1#ifndef _LINUX_FS_H 2#define _LINUX_FS_H 3 4/* 5 * This file has definitions for some important file table 6 * structures etc. 7 */ 8 9#include <linux/limits.h> 10#include <linux/ioctl.h> 11 12/* 13 * It's silly to have NR_OPEN bigger than NR_FILE, but you can change 14 * the file limit at runtime and only root can increase the per-process 15 * nr_file rlimit, so it's safe to set up a ridiculously high absolute 16 * upper limit on files-per-process. 17 * 18 * Some programs (notably those using select()) may have to be 19 * recompiled to take full advantage of the new limits.. 20 */ 21 22/* Fixed constants first: */ 23#undef NR_OPEN 24extern int sysctl_nr_open; 25#define INR_OPEN 1024 /* Initial setting for nfile rlimits */ 26 27#define BLOCK_SIZE_BITS 10 28#define BLOCK_SIZE (1<<BLOCK_SIZE_BITS) 29 30#define SEEK_SET 0 /* seek relative to beginning of file */ 31#define SEEK_CUR 1 /* seek relative to current file position */ 32#define SEEK_END 2 /* seek relative to end of file */ 33#define SEEK_MAX SEEK_END 34 35/* And dynamically-tunable limits and defaults: */ 36struct files_stat_struct { 37 int nr_files; /* read only */ 38 int nr_free_files; /* read only */ 39 int max_files; /* tunable */ 40}; 41extern struct files_stat_struct files_stat; 42extern int get_max_files(void); 43 44struct inodes_stat_t { 45 int nr_inodes; 46 int nr_unused; 47 int dummy[5]; /* padding for sysctl ABI compatibility */ 48}; 49extern struct inodes_stat_t inodes_stat; 50 51extern int leases_enable, lease_break_time; 52 53#ifdef CONFIG_DNOTIFY 54extern int dir_notify_enable; 55#endif 56 57#define NR_FILE 8192 /* this can well be larger on a larger system */ 58 59#define MAY_EXEC 1 60#define MAY_WRITE 2 61#define MAY_READ 4 62#define MAY_APPEND 8 63#define MAY_ACCESS 16 64#define MAY_OPEN 32 65 66#define FMODE_READ ((__force fmode_t)1) 67#define FMODE_WRITE ((__force fmode_t)2) 68 69/* Internal kernel extensions */ 70#define FMODE_LSEEK ((__force fmode_t)4) 71#define FMODE_PREAD ((__force fmode_t)8) 72#define FMODE_PWRITE FMODE_PREAD /* These go hand in hand */ 73 74/* File is being opened for execution. Primary users of this flag are 75 distributed filesystems that can use it to achieve correct ETXTBUSY 76 behavior for cross-node execution/opening_for_writing of files */ 77#define FMODE_EXEC ((__force fmode_t)16) 78 79#define FMODE_NDELAY ((__force fmode_t)32) 80#define FMODE_EXCL ((__force fmode_t)64) 81#define FMODE_WRITE_IOCTL ((__force fmode_t)128) 82#define FMODE_NDELAY_NOW ((__force fmode_t)256) 83 84#define RW_MASK 1 85#define RWA_MASK 2 86#define READ 0 87#define WRITE 1 88#define READA 2 /* read-ahead - don't block if no resources */ 89#define SWRITE 3 /* for ll_rw_block() - wait for buffer lock */ 90#define READ_SYNC (READ | (1 << BIO_RW_SYNC)) 91#define READ_META (READ | (1 << BIO_RW_META)) 92#define WRITE_SYNC (WRITE | (1 << BIO_RW_SYNC)) 93#define SWRITE_SYNC (SWRITE | (1 << BIO_RW_SYNC)) 94#define WRITE_BARRIER (WRITE | (1 << BIO_RW_BARRIER)) 95#define DISCARD_NOBARRIER (1 << BIO_RW_DISCARD) 96#define DISCARD_BARRIER ((1 << BIO_RW_DISCARD) | (1 << BIO_RW_BARRIER)) 97 98#define SEL_IN 1 99#define SEL_OUT 2 100#define SEL_EX 4 101 102/* public flags for file_system_type */ 103#define FS_REQUIRES_DEV 1 104#define FS_BINARY_MOUNTDATA 2 105#define FS_HAS_SUBTYPE 4 106#define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */ 107#define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() 108 * during rename() internally. 109 */ 110 111/* 112 * These are the fs-independent mount-flags: up to 32 flags are supported 113 */ 114#define MS_RDONLY 1 /* Mount read-only */ 115#define MS_NOSUID 2 /* Ignore suid and sgid bits */ 116#define MS_NODEV 4 /* Disallow access to device special files */ 117#define MS_NOEXEC 8 /* Disallow program execution */ 118#define MS_SYNCHRONOUS 16 /* Writes are synced at once */ 119#define MS_REMOUNT 32 /* Alter flags of a mounted FS */ 120#define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */ 121#define MS_DIRSYNC 128 /* Directory modifications are synchronous */ 122#define MS_NOATIME 1024 /* Do not update access times. */ 123#define MS_NODIRATIME 2048 /* Do not update directory access times */ 124#define MS_BIND 4096 125#define MS_MOVE 8192 126#define MS_REC 16384 127#define MS_VERBOSE 32768 /* War is peace. Verbosity is silence. 128 MS_VERBOSE is deprecated. */ 129#define MS_SILENT 32768 130#define MS_POSIXACL (1<<16) /* VFS does not apply the umask */ 131#define MS_UNBINDABLE (1<<17) /* change to unbindable */ 132#define MS_PRIVATE (1<<18) /* change to private */ 133#define MS_SLAVE (1<<19) /* change to slave */ 134#define MS_SHARED (1<<20) /* change to shared */ 135#define MS_RELATIME (1<<21) /* Update atime relative to mtime/ctime. */ 136#define MS_KERNMOUNT (1<<22) /* this is a kern_mount call */ 137#define MS_I_VERSION (1<<23) /* Update inode I_version field */ 138#define MS_ACTIVE (1<<30) 139#define MS_NOUSER (1<<31) 140 141/* 142 * Superblock flags that can be altered by MS_REMOUNT 143 */ 144#define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION) 145 146/* 147 * Old magic mount flag and mask 148 */ 149#define MS_MGC_VAL 0xC0ED0000 150#define MS_MGC_MSK 0xffff0000 151 152/* Inode flags - they have nothing to superblock flags now */ 153 154#define S_SYNC 1 /* Writes are synced at once */ 155#define S_NOATIME 2 /* Do not update access times */ 156#define S_APPEND 4 /* Append-only file */ 157#define S_IMMUTABLE 8 /* Immutable file */ 158#define S_DEAD 16 /* removed, but still open directory */ 159#define S_NOQUOTA 32 /* Inode is not counted to quota */ 160#define S_DIRSYNC 64 /* Directory modifications are synchronous */ 161#define S_NOCMTIME 128 /* Do not update file c/mtime */ 162#define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */ 163#define S_PRIVATE 512 /* Inode is fs-internal */ 164 165/* 166 * Note that nosuid etc flags are inode-specific: setting some file-system 167 * flags just means all the inodes inherit those flags by default. It might be 168 * possible to override it selectively if you really wanted to with some 169 * ioctl() that is not currently implemented. 170 * 171 * Exception: MS_RDONLY is always applied to the entire file system. 172 * 173 * Unfortunately, it is possible to change a filesystems flags with it mounted 174 * with files in use. This means that all of the inodes will not have their 175 * i_flags updated. Hence, i_flags no longer inherit the superblock mount 176 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org 177 */ 178#define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg)) 179 180#define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY) 181#define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \ 182 ((inode)->i_flags & S_SYNC)) 183#define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \ 184 ((inode)->i_flags & (S_SYNC|S_DIRSYNC))) 185#define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK) 186#define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME) 187#define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION) 188 189#define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA) 190#define IS_APPEND(inode) ((inode)->i_flags & S_APPEND) 191#define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE) 192#define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL) 193 194#define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD) 195#define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME) 196#define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE) 197#define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE) 198 199/* the read-only stuff doesn't really belong here, but any other place is 200 probably as bad and I don't want to create yet another include file. */ 201 202#define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */ 203#define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */ 204#define BLKRRPART _IO(0x12,95) /* re-read partition table */ 205#define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */ 206#define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */ 207#define BLKRASET _IO(0x12,98) /* set read ahead for block device */ 208#define BLKRAGET _IO(0x12,99) /* get current read ahead setting */ 209#define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */ 210#define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */ 211#define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */ 212#define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */ 213#define BLKSSZGET _IO(0x12,104)/* get block device sector size */ 214#if 0 215#define BLKPG _IO(0x12,105)/* See blkpg.h */ 216 217/* Some people are morons. Do not use sizeof! */ 218 219#define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */ 220#define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */ 221/* This was here just to show that the number is taken - 222 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */ 223#endif 224/* A jump here: 108-111 have been used for various private purposes. */ 225#define BLKBSZGET _IOR(0x12,112,size_t) 226#define BLKBSZSET _IOW(0x12,113,size_t) 227#define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */ 228#define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup) 229#define BLKTRACESTART _IO(0x12,116) 230#define BLKTRACESTOP _IO(0x12,117) 231#define BLKTRACETEARDOWN _IO(0x12,118) 232#define BLKDISCARD _IO(0x12,119) 233 234#define BMAP_IOCTL 1 /* obsolete - kept for compatibility */ 235#define FIBMAP _IO(0x00,1) /* bmap access */ 236#define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */ 237 238#define FS_IOC_GETFLAGS _IOR('f', 1, long) 239#define FS_IOC_SETFLAGS _IOW('f', 2, long) 240#define FS_IOC_GETVERSION _IOR('v', 1, long) 241#define FS_IOC_SETVERSION _IOW('v', 2, long) 242#define FS_IOC_FIEMAP _IOWR('f', 11, struct fiemap) 243#define FS_IOC32_GETFLAGS _IOR('f', 1, int) 244#define FS_IOC32_SETFLAGS _IOW('f', 2, int) 245#define FS_IOC32_GETVERSION _IOR('v', 1, int) 246#define FS_IOC32_SETVERSION _IOW('v', 2, int) 247 248/* 249 * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS) 250 */ 251#define FS_SECRM_FL 0x00000001 /* Secure deletion */ 252#define FS_UNRM_FL 0x00000002 /* Undelete */ 253#define FS_COMPR_FL 0x00000004 /* Compress file */ 254#define FS_SYNC_FL 0x00000008 /* Synchronous updates */ 255#define FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 256#define FS_APPEND_FL 0x00000020 /* writes to file may only append */ 257#define FS_NODUMP_FL 0x00000040 /* do not dump file */ 258#define FS_NOATIME_FL 0x00000080 /* do not update atime */ 259/* Reserved for compression usage... */ 260#define FS_DIRTY_FL 0x00000100 261#define FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 262#define FS_NOCOMP_FL 0x00000400 /* Don't compress */ 263#define FS_ECOMPR_FL 0x00000800 /* Compression error */ 264/* End compression flags --- maybe not all used */ 265#define FS_BTREE_FL 0x00001000 /* btree format dir */ 266#define FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 267#define FS_IMAGIC_FL 0x00002000 /* AFS directory */ 268#define FS_JOURNAL_DATA_FL 0x00004000 /* Reserved for ext3 */ 269#define FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 270#define FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 271#define FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 272#define FS_EXTENT_FL 0x00080000 /* Extents */ 273#define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */ 274#define FS_RESERVED_FL 0x80000000 /* reserved for ext2 lib */ 275 276#define FS_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */ 277#define FS_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */ 278 279 280#define SYNC_FILE_RANGE_WAIT_BEFORE 1 281#define SYNC_FILE_RANGE_WRITE 2 282#define SYNC_FILE_RANGE_WAIT_AFTER 4 283 284#ifdef __KERNEL__ 285 286#include <linux/linkage.h> 287#include <linux/wait.h> 288#include <linux/types.h> 289#include <linux/kdev_t.h> 290#include <linux/dcache.h> 291#include <linux/path.h> 292#include <linux/stat.h> 293#include <linux/cache.h> 294#include <linux/kobject.h> 295#include <linux/list.h> 296#include <linux/radix-tree.h> 297#include <linux/prio_tree.h> 298#include <linux/init.h> 299#include <linux/pid.h> 300#include <linux/mutex.h> 301#include <linux/capability.h> 302#include <linux/semaphore.h> 303#include <linux/fiemap.h> 304 305#include <asm/atomic.h> 306#include <asm/byteorder.h> 307 308struct export_operations; 309struct hd_geometry; 310struct iovec; 311struct nameidata; 312struct kiocb; 313struct pipe_inode_info; 314struct poll_table_struct; 315struct kstatfs; 316struct vm_area_struct; 317struct vfsmount; 318 319extern void __init inode_init(void); 320extern void __init inode_init_early(void); 321extern void __init files_init(unsigned long); 322 323struct buffer_head; 324typedef int (get_block_t)(struct inode *inode, sector_t iblock, 325 struct buffer_head *bh_result, int create); 326typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset, 327 ssize_t bytes, void *private); 328 329/* 330 * Attribute flags. These should be or-ed together to figure out what 331 * has been changed! 332 */ 333#define ATTR_MODE (1 << 0) 334#define ATTR_UID (1 << 1) 335#define ATTR_GID (1 << 2) 336#define ATTR_SIZE (1 << 3) 337#define ATTR_ATIME (1 << 4) 338#define ATTR_MTIME (1 << 5) 339#define ATTR_CTIME (1 << 6) 340#define ATTR_ATIME_SET (1 << 7) 341#define ATTR_MTIME_SET (1 << 8) 342#define ATTR_FORCE (1 << 9) /* Not a change, but a change it */ 343#define ATTR_ATTR_FLAG (1 << 10) 344#define ATTR_KILL_SUID (1 << 11) 345#define ATTR_KILL_SGID (1 << 12) 346#define ATTR_FILE (1 << 13) 347#define ATTR_KILL_PRIV (1 << 14) 348#define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */ 349#define ATTR_TIMES_SET (1 << 16) 350 351/* 352 * This is the Inode Attributes structure, used for notify_change(). It 353 * uses the above definitions as flags, to know which values have changed. 354 * Also, in this manner, a Filesystem can look at only the values it cares 355 * about. Basically, these are the attributes that the VFS layer can 356 * request to change from the FS layer. 357 * 358 * Derek Atkins <warlord@MIT.EDU> 94-10-20 359 */ 360struct iattr { 361 unsigned int ia_valid; 362 umode_t ia_mode; 363 uid_t ia_uid; 364 gid_t ia_gid; 365 loff_t ia_size; 366 struct timespec ia_atime; 367 struct timespec ia_mtime; 368 struct timespec ia_ctime; 369 370 /* 371 * Not an attribute, but an auxilary info for filesystems wanting to 372 * implement an ftruncate() like method. NOTE: filesystem should 373 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL). 374 */ 375 struct file *ia_file; 376}; 377 378/* 379 * Includes for diskquotas. 380 */ 381#include <linux/quota.h> 382 383/** 384 * enum positive_aop_returns - aop return codes with specific semantics 385 * 386 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has 387 * completed, that the page is still locked, and 388 * should be considered active. The VM uses this hint 389 * to return the page to the active list -- it won't 390 * be a candidate for writeback again in the near 391 * future. Other callers must be careful to unlock 392 * the page if they get this return. Returned by 393 * writepage(); 394 * 395 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has 396 * unlocked it and the page might have been truncated. 397 * The caller should back up to acquiring a new page and 398 * trying again. The aop will be taking reasonable 399 * precautions not to livelock. If the caller held a page 400 * reference, it should drop it before retrying. Returned 401 * by readpage(). 402 * 403 * address_space_operation functions return these large constants to indicate 404 * special semantics to the caller. These are much larger than the bytes in a 405 * page to allow for functions that return the number of bytes operated on in a 406 * given page. 407 */ 408 409enum positive_aop_returns { 410 AOP_WRITEPAGE_ACTIVATE = 0x80000, 411 AOP_TRUNCATED_PAGE = 0x80001, 412}; 413 414#define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */ 415#define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */ 416 417/* 418 * oh the beauties of C type declarations. 419 */ 420struct page; 421struct address_space; 422struct writeback_control; 423 424struct iov_iter { 425 const struct iovec *iov; 426 unsigned long nr_segs; 427 size_t iov_offset; 428 size_t count; 429}; 430 431size_t iov_iter_copy_from_user_atomic(struct page *page, 432 struct iov_iter *i, unsigned long offset, size_t bytes); 433size_t iov_iter_copy_from_user(struct page *page, 434 struct iov_iter *i, unsigned long offset, size_t bytes); 435void iov_iter_advance(struct iov_iter *i, size_t bytes); 436int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes); 437size_t iov_iter_single_seg_count(struct iov_iter *i); 438 439static inline void iov_iter_init(struct iov_iter *i, 440 const struct iovec *iov, unsigned long nr_segs, 441 size_t count, size_t written) 442{ 443 i->iov = iov; 444 i->nr_segs = nr_segs; 445 i->iov_offset = 0; 446 i->count = count + written; 447 448 iov_iter_advance(i, written); 449} 450 451static inline size_t iov_iter_count(struct iov_iter *i) 452{ 453 return i->count; 454} 455 456/* 457 * "descriptor" for what we're up to with a read. 458 * This allows us to use the same read code yet 459 * have multiple different users of the data that 460 * we read from a file. 461 * 462 * The simplest case just copies the data to user 463 * mode. 464 */ 465typedef struct { 466 size_t written; 467 size_t count; 468 union { 469 char __user *buf; 470 void *data; 471 } arg; 472 int error; 473} read_descriptor_t; 474 475typedef int (*read_actor_t)(read_descriptor_t *, struct page *, 476 unsigned long, unsigned long); 477 478struct address_space_operations { 479 int (*writepage)(struct page *page, struct writeback_control *wbc); 480 int (*readpage)(struct file *, struct page *); 481 void (*sync_page)(struct page *); 482 483 /* Write back some dirty pages from this mapping. */ 484 int (*writepages)(struct address_space *, struct writeback_control *); 485 486 /* Set a page dirty. Return true if this dirtied it */ 487 int (*set_page_dirty)(struct page *page); 488 489 int (*readpages)(struct file *filp, struct address_space *mapping, 490 struct list_head *pages, unsigned nr_pages); 491 492 /* 493 * ext3 requires that a successful prepare_write() call be followed 494 * by a commit_write() call - they must be balanced 495 */ 496 int (*prepare_write)(struct file *, struct page *, unsigned, unsigned); 497 int (*commit_write)(struct file *, struct page *, unsigned, unsigned); 498 499 int (*write_begin)(struct file *, struct address_space *mapping, 500 loff_t pos, unsigned len, unsigned flags, 501 struct page **pagep, void **fsdata); 502 int (*write_end)(struct file *, struct address_space *mapping, 503 loff_t pos, unsigned len, unsigned copied, 504 struct page *page, void *fsdata); 505 506 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */ 507 sector_t (*bmap)(struct address_space *, sector_t); 508 void (*invalidatepage) (struct page *, unsigned long); 509 int (*releasepage) (struct page *, gfp_t); 510 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov, 511 loff_t offset, unsigned long nr_segs); 512 int (*get_xip_mem)(struct address_space *, pgoff_t, int, 513 void **, unsigned long *); 514 /* migrate the contents of a page to the specified target */ 515 int (*migratepage) (struct address_space *, 516 struct page *, struct page *); 517 int (*launder_page) (struct page *); 518 int (*is_partially_uptodate) (struct page *, read_descriptor_t *, 519 unsigned long); 520}; 521 522/* 523 * pagecache_write_begin/pagecache_write_end must be used by general code 524 * to write into the pagecache. 