /* internal SB_I_* flags */ 1356 unsigned long s_magic; 1357 struct dentry *s_root; 1358 struct rw_semaphore s_umount; 1359 int s_count; 1360 atomic_t s_active; 1361 #ifdef CONFIG_SECURITY 1362 void *s_security; 1363 #endif 1364 const struct xattr_handler **s_xattr; 1365 1366 const struct fscrypt_operations *s_cop; 1367 1368 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */ 1369 struct list_head s_mounts; /* list of mounts; _not_ for fs use */ 1370 struct block_device *s_bdev; 1371 struct backing_dev_info *s_bdi; 1372 struct mtd_info *s_mtd; 1373 struct hlist_node s_instances; 1374 unsigned int s_quota_types; /* Bitmask of supported quota types */ 1375 struct quota_info s_dquot; /* Diskquota specific options */ 1376 1377 struct sb_writers s_writers; 1378 1379 char s_id[32]; /* Informational name */ 1380 uuid_t s_uuid; /* UUID */ 1381 1382 void *s_fs_info; /* Filesystem private info */ 1383 unsigned int s_max_links; 1384 fmode_t s_mode; 1385 1386 /* Granularity of c/m/atime in ns. 1387 Cannot be worse than a second */ 1388 u32 s_time_gran; 1389 1390 /* 1391 * The next field is for VFS *only*. No filesystems have any business 1392 * even looking at it. You had been warned. 1393 */ 1394 struct mutex s_vfs_rename_mutex; /* Kludge */ 1395 1396 /* 1397 * Filesystem subtype. If non-empty the filesystem type field 1398 * in /proc/mounts will be "type.subtype" 1399 */ 1400 char *s_subtype; 1401 1402 const struct dentry_operations *s_d_op; /* default d_op for dentries */ 1403 1404 /* 1405 * Saved pool identifier for cleancache (-1 means none) 1406 */ 1407 int cleancache_poolid; 1408 1409 struct shrinker s_shrink; /* per-sb shrinker handle */ 1410 1411 /* Number of inodes with nlink == 0 but still referenced */ 1412 atomic_long_t s_remove_count; 1413 1414 /* Being remounted read-only */ 1415 int s_readonly_remount; 1416 1417 /* AIO completions deferred from interrupt context */ 1418 struct workqueue_struct *s_dio_done_wq; 1419 struct hlist_head s_pins; 1420 1421 /* 1422 * Owning user namespace and default context in which to 1423 * interpret filesystem uids, gids, quotas, device nodes, 1424 * xattrs and security labels. 1425 */ 1426 struct user_namespace *s_user_ns; 1427 1428 /* 1429 * Keep the lru lists last in the structure so they always sit on their 1430 * own individual cachelines. 1431 */ 1432 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp; 1433 struct list_lru s_inode_lru ____cacheline_aligned_in_smp; 1434 struct rcu_head rcu; 1435 struct work_struct destroy_work; 1436 1437 struct mutex s_sync_lock; /* sync serialisation lock */ 1438 1439 /* 1440 * Indicates how deep in a filesystem stack this SB is 1441 */ 1442 int s_stack_depth; 1443 1444 /* s_inode_list_lock protects s_inodes */ 1445 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp; 1446 struct list_head s_inodes; /* all inodes */ 1447 1448 spinlock_t s_inode_wblist_lock; 1449 struct list_head s_inodes_wb; /* writeback inodes */ 1450 } __randomize_layout; 1451 1452 /* Helper functions so that in most cases filesystems will 1453 * not need to deal directly with kuid_t and kgid_t and can 1454 * instead deal with the raw numeric values that are stored 1455 * in the filesystem. 1456 */ 1457 static inline uid_t i_uid_read(const struct inode *inode) 1458 { 1459 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid); 1460 } 1461 1462 static inline gid_t i_gid_read(const struct inode *inode) 1463 { 1464 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid); 1465 } 1466 1467 static inline void i_uid_write(struct inode *inode, uid_t uid) 1468 { 1469 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid); 1470 } 1471 1472 static inline void i_gid_write(struct inode *inode, gid_t gid) 1473 { 1474 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid); 1475 } 1476 1477 extern struct timespec current_time(struct inode *inode); 1478 1479 /* 1480 * Snapshotting support. 1481 */ 1482 1483 void __sb_end_write(struct super_block *sb, int level); 1484 int __sb_start_write(struct super_block *sb, int level, bool wait); 1485 1486 #define __sb_writers_acquired(sb, lev) \ 1487 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_) 1488 #define __sb_writers_release(sb, lev) \ 1489 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_) 1490 1491 /** 1492 * sb_end_write - drop write access to a superblock 1493 * @sb: the super we wrote to 1494 * 1495 * Decrement number of writers to the filesystem. Wake up possible waiters 1496 * wanting to freeze the filesystem. 1497 */ 1498 static inline void sb_end_write(struct super_block *sb) 1499 { 1500 __sb_end_write(sb, SB_FREEZE_WRITE); 1501 } 1502 1503 /** 1504 * sb_end_pagefault - drop write access to a superblock from a page fault 1505 * @sb: the super we wrote to 1506 * 1507 * Decrement number of processes handling write page fault to the filesystem. 1508 * Wake up possible waiters wanting to freeze the filesystem. 1509 */ 1510 static inline void sb_end_pagefault(struct super_block *sb) 1511 { 1512 __sb_end_write(sb, SB_FREEZE_PAGEFAULT); 1513 } 1514 1515 /** 1516 * sb_end_intwrite - drop write access to a superblock for internal fs purposes 1517 * @sb: the super we wrote to 1518 * 1519 * Decrement fs-internal number of writers to the filesystem. Wake up possible 1520 * waiters wanting to freeze the filesystem. 1521 */ 1522 static inline void sb_end_intwrite(struct super_block *sb) 1523 { 1524 __sb_end_write(sb, SB_FREEZE_FS); 1525 } 1526 1527 /** 1528 * sb_start_write - get write access to a superblock 1529 * @sb: the super we write to 1530 * 1531 * When a process wants to write data or metadata to a file system (i.e. dirty 1532 * a page or an inode), it should embed the operation in a sb_start_write() - 1533 * sb_end_write() pair to get exclusion against file system freezing. This 1534 * function increments number of writers preventing freezing. If the file 1535 * system is already frozen, the function waits until the file system is 1536 * thawed. 1537 * 1538 * Since freeze protection behaves as a lock, users have to preserve 1539 * ordering of freeze protection and other filesystem locks. Generally, 1540 * freeze protection should be the outermost lock. In particular, we have: 1541 * 1542 * sb_start_write 1543 * -> i_mutex (write path, truncate, directory ops, ...) 1544 * -> s_umount (freeze_super, thaw_super) 1545 */ 1546 static inline void sb_start_write(struct super_block *sb) 1547 { 1548 __sb_start_write(sb, SB_FREEZE_WRITE, true); 1549 } 1550 1551 static inline int sb_start_write_trylock(struct super_block *sb) 1552 { 1553 return __sb_start_write(sb, SB_FREEZE_WRITE, false); 1554 } 1555 1556 /** 1557 * sb_start_pagefault - get write access to a superblock from a page fault 1558 * @sb: the super we write to 1559 * 1560 * When a process starts handling write page fault, it should embed the 1561 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get 1562 * exclusion against file system freezing. This is needed since the page fault 1563 * is going to dirty a page. This function increments number of running page 1564 * faults preventing freezing. If the file system is already frozen, the 1565 * function waits until the file system is thawed. 1566 * 1567 * Since page fault freeze protection behaves as a lock, users have to preserve 1568 * ordering of freeze protection and other filesystem locks. It is advised to 1569 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault 1570 * handling code implies lock dependency: 1571 * 1572 * mmap_sem 1573 * -> sb_start_pagefault 1574 */ 1575 static inline void sb_start_pagefault(struct super_block *sb) 1576 { 1577 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true); 1578 } 1579 1580 /* 1581 * sb_start_intwrite - get write access to a superblock for internal fs purposes 1582 * @sb: the super we write to 1583 * 1584 * This is the third level of protection against filesystem freezing. It is 1585 * free for use by a filesystem. The only requirement is that it must rank 1586 * below sb_start_pagefault. 1587 * 1588 * For example filesystem can call sb_start_intwrite() when starting a 1589 * transaction which somewhat eases handling of freezing for internal sources 1590 * of filesystem changes (internal fs threads, discarding preallocation on file 1591 * close, etc.). 1592 */ 1593 static inline void sb_start_intwrite(struct super_block *sb) 1594 { 1595 __sb_start_write(sb, SB_FREEZE_FS, true); 1596 } 1597 1598 1599 extern bool inode_owner_or_capable(const struct inode *inode); 1600 1601 /* 1602 * VFS helper functions.. 1603 */ 1604 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool); 1605 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t); 1606 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t); 1607 extern int vfs_symlink(struct inode *, struct dentry *, const char *); 1608 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **); 1609 extern int vfs_rmdir(struct inode *, struct dentry *); 1610 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **); 1611 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int); 1612 extern int vfs_whiteout(struct inode *, struct dentry *); 1613 1614 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode, 1615 int open_flag); 1616 1617 int vfs_mkobj(struct dentry *, umode_t, 1618 int (*f)(struct dentry *, umode_t, void *), 1619 void *); 1620 1621 /* 1622 * VFS file helper functions. 1623 */ 1624 extern void inode_init_owner(struct inode *inode, const struct inode *dir, 1625 umode_t mode); 1626 extern bool may_open_dev(const struct path *path); 1627 /* 1628 * VFS FS_IOC_FIEMAP helper definitions. 