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 24 #define NR_OPEN (1024*1024) /* Absolute upper limit on fd num */ 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 /* And dynamically-tunable limits and defaults: */ 31 struct files_stat_struct { 32 int nr_files; /* read only */ 33 int nr_free_files; /* read only */ 34 int max_files; /* tunable */ 35 }; 36 extern struct files_stat_struct files_stat; 37 extern int get_max_files(void); 38 39 struct inodes_stat_t { 40 int nr_inodes; 41 int nr_unused; 42 int dummy[5]; 43 }; 44 extern struct inodes_stat_t inodes_stat; 45 46 extern int leases_enable, lease_break_time; 47 48 #ifdef CONFIG_DNOTIFY 49 extern int dir_notify_enable; 50 #endif 51 52 #define NR_FILE 8192 /* this can well be larger on a larger system */ 53 54 #define MAY_EXEC 1 55 #define MAY_WRITE 2 56 #define MAY_READ 4 57 #define MAY_APPEND 8 58 59 #define FMODE_READ 1 60 #define FMODE_WRITE 2 61 62 /* Internal kernel extensions */ 63 #define FMODE_LSEEK 4 64 #define FMODE_PREAD 8 65 #define FMODE_PWRITE FMODE_PREAD /* These go hand in hand */ 66 67 /* File is being opened for execution. Primary users of this flag are 68 distributed filesystems that can use it to achieve correct ETXTBUSY 69 behavior for cross-node execution/opening_for_writing of files */ 70 #define FMODE_EXEC 16 71 72 #define RW_MASK 1 73 #define RWA_MASK 2 74 #define READ 0 75 #define WRITE 1 76 #define READA 2 /* read-ahead - don't block if no resources */ 77 #define SWRITE 3 /* for ll_rw_block() - wait for buffer lock */ 78 #define SPECIAL 4 /* For non-blockdevice requests in request queue */ 79 #define READ_SYNC (READ | (1 << BIO_RW_SYNC)) 80 #define WRITE_SYNC (WRITE | (1 << BIO_RW_SYNC)) 81 #define WRITE_BARRIER ((1 << BIO_RW) | (1 << BIO_RW_BARRIER)) 82 83 #define SEL_IN 1 84 #define SEL_OUT 2 85 #define SEL_EX 4 86 87 /* public flags for file_system_type */ 88 #define FS_REQUIRES_DEV 1 89 #define FS_BINARY_MOUNTDATA 2 90 #define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */ 91 #define FS_ODD_RENAME 32768 /* Temporary stuff; will go away as soon 92 * as nfs_rename() will be cleaned up 93 */ 94 /* 95 * These are the fs-independent mount-flags: up to 32 flags are supported 96 */ 97 #define MS_RDONLY 1 /* Mount read-only */ 98 #define MS_NOSUID 2 /* Ignore suid and sgid bits */ 99 #define MS_NODEV 4 /* Disallow access to device special files */ 100 #define MS_NOEXEC 8 /* Disallow program execution */ 101 #define MS_SYNCHRONOUS 16 /* Writes are synced at once */ 102 #define MS_REMOUNT 32 /* Alter flags of a mounted FS */ 103 #define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */ 104 #define MS_DIRSYNC 128 /* Directory modifications are synchronous */ 105 #define MS_NOATIME 1024 /* Do not update access times. */ 106 #define MS_NODIRATIME 2048 /* Do not update directory access times */ 107 #define MS_BIND 4096 108 #define MS_MOVE 8192 109 #define MS_REC 16384 110 #define MS_VERBOSE 32768 /* War is peace. Verbosity is silence. 111 MS_VERBOSE is deprecated. */ 112 #define MS_SILENT 32768 113 #define MS_POSIXACL (1<<16) /* VFS does not apply the umask */ 114 #define MS_UNBINDABLE (1<<17) /* change to unbindable */ 115 #define MS_PRIVATE (1<<18) /* change to private */ 116 #define MS_SLAVE (1<<19) /* change to slave */ 117 #define MS_SHARED (1<<20) /* change to shared */ 118 #define MS_ACTIVE (1<<30) 119 #define MS_NOUSER (1<<31) 120 121 /* 122 * Superblock flags that can be altered by MS_REMOUNT 123 */ 124 #define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK) 125 126 /* 127 * Old magic mount flag and mask 128 */ 129 #define MS_MGC_VAL 0xC0ED0000 130 #define MS_MGC_MSK 0xffff0000 131 132 /* Inode flags - they have nothing to superblock flags now */ 133 134 #define S_SYNC 1 /* Writes are synced at once */ 135 #define S_NOATIME 2 /* Do not update access times */ 136 #define S_APPEND 4 /* Append-only file */ 137 #define S_IMMUTABLE 8 /* Immutable file */ 138 #define S_DEAD 16 /* removed, but still open directory */ 139 #define S_NOQUOTA 32 /* Inode is not counted to quota */ 140 #define S_DIRSYNC 64 /* Directory modifications are synchronous */ 141 #define S_NOCMTIME 128 /* Do not update file c/mtime */ 142 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */ 143 #define S_PRIVATE 512 /* Inode is fs-internal */ 144 145 /* 146 * Note that nosuid etc flags are inode-specific: setting some file-system 147 * flags just means all the inodes inherit those flags by default. It might be 148 * possible to override it selectively if you really wanted to with some 149 * ioctl() that is not currently implemented. 150 * 151 * Exception: MS_RDONLY is always applied to the entire file system. 152 * 153 * Unfortunately, it is possible to change a filesystems flags with it mounted 154 * with files in use. This means that all of the inodes will not have their 155 * i_flags updated. Hence, i_flags no longer inherit the superblock mount 156 * flags, so these have to be checked separately. -- [email protected] 157 */ 158 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg)) 159 160 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY) 161 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \ 162 ((inode)->i_flags & S_SYNC)) 163 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \ 164 ((inode)->i_flags & (S_SYNC|S_DIRSYNC))) 165 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK) 166 167 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA) 168 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND) 169 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE) 170 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL) 171 172 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD) 173 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME) 174 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE) 175 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE) 176 177 /* the read-only stuff doesn't really belong here, but any other place is 178 probably as bad and I don't want to create yet another include file. */ 179 180 #define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */ 181 #define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */ 182 #define BLKRRPART _IO(0x12,95) /* re-read partition table */ 183 #define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */ 184 #define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */ 185 #define BLKRASET _IO(0x12,98) /* set read ahead for block device */ 186 #define BLKRAGET _IO(0x12,99) /* get current read ahead setting */ 187 #define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */ 188 #define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */ 189 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */ 190 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */ 191 #define BLKSSZGET _IO(0x12,104)/* get block device sector size */ 192 #if 0 193 #define BLKPG _IO(0x12,105)/* See blkpg.h */ 194 195 /* Some people are morons. Do not use sizeof! */ 196 197 #define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */ 198 #define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */ 199 /* This was here just to show that the number is taken - 200 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */ 201 #endif 202 /* A jump here: 108-111 have been used for various private purposes. */ 203 #define BLKBSZGET _IOR(0x12,112,size_t) 204 #define BLKBSZSET _IOW(0x12,113,size_t) 205 #define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */ 206 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup) 207 #define BLKTRACESTART _IO(0x12,116) 208 #define BLKTRACESTOP _IO(0x12,117) 209 #define BLKTRACETEARDOWN _IO(0x12,118) 210 211 #define BMAP_IOCTL 1 /* obsolete - kept for compatibility */ 212 #define FIBMAP _IO(0x00,1) /* bmap access */ 213 #define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */ 214 215 #define SYNC_FILE_RANGE_WAIT_BEFORE 1 216 #define SYNC_FILE_RANGE_WRITE 2 217 #define SYNC_FILE_RANGE_WAIT_AFTER 4 218 219 #ifdef __KERNEL__ 220 221 #include <linux/linkage.h> 222 #include <linux/wait.h> 223 #include <linux/types.h> 224 #include <linux/kdev_t.h> 225 #include <linux/dcache.h> 226 #include <linux/stat.h> 227 #include <linux/cache.h> 228 #include <linux/kobject.h> 229 #include <linux/list.h> 230 #include <linux/radix-tree.h> 231 #include <linux/prio_tree.h> 232 #include <linux/init.h> 233 #include <linux/sched.h> 234 #include <linux/mutex.h> 235 236 #include <asm/atomic.h> 237 #include <asm/semaphore.h> 238 #include <asm/byteorder.h> 239 240 struct hd_geometry; 241 struct iovec; 242 struct nameidata; 243 struct kiocb; 244 struct pipe_inode_info; 245 struct poll_table_struct; 246 struct kstatfs; 247 struct vm_area_struct; 248 struct vfsmount; 249 250 extern void __init inode_init(unsigned long); 251 extern void __init inode_init_early(void); 252 extern void __init mnt_init(unsigned long); 253 extern void __init files_init(unsigned long); 254 255 struct buffer_head; 256 typedef int (get_block_t)(struct inode *inode, sector_t iblock, 257 struct buffer_head *bh_result, int create); 258 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset, 259 ssize_t bytes, void *private); 260 261 /* 262 * Attribute flags. These should be or-ed together to figure out what 263 * has been changed! 264 */ 265 #define ATTR_MODE 1 266 #define ATTR_UID 2 267 #define ATTR_GID 4 268 #define ATTR_SIZE 8 269 #define ATTR_ATIME 16 270 #define ATTR_MTIME 32 271 #define ATTR_CTIME 64 272 #define ATTR_ATIME_SET 128 273 #define ATTR_MTIME_SET 256 274 #define ATTR_FORCE 512 /* Not a change, but a change it */ 275 #define ATTR_ATTR_FLAG 1024 276 #define ATTR_KILL_SUID 2048 277 #define ATTR_KILL_SGID 4096 278 #define ATTR_FILE 8192 279 280 /* 281 * This is the Inode Attributes structure, used for notify_change(). It 282 * uses the above definitions as flags, to know which values have changed. 283 * Also, in this manner, a Filesystem can look at only the values it cares 284 * about. Basically, these are the attributes that the VFS layer can 285 * request to change from the FS layer. 286 * 287 * Derek Atkins <[email protected]> 94-10-20 288 */ 289 struct iattr { 290 unsigned int ia_valid; 291 umode_t ia_mode; 292 uid_t ia_uid; 293 gid_t ia_gid; 294 loff_t ia_size; 295 struct timespec ia_atime; 296 struct timespec ia_mtime; 297 struct timespec ia_ctime; 298 299 /* 300 * Not an attribute, but an auxilary info for filesystems wanting to 301 * implement an ftruncate() like method. NOTE: filesystem should 302 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL). 303 */ 304 struct file *ia_file; 305 }; 306 307 /* 308 * Includes for diskquotas. 309 */ 310 #include <linux/quota.h> 311 312 /** 313 * enum positive_aop_returns - aop return codes with specific semantics 314 * 315 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has 316 * completed, that the page is still locked, and 317 * should be considered active. The VM uses this hint 318 * to return the page to the active list -- it won't 319 * be a candidate for writeback again in the near 320 * future. Other callers must be careful to unlock 321 * the page if they get this return. Returned by 322 * writepage(); 323 * 324 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has 325 * unlocked it and the page might have been truncated. 