525 */ 526int pagecache_write_begin(struct file *, struct address_space *mapping, 527 loff_t pos, unsigned len, unsigned flags, 528 struct page **pagep, void **fsdata); 529 530int pagecache_write_end(struct file *, struct address_space *mapping, 531 loff_t pos, unsigned len, unsigned copied, 532 struct page *page, void *fsdata); 533 534struct backing_dev_info; 535struct address_space { 536 struct inode *host; /* owner: inode, block_device */ 537 struct radix_tree_root page_tree; /* radix tree of all pages */ 538 spinlock_t tree_lock; /* and lock protecting it */ 539 unsigned int i_mmap_writable;/* count VM_SHARED mappings */ 540 struct prio_tree_root i_mmap; /* tree of private and shared mappings */ 541 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */ 542 spinlock_t i_mmap_lock; /* protect tree, count, list */ 543 unsigned int truncate_count; /* Cover race condition with truncate */ 544 unsigned long nrpages; /* number of total pages */ 545 pgoff_t writeback_index;/* writeback starts here */ 546 const struct address_space_operations *a_ops; /* methods */ 547 unsigned long flags; /* error bits/gfp mask */ 548 struct backing_dev_info *backing_dev_info; /* device readahead, etc */ 549 spinlock_t private_lock; /* for use by the address_space */ 550 struct list_head private_list; /* ditto */ 551 struct address_space *assoc_mapping; /* ditto */ 552} __attribute__((aligned(sizeof(long)))); 553 /* 554 * On most architectures that alignment is already the case; but 555 * must be enforced here for CRIS, to let the least signficant bit 556 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON. 557 */ 558 559struct block_device { 560 dev_t bd_dev; /* not a kdev_t - it's a search key */ 561 struct inode * bd_inode; /* will die */ 562 int bd_openers; 563 struct mutex bd_mutex; /* open/close mutex */ 564 struct semaphore bd_mount_sem; 565 struct list_head bd_inodes; 566 void * bd_holder; 567 int bd_holders; 568#ifdef CONFIG_SYSFS 569 struct list_head bd_holder_list; 570#endif 571 struct block_device * bd_contains; 572 unsigned bd_block_size; 573 struct hd_struct * bd_part; 574 /* number of times partitions within this device have been opened. */ 575 unsigned bd_part_count; 576 int bd_invalidated; 577 struct gendisk * bd_disk; 578 struct list_head bd_list; 579 struct backing_dev_info *bd_inode_backing_dev_info; 580 /* 581 * Private data. You must have bd_claim'ed the block_device 582 * to use this. NOTE: bd_claim allows an owner to claim 583 * the same device multiple times, the owner must take special 584 * care to not mess up bd_private for that case. 585 */ 586 unsigned long bd_private; 587}; 588 589/* 590 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache 591 * radix trees 592 */ 593#define PAGECACHE_TAG_DIRTY 0 594#define PAGECACHE_TAG_WRITEBACK 1 595 596int mapping_tagged(struct address_space *mapping, int tag); 597 598/* 599 * Might pages of this file be mapped into userspace? 600 */ 601static inline int mapping_mapped(struct address_space *mapping) 602{ 603 return !prio_tree_empty(&mapping->i_mmap) || 604 !list_empty(&mapping->i_mmap_nonlinear); 605} 606 607/* 608 * Might pages of this file have been modified in userspace? 609 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff 610 * marks vma as VM_SHARED if it is shared, and the file was opened for 611 * writing i.e. vma may be mprotected writable even if now readonly. 612 */ 613static inline int mapping_writably_mapped(struct address_space *mapping) 614{ 615 return mapping->i_mmap_writable != 0; 616} 617 618/* 619 * Use sequence counter to get consistent i_size on 32-bit processors. 620 */ 621#if BITS_PER_LONG==32 && defined(CONFIG_SMP) 622#include <linux/seqlock.h> 623#define __NEED_I_SIZE_ORDERED 624#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount) 625#else 626#define i_size_ordered_init(inode) do { } while (0) 627#endif 628 629struct inode { 630 struct hlist_node i_hash; 631 struct list_head i_list; 632 struct list_head i_sb_list; 633 struct list_head i_dentry; 634 unsigned long i_ino; 635 atomic_t i_count; 636 unsigned int i_nlink; 637 uid_t i_uid; 638 gid_t i_gid; 639 dev_t i_rdev; 640 u64 i_version; 641 loff_t i_size; 642#ifdef __NEED_I_SIZE_ORDERED 643 seqcount_t i_size_seqcount; 644#endif 645 struct timespec i_atime; 646 struct timespec i_mtime; 647 struct timespec i_ctime; 648 unsigned int i_blkbits; 649 blkcnt_t i_blocks; 650 unsigned short i_bytes; 651 umode_t i_mode; 652 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */ 653 struct mutex i_mutex; 654 struct rw_semaphore i_alloc_sem; 655 const struct inode_operations *i_op; 656 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */ 657 struct super_block *i_sb; 658 struct file_lock *i_flock; 659 struct address_space *i_mapping; 660 struct address_space i_data; 661#ifdef CONFIG_QUOTA 662 struct dquot *i_dquot[MAXQUOTAS]; 663#endif 664 struct list_head i_devices; 665 union { 666 struct pipe_inode_info *i_pipe; 667 struct block_device *i_bdev; 668 struct cdev *i_cdev; 669 }; 670 int i_cindex; 671 672 __u32 i_generation; 673 674#ifdef CONFIG_DNOTIFY 675 unsigned long i_dnotify_mask; /* Directory notify events */ 676 struct dnotify_struct *i_dnotify; /* for directory notifications */ 677#endif 678 679#ifdef CONFIG_INOTIFY 680 struct list_head inotify_watches; /* watches on this inode */ 681 struct mutex inotify_mutex; /* protects the watches list */ 682#endif 683 684 unsigned long i_state; 685 unsigned long dirtied_when; /* jiffies of first dirtying */ 686 687 unsigned int i_flags; 688 689 atomic_t i_writecount; 690#ifdef CONFIG_SECURITY 691 void *i_security; 692#endif 693 void *i_private; /* fs or device private pointer */ 694}; 695 696/* 697 * inode->i_mutex nesting subclasses for the lock validator: 698 * 699 * 0: the object of the current VFS operation 700 * 1: parent 701 * 2: child/target 702 * 3: quota file 703 * 704 * The locking order between these classes is 705 * parent -> child -> normal -> xattr -> quota 706 */ 707enum inode_i_mutex_lock_class 708{ 709 I_MUTEX_NORMAL, 710 I_MUTEX_PARENT, 711 I_MUTEX_CHILD, 712 I_MUTEX_XATTR, 713 I_MUTEX_QUOTA 714}; 715 716extern void inode_double_lock(struct inode *inode1, struct inode *inode2); 717extern void inode_double_unlock(struct inode *inode1, struct inode *inode2); 718 719/* 720 * NOTE: in a 32bit arch with a preemptable kernel and 721 * an UP compile the i_size_read/write must be atomic 722 * with respect to the local cpu (unlike with preempt disabled), 723 * but they don't need to be atomic with respect to other cpus like in 724 * true SMP (so they need either to either locally disable irq around 725 * the read or for example on x86 they can be still implemented as a 726 * cmpxchg8b without the need of the lock prefix). For SMP compiles 727 * and 64bit archs it makes no difference if preempt is enabled or not. 728 */ 729static inline loff_t i_size_read(const struct inode *inode) 730{ 731#if BITS_PER_LONG==32 && defined(CONFIG_SMP) 732 loff_t i_size; 733 unsigned int seq; 734 735 do { 736 seq = read_seqcount_begin(&inode->i_size_seqcount); 737 i_size = inode->i_size; 738 } while (read_seqcount_retry(&inode->i_size_seqcount, seq)); 739 return i_size; 740#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT) 741 loff_t i_size; 742 743 preempt_disable(); 744 i_size = inode->i_size; 745 preempt_enable(); 746 return i_size; 747#else 748 return inode->i_size; 749#endif 750} 751 752/* 753 * NOTE: unlike i_size_read(), i_size_write() does need locking around it 754 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount 755 * can be lost, resulting in subsequent i_size_read() calls spinning forever. 756 */ 757static inline void i_size_write(struct inode *inode, loff_t i_size) 758{ 759#if BITS_PER_LONG==32 && defined(CONFIG_SMP) 760 write_seqcount_begin(&inode->i_size_seqcount); 761 inode->i_size = i_size; 762 write_seqcount_end(&inode->i_size_seqcount); 763#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT) 764 preempt_disable(); 765 inode->i_size = i_size; 766 preempt_enable(); 767#else 768 inode->i_size = i_size; 769#endif 770} 771 772static inline unsigned iminor(const struct inode *inode) 773{ 774 return MINOR(inode->i_rdev); 775} 776 777static inline unsigned imajor(const struct inode *inode) 778{ 779 return MAJOR(inode->i_rdev); 780} 781 782extern struct block_device *I_BDEV(struct inode *inode); 783 784struct fown_struct { 785 rwlock_t lock; /* protects pid, uid, euid fields */ 786 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */ 787 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */ 788 uid_t uid, euid; /* uid/euid of process setting the owner */ 789 int signum; /* posix.1b rt signal to be delivered on IO */ 790}; 791 792/* 793 * Track a single file's readahead state 794 */ 795struct file_ra_state { 796 pgoff_t start; /* where readahead started */ 797 unsigned int size; /* # of readahead pages */ 798 unsigned int async_size; /* do asynchronous readahead when 799 there are only # of pages ahead */ 800 801 unsigned int ra_pages; /* Maximum readahead window */ 802 int mmap_miss; /* Cache miss stat for mmap accesses */ 803 loff_t prev_pos; /* Cache last read() position */ 804}; 805 806/* 807 * Check if @index falls in the readahead windows. 808 */ 809static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index) 810{ 811 return (index >= ra->start && 812 index < ra->start + ra->size); 813} 814 815#define FILE_MNT_WRITE_TAKEN 1 816#define FILE_MNT_WRITE_RELEASED 2 817 818struct file { 819 /* 820 * fu_list becomes invalid after file_free is called and queued via 821 * fu_rcuhead for RCU freeing 822 */ 823 union { 824 struct list_head fu_list; 825 struct rcu_head fu_rcuhead; 826 } f_u; 827 struct path f_path; 828#define f_dentry f_path.dentry 829#define f_vfsmnt f_path.mnt 830 const struct file_operations *f_op; 831 atomic_long_t f_count; 832 unsigned int f_flags; 833 fmode_t f_mode; 834 loff_t f_pos; 835 struct fown_struct f_owner; 836 unsigned int f_uid, f_gid; 837 struct file_ra_state f_ra; 838 839 u64 f_version; 840#ifdef CONFIG_SECURITY 841 void *f_security; 842#endif 843 /* needed for tty driver, and maybe others */ 844 void *private_data; 845 846#ifdef CONFIG_EPOLL 847 /* Used by fs/eventpoll.c to link all the hooks to this file */ 848 struct list_head f_ep_links; 849 spinlock_t f_ep_lock; 850#endif /* #ifdef CONFIG_EPOLL */ 851 struct address_space *f_mapping; 852#ifdef CONFIG_DEBUG_WRITECOUNT 853 unsigned long f_mnt_write_state; 854#endif 855}; 856extern spinlock_t files_lock; 857#define file_list_lock() spin_lock(&files_lock); 858#define file_list_unlock() spin_unlock(&files_lock); 859 860#define get_file(x) atomic_long_inc(&(x)->f_count) 861#define file_count(x) atomic_long_read(&(x)->f_count) 862 863#ifdef CONFIG_DEBUG_WRITECOUNT 864static inline void file_take_write(struct file *f) 865{ 866 WARN_ON(f->f_mnt_write_state != 0); 867 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN; 868} 869static inline void file_release_write(struct file *f) 870{ 871 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED; 872} 873static inline void file_reset_write(struct file *f) 874{ 875 f->f_mnt_write_state = 0; 876} 877static inline void file_check_state(struct file *f) 878{ 879 /* 880 * At this point, either both or neither of these bits 881 * should be set. 882 */ 883 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN); 884 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED); 885} 886static inline int file_check_writeable(struct file *f) 887{ 888 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN) 889 return 0; 890 printk(KERN_WARNING "writeable file with no " 891 "mnt_want_write()\n"); 892 WARN_ON(1); 893 return -EINVAL; 894} 895#else /* !CONFIG_DEBUG_WRITECOUNT */ 896static inline void file_take_write(struct file *filp) {} 897static inline void file_release_write(struct file *filp) {} 898static inline void file_reset_write(struct file *filp) {} 899static inline void file_check_state(struct file *filp) {} 900static inline int file_check_writeable(struct file *filp) 901{ 902 return 0; 903} 904#endif /* CONFIG_DEBUG_WRITECOUNT */ 905 906#define MAX_NON_LFS ((1UL<<31) - 1) 907 908/* Page cache limit. The filesystems should put that into their s_maxbytes 909 limits, otherwise bad things can happen in VM. */ 910#if BITS_PER_LONG==32 911#define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1) 912#elif BITS_PER_LONG==64 913#define MAX_LFS_FILESIZE 0x7fffffffffffffffUL 914#endif 915 916#define FL_POSIX 1 917#define FL_FLOCK 2 918#define FL_ACCESS 8 /* not trying to lock, just looking */ 919#define FL_EXISTS 16 /* when unlocking, test for existence */ 920#define FL_LEASE 32 /* lease held on this file */ 921#define FL_CLOSE 64 /* unlock on close */ 922#define FL_SLEEP 128 /* A blocking lock */ 923 924/* 925 * Special return value from posix_lock_file() and vfs_lock_file() for 926 * asynchronous locking. 