1629 */ 1630 struct fiemap_extent_info { 1631 unsigned int fi_flags; /* Flags as passed from user */ 1632 unsigned int fi_extents_mapped; /* Number of mapped extents */ 1633 unsigned int fi_extents_max; /* Size of fiemap_extent array */ 1634 struct fiemap_extent __user *fi_extents_start; /* Start of 1635 fiemap_extent array */ 1636 }; 1637 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical, 1638 u64 phys, u64 len, u32 flags); 1639 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags); 1640 1641 /* 1642 * File types 1643 * 1644 * NOTE! These match bits 12..15 of stat.st_mode 1645 * (ie "(i_mode >> 12) & 15"). 1646 */ 1647 #define DT_UNKNOWN 0 1648 #define DT_FIFO 1 1649 #define DT_CHR 2 1650 #define DT_DIR 4 1651 #define DT_BLK 6 1652 #define DT_REG 8 1653 #define DT_LNK 10 1654 #define DT_SOCK 12 1655 #define DT_WHT 14 1656 1657 /* 1658 * This is the "filldir" function type, used by readdir() to let 1659 * the kernel specify what kind of dirent layout it wants to have. 1660 * This allows the kernel to read directories into kernel space or 1661 * to have different dirent layouts depending on the binary type. 1662 */ 1663 struct dir_context; 1664 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64, 1665 unsigned); 1666 1667 struct dir_context { 1668 const filldir_t actor; 1669 loff_t pos; 1670 }; 1671 1672 struct block_device_operations; 1673 1674 /* These macros are for out of kernel modules to test that 1675 * the kernel supports the unlocked_ioctl and compat_ioctl 1676 * fields in struct file_operations. */ 1677 #define HAVE_COMPAT_IOCTL 1 1678 #define HAVE_UNLOCKED_IOCTL 1 1679 1680 /* 1681 * These flags let !MMU mmap() govern direct device mapping vs immediate 1682 * copying more easily for MAP_PRIVATE, especially for ROM filesystems. 1683 * 1684 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE) 1685 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED) 1686 * NOMMU_MAP_READ: Can be mapped for reading 1687 * NOMMU_MAP_WRITE: Can be mapped for writing 1688 * NOMMU_MAP_EXEC: Can be mapped for execution 1689 */ 1690 #define NOMMU_MAP_COPY 0x00000001 1691 #define NOMMU_MAP_DIRECT 0x00000008 1692 #define NOMMU_MAP_READ VM_MAYREAD 1693 #define NOMMU_MAP_WRITE VM_MAYWRITE 1694 #define NOMMU_MAP_EXEC VM_MAYEXEC 1695 1696 #define NOMMU_VMFLAGS \ 1697 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC) 1698 1699 1700 struct iov_iter; 1701 1702 struct file_operations { 1703 struct module *owner; 1704 loff_t (*llseek) (struct file *, loff_t, int); 1705 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *); 1706 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *); 1707 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *); 1708 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *); 1709 int (*iterate) (struct file *, struct dir_context *); 1710 int (*iterate_shared) (struct file *, struct dir_context *); 1711 __poll_t (*poll) (struct file *, struct poll_table_struct *); 1712 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long); 1713 long (*compat_ioctl) (struct file *, unsigned int, unsigned long); 1714 int (*mmap) (struct file *, struct vm_area_struct *); 1715 unsigned long mmap_supported_flags; 1716 int (*open) (struct inode *, struct file *); 1717 int (*flush) (struct file *, fl_owner_t id); 1718 int (*release) (struct inode *, struct file *); 1719 int (*fsync) (struct file *, loff_t, loff_t, int datasync); 1720 int (*fasync) (int, struct file *, int); 1721 int (*lock) (struct file *, int, struct file_lock *); 1722 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int); 1723 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); 1724 int (*check_flags)(int); 1725 int (*flock) (struct file *, int, struct file_lock *); 1726 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int); 1727 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int); 1728 int (*setlease)(struct file *, long, struct file_lock **, void **); 1729 long (*fallocate)(struct file *file, int mode, loff_t offset, 1730 loff_t len); 1731 void (*show_fdinfo)(struct seq_file *m, struct file *f); 1732 #ifndef CONFIG_MMU 1733 unsigned (*mmap_capabilities)(struct file *); 1734 #endif 1735 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *, 1736 loff_t, size_t, unsigned int); 1737 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t, 1738 u64); 1739 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *, 1740 u64); 1741 } __randomize_layout; 1742 1743 struct inode_operations { 1744 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int); 1745 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *); 1746 int (*permission) (struct inode *, int); 1747 struct posix_acl * (*get_acl)(struct inode *, int); 1748 1749 int (*readlink) (struct dentry *, char __user *,int); 1750 1751 int (*create) (struct inode *,struct dentry *, umode_t, bool); 1752 int (*link) (struct dentry *,struct inode *,struct dentry *); 1753 int (*unlink) (struct inode *,struct dentry *); 1754 int (*symlink) (struct inode *,struct dentry *,const char *); 1755 int (*mkdir) (struct inode *,struct dentry *,umode_t); 1756 int (*rmdir) (struct inode *,struct dentry *); 1757 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t); 1758 int (*rename) (struct inode *, struct dentry *, 1759 struct inode *, struct dentry *, unsigned int); 1760 int (*setattr) (struct dentry *, struct iattr *); 1761 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int); 1762 ssize_t (*listxattr) (struct dentry *, char *, size_t); 1763 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, 1764 u64 len); 1765 int (*update_time)(struct inode *, struct timespec *, int); 1766 int (*atomic_open)(struct inode *, struct dentry *, 1767 struct file *, unsigned open_flag, 1768 umode_t create_mode, int *opened); 1769 int (*tmpfile) (struct inode *, struct dentry *, umode_t); 1770 int (*set_acl)(struct inode *, struct posix_acl *, int); 1771 } ____cacheline_aligned; 1772 1773 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio, 1774 struct iov_iter *iter) 1775 { 1776 return file->f_op->read_iter(kio, iter); 1777 } 1778 1779 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio, 1780 struct iov_iter *iter) 1781 { 1782 return file->f_op->write_iter(kio, iter); 1783 } 1784 1785 static inline int call_mmap(struct file *file, struct vm_area_struct *vma) 1786 { 1787 return file->f_op->mmap(file, vma); 1788 } 1789 1790 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector, 1791 unsigned long nr_segs, unsigned long fast_segs, 1792 struct iovec *fast_pointer, 1793 struct iovec **ret_pointer); 1794 1795 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *); 1796 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *); 1797 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *); 1798 extern ssize_t vfs_readv(struct file *, const struct iovec __user *, 1799 unsigned long, loff_t *, rwf_t); 1800 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *, 1801 loff_t, size_t, unsigned int); 1802 extern int vfs_clone_file_prep_inodes(struct inode *inode_in, loff_t pos_in, 1803 struct inode *inode_out, loff_t pos_out, 1804 u64 *len, bool is_dedupe); 1805 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in, 1806 struct file *file_out, loff_t pos_out, u64 len); 1807 extern int vfs_dedupe_file_range_compare(struct inode *src, loff_t srcoff, 1808 struct inode *dest, loff_t destoff, 1809 loff_t len, bool *is_same); 1810 extern int vfs_dedupe_file_range(struct file *file, 1811 struct file_dedupe_range *same); 1812 1813 struct super_operations { 1814 struct inode *(*alloc_inode)(struct super_block *sb); 1815 void (*destroy_inode)(struct inode *); 1816 1817 void (*dirty_inode) (struct inode *, int flags); 1818 int (*write_inode) (struct inode *, struct writeback_control *wbc); 1819 int (*drop_inode) (struct inode *); 1820 void (*evict_inode) (struct inode *); 1821 void (*put_super) (struct super_block *); 1822 int (*sync_fs)(struct super_block *sb, int wait); 1823 int (*freeze_super) (struct super_block *); 1824 int (*freeze_fs) (struct super_block *); 1825 int (*thaw_super) (struct super_block *); 1826 int (*unfreeze_fs) (struct super_block *); 1827 int (*statfs) (struct dentry *, struct kstatfs *); 1828 int (*remount_fs) (struct super_block *, int *, char *); 1829 void (*umount_begin) (struct super_block *); 1830 1831 int (*show_options)(struct seq_file *, struct dentry *); 1832 int (*show_devname)(struct seq_file *, struct dentry *); 1833 int (*show_path)(struct seq_file *, struct dentry *); 1834 int (*show_stats)(struct seq_file *, struct dentry *); 1835 #ifdef CONFIG_QUOTA 1836 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t); 1837 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t); 1838 struct dquot **(*get_dquots)(struct inode *); 1839 #endif 1840 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t); 1841 long (*nr_cached_objects)(struct super_block *, 1842 struct shrink_control *); 1843 long (*free_cached_objects)(struct super_block *, 1844 struct shrink_control *); 1845 }; 1846 1847 /* 1848 * Inode flags - they have no relation to superblock flags now 1849 */ 1850 #define S_SYNC 1 /* Writes are synced at once */ 1851 #define S_NOATIME 2 /* Do not update access times */ 1852 #define S_APPEND 4 /* Append-only file */ 1853 #define S_IMMUTABLE 8 /* Immutable file */ 1854 #define S_DEAD 16 /* removed, but still open directory */ 1855 #define S_NOQUOTA 32 /* Inode is not counted to quota */ 1856 #define S_DIRSYNC 64 /* Directory modifications are synchronous */ 1857 #define S_NOCMTIME 128 /* Do not update file c/mtime */ 1858 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */ 1859 #define S_PRIVATE 512 /* Inode is fs-internal */ 1860 #define S_IMA 1024 /* Inode has an associated IMA struct */ 1861 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */ 1862 #define S_NOSEC 4096 /* no suid or xattr security attributes */ 1863 #ifdef CONFIG_FS_DAX 1864 #define S_DAX 8192 /* Direct Access, avoiding the page cache */ 1865 #else 1866 #define S_DAX 0 /* Make all the DAX code disappear */ 1867 #endif 1868 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */ 1869 1870 /* 1871 * Note that nosuid etc flags are inode-specific: setting some file-system 1872 * flags just means all the inodes inherit those flags by default. It might be 1873 * possible to override it selectively if you really wanted to with some 1874 * ioctl() that is not currently implemented. 