326 * The caller should back up to acquiring a new page and 327 * trying again. The aop will be taking reasonable 328 * precautions not to livelock. If the caller held a page 329 * reference, it should drop it before retrying. Returned 330 * by readpage(), prepare_write(), and commit_write(). 331 * 332 * address_space_operation functions return these large constants to indicate 333 * special semantics to the caller. These are much larger than the bytes in a 334 * page to allow for functions that return the number of bytes operated on in a 335 * given page. 336 */ 337 338 enum positive_aop_returns { 339 AOP_WRITEPAGE_ACTIVATE = 0x80000, 340 AOP_TRUNCATED_PAGE = 0x80001, 341 }; 342 343 /* 344 * oh the beauties of C type declarations. 345 */ 346 struct page; 347 struct address_space; 348 struct writeback_control; 349 350 struct address_space_operations { 351 int (*writepage)(struct page *page, struct writeback_control *wbc); 352 int (*readpage)(struct file *, struct page *); 353 void (*sync_page)(struct page *); 354 355 /* Write back some dirty pages from this mapping. */ 356 int (*writepages)(struct address_space *, struct writeback_control *); 357 358 /* Set a page dirty. Return true if this dirtied it */ 359 int (*set_page_dirty)(struct page *page); 360 361 int (*readpages)(struct file *filp, struct address_space *mapping, 362 struct list_head *pages, unsigned nr_pages); 363 364 /* 365 * ext3 requires that a successful prepare_write() call be followed 366 * by a commit_write() call - they must be balanced 367 */ 368 int (*prepare_write)(struct file *, struct page *, unsigned, unsigned); 369 int (*commit_write)(struct file *, struct page *, unsigned, unsigned); 370 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */ 371 sector_t (*bmap)(struct address_space *, sector_t); 372 void (*invalidatepage) (struct page *, unsigned long); 373 int (*releasepage) (struct page *, gfp_t); 374 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov, 375 loff_t offset, unsigned long nr_segs); 376 struct page* (*get_xip_page)(struct address_space *, sector_t, 377 int); 378 /* migrate the contents of a page to the specified target */ 379 int (*migratepage) (struct address_space *, 380 struct page *, struct page *); 381 }; 382 383 struct backing_dev_info; 384 struct address_space { 385 struct inode *host; /* owner: inode, block_device */ 386 struct radix_tree_root page_tree; /* radix tree of all pages */ 387 rwlock_t tree_lock; /* and rwlock protecting it */ 388 unsigned int i_mmap_writable;/* count VM_SHARED mappings */ 389 struct prio_tree_root i_mmap; /* tree of private and shared mappings */ 390 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */ 391 spinlock_t i_mmap_lock; /* protect tree, count, list */ 392 unsigned int truncate_count; /* Cover race condition with truncate */ 393 unsigned long nrpages; /* number of total pages */ 394 pgoff_t writeback_index;/* writeback starts here */ 395 const struct address_space_operations *a_ops; /* methods */ 396 unsigned long flags; /* error bits/gfp mask */ 397 struct backing_dev_info *backing_dev_info; /* device readahead, etc */ 398 spinlock_t private_lock; /* for use by the address_space */ 399 struct list_head private_list; /* ditto */ 400 struct address_space *assoc_mapping; /* ditto */ 401 } __attribute__((aligned(sizeof(long)))); 402 /* 403 * On most architectures that alignment is already the case; but 404 * must be enforced here for CRIS, to let the least signficant bit 405 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON. 406 */ 407 408 struct block_device { 409 dev_t bd_dev; /* not a kdev_t - it's a search key */ 410 struct inode * bd_inode; /* will die */ 411 int bd_openers; 412 struct mutex bd_mutex; /* open/close mutex */ 413 struct mutex bd_mount_mutex; /* mount mutex */ 414 struct list_head bd_inodes; 415 void * bd_holder; 416 int bd_holders; 417 #ifdef CONFIG_SYSFS 418 struct