927 */ 928#define FILE_LOCK_DEFERRED 1 929 930/* 931 * The POSIX file lock owner is determined by 932 * the "struct files_struct" in the thread group 933 * (or NULL for no owner - BSD locks). 934 * 935 * Lockd stuffs a "host" pointer into this. 936 */ 937typedef struct files_struct *fl_owner_t; 938 939struct file_lock_operations { 940 void (*fl_copy_lock)(struct file_lock *, struct file_lock *); 941 void (*fl_release_private)(struct file_lock *); 942}; 943 944struct lock_manager_operations { 945 int (*fl_compare_owner)(struct file_lock *, struct file_lock *); 946 void (*fl_notify)(struct file_lock *); /* unblock callback */ 947 int (*fl_grant)(struct file_lock *, struct file_lock *, int); 948 void (*fl_copy_lock)(struct file_lock *, struct file_lock *); 949 void (*fl_release_private)(struct file_lock *); 950 void (*fl_break)(struct file_lock *); 951 int (*fl_mylease)(struct file_lock *, struct file_lock *); 952 int (*fl_change)(struct file_lock **, int); 953}; 954 955struct lock_manager { 956 struct list_head list; 957}; 958 959void locks_start_grace(struct lock_manager *); 960void locks_end_grace(struct lock_manager *); 961int locks_in_grace(void); 962 963/* that will die - we need it for nfs_lock_info */ 964#include <linux/nfs_fs_i.h> 965 966struct file_lock { 967 struct file_lock *fl_next; /* singly linked list for this inode */ 968 struct list_head fl_link; /* doubly linked list of all locks */ 969 struct list_head fl_block; /* circular list of blocked processes */ 970 fl_owner_t fl_owner; 971 unsigned char fl_flags; 972 unsigned char fl_type; 973 unsigned int fl_pid; 974 struct pid *fl_nspid; 975 wait_queue_head_t fl_wait; 976 struct file *fl_file; 977 loff_t fl_start; 978 loff_t fl_end; 979 980 struct fasync_struct * fl_fasync; /* for lease break notifications */ 981 unsigned long fl_break_time; /* for nonblocking lease breaks */ 982 983 struct file_lock_operations *fl_ops; /* Callbacks for filesystems */ 984 struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */ 985 union { 986 struct nfs_lock_info nfs_fl; 987 struct nfs4_lock_info nfs4_fl; 988 struct { 989 struct list_head link; /* link in AFS vnode's pending_locks list */ 990 int state; /* state of grant or error if -ve */ 991 } afs; 992 } fl_u; 993}; 994 995/* The following constant reflects the upper bound of the file/locking space */ 996#ifndef OFFSET_MAX 997#define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1))) 998#define OFFSET_MAX INT_LIMIT(loff_t) 999#define OFFT_OFFSET_MAX INT_LIMIT(off_t) 1000#endif 1001 1002#include <linux/fcntl.h> 1003 1004extern void send_sigio(struct fown_struct *fown, int fd, int band); 1005 1006/* fs/sync.c */ 1007extern int do_sync_mapping_range(struct address_space *mapping, loff_t offset, 1008 loff_t endbyte, unsigned int flags); 1009 1010#ifdef CONFIG_FILE_LOCKING 1011extern int fcntl_getlk(struct file *, struct flock __user *); 1012extern int fcntl_setlk(unsigned int, struct file *, unsigned int, 1013 struct flock __user *); 1014 1015#if BITS_PER_LONG == 32 1016extern int fcntl_getlk64(struct file *, struct flock64 __user *); 1017extern int fcntl_setlk64(unsigned int, struct file *, unsigned int, 1018 struct flock64 __user *); 1019#endif 1020 1021extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg); 1022extern int fcntl_getlease(struct file *filp); 1023 1024/* fs/locks.c */ 1025extern void locks_init_lock(struct file_lock *); 1026extern void locks_copy_lock(struct file_lock *, struct file_lock *); 1027extern void __locks_copy_lock(struct file_lock *, const struct file_lock *); 1028extern void locks_remove_posix(struct file *, fl_owner_t); 1029extern void locks_remove_flock(struct file *); 1030extern void posix_test_lock(struct file *, struct file_lock *); 1031extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *); 1032extern int posix_lock_file_wait(struct file *, struct file_lock *); 1033extern int posix_unblock_lock(struct file *, struct file_lock *); 1034extern int vfs_test_lock(struct file *, struct file_lock *); 1035extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *); 1036extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl); 1037extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl); 1038extern int __break_lease(struct inode *inode, unsigned int flags); 1039extern void lease_get_mtime(struct inode *, struct timespec *time); 1040extern int generic_setlease(struct file *, long, struct file_lock **); 1041extern int vfs_setlease(struct file *, long, struct file_lock **); 1042extern int lease_modify(struct file_lock **, int); 1043extern int lock_may_read(struct inode *, loff_t start, unsigned long count); 1044extern int lock_may_write(struct inode *, loff_t start, unsigned long count); 1045#else /* !CONFIG_FILE_LOCKING */ 1046#define fcntl_getlk(a, b) ({ -EINVAL; }) 1047#define fcntl_setlk(a, b, c, d) ({ -EACCES; }) 1048#if BITS_PER_LONG == 32 1049#define fcntl_getlk64(a, b) ({ -EINVAL; }) 1050#define fcntl_setlk64(a, b, c, d) ({ -EACCES; }) 1051#endif 1052#define fcntl_setlease(a, b, c) ({ 0; }) 1053#define fcntl_getlease(a) ({ 0; }) 1054#define locks_init_lock(a) ({ }) 1055#define __locks_copy_lock(a, b) ({ }) 1056#define locks_copy_lock(a, b) ({ }) 1057#define locks_remove_posix(a, b) ({ }) 1058#define locks_remove_flock(a) ({ }) 1059#define posix_test_lock(a, b) ({ 0; }) 1060#define posix_lock_file(a, b, c) ({ -ENOLCK; }) 1061#define posix_lock_file_wait(a, b) ({ -ENOLCK; }) 1062#define posix_unblock_lock(a, b) (-ENOENT) 1063#define vfs_test_lock(a, b) ({ 0; }) 1064#define vfs_lock_file(a, b, c, d) (-ENOLCK) 1065#define vfs_cancel_lock(a, b) ({ 0; }) 1066#define flock_lock_file_wait(a, b) ({ -ENOLCK; }) 1067#define __break_lease(a, b) ({ 0; }) 1068#define lease_get_mtime(a, b) ({ }) 1069#define generic_setlease(a, b, c) ({ -EINVAL; }) 1070#define vfs_setlease(a, b, c) ({ -EINVAL; }) 1071#define lease_modify(a, b) ({ -EINVAL; }) 1072#define lock_may_read(a, b, c) ({ 1; }) 1073#define lock_may_write(a, b, c) ({ 1; }) 1074#endif /* !CONFIG_FILE_LOCKING */ 1075 1076 1077struct fasync_struct { 1078 int magic; 1079 int fa_fd; 1080 struct fasync_struct *fa_next; /* singly linked list */ 1081 struct file *fa_file; 1082}; 1083 1084#define FASYNC_MAGIC 0x4601 1085 1086/* SMP safe fasync helpers: */ 1087extern int fasync_helper(int, struct file *, int, struct fasync_struct **); 1088/* can be called from interrupts */ 1089extern void kill_fasync(struct fasync_struct **, int, int); 1090/* only for net: no internal synchronization */ 1091extern void __kill_fasync(struct fasync_struct *, int, int); 1092 1093extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force); 1094extern int f_setown(struct file *filp, unsigned long arg, int force); 1095extern void f_delown(struct file *filp); 1096extern pid_t f_getown(struct file *filp); 1097extern int send_sigurg(struct fown_struct *fown); 1098 1099/* 1100 * Umount options 1101 */ 1102 1103#define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */ 1104#define MNT_DETACH 0x00000002 /* Just detach from the tree */ 1105#define MNT_EXPIRE 0x00000004 /* Mark for expiry */ 1106 1107extern struct list_head super_blocks; 1108extern spinlock_t sb_lock; 1109 1110#define sb_entry(list) list_entry((list), struct super_block, s_list) 1111#define S_BIAS (1<<30) 1112struct super_block { 1113 struct list_head s_list; /* Keep this first */ 1114 dev_t s_dev; /* search index; _not_ kdev_t */ 1115 unsigned long s_blocksize; 1116 unsigned char s_blocksize_bits; 1117 unsigned char s_dirt; 1118 unsigned long long s_maxbytes; /* Max file size */ 1119 struct file_system_type *s_type; 1120 const struct super_operations *s_op; 1121 struct dquot_operations *dq_op; 1122 struct quotactl_ops *s_qcop; 1123 const struct export_operations *s_export_op; 1124 unsigned long s_flags; 1125 unsigned long s_magic; 1126 struct dentry *s_root; 1127 struct rw_semaphore s_umount; 1128 struct mutex s_lock; 1129 int s_count; 1130 int s_syncing; 1131 int s_need_sync_fs; 1132 atomic_t s_active; 1133#ifdef CONFIG_SECURITY 1134 void *s_security; 1135#endif 1136 struct xattr_handler **s_xattr; 1137 1138 struct list_head s_inodes; /* all inodes */ 1139 struct list_head s_dirty; /* dirty inodes */ 1140 struct list_head s_io; /* parked for writeback */ 1141 struct list_head s_more_io; /* parked for more writeback */ 1142 struct hlist_head s_anon; /* anonymous dentries for (nfs) exporting */ 1143 struct list_head s_files; 1144 /* s_dentry_lru and s_nr_dentry_unused are protected by dcache_lock */ 1145 struct list_head s_dentry_lru; /* unused dentry lru */ 1146 int s_nr_dentry_unused; /* # of dentry on lru */ 1147 1148 struct block_device *s_bdev; 1149 struct mtd_info *s_mtd; 1150 struct list_head s_instances; 1151 struct quota_info s_dquot; /* Diskquota specific options */ 1152 1153 int s_frozen; 1154 wait_queue_head_t s_wait_unfrozen; 1155 1156 char s_id[32]; /* Informational name */ 1157 1158 void *s_fs_info; /* Filesystem private info */ 1159 fmode_t s_mode; 1160 1161 /* 1162 * The next field is for VFS *only*. No filesystems have any business 1163 * even looking at it. You had been warned. 1164 */ 1165 struct mutex s_vfs_rename_mutex; /* Kludge */ 1166 1167 /* Granularity of c/m/atime in ns. 1168 Cannot be worse than a second */ 1169 u32 s_time_gran; 1170 1171 /* 1172 * Filesystem subtype. If non-empty the filesystem type field 1173 * in /proc/mounts will be "type.subtype" 1174 */ 1175 char *s_subtype; 1176 1177 /* 1178 * Saved mount options for lazy filesystems using 1179 * generic_show_options() 1180 */ 1181 char *s_options; 1182}; 1183 1184extern struct timespec current_fs_time(struct super_block *sb); 1185 1186/* 1187 * Snapshotting support. 