1875 * 1876 * Exception: SB_RDONLY is always applied to the entire file system. 1877 * 1878 * Unfortunately, it is possible to change a filesystems flags with it mounted 1879 * with files in use. This means that all of the inodes will not have their 1880 * i_flags updated. Hence, i_flags no longer inherit the superblock mount 1881 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org 1882 */ 1883 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg)) 1884 1885 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; } 1886 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb) 1887 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \ 1888 ((inode)->i_flags & S_SYNC)) 1889 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \ 1890 ((inode)->i_flags & (S_SYNC|S_DIRSYNC))) 1891 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK) 1892 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME) 1893 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION) 1894 1895 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA) 1896 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND) 1897 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE) 1898 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL) 1899 1900 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD) 1901 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME) 1902 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE) 1903 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE) 1904 #define IS_IMA(inode) ((inode)->i_flags & S_IMA) 1905 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT) 1906 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC) 1907 #define IS_DAX(inode) ((inode)->i_flags & S_DAX) 1908 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED) 1909 1910 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \ 1911 (inode)->i_rdev == WHITEOUT_DEV) 1912 1913 static inline bool HAS_UNMAPPED_ID(struct inode *inode) 1914 { 1915 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid); 1916 } 1917 1918 static inline enum rw_hint file_write_hint(struct file *file) 1919 { 1920 if (file->f_write_hint != WRITE_LIFE_NOT_SET) 1921 return file->f_write_hint; 1922 1923 return file_inode(file)->i_write_hint; 1924 } 1925 1926 static inline int iocb_flags(struct file *file); 1927 1928 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp) 1929 { 1930 *kiocb = (struct kiocb) { 1931 .ki_filp = filp, 1932 .ki_flags = iocb_flags(filp), 1933 .ki_hint = file_write_hint(filp), 1934 }; 1935 } 1936 1937 /* 1938 * Inode state bits. Protected by inode->i_lock 1939 * 1940 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC, 1941 * I_DIRTY_DATASYNC and I_DIRTY_PAGES. 1942 * 1943 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW, 1944 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at 1945 * various stages of removing an inode. 1946 * 1947 * Two bits are used for locking and completion notification, I_NEW and I_SYNC. 1948 * 1949 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on 1950 * fdatasync(). i_atime is the usual cause. 1951 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of 1952 * these changes separately from I_DIRTY_SYNC so that we 1953 * don't have to write inode on fdatasync() when only 1954 * mtime has changed in it. 1955 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean. 1956 * I_NEW Serves as both a mutex and completion notification. 1957 * New inodes set I_NEW. If two processes both create 1958 * the same inode, one of them will release its inode and 1959 * wait for I_NEW to be released before returning. 1960 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can 1961 * also cause waiting on I_NEW, without I_NEW actually 1962 * being set. find_inode() uses this to prevent returning 1963 * nearly-dead inodes. 1964 * I_WILL_FREE Must be set when calling write_inode_now() if i_count 1965 * is zero. I_FREEING must be set when I_WILL_FREE is 1966 * cleared. 1967 * I_FREEING Set when inode is about to be freed but still has dirty 1968 * pages or buffers attached or the inode itself is still 1969 * dirty. 1970 * I_CLEAR Added by clear_inode(). In this state the inode is 1971 * clean and can be destroyed. Inode keeps I_FREEING. 1972 * 1973 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are 1974 * prohibited for many purposes. iget() must wait for 1975 * the inode to be completely released, then create it 1976 * anew. Other functions will just ignore such inodes, 1977 * if appropriate. I_NEW is used for waiting. 1978 * 1979 * I_SYNC Writeback of inode is running. The bit is set during 1980 * data writeback, and cleared with a wakeup on the bit 1981 * address once it is done. The bit is also used to pin 1982 * the inode in memory for flusher thread. 1983 * 1984 * I_REFERENCED Marks the inode as recently references on the LRU list. 1985 * 1986 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit(). 1987 * 1988 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to 1989 * synchronize competing switching instances and to tell 1990 * wb stat updates to grab mapping->tree_lock. See 1991 * inode_switch_wb_work_fn() for details. 1992 * 1993 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper 1994 * and work dirs among overlayfs mounts. 1995 * 1996 * Q: What is the difference between I_WILL_FREE and I_FREEING? 1997 */ 1998 #define I_DIRTY_SYNC (1 << 0) 1999 #define I_DIRTY_DATASYNC (1 << 1) 2000 #define I_DIRTY_PAGES (1 << 2) 2001 #define __I_NEW 3 2002 #define I_NEW (1 << __I_NEW) 2003 #define I_WILL_FREE (1 << 4) 2004 #define I_FREEING (1 << 5) 2005 #define I_CLEAR (1 << 6) 2006 #define __I_SYNC 7 2007 #define I_SYNC (1 << __I_SYNC) 2008 #define I_REFERENCED (1 << 8) 2009 #define __I_DIO_WAKEUP 9 2010 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP) 2011 #define I_LINKABLE (1 << 10) 2012 #define I_DIRTY_TIME (1 << 11) 2013 #define __I_DIRTY_TIME_EXPIRED 12 2014 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED) 2015 #define I_WB_SWITCH (1 << 13) 2016 #define I_OVL_INUSE (1 << 14) 2017 2018 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES) 2019 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME) 2020 2021 extern void __mark_inode_dirty(struct inode *, int); 2022 static inline void mark_inode_dirty(struct inode *inode) 2023 { 2024 __mark_inode_dirty(inode, I_DIRTY); 2025 } 2026 2027 static inline void mark_inode_dirty_sync(struct inode *inode) 2028 { 2029 __mark_inode_dirty(inode, I_DIRTY_SYNC); 2030 } 2031 2032 extern void inc_nlink(struct inode *inode); 2033 extern void drop_nlink(struct inode *inode); 2034 extern void clear_nlink(struct inode *inode); 2035 extern void set_nlink(struct inode *inode, unsigned int nlink); 2036 2037 static inline void inode_inc_link_count(struct inode *inode) 2038 { 2039 inc_nlink(inode); 2040 mark_inode_dirty(inode); 2041 } 2042 2043 static inline void inode_dec_link_count(struct inode *inode) 2044 { 2045 drop_nlink(inode); 2046 mark_inode_dirty(inode); 2047 } 2048 2049 enum file_time_flags { 2050 S_ATIME = 1, 2051 S_MTIME = 2, 2052 S_CTIME = 4, 2053 S_VERSION = 8, 2054 }; 2055 2056 extern void touch_atime(const struct path *); 2057 static inline void file_accessed(struct file *file) 2058 { 2059 if (!(file->f_flags & O_NOATIME)) 2060 touch_atime(&file->f_path); 2061 } 2062 2063 int sync_inode(struct inode *inode, struct writeback_control *wbc); 2064 int sync_inode_metadata(struct inode *inode, int wait); 2065 2066 struct file_system_type { 2067 const char *name; 2068 int fs_flags; 2069 #define FS_REQUIRES_DEV 1 2070 #define FS_BINARY_MOUNTDATA 2 2071 #define FS_HAS_SUBTYPE 4 2072 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */ 2073 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */ 2074 struct dentry *(*mount) (struct file_system_type *, int, 2075 const char *, void *); 2076 void (*kill_sb) (struct super_block *); 2077 struct module *owner; 2078 struct file_system_type * next; 2079 struct hlist_head fs_supers; 2080 2081 struct lock_class_key s_lock_key; 2082 struct lock_class_key s_umount_key; 2083 struct lock_class_key s_vfs_rename_key; 2084 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS]; 2085 2086 struct lock_class_key i_lock_key; 2087 struct lock_class_key i_mutex_key; 2088 struct lock_class_key i_mutex_dir_key; 2089 }; 2090 2091 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME) 2092 2093 extern struct dentry *mount_ns(struct file_system_type *fs_type, 2094 int flags, void *data, void *ns, struct user_namespace *user_ns, 2095 int (*fill_super)(struct super_block *, void *, int)); 2096 #ifdef CONFIG_BLOCK 2097 extern struct dentry *mount_bdev(struct file_system_type *fs_type, 2098 int flags, const char *dev_name, void *data, 2099 int (*fill_super)(struct super_block *, void *, int)); 2100 #else 2101 static inline struct dentry *mount_bdev(struct file_system_type *fs_type, 2102 int flags, const char *dev_name, void *data, 2103 int (*fill_super)(struct super_block *, void *, int)) 2104 { 2105 return ERR_PTR(-ENODEV); 2106 } 2107 #endif 2108 extern struct dentry *mount_single(struct file_system_type *fs_type, 2109 int flags, void *data, 2110 int (*fill_super)(struct super_block *, void *, int)); 2111 extern struct dentry *mount_nodev(struct file_system_type *fs_type, 2112 int flags, void *data, 2113 int (*fill_super)(struct super_block *, void *, int)); 2114 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path); 2115 void generic_shutdown_super(struct super_block *sb); 2116 #ifdef CONFIG_BLOCK 2117 void kill_block_super(struct super_block *sb); 2118 #else 2119 static inline void kill_block_super(struct super_block *sb) 2120 { 2121 BUG(); 2122 } 2123 #endif 2124 void kill_anon_super(struct super_block *sb); 2125 void kill_litter_super(struct super_block *sb); 2126 void deactivate_super(struct super_block *sb); 2127 void deactivate_locked_super(struct super_block *sb); 2128 int set_anon_super(struct super_block *s, void *data); 2129 int get_anon_bdev(dev_t *); 2130 void free_anon_bdev(dev_t); 2131 struct super_block *sget_userns(struct file_system_type *type, 2132 int (*test)(struct super_block *,void *), 2133 int (*set)(struct super_block *,void *), 2134 int flags, struct user_namespace *user_ns, 2135 void *data); 2136 struct super_block *sget(struct file_system_type *type, 2137 int (*test)(struct super_block *,void *), 2138 int (*set)(struct super_block *,void *), 2139 int flags, void *data); 2140 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *, 2141 const struct super_operations *ops, 2142 const struct xattr_handler **xattr, 2143 const struct dentry_operations *dops, 2144 unsigned long); 2145 2146 static inline struct dentry * 2147 mount_pseudo(struct file_system_type *fs_type, char *name, 2148 const struct super_operations *ops, 2149 const struct dentry_operations *dops, unsigned long magic) 2150 { 2151 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic); 2152 } 2153 2154 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */ 2155 #define fops_get(fops) \ 2156 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL)) 2157 #define fops_put(fops) \ 2158 do { if (fops) module_put((fops)->owner); } while(0) 2159 /* 2160 * This one is to be used *ONLY* from ->open() instances. 2161 * fops must be non-NULL, pinned down *and* module dependencies 2162 * should be sufficient to pin the caller down as well. 2163 */ 2164 #define replace_fops(f, fops) \ 2165 do { \ 2166 struct file *__file = (f); \ 2167 fops_put(__file->f_op); \ 2168 BUG_ON(!(__file->f_op = (fops))); \ 2169 } while(0) 2170 2171 extern int register_filesystem(struct file_system_type *); 2172 extern int unregister_filesystem(struct file_system_type *); 2173 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data); 2174 #define kern_mount(type) kern_mount_data(type, NULL) 2175 extern void kern_unmount(struct vfsmount *mnt); 2176 extern int may_umount_tree(struct vfsmount *); 2177 extern int may_umount(struct vfsmount *); 2178 extern long do_mount(const char *, const char __user *, 2179 const char *, unsigned long, void *); 2180 extern struct vfsmount *collect_mounts(const struct path *); 2181 extern void drop_collected_mounts(struct vfsmount *); 2182 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *, 2183 struct vfsmount *); 2184 extern int vfs_statfs(const struct path *, struct kstatfs *); 2185 extern int user_statfs(const char __user *, struct kstatfs *); 2186 extern int fd_statfs(int, struct kstatfs *); 2187 extern int freeze_super(struct super_block *super); 2188 extern int thaw_super(struct super_block *super); 2189 extern bool our_mnt(struct vfsmount *mnt); 2190 extern __printf(2, 3) 2191 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...); 2192 extern int super_setup_bdi(struct super_block *sb); 2193 2194 extern int current_umask(void); 2195 2196 extern void ihold(struct inode * inode); 2197 extern void iput(struct inode *); 2198 extern int generic_update_time(struct inode *, struct timespec *, int); 2199 2200 /* /sys/fs */ 2201 extern struct kobject *fs_kobj; 2202 2203 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK) 2204 2205 #ifdef CONFIG_MANDATORY_FILE_LOCKING 2206 extern int locks_mandatory_locked(struct file *); 2207 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char); 2208 2209 /* 2210 * Candidates for mandatory locking have the setgid bit set 2211 * but no group execute bit - an otherwise meaningless combination. 2212 */ 2213 2214 static inline int __mandatory_lock(struct inode *ino) 2215 { 2216 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID; 2217 } 2218 2219 /* 2220 * ... and these candidates should be on SB_MANDLOCK mounted fs, 2221 * otherwise these will be advisory locks 2222 */ 2223 2224 static inline int mandatory_lock(struct inode *ino) 2225 { 2226 return IS_MANDLOCK(ino) && __mandatory_lock(ino); 2227 } 2228 2229 static inline int locks_verify_locked(struct file *file) 2230 { 2231 if (mandatory_lock(locks_inode(file))) 2232 return locks_mandatory_locked(file); 2233 return 0; 2234 } 2235 2236 static inline int locks_verify_truncate(struct inode *inode, 2237 struct file *f, 2238 loff_t size) 2239 { 2240 if (!inode->i_flctx || !mandatory_lock(inode)) 2241 return 0; 2242 2243 if (size < inode->i_size) { 2244 return locks_mandatory_area(inode, f, size, inode->i_size - 1, 2245 F_WRLCK); 2246 } else { 2247 return locks_mandatory_area(inode, f, inode->i_size, size - 1, 2248 F_WRLCK); 2249 } 2250 } 2251 2252 #else /* !CONFIG_MANDATORY_FILE_LOCKING */ 2253 2254 static inline int locks_mandatory_locked(struct file *file) 2255 { 2256 return 0; 2257 } 2258 2259 static inline int locks_mandatory_area(struct inode *inode, struct file *filp, 2260 loff_t start, loff_t end, unsigned char type) 2261 { 2262 return 0; 2263 } 2264 2265 static inline int __mandatory_lock(struct inode *inode) 2266 { 2267 return 0; 2268 } 2269 2270 static inline int mandatory_lock(struct inode *inode) 2271 { 2272 return 0; 2273 } 2274 2275 static inline int locks_verify_locked(struct file *file) 2276 { 2277 return 0; 2278 } 2279 2280 static inline int locks_verify_truncate(struct inode *inode, struct file *filp, 2281 size_t size) 2282 { 2283 return 0; 2284 } 2285 2286 #endif /* CONFIG_MANDATORY_FILE_LOCKING */ 2287 2288 2289 #ifdef CONFIG_FILE_LOCKING 2290 static inline int break_lease(struct inode *inode, unsigned int mode) 2291 { 2292 /* 2293 * Since this check is lockless, we must ensure that any refcounts 2294 * taken are done before checking i_flctx->flc_lease. Otherwise, we 2295 * could end up racing with tasks trying to set a new lease on this 2296 * file. 2297 */ 2298 smp_mb(); 2299 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease)) 2300 return __break_lease(inode, mode, FL_LEASE); 2301 return 0; 2302 } 2303 2304 static inline int break_deleg(struct inode *inode, unsigned int mode) 2305 { 2306 /* 2307 * Since this check is lockless, we must ensure that any refcounts 2308 * taken are done before checking i_flctx->flc_lease. Otherwise, we 2309 * could end up racing with tasks trying to set a new lease on this 2310 * file. 2311 */ 2312 smp_mb(); 2313 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease)) 2314 return __break_lease(inode, mode, FL_DELEG); 2315 return 0; 2316 } 2317 2318 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode) 2319 { 2320 int ret; 2321 2322 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK); 2323 if (ret == -EWOULDBLOCK && delegated_inode) { 2324 *delegated_inode = inode; 2325 ihold(inode); 2326 } 2327 return ret; 2328 } 2329 2330 static inline int break_deleg_wait(struct inode **delegated_inode) 2331 { 2332 int ret; 2333 2334 ret = break_deleg(*delegated_inode, O_WRONLY); 2335 iput(*delegated_inode); 2336 *delegated_inode = NULL; 2337 return ret; 2338 } 2339 2340 static inline int break_layout(struct inode *inode, bool wait) 2341 { 2342 smp_mb(); 2343 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease)) 2344 return __break_lease(inode, 2345 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK, 2346 FL_LAYOUT); 2347 return 0; 2348 } 2349 2350 #else /* !CONFIG_FILE_LOCKING */ 2351 static inline int break_lease(struct inode *inode, unsigned int mode) 2352 { 2353 return 0; 2354 } 2355 2356 static inline int break_deleg(struct inode *inode, unsigned int mode) 2357 { 2358 return 0; 2359 } 2360 2361 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode) 2362 { 2363 return 0; 2364 } 2365 2366 static inline int break_deleg_wait(struct inode **delegated_inode) 2367 { 2368 BUG(); 2369 return 0; 2370 } 2371 2372 static inline int break_layout(struct inode *inode, bool wait) 2373 { 2374 return 0; 2375 } 2376 2377 #endif /* CONFIG_FILE_LOCKING */ 2378 2379 /* fs/open.c */ 2380 struct audit_names; 2381 struct filename { 2382 const char *name; /* pointer to actual string */ 2383 const __user char *uptr; /* original userland pointer */ 2384 struct audit_names *aname; 2385 int refcnt; 2386 const char iname[]; 2387 }; 2388 2389 extern long vfs_truncate(const struct path *, loff_t); 2390 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs, 2391 struct file *filp); 2392 extern int vfs_fallocate(struct file *file, int mode, loff_t offset, 2393 loff_t len); 2394 extern long do_sys_open(int dfd, const char __user *filename, int flags, 2395 umode_t mode); 2396 extern struct file *file_open_name(struct filename *, int, umode_t); 2397 extern struct file *filp_open(const char *, int, umode_t); 2398 extern struct file *file_open_root(struct dentry *, struct vfsmount *, 2399 const char *, int, umode_t); 2400 extern struct file * dentry_open(const struct path *, int, const