list_head bd_holder_list; 419 #endif 420 struct block_device * bd_contains; 421 unsigned bd_block_size; 422 struct hd_struct * bd_part; 423 /* number of times partitions within this device have been opened. */ 424 unsigned bd_part_count; 425 int bd_invalidated; 426 struct gendisk * bd_disk; 427 struct list_head bd_list; 428 struct backing_dev_info *bd_inode_backing_dev_info; 429 /* 430 * Private data. You must have bd_claim'ed the block_device 431 * to use this. NOTE: bd_claim allows an owner to claim 432 * the same device multiple times, the owner must take special 433 * care to not mess up bd_private for that case. 434 */ 435 unsigned long bd_private; 436 }; 437 438 /* 439 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache 440 * radix trees 441 */ 442 #define PAGECACHE_TAG_DIRTY 0 443 #define PAGECACHE_TAG_WRITEBACK 1 444 445 int mapping_tagged(struct address_space *mapping, int tag); 446 447 /* 448 * Might pages of this file be mapped into userspace? 449 */ 450 static inline int mapping_mapped(struct address_space *mapping) 451 { 452 return !prio_tree_empty(&mapping->i_mmap) || 453 !list_empty(&mapping->i_mmap_nonlinear); 454 } 455 456 /* 457 * Might pages of this file have been modified in userspace? 458 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff 459 * marks vma as VM_SHARED if it is shared, and the file was opened for 460 * writing i.e. vma may be mprotected writable even if now readonly. 461 */ 462 static inline int mapping_writably_mapped(struct address_space *mapping) 463 { 464 return mapping->i_mmap_writable != 0; 465 } 466 467 /* 468 * Use sequence counter to get consistent i_size on 32-bit processors. 469 */ 470 #if BITS_PER_LONG==32 && defined(CONFIG_SMP) 471 #include <linux/seqlock.h> 472 #define __NEED_I_SIZE_ORDERED 473 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount) 474 #else 475 #define i_size_ordered_init(inode) do { } while (0) 476 #endif 477 478 struct inode { 479 struct hlist_node i_hash; 480 struct list_head i_list; 481 struct list_head i_sb_list; 482 struct list_head i_dentry; 483 unsigned long i_ino; 484 atomic_t i_count; 485 umode_t i_mode; 486 unsigned int i_nlink; 487 uid_t i_uid; 488 gid_t i_gid; 489 dev_t i_rdev; 490 loff_t i_size; 491 struct timespec i_atime; 492 struct timespec i_mtime; 493 struct timespec i_ctime; 494 unsigned int i_blkbits; 495 unsigned long i_blksize; 496 unsigned long i_version; 497 blkcnt_t i_blocks; 498 unsigned short i_bytes; 499 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */ 500 struct mutex i_mutex; 501 struct rw_semaphore i_alloc_sem; 502 struct inode_operations *i_op; 503 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */ 504 struct super_block *i_sb; 505 struct file_lock *i_flock; 506 struct address_space *i_mapping; 507 struct address_space i_data; 508 #ifdef CONFIG_QUOTA 509 struct dquot *i_dquot[MAXQUOTAS]; 510 #endif 511 /* These three should probably be a union */ 512 struct list_head i_devices; 513 struct pipe_inode_info *i_pipe; 514 struct block_device *i_bdev; 515 struct cdev *i_cdev; 516 int i_cindex; 517 518 __u32 i_generation; 519 520 #ifdef CONFIG_DNOTIFY 521 unsigned long i_dnotify_mask; /* Directory notify events */ 522 struct dnotify_struct *i_dnotify; /* for directory notifications */ 523 #endif 524 525 #ifdef CONFIG_INOTIFY 526 struct list_head inotify_watches; /* watches on this inode */ 527 struct mutex inotify_mutex; /* protects the watches list */ 528 #endif 529 530 unsigned long i_state; 531 unsigned long dirtied_when; /* jiffies of first dirtying */ 532 533 unsigned int i_flags; 534 535 atomic_t i_writecount; 536 void *i_security; 537 union { 538 void *generic_ip; 539 } u; 540 #ifdef __NEED_I_SIZE_ORDERED 541 seqcount_t i_size_seqcount; 542 #endif 543 }; 544 545 /* 546 * NOTE: in a 32bit arch