1188 */ 1189enum { 1190 SB_UNFROZEN = 0, 1191 SB_FREEZE_WRITE = 1, 1192 SB_FREEZE_TRANS = 2, 1193}; 1194 1195#define vfs_check_frozen(sb, level) \ 1196 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level))) 1197 1198#define get_fs_excl() atomic_inc(&current->fs_excl) 1199#define put_fs_excl() atomic_dec(&current->fs_excl) 1200#define has_fs_excl() atomic_read(&current->fs_excl) 1201 1202#define is_owner_or_cap(inode) \ 1203 ((current->fsuid == (inode)->i_uid) || capable(CAP_FOWNER)) 1204 1205/* not quite ready to be deprecated, but... */ 1206extern void lock_super(struct super_block *); 1207extern void unlock_super(struct super_block *); 1208 1209/* 1210 * VFS helper functions.. 1211 */ 1212extern int vfs_permission(struct nameidata *, int); 1213extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *); 1214extern int vfs_mkdir(struct inode *, struct dentry *, int); 1215extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t); 1216extern int vfs_symlink(struct inode *, struct dentry *, const char *); 1217extern int vfs_link(struct dentry *, struct inode *, struct dentry *); 1218extern int vfs_rmdir(struct inode *, struct dentry *); 1219extern int vfs_unlink(struct inode *, struct dentry *); 1220extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *); 1221 1222/* 1223 * VFS dentry helper functions. 1224 */ 1225extern void dentry_unhash(struct dentry *dentry); 1226 1227/* 1228 * VFS file helper functions. 1229 */ 1230extern int file_permission(struct file *, int); 1231 1232/* 1233 * VFS FS_IOC_FIEMAP helper definitions. 1234 */ 1235struct fiemap_extent_info { 1236 unsigned int fi_flags; /* Flags as passed from user */ 1237 unsigned int fi_extents_mapped; /* Number of mapped extents */ 1238 unsigned int fi_extents_max; /* Size of fiemap_extent array */ 1239 struct fiemap_extent *fi_extents_start; /* Start of fiemap_extent 1240 * array */ 1241}; 1242int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical, 1243 u64 phys, u64 len, u32 flags); 1244int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags); 1245 1246/* 1247 * File types 1248 * 1249 * NOTE! These match bits 12..15 of stat.st_mode 1250 * (ie "(i_mode >> 12) & 15"). 1251 */ 1252#define DT_UNKNOWN 0 1253#define DT_FIFO 1 1254#define DT_CHR 2 1255#define DT_DIR 4 1256#define DT_BLK 6 1257#define DT_REG 8 1258#define DT_LNK 10 1259#define DT_SOCK 12 1260#define DT_WHT 14 1261 1262#define OSYNC_METADATA (1<<0) 1263#define OSYNC_DATA (1<<1) 1264#define OSYNC_INODE (1<<2) 1265int generic_osync_inode(struct inode *, struct address_space *, int); 1266 1267/* 1268 * This is the "filldir" function type, used by readdir() to let 1269 * the kernel specify what kind of dirent layout it wants to have. 1270 * This allows the kernel to read directories into kernel space or 1271 * to have different dirent layouts depending on the binary type. 1272 */ 1273typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned); 1274struct block_device_operations; 1275 1276/* These macros are for out of kernel modules to test that 1277 * the kernel supports the unlocked_ioctl and compat_ioctl 1278 * fields in struct file_operations. */ 1279#define HAVE_COMPAT_IOCTL 1 1280#define HAVE_UNLOCKED_IOCTL 1 1281 1282/* 1283 * NOTE: 1284 * read, write, poll, fsync, readv, writev, unlocked_ioctl and compat_ioctl 1285 * can be called without the big kernel lock held in all filesystems. 1286 */ 1287struct file_operations { 1288 struct module *owner; 1289 loff_t (*llseek) (struct file *, loff_t, int); 1290 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *); 1291 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *); 1292 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t); 1293 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t); 1294 int (*readdir) (struct file *, void *, filldir_t); 1295 unsigned int (*poll) (struct file *, struct poll_table_struct *); 1296 int (*ioctl) (struct inode *, struct file *, unsigned int, unsigned long); 1297 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long); 1298 long (*compat_ioctl) (struct file *, unsigned int, unsigned long); 1299 int (*mmap) (struct file *, struct vm_area_struct *); 1300 int (*open) (struct inode *, struct file *); 1301 int (*flush) (struct file *, fl_owner_t id); 1302 int (*release) (struct inode *, struct file *); 1303 int (*fsync) (struct file *, struct dentry *, int datasync); 1304 int (*aio_fsync) (struct kiocb *, int datasync); 1305 int (*fasync) (int, struct file *, int); 1306 int (*lock) (struct file *, int, struct file_lock *); 1307 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int); 1308 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); 1309 int (*check_flags)(int); 1310 int (*dir_notify)(struct file *filp, unsigned long arg); 1311 int (*flock) (struct file *, int, struct file_lock *); 1312 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int); 1313 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int); 1314 int (*setlease)(struct file *, long, struct file_lock **); 1315}; 1316 1317struct inode_operations { 1318 int (*create) (struct inode *,struct dentry *,int, struct nameidata *); 1319 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *); 1320 int (*link) (struct dentry *,struct inode *,struct dentry *); 1321 int (*unlink) (struct inode *,struct dentry *); 1322 int (*symlink) (struct inode *,struct dentry *,const char *); 1323 int (*mkdir) (struct inode *,struct dentry *,int); 1324 int (*rmdir) (struct inode *,struct dentry *); 1325 int (*mknod) (struct inode *,struct dentry *,int,dev_t); 1326 int (*rename) (struct inode *, struct dentry *, 1327 struct inode *, struct dentry *); 1328 int (*readlink) (struct dentry *, char __user *,int); 1329 void * (*follow_link) (struct dentry *, struct nameidata *); 1330 void (*put_link) (struct dentry *, struct nameidata *, void *); 1331 void (*truncate) (struct inode *); 1332 int (*permission) (struct inode *, int); 1333 int (*setattr) (struct dentry *, struct iattr *); 1334 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *); 1335 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int); 1336 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t); 1337 ssize_t (*listxattr) (struct dentry *, char *, size_t); 1338 int (*removexattr) (struct dentry *, const char *); 1339 void (*truncate_range)(struct inode *, loff_t, loff_t); 1340 long (*fallocate)(struct inode *inode, int mode, loff_t offset, 1341 loff_t len); 1342 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, 1343 u64 len); 1344}; 1345 1346struct seq_file; 1347 1348ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector, 1349 unsigned long nr_segs, unsigned long fast_segs, 1350 struct iovec *fast_pointer, 1351 struct iovec **ret_pointer); 1352 1353extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *); 1354extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *); 1355extern ssize_t vfs_readv(struct file *, const struct iovec __user *, 1356 unsigned long, loff_t *); 1357extern ssize_t vfs_writev(struct file *, const struct iovec __user *, 1358 unsigned long, loff_t *); 1359 1360struct super_operations { 1361 struct inode *(*alloc_inode)(struct super_block *sb); 1362 void (*destroy_inode)(struct inode *); 1363 1364 void (*dirty_inode) (struct inode *); 1365 int (*write_inode) (struct inode *, int); 1366 void (*drop_inode) (struct inode *); 1367 void (*delete_inode) (struct inode *); 1368 void (*put_super) (struct super_block *); 1369 void (*write_super) (struct super_block *); 1370 int (*sync_fs)(struct super_block *sb, int wait); 1371 void (*write_super_lockfs) (struct super_block *); 1372 void (*unlockfs) (struct super_block *); 1373 int (*statfs) (struct dentry *, struct kstatfs *); 1374 int (*remount_fs) (struct super_block *, int *, char *); 1375 void (*clear_inode) (struct inode *); 1376 void (*umount_begin) (struct super_block *); 1377 1378 int (*show_options)(struct seq_file *, struct vfsmount *); 1379 int (*show_stats)(struct seq_file *, struct vfsmount *); 1380#ifdef CONFIG_QUOTA 1381 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t); 1382 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t); 1383#endif 1384}; 1385 1386/* 1387 * Inode state bits. Protected by inode_lock. 1388 * 1389 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC, 1390 * I_DIRTY_DATASYNC and I_DIRTY_PAGES. 1391 * 1392 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW, 1393 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at 1394 * various stages of removing an inode. 1395 * 1396 * Two bits are used for locking and completion notification, I_LOCK and I_SYNC. 1397 * 1398 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on 1399 * fdatasync(). i_atime is the usual cause. 1400 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of 1401 * these changes separately from I_DIRTY_SYNC so that we 1402 * don't have to write inode on fdatasync() when only 1403 * mtime has changed in it. 1404 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean. 1405 * I_NEW get_new_inode() sets i_state to I_LOCK|I_NEW. Both 1406 * are cleared by unlock_new_inode(), called from iget(). 1407 * I_WILL_FREE Must be set when calling write_inode_now() if i_count 1408 * is zero. I_FREEING must be set when I_WILL_FREE is 1409 * cleared. 1410 * I_FREEING Set when inode is about to be freed but still has dirty 1411 * pages or buffers attached or the inode itself is still 1412 * dirty. 1413 * I_CLEAR Set by clear_inode(). In this state the inode is clean 1414 * and can be destroyed. 1415 * 1416 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are 1417 * prohibited for many purposes. iget() must wait for 1418 * the inode to be completely released, then create it 1419 * anew. Other functions will just ignore such inodes, 1420 * if appropriate. I_LOCK is used for waiting. 1421 * 1422 * I_LOCK Serves as both a mutex and completion notification. 1423 * New inodes set I_LOCK. If two processes both create 1424 * the same inode, one of them will release its inode and 1425 * wait for I_LOCK to be released before returning. 1426 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can 1427 * also cause waiting on I_LOCK, without I_LOCK actually 1428 * being set. find_inode() uses this to prevent returning 1429 * nearly-dead inodes. 