struct cred *); 2401 extern int filp_close(struct file *, fl_owner_t id); 2402 2403 extern struct filename *getname_flags(const char __user *, int, int *); 2404 extern struct filename *getname(const char __user *); 2405 extern struct filename *getname_kernel(const char *); 2406 extern void putname(struct filename *name); 2407 2408 enum { 2409 FILE_CREATED = 1, 2410 FILE_OPENED = 2 2411 }; 2412 extern int finish_open(struct file *file, struct dentry *dentry, 2413 int (*open)(struct inode *, struct file *), 2414 int *opened); 2415 extern int finish_no_open(struct file *file, struct dentry *dentry); 2416 2417 /* fs/ioctl.c */ 2418 2419 extern int ioctl_preallocate(struct file *filp, void __user *argp); 2420 2421 /* fs/dcache.c */ 2422 extern void __init vfs_caches_init_early(void); 2423 extern void __init vfs_caches_init(void); 2424 2425 extern struct kmem_cache *names_cachep; 2426 2427 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL) 2428 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name)) 2429 2430 #ifdef CONFIG_BLOCK 2431 extern int register_blkdev(unsigned int, const char *); 2432 extern void unregister_blkdev(unsigned int, const char *); 2433 extern void bdev_unhash_inode(dev_t dev); 2434 extern struct block_device *bdget(dev_t); 2435 extern struct block_device *bdgrab(struct block_device *bdev); 2436 extern void bd_set_size(struct block_device *, loff_t size); 2437 extern void bd_forget(struct inode *inode); 2438 extern void bdput(struct block_device *); 2439 extern void invalidate_bdev(struct block_device *); 2440 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *); 2441 extern int sync_blockdev(struct block_device *bdev); 2442 extern void kill_bdev(struct block_device *); 2443 extern struct super_block *freeze_bdev(struct block_device *); 2444 extern void emergency_thaw_all(void); 2445 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb); 2446 extern int fsync_bdev(struct block_device *); 2447 2448 extern struct super_block *blockdev_superblock; 2449 2450 static inline bool sb_is_blkdev_sb(struct super_block *sb) 2451 { 2452 return sb == blockdev_superblock; 2453 } 2454 #else 2455 static inline void bd_forget(struct inode *inode) {} 2456 static inline int sync_blockdev(struct block_device *bdev) { return 0; } 2457 static inline void kill_bdev(struct block_device *bdev) {} 2458 static inline void invalidate_bdev(struct block_device *bdev) {} 2459 2460 static inline struct super_block *freeze_bdev(struct block_device *sb) 2461 { 2462 return NULL; 2463 } 2464 2465 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb) 2466 { 2467 return 0; 2468 } 2469 2470 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg) 2471 { 2472 } 2473 2474 static inline bool sb_is_blkdev_sb(struct super_block *sb) 2475 { 2476 return false; 2477 } 2478 #endif 2479 extern int sync_filesystem(struct super_block *); 2480 extern const struct file_operations def_blk_fops; 2481 extern const struct file_operations def_chr_fops; 2482 #ifdef CONFIG_BLOCK 2483 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long); 2484 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long); 2485 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long); 2486 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder); 2487 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode, 2488 void *holder); 2489 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, 2490 void *holder); 2491 extern void blkdev_put(struct block_device *bdev, fmode_t mode); 2492 extern int __blkdev_reread_part(struct block_device *bdev); 2493 extern int blkdev_reread_part(struct block_device *bdev); 2494 2495 #ifdef CONFIG_SYSFS 2496 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk); 2497 extern void bd_unlink_disk_holder(struct block_device *bdev, 2498 struct gendisk *disk); 2499 #else 2500 static inline int bd_link_disk_holder(struct block_device *bdev, 2501 struct gendisk *disk) 2502 { 2503 return 0; 2504 } 2505 static inline void bd_unlink_disk_holder(struct block_device *bdev, 2506 struct gendisk *disk) 2507 { 2508 } 2509 #endif 2510 #endif 2511 2512 /* fs/char_dev.c */ 2513 #define CHRDEV_MAJOR_MAX 512 2514 /* Marks the bottom of the first segment of free char majors */ 2515 #define CHRDEV_MAJOR_DYN_END 234 2516 /* Marks the top and bottom of the second segment of free char majors */ 2517 #define CHRDEV_MAJOR_DYN_EXT_START 511 2518 #define CHRDEV_MAJOR_DYN_EXT_END 384 2519 2520 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *); 2521 extern int register_chrdev_region(dev_t, unsigned, const char *); 2522 extern int __register_chrdev(unsigned int major, unsigned int baseminor, 2523 unsigned int count, const char *name, 2524 const struct file_operations *fops); 2525 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor, 2526 unsigned int count, const char *name); 2527 extern void unregister_chrdev_region(dev_t, unsigned); 2528 extern void chrdev_show(struct seq_file *,off_t); 2529 2530 static inline int register_chrdev(unsigned int major, const char *name, 2531 const struct file_operations *fops) 2532 { 2533 return __register_chrdev(major, 0, 256, name, fops); 2534 } 2535 2536 static inline void unregister_chrdev(unsigned int major, const char *name) 2537 { 2538 __unregister_chrdev(major, 0, 256, name); 2539 } 2540 2541 /* fs/block_dev.c */ 2542 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */ 2543 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */ 2544 2545 #ifdef CONFIG_BLOCK 2546 #define BLKDEV_MAJOR_MAX 512 2547 extern const char *__bdevname(dev_t, char *buffer); 2548 extern const char *bdevname(struct block_device *bdev, char *buffer); 2549 extern struct block_device *lookup_bdev(const char *); 2550 extern void blkdev_show(struct seq_file *,off_t); 2551 2552 #else 2553 #define BLKDEV_MAJOR_MAX 0 2554 #endif 2555 2556 extern void init_special_inode(struct inode *, umode_t, dev_t); 2557 2558 /* Invalid inode operations -- fs/bad_inode.c */ 2559 extern void make_bad_inode(struct inode *); 2560 extern bool is_bad_inode(struct inode *); 2561 2562 #ifdef CONFIG_BLOCK 2563 extern void check_disk_size_change(struct gendisk *disk, 2564 struct block_device *bdev); 2565 extern int revalidate_disk(struct gendisk *); 2566 extern int check_disk_change(struct block_device *); 2567 extern int __invalidate_device(struct block_device *, bool); 2568 extern int invalidate_partition(struct gendisk *, int); 2569 #endif 2570 unsigned long invalidate_mapping_pages(struct address_space *mapping, 2571 pgoff_t start, pgoff_t end); 2572 2573 static inline void invalidate_remote_inode(struct inode *inode) 2574 { 2575 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) || 2576 S_ISLNK(inode->i_mode)) 2577 invalidate_mapping_pages(inode->i_mapping, 0, -1); 2578 } 2579 extern int invalidate_inode_pages2(struct address_space *mapping); 2580 extern int invalidate_inode_pages2_range(struct address_space *mapping, 2581 pgoff_t start, pgoff_t end); 2582 extern int write_inode_now(struct inode *, int); 2583 extern int filemap_fdatawrite(struct address_space *); 2584 extern int filemap_flush(struct address_space *); 2585 extern int filemap_fdatawait_keep_errors(struct address_space *mapping); 2586 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart, 2587 loff_t lend); 2588 2589 static inline int filemap_fdatawait(struct address_space *mapping) 2590 { 2591 return filemap_fdatawait_range(mapping, 0, LLONG_MAX); 2592 } 2593 2594 extern bool filemap_range_has_page(struct address_space *, loff_t lstart, 2595 loff_t lend); 2596 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart, 2597 loff_t lend); 2598 extern int filemap_write_and_wait(struct address_space *mapping); 2599 extern int filemap_write_and_wait_range(struct address_space *mapping, 2600 loff_t lstart, loff_t lend); 2601 extern int __filemap_fdatawrite_range(struct address_space *mapping, 2602 loff_t start, loff_t end, int sync_mode); 2603 extern int filemap_fdatawrite_range(struct address_space *mapping, 2604 loff_t start, loff_t end); 2605 extern int filemap_check_errors(struct address_space *mapping); 2606 extern void __filemap_set_wb_err(struct address_space *mapping, int err); 2607 2608 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart, 2609 loff_t lend); 2610 extern int __must_check file_check_and_advance_wb_err(struct file *file); 2611 extern int __must_check file_write_and_wait_range(struct file *file, 2612 loff_t start, loff_t end); 2613 2614 static inline int file_write_and_wait(struct file *file) 2615 { 2616 return file_write_and_wait_range(file, 0, LLONG_MAX); 2617 } 2618 2619 /** 2620 * filemap_set_wb_err - set a writeback error on an address_space 2621 * @mapping: mapping in which to set writeback error 2622 * @err: error to be set in mapping 2623 * 2624 * When writeback fails in some way, we must record that error so that 2625 * userspace can be informed when fsync and the like are called. We endeavor 2626 * to report errors on any file that was open at the time of the error. Some 2627 * internal callers also need to know when writeback errors have occurred. 2628 * 2629 * When a writeback error occurs, most filesystems will want to call 2630 * filemap_set_wb_err to record the error in the mapping so that it will be 2631 * automatically reported whenever fsync is called on the file. 2632 */ 2633 static inline void filemap_set_wb_err(struct address_space *mapping, int err) 2634 { 2635 /* Fastpath for common case of no error */ 2636 if (unlikely(err)) 2637 __filemap_set_wb_err(mapping, err); 2638 } 2639 2640 /** 2641 * filemap_check_wb_error - has an error occurred since the mark was sampled? 2642 * @mapping: mapping to check for writeback errors 2643 * @since: previously-sampled errseq_t 2644 * 2645 * Grab the errseq_t value from the mapping, and see if it has changed "since" 2646 * the given value was sampled. 2647 * 2648 * If it has then report the latest error set, otherwise return 0. 2649 */ 2650 static inline int filemap_check_wb_err(struct address_space *mapping, 2651 errseq_t since) 2652 { 2653 return errseq_check(&mapping->wb_err, since); 2654 } 2655 2656 /** 2657 * filemap_sample_wb_err - sample the current errseq_t to test for later errors 2658 * @mapping: mapping to be sampled 2659 * 2660 * Writeback errors are always reported relative to a particular sample point 2661 * in the past. This function provides those sample points. 2662 */ 2663 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping) 2664 { 2665 return errseq_sample(&mapping->wb_err); 2666 } 2667 2668 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end, 2669 int datasync); 2670 extern int vfs_fsync(struct file *file, int datasync); 2671 2672 /* 2673 * Sync the bytes written if this was a synchronous write. Expect ki_pos 2674 * to already be updated for the write, and will return either the amount 2675 * of bytes passed in, or an error if syncing the file failed. 2676 */ 2677 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count) 2678 { 2679 if (iocb->ki_flags & IOCB_DSYNC) { 2680 int ret = vfs_fsync_range(iocb->ki_filp, 2681 iocb->ki_pos - count, iocb->ki_pos - 1, 2682 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1); 2683 if (ret) 2684 return ret; 2685 } 2686 2687 return count; 2688 } 2689 2690 extern void emergency_sync(void); 2691 extern void emergency_remount(void); 2692 #ifdef CONFIG_BLOCK 2693 extern sector_t bmap(struct inode *, sector_t); 2694 #endif 2695 extern int notify_change(struct dentry *, struct iattr *, struct inode **); 2696 extern int inode_permission(struct inode *, int); 2697 extern int generic_permission(struct inode *, int); 2698 extern int __check_sticky(struct inode *dir, struct inode *inode); 2699 2700 static inline bool execute_ok(struct inode *inode) 2701 { 2702 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode); 2703 } 2704 2705 static inline void file_start_write(struct file *file) 2706 { 2707 if (!S_ISREG(file_inode(file)->i_mode)) 2708 return; 2709 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true); 2710 } 2711 2712 static inline bool file_start_write_trylock(struct file *file) 2713 { 2714 if (!S_ISREG(file_inode(file)->i_mode)) 2715 return true; 2716 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false); 2717 } 2718 2719 static inline void file_end_write(struct file *file) 2720 { 2721 if (!S_ISREG(file_inode(file)->i_mode)) 2722 return; 2723 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE); 2724 } 2725 2726 static inline int do_clone_file_range(struct file *file_in, loff_t pos_in, 2727 struct file *file_out, loff_t pos_out, 2728 u64 len) 2729 { 2730 int ret; 2731 2732 file_start_write(file_out); 2733 ret = vfs_clone_file_range(file_in, pos_in, file_out, pos_out, len); 2734 file_end_write(file_out); 2735 2736 return ret; 2737 } 2738 2739 /* 2740 * get_write_access() gets write permission for a file. 2741 * put_write_access() releases this write permission. 2742 * This is used for regular files. 2743 * We cannot support write (and maybe mmap read-write shared) accesses and 2744 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode 2745 * can have the following values: 2746 * 0: no writers, no VM_DENYWRITE mappings 2747 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist 2748 * > 0: (i_writecount) users are writing to the file. 2749 * 2750 * Normally we operate on that counter with atomic_{inc,dec} and it's safe 2751 * except for the cases where we don't hold i_writecount yet. Then we need to 2752 * use {get,deny}_write_access() - these functions check the sign and refuse 2753 * to do the change if sign is wrong. 2754 */ 2755 static inline int get_write_access(struct inode *inode) 2756 { 2757 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY; 2758 } 2759 static inline int deny_write_access(struct file *file) 2760 { 2761 struct inode *inode = file_inode(file); 2762 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY; 2763 } 2764 static inline void put_write_access(struct inode * inode) 2765 { 2766 atomic_dec(&inode->i_writecount); 2767 } 2768 static inline void allow_write_access(struct file *file) 2769 { 2770 if (file) 2771 atomic_inc(&file_inode(file)->i_writecount); 2772 } 2773 static inline bool inode_is_open_for_write(const struct inode *inode) 2774 { 2775 return atomic_read(&inode->i_writecount) > 0; 2776 } 2777 2778 #ifdef CONFIG_IMA 2779 static inline void i_readcount_dec(struct inode *inode) 2780 { 2781 BUG_ON(!atomic_read(&inode->i_readcount)); 2782 atomic_dec(&inode->i_readcount); 2783 } 2784 static inline void i_readcount_inc(struct inode *inode) 2785 { 2786 atomic_inc(&inode->i_readcount); 2787 } 2788 #else 2789 static inline void i_readcount_dec(struct inode *inode) 2790 { 2791 return; 2792 } 2793 static inline void i_readcount_inc(struct inode *inode) 2794 { 2795 return; 2796 } 2797 #endif 2798 extern int do_pipe_flags(int *, int); 2799 2800 #define __kernel_read_file_id(id) \ 2801 id(UNKNOWN, unknown) \ 2802 id(FIRMWARE, firmware) \ 2803 id(FIRMWARE_PREALLOC_BUFFER, firmware) \ 2804 id(MODULE, kernel-module) \ 2805 id(KEXEC_IMAGE, kexec-image) \ 2806 id(KEXEC_INITRAMFS, kexec-initramfs) \ 2807 id(POLICY, security-policy) \ 2808 id(X509_CERTIFICATE, x509-certificate) \ 2809 id(MAX_ID, ) 2810 2811 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM, 2812 #define __fid_stringify(dummy, str) #str, 2813 2814 enum kernel_read_file_id { 2815 __kernel_read_file_id(__fid_enumify) 2816 }; 2817 2818 static const char * const kernel_read_file_str[] = { 2819 __kernel_read_file_id(__fid_stringify) 2820 }; 2821 2822 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id) 2823 { 2824 if ((unsigned)id >= READING_MAX_ID) 2825 return kernel_read_file_str[READING_UNKNOWN]; 2826 2827 return kernel_read_file_str[id]; 2828 } 2829 2830 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t, 2831 enum kernel_read_file_id); 2832 extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t, 2833 enum kernel_read_file_id); 2834 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t, 2835 enum kernel_read_file_id); 2836 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *); 2837 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *); 2838 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *); 2839 extern struct file * open_exec(const char *); 2840 2841 /* fs/dcache.c -- generic fs support functions */ 2842 extern bool is_subdir(struct dentry *, struct dentry *); 2843 extern bool path_is_under(const struct path *, const struct path *); 2844 2845 extern char *file_path(struct file *, char *, int); 2846 2847 #include <linux/err.h> 2848 2849 /* needed for stackable file system support */ 2850 extern loff_t default_llseek(struct file *file, loff_t offset, int whence); 2851 2852 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence); 2853 2854 extern int inode_init_always(struct super_block *, struct inode *); 2855 extern void inode_init_once(struct inode *); 2856 extern void address_space_init_once(struct address_space *mapping); 2857 extern struct inode * igrab(struct inode *); 2858 extern ino_t iunique(struct super_block *, ino_t); 2859 extern int inode_needs_sync(struct inode *inode); 2860 extern int generic_delete_inode(struct inode *inode); 2861 static inline int generic_drop_inode(struct inode *inode) 2862 { 2863 return !inode->i_nlink || inode_unhashed(inode); 2864 } 2865 2866 extern struct inode *ilookup5_nowait(struct super_block *sb, 2867 unsigned long hashval, int (*test)(struct inode *, void *), 2868 void *data); 2869 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval, 2870 int (*test)(struct inode *, void *), void *data); 2871 extern struct inode *ilookup(struct super_block *sb, unsigned long ino); 2872 2873 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *); 2874 extern struct inode * iget_locked(struct super_block *, unsigned long); 2875 extern struct inode *find_inode_nowait(struct super_block *, 2876 unsigned long, 2877 int (*match)(struct inode *, 2878 unsigned long, void *), 2879 void *data); 2880 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *); 2881 extern int insert_inode_locked(struct inode *); 2882 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2883 extern void lockdep_annotate_inode_mutex_key(struct inode *inode); 2884 #else 2885 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { }; 2886 #endif 2887 extern void unlock_new_inode(struct inode *); 2888 extern unsigned int get_next_ino(void); 2889 extern void evict_inodes(struct super_block *sb); 2890 2891 extern void __iget(struct inode * inode); 2892 extern void iget_failed(struct inode *); 2893 extern void clear_inode(struct inode *); 2894 extern void __destroy_inode(struct inode *); 2895 extern struct inode *new_inode_pseudo(struct super_block *sb); 2896 extern struct inode *new_inode(struct super_block *sb); 2897 extern void free_inode_nonrcu(struct inode *inode); 2898 extern int should_remove_suid(struct dentry *); 2899 extern int file_remove_privs(struct file *); 2900 2901 extern void __insert_inode_hash(struct inode *, unsigned long hashval); 2902 static inline void insert_inode_hash(struct inode *inode) 2903 { 2904 __insert_inode_hash(inode, inode->i_ino); 2905 } 2906 2907 extern void __remove_inode_hash(struct inode *); 2908 static inline void remove_inode_hash(struct inode *inode) 2909 { 