1430 * I_SYNC Similar to I_LOCK, but limited in scope to writeback 1431 * of inode dirty data. Having a separate lock for this 1432 * purpose reduces latency and prevents some filesystem- 1433 * specific deadlocks. 1434 * 1435 * Q: What is the difference between I_WILL_FREE and I_FREEING? 1436 * Q: igrab() only checks on (I_FREEING|I_WILL_FREE). Should it also check on 1437 * I_CLEAR? If not, why? 1438 */ 1439#define I_DIRTY_SYNC 1 1440#define I_DIRTY_DATASYNC 2 1441#define I_DIRTY_PAGES 4 1442#define I_NEW 8 1443#define I_WILL_FREE 16 1444#define I_FREEING 32 1445#define I_CLEAR 64 1446#define __I_LOCK 7 1447#define I_LOCK (1 << __I_LOCK) 1448#define __I_SYNC 8 1449#define I_SYNC (1 << __I_SYNC) 1450 1451#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES) 1452 1453extern void __mark_inode_dirty(struct inode *, int); 1454static inline void mark_inode_dirty(struct inode *inode) 1455{ 1456 __mark_inode_dirty(inode, I_DIRTY); 1457} 1458 1459static inline void mark_inode_dirty_sync(struct inode *inode) 1460{ 1461 __mark_inode_dirty(inode, I_DIRTY_SYNC); 1462} 1463 1464/** 1465 * inc_nlink - directly increment an inode's link count 1466 * @inode: inode 1467 * 1468 * This is a low-level filesystem helper to replace any 1469 * direct filesystem manipulation of i_nlink. Currently, 1470 * it is only here for parity with dec_nlink(). 1471 */ 1472static inline void inc_nlink(struct inode *inode) 1473{ 1474 inode->i_nlink++; 1475} 1476 1477static inline void inode_inc_link_count(struct inode *inode) 1478{ 1479 inc_nlink(inode); 1480 mark_inode_dirty(inode); 1481} 1482 1483/** 1484 * drop_nlink - directly drop an inode's link count 1485 * @inode: inode 1486 * 1487 * This is a low-level filesystem helper to replace any 1488 * direct filesystem manipulation of i_nlink. In cases 1489 * where we are attempting to track writes to the 1490 * filesystem, a decrement to zero means an imminent 1491 * write when the file is truncated and actually unlinked 1492 * on the filesystem. 1493 */ 1494static inline void drop_nlink(struct inode *inode) 1495{ 1496 inode->i_nlink--; 1497} 1498 1499/** 1500 * clear_nlink - directly zero an inode's link count 1501 * @inode: inode 1502 * 1503 * This is a low-level filesystem helper to replace any 1504 * direct filesystem manipulation of i_nlink. See 1505 * drop_nlink() for why we care about i_nlink hitting zero. 1506 */ 1507static inline void clear_nlink(struct inode *inode) 1508{ 1509 inode->i_nlink = 0; 1510} 1511 1512static inline void inode_dec_link_count(struct inode *inode) 1513{ 1514 drop_nlink(inode); 1515 mark_inode_dirty(inode); 1516} 1517 1518/** 1519 * inode_inc_iversion - increments i_version 1520 * @inode: inode that need to be updated 1521 * 1522 * Every time the inode is modified, the i_version field will be incremented. 1523 * The filesystem has to be mounted with i_version flag 1524 */ 1525 1526static inline void inode_inc_iversion(struct inode *inode) 1527{ 1528 spin_lock(&inode->i_lock); 1529 inode->i_version++; 1530 spin_unlock(&inode->i_lock); 1531} 1532 1533extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry); 1534static inline void file_accessed(struct file *file) 1535{ 1536 if (!(file->f_flags & O_NOATIME)) 1537 touch_atime(file->f_path.mnt, file->f_path.dentry); 1538} 1539 1540int sync_inode(struct inode *inode, struct writeback_control *wbc); 1541 1542struct file_system_type { 1543 const char *name; 1544 int fs_flags; 1545 int (*get_sb) (struct file_system_type *, int, 1546 const char *, void *, struct vfsmount *); 1547 void (*kill_sb) (struct super_block *); 1548 struct module *owner; 1549 struct file_system_type * next; 1550 struct list_head fs_supers; 1551 1552 struct lock_class_key s_lock_key; 1553 struct lock_class_key s_umount_key; 1554 1555 struct lock_class_key i_lock_key; 1556 struct lock_class_key i_mutex_key; 1557 struct lock_class_key i_mutex_dir_key; 1558 struct lock_class_key i_alloc_sem_key; 1559}; 1560 1561extern int get_sb_bdev(struct file_system_type *fs_type, 1562 int flags, const char *dev_name, void *data, 1563 int (*fill_super)(struct super_block *, void *, int), 1564 struct vfsmount *mnt); 1565extern int get_sb_single(struct file_system_type *fs_type, 1566 int flags, void *data, 1567 int (*fill_super)(struct super_block *, void *, int), 1568 struct vfsmount *mnt); 1569extern int get_sb_nodev(struct file_system_type *fs_type, 1570 int flags, void *data, 1571 int (*fill_super)(struct super_block *, void *, int), 1572 struct vfsmount *mnt); 1573void generic_shutdown_super(struct super_block *sb); 1574void kill_block_super(struct super_block *sb); 1575void kill_anon_super(struct super_block *sb); 1576void kill_litter_super(struct super_block *sb); 1577void deactivate_super(struct super_block *sb); 1578int set_anon_super(struct super_block *s, void *data); 1579struct super_block *sget(struct file_system_type *type, 1580 int (*test)(struct super_block *,void *), 1581 int (*set)(struct super_block *,void *), 1582 void *data); 1583extern int get_sb_pseudo(struct file_system_type *, char *, 1584 const struct super_operations *ops, unsigned long, 1585 struct vfsmount *mnt); 1586extern int simple_set_mnt(struct vfsmount *mnt, struct super_block *sb); 1587int __put_super_and_need_restart(struct super_block *sb); 1588 1589/* Alas, no aliases. Too much hassle with bringing module.h everywhere */ 1590#define fops_get(fops) \ 1591 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL)) 1592#define fops_put(fops) \ 1593 do { if (fops) module_put((fops)->owner); } while(0) 1594 1595extern int register_filesystem(struct file_system_type *); 1596extern int unregister_filesystem(struct file_system_type *); 1597extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data); 1598#define kern_mount(type) kern_mount_data(type, NULL) 1599extern int may_umount_tree(struct vfsmount *); 1600extern int may_umount(struct vfsmount *); 1601extern long do_mount(char *, char *, char *, unsigned long, void *); 1602extern struct vfsmount *collect_mounts(struct vfsmount *, struct dentry *); 1603extern void drop_collected_mounts(struct vfsmount *); 1604 1605extern int vfs_statfs(struct dentry *, struct kstatfs *); 1606 1607/* /sys/fs */ 1608extern struct kobject *fs_kobj; 1609 1610extern int rw_verify_area(int, struct file *, loff_t *, size_t); 1611 1612#define FLOCK_VERIFY_READ 1 1613#define FLOCK_VERIFY_WRITE 2 1614 1615#ifdef CONFIG_FILE_LOCKING 1616extern int locks_mandatory_locked(struct inode *); 1617extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t); 1618 1619/* 1620 * Candidates for mandatory locking have the setgid bit set 1621 * but no group execute bit - an otherwise meaningless combination. 1622 */ 1623 1624static inline int __mandatory_lock(struct inode *ino) 1625{ 1626 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID; 1627} 1628 1629/* 1630 * ... and these candidates should be on MS_MANDLOCK mounted fs, 1631 * otherwise these will be advisory locks 1632 */ 1633 1634static inline int mandatory_lock(struct inode *ino) 1635{ 1636 return IS_MANDLOCK(ino) && __mandatory_lock(ino); 1637} 1638 1639static inline int locks_verify_locked(struct inode *inode) 1640{ 1641 if (mandatory_lock(inode)) 1642 return locks_mandatory_locked(inode); 1643 return 0; 1644} 1645 1646static inline int locks_verify_truncate(struct inode *inode, 1647 struct file *filp, 1648 loff_t size) 1649{ 1650 if (inode->i_flock && mandatory_lock(inode)) 1651 return locks_mandatory_area( 1652 FLOCK_VERIFY_WRITE, inode, filp, 1653 size < inode->i_size ? size : inode->i_size, 1654 (size < inode->i_size ? inode->i_size - size 1655 : size - inode->i_size) 1656 ); 1657 return 0; 1658} 1659 1660static inline int break_lease(struct inode *inode, unsigned int mode) 1661{ 1662 if (inode->i_flock) 1663 return __break_lease(inode, mode); 1664 return 0; 1665} 1666#else /* !CONFIG_FILE_LOCKING */ 1667#define locks_mandatory_locked(a) ({ 0; }) 1668#define locks_mandatory_area(a, b, c, d, e) ({ 0; }) 1669#define __mandatory_lock(a) ({ 0; }) 1670#define mandatory_lock(a) ({ 0; }) 1671#define locks_verify_locked(a) ({ 0; }) 1672#define locks_verify_truncate(a, b, c) ({ 0; }) 1673#define break_lease(a, b) ({ 0; }) 1674#endif /* CONFIG_FILE_LOCKING */ 1675 1676/* fs/open.c */ 1677 1678extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs, 1679 struct file *filp); 1680extern long do_sys_open(int dfd, const char __user *filename, int flags, 1681 int mode); 1682extern struct file *filp_open(const char *, int, int); 1683extern struct file * dentry_open(struct dentry *, struct vfsmount *, int); 1684extern int filp_close(struct file *, fl_owner_t id); 1685extern char * getname(const char __user *); 1686 1687/* fs/dcache.c */ 1688extern void __init vfs_caches_init_early(void); 1689extern void __init vfs_caches_init(unsigned long); 1690 1691extern struct kmem_cache *names_cachep; 1692 1693#define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL) 1694#define __putname(name) kmem_cache_free(names_cachep, (void *)(name)) 1695#ifndef CONFIG_AUDITSYSCALL 1696#define putname(name) __putname(name) 1697#else 1698extern void putname(const char *name); 1699#endif 1700 1701#ifdef CONFIG_BLOCK 1702extern int register_blkdev(unsigned int, const char *); 1703extern void unregister_blkdev(unsigned int, const char *); 1704extern struct block_device *bdget(dev_t); 1705extern void bd_set_size(struct block_device *, loff_t size); 1706extern void bd_forget(struct inode *inode); 1707extern void bdput(struct block_device *); 1708extern struct block_device *open_by_devnum(dev_t, fmode_t); 1709#else 1710static inline void bd_forget(struct inode *inode) {} 1711#endif 1712extern const struct file_operations def_blk_fops; 1713extern const struct file_operations def_chr_fops; 1714extern const struct file_operations bad_sock_fops; 1715extern const struct file_operations def_fifo_fops; 1716#ifdef CONFIG_BLOCK 1717extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long); 1718extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long); 1719extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long); 1720extern int blkdev_get(struct block_device *, fmode_t); 1721extern int blkdev_put(struct block_device *, fmode_t); 1722extern int bd_claim(struct block_device *, void *); 1723extern void bd_release(struct block_device *); 1724#ifdef CONFIG_SYSFS 1725extern int bd_claim_by_disk(struct block_device *, void *, struct gendisk *); 1726extern void bd_release_from_disk(struct block_device *, struct gendisk *); 1727#else 1728#define