2910 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash)) 2911 __remove_inode_hash(inode); 2912 } 2913 2914 extern void inode_sb_list_add(struct inode *inode); 2915 2916 #ifdef CONFIG_BLOCK 2917 extern int bdev_read_only(struct block_device *); 2918 #endif 2919 extern int set_blocksize(struct block_device *, int); 2920 extern int sb_set_blocksize(struct super_block *, int); 2921 extern int sb_min_blocksize(struct super_block *, int); 2922 2923 extern int generic_file_mmap(struct file *, struct vm_area_struct *); 2924 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *); 2925 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *); 2926 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *); 2927 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *); 2928 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *); 2929 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *); 2930 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t); 2931 2932 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos, 2933 rwf_t flags); 2934 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos, 2935 rwf_t flags); 2936 2937 /* fs/block_dev.c */ 2938 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to); 2939 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from); 2940 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end, 2941 int datasync); 2942 extern void block_sync_page(struct page *page); 2943 2944 /* fs/splice.c */ 2945 extern ssize_t generic_file_splice_read(struct file *, loff_t *, 2946 struct pipe_inode_info *, size_t, unsigned int); 2947 extern ssize_t iter_file_splice_write(struct pipe_inode_info *, 2948 struct file *, loff_t *, size_t, unsigned int); 2949 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe, 2950 struct file *out, loff_t *, size_t len, unsigned int flags); 2951 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out, 2952 loff_t *opos, size_t len, unsigned int flags); 2953 2954 2955 extern void 2956 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping); 2957 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence); 2958 extern loff_t no_llseek(struct file *file, loff_t offset, int whence); 2959 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize); 2960 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence); 2961 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset, 2962 int whence, loff_t maxsize, loff_t eof); 2963 extern loff_t fixed_size_llseek(struct file *file, loff_t offset, 2964 int whence, loff_t size); 2965 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t); 2966 extern loff_t no_seek_end_llseek(struct file *, loff_t, int); 2967 extern int generic_file_open(struct inode * inode, struct file * filp); 2968 extern int nonseekable_open(struct inode * inode, struct file * filp); 2969 2970 #ifdef CONFIG_BLOCK 2971 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode, 2972 loff_t file_offset); 2973 2974 enum { 2975 /* need locking between buffered and direct access */ 2976 DIO_LOCKING = 0x01, 2977 2978 /* filesystem does not support filling holes */ 2979 DIO_SKIP_HOLES = 0x02, 2980 2981 /* filesystem can handle aio writes beyond i_size */ 2982 DIO_ASYNC_EXTEND = 0x04, 2983 2984 /* inode/fs/bdev does not need truncate protection */ 2985 DIO_SKIP_DIO_COUNT = 0x08, 2986 }; 2987 2988 void dio_end_io(struct bio *bio); 2989 void dio_warn_stale_pagecache(struct file *filp); 2990 2991 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode, 2992 struct block_device *bdev, struct iov_iter *iter, 2993 get_block_t get_block, 2994 dio_iodone_t end_io, dio_submit_t submit_io, 2995 int flags); 2996 2997 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb, 2998 struct inode *inode, 2999 struct iov_iter *iter, 3000 get_block_t get_block) 3001 { 3002 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter, 3003 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES); 3004 } 3005 #endif 3006 3007 void inode_dio_wait(struct inode *inode); 3008 3009 /* 3010 * inode_dio_begin - signal start of a direct I/O requests 3011 * @inode: inode the direct I/O happens on 3012 * 3013 * This is called once we've finished processing a direct I/O request, 3014 * and is used to wake up callers waiting for direct I/O to be quiesced. 3015 */ 3016 static inline void inode_dio_begin(struct inode *inode) 3017 { 3018 atomic_inc(&inode->i_dio_count); 3019 } 3020 3021 /* 3022 * inode_dio_end - signal finish of a direct I/O requests 3023 * @inode: inode the direct I/O happens on 3024 * 3025 * This is called once we've finished processing a direct I/O request, 3026 * and is used to wake up callers waiting for direct I/O to be quiesced. 3027 */ 3028 static inline void inode_dio_end(struct inode *inode) 3029 { 3030 if (atomic_dec_and_test(&inode->i_dio_count)) 3031 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP); 3032 } 3033 3034 extern void inode_set_flags(struct inode *inode, unsigned int flags, 3035 unsigned int mask); 3036 3037 extern const struct file_operations generic_ro_fops; 3038 3039 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m)) 3040 3041 extern int readlink_copy(char __user *, int, const char *); 3042 extern int page_readlink(struct dentry *, char __user *, int); 3043 extern const char *page_get_link(struct dentry *, struct inode *, 3044 struct delayed_call *); 3045 extern void page_put_link(void *); 3046 extern int __page_symlink(struct inode *inode, const char *symname, int len, 3047 int nofs); 3048 extern int page_symlink(struct inode *inode, const char *symname, int len); 3049 extern const struct inode_operations page_symlink_inode_operations; 3050 extern void kfree_link(void *); 3051 extern void generic_fillattr(struct inode *, struct kstat *); 3052 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int); 3053 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int); 3054 void __inode_add_bytes(struct inode *inode, loff_t bytes); 3055 void inode_add_bytes(struct inode *inode, loff_t bytes); 3056 void __inode_sub_bytes(struct inode *inode, loff_t bytes); 3057 void inode_sub_bytes(struct inode *inode, loff_t bytes); 3058 static inline loff_t __inode_get_bytes(struct inode *inode) 3059 { 3060 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes; 3061 } 3062 loff_t inode_get_bytes(struct inode *inode); 3063 void inode_set_bytes(struct inode *inode, loff_t bytes); 3064 const char *simple_get_link(struct dentry *, struct inode *, 3065 struct delayed_call *); 3066 extern const struct inode_operations simple_symlink_inode_operations; 3067 3068 extern int iterate_dir(struct file *, struct dir_context *); 3069 3070 extern int vfs_statx(int, const char __user *, int, struct kstat *, u32); 3071 extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int); 3072 3073 static inline int vfs_stat(const char __user *filename, struct kstat *stat) 3074 { 3075 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT, 3076 stat, STATX_BASIC_STATS); 3077 } 3078 static inline int vfs_lstat(const char __user *name, struct kstat *stat) 3079 { 3080 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT, 3081 stat, STATX_BASIC_STATS); 3082 } 3083 static inline int vfs_fstatat(int dfd, const char __user *filename, 3084 struct kstat *stat, int flags) 3085 { 3086 return vfs_statx(dfd, filename, flags | AT_NO_AUTOMOUNT, 3087 stat, STATX_BASIC_STATS); 3088 } 3089 static inline int vfs_fstat(int fd, struct kstat *stat) 3090 { 3091 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0); 3092 } 3093 3094 3095 extern const char *vfs_get_link(struct dentry *, struct delayed_call *); 3096 extern int vfs_readlink(struct dentry *, char __user *, int); 3097 3098 extern int __generic_block_fiemap(struct inode *inode, 3099 struct fiemap_extent_info *fieinfo, 3100 loff_t start, loff_t len, 3101 get_block_t *get_block); 3102 extern int generic_block_fiemap(struct inode *inode, 3103 struct fiemap_extent_info *fieinfo, u64 start, 3104 u64 len, get_block_t *get_block); 3105 3106 extern struct file_system_type *get_filesystem(struct file_system_type *fs); 3107 extern void put_filesystem(struct file_system_type *fs); 3108 extern struct file_system_type *get_fs_type(const char *name); 3109 extern struct super_block *get_super(struct block_device *); 3110 extern struct super_block *get_super_thawed(struct block_device *); 3111 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev); 3112 extern struct super_block *get_active_super(struct block_device *bdev); 3113 extern void drop_super(struct super_block *sb); 3114 extern void drop_super_exclusive(struct super_block *sb); 3115 extern void iterate_supers(void (*)(struct super_block *, void *), void *); 3116 extern void iterate_supers_type(struct file_system_type *, 3117 void (*)(struct super_block *, void *), void *); 3118 3119 extern int dcache_dir_open(struct inode *, struct file *); 3120 extern int dcache_dir_close(struct inode *, struct file *); 3121 extern loff_t dcache_dir_lseek(struct file *, loff_t, int); 3122 extern int dcache_readdir(struct file *, struct dir_context *); 3123 extern int simple_setattr(struct dentry *, struct iattr *); 3124 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int); 3125 extern int simple_statfs(struct dentry *, struct kstatfs *); 3126 extern int simple_open(struct inode *inode, struct file *file); 3127 extern int simple_link(struct dentry *, struct inode *, struct dentry *); 3128 extern int simple_unlink(struct inode *, struct dentry *); 3129 extern int simple_rmdir(struct inode *, struct dentry *); 3130 extern int simple_rename(struct inode *, struct dentry *, 3131 struct inode *, struct dentry *, unsigned int); 3132 extern int noop_fsync(struct