bd_claim_by_disk(bdev, holder, disk) bd_claim(bdev, holder) 1729#define bd_release_from_disk(bdev, disk) bd_release(bdev) 1730#endif 1731#endif 1732 1733/* fs/char_dev.c */ 1734#define CHRDEV_MAJOR_HASH_SIZE 255 1735extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *); 1736extern int register_chrdev_region(dev_t, unsigned, const char *); 1737extern int register_chrdev(unsigned int, const char *, 1738 const struct file_operations *); 1739extern void unregister_chrdev(unsigned int, const char *); 1740extern void unregister_chrdev_region(dev_t, unsigned); 1741extern void chrdev_show(struct seq_file *,off_t); 1742 1743/* fs/block_dev.c */ 1744#define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */ 1745#define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */ 1746 1747#ifdef CONFIG_BLOCK 1748#define BLKDEV_MAJOR_HASH_SIZE 255 1749extern const char *__bdevname(dev_t, char *buffer); 1750extern const char *bdevname(struct block_device *bdev, char *buffer); 1751extern struct block_device *lookup_bdev(const char *); 1752extern struct block_device *open_bdev_exclusive(const char *, fmode_t, void *); 1753extern void close_bdev_exclusive(struct block_device *, fmode_t); 1754extern void blkdev_show(struct seq_file *,off_t); 1755 1756#else 1757#define BLKDEV_MAJOR_HASH_SIZE 0 1758#endif 1759 1760extern void init_special_inode(struct inode *, umode_t, dev_t); 1761 1762/* Invalid inode operations -- fs/bad_inode.c */ 1763extern void make_bad_inode(struct inode *); 1764extern int is_bad_inode(struct inode *); 1765 1766extern const struct file_operations read_pipefifo_fops; 1767extern const struct file_operations write_pipefifo_fops; 1768extern const struct file_operations rdwr_pipefifo_fops; 1769 1770extern int fs_may_remount_ro(struct super_block *); 1771 1772#ifdef CONFIG_BLOCK 1773/* 1774 * return READ, READA, or WRITE 1775 */ 1776#define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK)) 1777 1778/* 1779 * return data direction, READ or WRITE 1780 */ 1781#define bio_data_dir(bio) ((bio)->bi_rw & 1) 1782 1783extern void check_disk_size_change(struct gendisk *disk, 1784 struct block_device *bdev); 1785extern int revalidate_disk(struct gendisk *); 1786extern int check_disk_change(struct block_device *); 1787extern int __invalidate_device(struct block_device *); 1788extern int invalidate_partition(struct gendisk *, int); 1789#endif 1790extern int invalidate_inodes(struct super_block *); 1791unsigned long __invalidate_mapping_pages(struct address_space *mapping, 1792 pgoff_t start, pgoff_t end, 1793 bool be_atomic); 1794unsigned long invalidate_mapping_pages(struct address_space *mapping, 1795 pgoff_t start, pgoff_t end); 1796 1797static inline unsigned long __deprecated 1798invalidate_inode_pages(struct address_space *mapping) 1799{ 1800 return invalidate_mapping_pages(mapping, 0, ~0UL); 1801} 1802 1803static inline void invalidate_remote_inode(struct inode *inode) 1804{ 1805 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) || 1806 S_ISLNK(inode->i_mode)) 1807 invalidate_mapping_pages(inode->i_mapping, 0, -1); 1808} 1809extern int invalidate_inode_pages2(struct address_space *mapping); 1810extern int invalidate_inode_pages2_range(struct address_space *mapping, 1811 pgoff_t start, pgoff_t end); 1812extern void generic_sync_sb_inodes(struct super_block *sb, 1813 struct writeback_control *wbc); 1814extern int write_inode_now(struct inode *, int); 1815extern int filemap_fdatawrite(struct address_space *); 1816extern int filemap_flush(struct address_space *); 1817extern int filemap_fdatawait(struct address_space *); 1818extern int filemap_write_and_wait(struct address_space *mapping); 1819extern int filemap_write_and_wait_range(struct address_space *mapping, 1820 loff_t lstart, loff_t lend); 1821extern int wait_on_page_writeback_range(struct address_space *mapping, 1822 pgoff_t start, pgoff_t end); 1823extern int __filemap_fdatawrite_range(struct address_space *mapping, 1824 loff_t start, loff_t end, int sync_mode); 1825extern int filemap_fdatawrite_range(struct address_space *mapping, 1826 loff_t start, loff_t end); 1827 1828extern long do_fsync(struct file *file, int datasync); 1829extern void sync_supers(void); 1830extern void sync_filesystems(int wait); 1831extern void __fsync_super(struct super_block *sb); 1832extern void emergency_sync(void); 1833extern void emergency_remount(void); 1834extern int do_remount_sb(struct super_block *sb, int flags, 1835 void *data, int force); 1836#ifdef CONFIG_BLOCK 1837extern sector_t bmap(struct inode *, sector_t); 1838#endif 1839extern int notify_change(struct dentry *, struct iattr *); 1840extern int inode_permission(struct inode *, int); 1841extern int generic_permission(struct inode *, int, 1842 int (*check_acl)(struct inode *, int)); 1843 1844static inline bool execute_ok(struct inode *inode) 1845{ 1846 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode); 1847} 1848 1849extern int get_write_access(struct inode *); 1850extern int deny_write_access(struct file *); 1851static inline void put_write_access(struct inode * inode) 1852{ 1853 atomic_dec(&inode->i_writecount); 1854} 1855static inline void allow_write_access(struct file *file) 1856{ 1857 if (file) 1858 atomic_inc(&file->f_path.dentry->d_inode->i_writecount); 1859} 1860extern int do_pipe(int *); 1861extern int do_pipe_flags(int *, int); 1862extern struct file *create_read_pipe(struct file *f, int flags); 1863extern struct file *create_write_pipe(int flags); 1864extern void free_write_pipe(struct file *); 1865 1866extern struct file *do_filp_open(int dfd, const char *pathname, 1867 int open_flag, int mode); 1868extern int may_open(struct nameidata *, int, int); 1869 1870extern int kernel_read(struct file *, unsigned long, char *, unsigned long); 1871extern struct file * open_exec(const char *); 1872 1873/* fs/dcache.c -- generic fs support functions */ 1874extern int is_subdir(struct dentry *, struct dentry *); 1875extern ino_t find_inode_number(struct dentry *, struct qstr *); 1876 1877#include <linux/err.h> 1878 1879/* needed for stackable file system support */ 1880extern loff_t default_llseek(struct file *file, loff_t offset, int origin); 1881 1882extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin); 1883 1884extern void inode_init_once(struct inode *); 1885extern void iput(struct inode *); 1886extern struct inode * igrab(struct inode *); 1887extern ino_t iunique(struct super_block *, ino_t); 1888extern int inode_needs_sync(struct inode *inode); 1889extern void generic_delete_inode(struct inode *inode); 1890extern void generic_drop_inode(struct inode *inode); 1891 1892extern struct inode *ilookup5_nowait(struct super_block *sb, 1893 unsigned long hashval, int (*test)(struct inode *, void *), 1894 void *data); 1895extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval, 1896 int (*test)(struct inode *, void *), void *data); 1897extern struct inode *ilookup(struct super_block *sb, unsigned long ino); 1898 1899extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *); 1900extern struct inode * iget_locked(struct super_block *, unsigned long); 1901extern void unlock_new_inode(struct inode *); 1902 1903extern void __iget(struct inode * inode); 1904extern void iget_failed(struct inode *); 1905extern void clear_inode(struct inode *); 1906extern void destroy_inode(struct inode *); 1907extern struct inode *new_inode(struct super_block *); 1908extern int should_remove_suid(struct dentry *); 1909extern int file_remove_suid(struct file *); 1910 1911extern void __insert_inode_hash(struct inode *, unsigned long hashval); 1912extern void remove_inode_hash(struct inode *); 1913static inline void insert_inode_hash(struct inode *inode) { 1914 __insert_inode_hash(inode, inode->i_ino); 1915} 1916 1917extern struct file * get_empty_filp(void); 1918extern void file_move(struct file *f, struct list_head *list); 1919extern void file_kill(struct file *f); 1920#ifdef CONFIG_BLOCK 1921struct bio; 1922extern void submit_bio(int, struct bio *); 1923extern int bdev_read_only(struct block_device *); 1924#endif 1925extern int set_blocksize(struct block_device *, int); 1926extern int sb_set_blocksize(struct super_block *, int); 1927extern int sb_min_blocksize(struct super_block *, int); 1928extern int sb_has_dirty_inodes(struct super_block *); 1929 1930extern int generic_file_mmap(struct file *, struct vm_area_struct *); 1931extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *); 1932extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size); 1933int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk); 1934extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t); 1935extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t); 1936extern ssize_t generic_file_aio_write_nolock(struct kiocb *, const struct iovec *, 1937 unsigned long, loff_t); 1938extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *, 1939 unsigned long *, loff_t, loff_t *, size_t, size_t); 1940extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *, 1941 unsigned long, loff_t, loff_t *, size_t, ssize_t); 1942extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos); 1943extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos); 1944extern int generic_segment_checks(const struct iovec *iov, 1945 unsigned long *nr_segs, size_t *count, int access_flags); 1946 1947/* fs/splice.c */ 1948extern ssize_t generic_file_splice_read(struct file *, loff_t *, 1949 struct pipe_inode_info *, size_t, unsigned int); 1950extern ssize_t generic_file_splice_write(struct pipe_inode_info *, 1951 struct file *, loff_t *, size_t, unsigned int); 1952extern ssize_t generic_file_splice_write_nolock(struct pipe_inode_info *, 1953 struct file *, loff_t *, size_t, unsigned int); 1954extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe, 1955 struct file *out, loff_t *, size_t len, unsigned int flags); 1956extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out, 1957 size_t len, unsigned int flags); 1958 1959extern void 1960file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping); 1961extern loff_t no_llseek(struct file *file, loff_t offset, int origin); 1962extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin); 1963extern loff_t generic_file_llseek_unlocked(struct file *file, loff_t offset, 1964 int origin); 1965extern int generic_file_open(struct inode * inode, struct file * filp); 1966extern int nonseekable_open(struct inode * inode, struct file * filp); 1967 1968#ifdef CONFIG_FS_XIP 1969extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len, 