file *, loff_t, loff_t, int); 3133 extern int simple_empty(struct dentry *); 3134 extern int simple_readpage(struct file *file, struct page *page); 3135 extern int simple_write_begin(struct file *file, struct address_space *mapping, 3136 loff_t pos, unsigned len, unsigned flags, 3137 struct page **pagep, void **fsdata); 3138 extern int simple_write_end(struct file *file, struct address_space *mapping, 3139 loff_t pos, unsigned len, unsigned copied, 3140 struct page *page, void *fsdata); 3141 extern int always_delete_dentry(const struct dentry *); 3142 extern struct inode *alloc_anon_inode(struct super_block *); 3143 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **); 3144 extern const struct dentry_operations simple_dentry_operations; 3145 3146 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags); 3147 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *); 3148 extern const struct file_operations simple_dir_operations; 3149 extern const struct inode_operations simple_dir_inode_operations; 3150 extern void make_empty_dir_inode(struct inode *inode); 3151 extern bool is_empty_dir_inode(struct inode *inode); 3152 struct tree_descr { const char *name; const struct file_operations *ops; int mode; }; 3153 struct dentry *d_alloc_name(struct dentry *, const char *); 3154 extern int simple_fill_super(struct super_block *, unsigned long, 3155 const struct tree_descr *); 3156 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count); 3157 extern void simple_release_fs(struct vfsmount **mount, int *count); 3158 3159 extern ssize_t simple_read_from_buffer(void __user *to, size_t count, 3160 loff_t *ppos, const void *from, size_t available); 3161 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos, 3162 const void __user *from, size_t count); 3163 3164 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int); 3165 extern int generic_file_fsync(struct file *, loff_t, loff_t, int); 3166 3167 extern int generic_check_addressable(unsigned, u64); 3168 3169 #ifdef CONFIG_MIGRATION 3170 extern int buffer_migrate_page(struct address_space *, 3171 struct page *, struct page *, 3172 enum migrate_mode); 3173 #else 3174 #define buffer_migrate_page NULL 3175 #endif 3176 3177 extern int setattr_prepare(struct dentry *, struct iattr *); 3178 extern int inode_newsize_ok(const struct inode *, loff_t offset); 3179 extern void setattr_copy(struct inode *inode, const struct iattr *attr); 3180 3181 extern int file_update_time(struct file *file); 3182 3183 static inline bool io_is_direct(struct file *filp) 3184 { 3185 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host); 3186 } 3187 3188 static inline bool vma_is_dax(struct vm_area_struct *vma) 3189 { 3190 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host); 3191 } 3192 3193 static inline bool vma_is_fsdax(struct vm_area_struct *vma) 3194 { 3195 struct inode *inode; 3196 3197 if (!vma->vm_file) 3198 return false; 3199 if (!vma_is_dax(vma)) 3200 return false; 3201 inode = file_inode(vma->vm_file); 3202 if (S_ISCHR(inode->i_mode)) 3203 return false; /* device-dax */ 3204 return true; 3205 } 3206 3207 static inline int iocb_flags(struct file *file) 3208 { 3209 int res = 0; 3210 if (file->f_flags & O_APPEND) 3211 res |= IOCB_APPEND; 3212 if (io_is_direct(file)) 3213 res |= IOCB_DIRECT; 3214 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host)) 3215 res |= IOCB_DSYNC; 3216 if (file->f_flags & __O_SYNC) 3217 res |= IOCB_SYNC; 3218 return res; 3219 } 3220 3221 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags) 3222 { 3223 if (unlikely(flags & ~RWF_SUPPORTED)) 3224 return -EOPNOTSUPP; 3225 3226 if (flags & RWF_NOWAIT) { 3227 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT)) 3228 return -EOPNOTSUPP; 3229 ki->ki_flags |= IOCB_NOWAIT; 3230 } 3231 if (flags & RWF_HIPRI) 3232 ki->ki_flags |= IOCB_HIPRI; 3233 if (flags & RWF_DSYNC) 3234 ki->ki_flags |= IOCB_DSYNC; 3235 if (flags & RWF_SYNC) 3236 ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC); 3237 if (flags & RWF_APPEND) 3238 ki->ki_flags |= IOCB_APPEND; 3239 return 0; 3240 } 3241 3242 static inline ino_t parent_ino(struct dentry *dentry) 3243 { 3244 ino_t res; 3245 3246 /* 3247 * Don't strictly need d_lock here? If the parent ino could change 3248 * then surely we'd have a deeper race in the caller? 3249 */ 3250 spin_lock(&dentry->d_lock); 3251 res = dentry->d_parent->d_inode->i_ino; 3252 spin_unlock(&dentry->d_lock); 3253 return res; 3254 } 3255 3256 /* Transaction based IO helpers */ 3257 3258 /* 3259 * An argresp is stored in an allocated page and holds the 3260 * size of the argument or response, along with its content 3261 */ 3262 struct simple_transaction_argresp { 3263 ssize_t size; 3264 char data[0]; 3265 }; 3266 3267 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp)) 3268 3269 char *simple_transaction_get(struct file *file, const char __user *buf, 3270 size_t size); 3271 ssize_t simple_transaction_read(struct file *file, char __user *buf, 3272 size_t size, loff_t *pos); 3273 int simple_transaction_release(struct inode *inode, struct file *file); 3274 3275 void simple_transaction_set(struct file *file, size_t n); 3276 3277 /* 3278 * simple attribute files 3279 * 3280 * These attributes behave similar to those in sysfs: 3281 * 3282 * Writing to an attribute immediately sets a value, an open file can be 3283 * written to multiple times. 3284 * 3285 * Reading from an attribute creates a buffer from the value that might get 3286 * read with multiple read calls. When the attribute has been read 3287 * completely, no further read calls are possible until the file is opened 3288 * again. 3289 * 3290 * All attributes contain a text representation of a numeric value 3291 * that are accessed with the get() and set() functions. 3292 */ 3293 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \ 3294 static int __fops ## _open(struct inode *inode, struct file *file) \ 3295 { \ 3296 __simple_attr_check_format(__fmt, 0ull); \ 3297 return simple_attr_open(inode, file, __get, __set, __fmt); \ 3298 } \ 3299 static const struct file_operations __fops = { \ 3300 .owner = THIS_MODULE, \ 3301 .open = __fops ## _open, \ 3302 .release = simple_attr_release, \ 3303 .read = simple_attr_read, \ 3304 .write = simple_attr_write, \ 3305 .llseek = generic_file_llseek, \ 3306 } 3307 3308 static inline __printf(1, 2) 3309 void __simple_attr_check_format(const char *fmt, ...) 3310 { 3311 /* don't do anything, just let the compiler check the arguments; */ 3312 } 3313 3314 int simple_attr_open(struct inode *inode, struct file *file, 3315 int (*get)(void *, u64 *), int (*set)(void *, u64), 3316 const char *fmt); 3317 int simple_attr_release(struct inode *inode, struct file *file); 3318 ssize_t simple_attr_read(struct file *file, char __user *buf, 3319 size_t len, loff_t *ppos); 3320 ssize_t simple_attr_write(struct file *file, const char __user *buf, 3321 size_t len, loff_t *ppos); 3322 3323 struct ctl_table; 3324 int proc_nr_files(struct ctl_table *table, int write, 3325 void __user *buffer, size_t *lenp, loff_t *ppos); 3326 int proc_nr_dentry(struct ctl_table *table, int write, 3327 void __user *buffer, size_t *lenp, loff_t *ppos); 3328 int proc_nr_inodes(struct ctl_table *table, int write, 3329 void __user *buffer, size_t *lenp, loff_t *ppos); 3330 int __init get_filesystem_list(char *buf); 3331 3332 #define __FMODE_EXEC ((__force int) FMODE_EXEC) 3333 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY) 3334 3335 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE]) 3336 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \ 3337 (flag & __FMODE_NONOTIFY))) 3338 3339 static inline bool is_sxid(umode_t mode) 3340 { 3341 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP)); 3342 } 3343 3344 static inline int check_sticky(struct inode *dir, struct inode *inode) 3345 { 3346 if (!(dir->i_mode & S_ISVTX)) 3347 return 0; 3348 3349 return __check_sticky(dir, inode); 3350 } 3351 3352 static inline void inode_has_no_xattr(struct inode *inode) 3353 { 3354 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC)) 3355 inode->i_flags |= S_NOSEC; 3356 } 3357 3358 static inline bool is_root_inode(struct inode *inode) 3359 { 3360 return inode == inode->i_sb->s_root->d_inode; 3361 } 3362 3363 static inline bool dir_emit(struct dir_context *ctx, 3364 const char *name, int namelen, 3365 u64 ino, unsigned type) 3366 { 3367 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0; 3368 } 3369 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx) 3370 { 3371 return ctx->actor(ctx, ".", 1, ctx->pos, 3372 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0; 3373 } 3374 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx) 3375 { 3376 return ctx->actor(ctx, "..", 2, ctx->pos, 3377 parent_ino(file->f_path.dentry), DT_DIR) == 0; 3378 } 3379 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx) 3380 { 3381 if (ctx->pos == 0) { 3382 if (!dir_emit_dot(file, ctx)) 3383 return false; 3384 ctx->pos = 1; 3385 } 3386 if (ctx->pos == 1) { 3387 if (!dir_emit_dotdot(file, ctx)) 3388 return false; 3389 ctx->pos = 2; 3390 } 3391 return true; 3392 } 3393 static inline bool dir_relax(struct inode *inode) 3394 { 3395 inode_unlock(inode); 3396 inode_lock(inode); 3397 return !IS_DEADDIR(inode); 3398 } 3399 3400 static inline bool dir_relax_shared(struct inode *inode) 3401 { 3402 inode_unlock_shared(inode); 3403 inode_lock_shared(inode); 3404 return !IS_DEADDIR(inode); 3405 } 3406 3407 extern bool path_noexec(const struct path *path); 3408 extern void inode_nohighmem(struct inode *inode); 3409 3410 #endif /* _LINUX_FS_H */ 3411
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.