1970 loff_t *ppos); 1971extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma); 1972extern ssize_t xip_file_write(struct file *filp, const char __user *buf, 1973 size_t len, loff_t *ppos); 1974extern int xip_truncate_page(struct address_space *mapping, loff_t from); 1975#else 1976static inline int xip_truncate_page(struct address_space *mapping, loff_t from) 1977{ 1978 return 0; 1979} 1980#endif 1981 1982#ifdef CONFIG_BLOCK 1983ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode, 1984 struct block_device *bdev, const struct iovec *iov, loff_t offset, 1985 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io, 1986 int lock_type); 1987 1988enum { 1989 DIO_LOCKING = 1, /* need locking between buffered and direct access */ 1990 DIO_NO_LOCKING, /* bdev; no locking at all between buffered/direct */ 1991 DIO_OWN_LOCKING, /* filesystem locks buffered and direct internally */ 1992}; 1993 1994static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb, 1995 struct inode *inode, struct block_device *bdev, const struct iovec *iov, 1996 loff_t offset, unsigned long nr_segs, get_block_t get_block, 1997 dio_iodone_t end_io) 1998{ 1999 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset, 2000 nr_segs, get_block, end_io, DIO_LOCKING); 2001} 2002 2003static inline ssize_t blockdev_direct_IO_no_locking(int rw, struct kiocb *iocb, 2004 struct inode *inode, struct block_device *bdev, const struct iovec *iov, 2005 loff_t offset, unsigned long nr_segs, get_block_t get_block, 2006 dio_iodone_t end_io) 2007{ 2008 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset, 2009 nr_segs, get_block, end_io, DIO_NO_LOCKING); 2010} 2011 2012static inline ssize_t blockdev_direct_IO_own_locking(int rw, struct kiocb *iocb, 2013 struct inode *inode, struct block_device *bdev, const struct iovec *iov, 2014 loff_t offset, unsigned long nr_segs, get_block_t get_block, 2015 dio_iodone_t end_io) 2016{ 2017 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset, 2018 nr_segs, get_block, end_io, DIO_OWN_LOCKING); 2019} 2020#endif 2021 2022extern const struct file_operations generic_ro_fops; 2023 2024#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m)) 2025 2026extern int vfs_readlink(struct dentry *, char __user *, int, const char *); 2027extern int vfs_follow_link(struct nameidata *, const char *); 2028extern int page_readlink(struct dentry *, char __user *, int); 2029extern void *page_follow_link_light(struct dentry *, struct nameidata *); 2030extern void page_put_link(struct dentry *, struct nameidata *, void *); 2031extern int __page_symlink(struct inode *inode, const char *symname, int len, 2032 gfp_t gfp_mask); 2033extern int page_symlink(struct inode *inode, const char *symname, int len); 2034extern const struct inode_operations page_symlink_inode_operations; 2035extern int generic_readlink(struct dentry *, char __user *, int); 2036extern void generic_fillattr(struct inode *, struct kstat *); 2037extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *); 2038void inode_add_bytes(struct inode *inode, loff_t bytes); 2039void inode_sub_bytes(struct inode *inode, loff_t bytes); 2040loff_t inode_get_bytes(struct inode *inode); 2041void inode_set_bytes(struct inode *inode, loff_t bytes); 2042 2043extern int vfs_readdir(struct file *, filldir_t, void *); 2044 2045extern int vfs_stat(char __user *, struct kstat *); 2046extern int vfs_lstat(char __user *, struct kstat *); 2047extern int vfs_stat_fd(int dfd, char __user *, struct kstat *); 2048extern int vfs_lstat_fd(int dfd, char __user *, struct kstat *); 2049extern int vfs_fstat(unsigned int, struct kstat *); 2050 2051extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd, 2052 unsigned long arg); 2053extern int generic_block_fiemap(struct inode *inode, 2054 struct fiemap_extent_info *fieinfo, u64 start, 2055 u64 len, get_block_t *get_block); 2056 2057extern void get_filesystem(struct file_system_type *fs); 2058extern void put_filesystem(struct file_system_type *fs); 2059extern struct file_system_type *get_fs_type(const char *name); 2060extern struct super_block *get_super(struct block_device *); 2061extern struct super_block *user_get_super(dev_t); 2062extern void drop_super(struct super_block *sb); 2063 2064extern int dcache_dir_open(struct inode *, struct file *); 2065extern int dcache_dir_close(struct inode *, struct file *); 2066extern loff_t dcache_dir_lseek(struct file *, loff_t, int); 2067extern int dcache_readdir(struct file *, void *, filldir_t); 2068extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *); 2069extern int simple_statfs(struct dentry *, struct kstatfs *); 2070extern int simple_link(struct dentry *, struct inode *, struct dentry *); 2071extern int simple_unlink(struct inode *, struct dentry *); 2072extern int simple_rmdir(struct inode *, struct dentry *); 2073extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *); 2074extern int simple_sync_file(struct file *, struct dentry *, int); 2075extern int simple_empty(struct dentry *); 2076extern int simple_readpage(struct file *file, struct page *page); 2077extern int simple_prepare_write(struct file *file, struct page *page, 2078 unsigned offset, unsigned to); 2079extern int simple_write_begin(struct file *file, struct address_space *mapping, 2080 loff_t pos, unsigned len, unsigned flags, 2081 struct page **pagep, void **fsdata); 2082extern int simple_write_end(struct file *file, struct address_space *mapping, 2083 loff_t pos, unsigned len, unsigned copied, 2084 struct page *page, void *fsdata); 2085 2086extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *); 2087extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *); 2088extern const struct file_operations simple_dir_operations; 2089extern const struct inode_operations simple_dir_inode_operations; 2090struct tree_descr { char *name; const struct file_operations *ops; int mode; }; 2091struct dentry *d_alloc_name(struct dentry *, const char *); 2092extern int simple_fill_super(struct super_block *, int, struct tree_descr *); 2093extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count); 2094extern void simple_release_fs(struct vfsmount **mount, int *count); 2095 2096extern ssize_t simple_read_from_buffer(void __user *to, size_t count, 2097 loff_t *ppos, const void *from, size_t available); 2098 2099#ifdef CONFIG_MIGRATION 2100extern int buffer_migrate_page(struct address_space *, 2101 struct page *, struct page *); 2102#else 2103#define buffer_migrate_page NULL 2104#endif 2105 2106extern int inode_change_ok(struct inode *, struct iattr *); 2107extern int __must_check inode_setattr(struct inode *, struct iattr *); 2108 2109extern void file_update_time(struct file *file); 2110 2111extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt); 2112extern void save_mount_options(struct super_block *sb, char *options); 2113 2114static inline ino_t parent_ino(struct dentry *dentry) 2115{ 2116 ino_t res; 2117 2118 spin_lock(&dentry->d_lock); 2119 res = dentry->d_parent->d_inode->i_ino; 2120 spin_unlock(&dentry->d_lock); 2121 return res; 2122} 2123 2124/* Transaction based IO helpers */ 2125 2126/* 2127 * An argresp is stored in an allocated page and holds the 2128 * size of the argument or response, along with its content 2129 */ 2130struct simple_transaction_argresp { 2131 ssize_t size; 2132 char data[0]; 2133}; 2134 2135#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp)) 2136 2137char *simple_transaction_get(struct file *file, const char __user *buf, 2138 size_t size); 2139ssize_t simple_transaction_read(struct file *file, char __user *buf, 2140 size_t size, loff_t *pos); 2141int simple_transaction_release(struct inode *inode, struct file *file); 2142 2143static inline void simple_transaction_set(struct file *file, size_t n) 2144{ 2145 struct simple_transaction_argresp *ar = file->private_data; 2146 2147 BUG_ON(n > SIMPLE_TRANSACTION_LIMIT); 2148 2149 /* 2150 * The barrier ensures that ar->size will really remain zero until 2151 * ar->data is ready for reading. 2152 */ 2153 smp_mb(); 2154 ar->size = n; 2155} 2156 2157/* 2158 * simple attribute files 2159 * 2160 * These attributes behave similar to those in sysfs: 2161 * 2162 * Writing to an attribute immediately sets a value, an open file can be 2163 * written to multiple times. 2164 * 2165 * Reading from an attribute creates a buffer from the value that might get 2166 * read with multiple read calls. When the attribute has been read 2167 * completely, no further read calls are possible until the file is opened 2168 * again. 2169 * 2170 * All attributes contain a text representation of a numeric value 2171 * that are accessed with the get() and set() functions. 2172 */ 2173#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \ 2174static int __fops ## _open(struct inode *inode, struct file *file) \ 2175{ \ 2176 __simple_attr_check_format(__fmt, 0ull); \ 2177 return simple_attr_open(inode, file, __get, __set, __fmt); \ 2178} \ 2179static struct file_operations __fops = { \ 2180 .owner = THIS_MODULE, \ 2181 .open = __fops ## _open, \ 2182 .release = simple_attr_release, \ 2183 .read = simple_attr_read, \ 2184 .write = simple_attr_write, \ 2185}; 2186 2187static inline void __attribute__((format(printf, 1, 2))) 2188__simple_attr_check_format(const char *fmt, ...) 2189{ 2190 /* don't do anything, just let the compiler check the arguments; */ 2191} 2192 2193int simple_attr_open(struct inode *inode, struct file *file, 2194 int (*get)(void *, u64 *), int (*set)(void *, u64), 2195 const char *fmt); 2196int simple_attr_release(struct inode *inode, struct file *file); 2197ssize_t simple_attr_read(struct file *file, char __user *buf, 2198 size_t len, loff_t *ppos); 2199ssize_t simple_attr_write(struct file *file, const char __user *buf, 2200 size_t len, loff_t *ppos); 2201 2202 2203#ifdef CONFIG_SECURITY 2204static inline char *alloc_secdata(void) 2205{ 2206 return (char *)get_zeroed_page(GFP_KERNEL); 2207} 2208 2209static inline void free_secdata(void *secdata) 2210{ 2211 free_page((unsigned long)secdata); 2212} 2213#else 2214static inline char *alloc_secdata(void) 2215{ 2216 return (char *)1; 2217} 2218 2219static inline void free_secdata(void *secdata) 2220{ } 2221#endif /* CONFIG_SECURITY */ 2222 2223struct ctl_table; 2224int proc_nr_files(struct ctl_table *table, int write, struct file *filp, 2225 void __user *buffer, size_t *lenp, loff_t *ppos); 2226 2227int get_filesystem_list(char * buf); 2228 2229#endif /* __KERNEL__ */ 2230#endif /* _LINUX_FS_H */