with a preemptable kernel and 547 * an UP compile the i_size_read/write must be atomic 548 * with respect to the local cpu (unlike with preempt disabled), 549 * but they don't need to be atomic with respect to other cpus like in 550 * true SMP (so they need either to either locally disable irq around 551 * the read or for example on x86 they can be still implemented as a 552 * cmpxchg8b without the need of the lock prefix). For SMP compiles 553 * and 64bit archs it makes no difference if preempt is enabled or not. 554 */ 555 static inline loff_t i_size_read(struct inode *inode) 556 { 557 #if BITS_PER_LONG==32 && defined(CONFIG_SMP) 558 loff_t i_size; 559 unsigned int seq; 560 561 do { 562 seq = read_seqcount_begin(&inode->i_size_seqcount); 563 i_size = inode->i_size; 564 } while (read_seqcount_retry(&inode->i_size_seqcount, seq)); 565 return i_size; 566 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT) 567 loff_t i_size; 568 569 preempt_disable(); 570 i_size = inode->i_size; 571 preempt_enable(); 572 return i_size; 573 #else 574 return inode->i_size; 575 #endif 576 } 577 578 579 static inline void i_size_write(struct inode *inode, loff_t i_size) 580 { 581 #if BITS_PER_LONG==32 && defined(CONFIG_SMP) 582 write_seqcount_begin(&inode->i_size_seqcount); 583 inode->i_size = i_size; 584 write_seqcount_end(&inode->i_size_seqcount); 585 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT) 586 preempt_disable(); 587 inode->i_size = i_size; 588 preempt_enable(); 589 #else 590 inode->i_size = i_size; 591 #endif 592 } 593 594 static inline unsigned iminor(struct inode *inode) 595 { 596 return MINOR(inode->i_rdev); 597 } 598 599 static inline unsigned imajor(struct inode *inode) 600 { 601 return MAJOR(inode->i_rdev); 602 } 603 604 extern struct block_device *I_BDEV(struct inode *inode); 605 606 struct fown_struct { 607 rwlock_t lock; /* protects pid, uid, euid fields */ 608 int pid; /* pid or -pgrp where SIGIO should be sent */ 609 uid_t uid, euid; /* uid/euid of process setting the owner */ 610 void *security; 611 int signum; /* posix.1b rt signal to be delivered on IO */ 612 }; 613 614 /* 615 * Track a single file's readahead state 616 */ 617 struct file_ra_state { 618 unsigned long start; /* Current window */ 619 unsigned long size; 620 unsigned long flags; /* ra flags RA_FLAG_xxx*/ 621 unsigned long cache_hit; /* cache hit count*/ 622 unsigned long prev_page; /* Cache last read() position */ 623 unsigned long ahead_start; /* Ahead window */ 624 unsigned long ahead_size; 625 unsigned long ra_pages; /* Maximum readahead window */ 626 unsigned long mmap_hit; /* Cache hit stat for mmap accesses */ 627 unsigned long mmap_miss; /* Cache miss stat for mmap accesses */ 628 }; 629 #define RA_FLAG_MISS 0x01 /* a cache miss occured against this file */ 630 #define RA_FLAG_INCACHE 0x02 /* file is already in cache */ 631 632 struct file { 633 /* 634 * fu_list becomes invalid after file_free is called and queued via 635 * fu_rcuhead for RCU freeing 636 */ 637 union { 638 struct list_head fu_list; 639 struct rcu_head fu_rcuhead; 640 } f_u; 641 struct dentry *f_dentry; 642 struct vfsmount *f_vfsmnt; 643 const struct file_operations *f_op; 644 atomic_t f_count; 645 unsigned int f_flags; 646 mode_t f_mode; 647 loff_t f_pos; 648 struct fown_struct f_owner; 649 unsigned int f_uid, f_gid; 650 struct file_ra_state f_ra; 651 652 unsigned long f_version; 653 void *f_security; 654 655 /* needed for tty driver, and maybe others */ 656 void *private_data; 657 658 #ifdef CONFIG_EPOLL 659 /* Used by fs/eventpoll.c to link all the hooks to this file */ 660 struct list_head f_ep_links; 661 spinlock_t f_ep_lock; 662 #endif /* #ifdef CONFIG_EPOLL */ 663 struct address_space *f_mapping; 664 }; 665 extern spinlock_t files_lock; 666 #define file_list_lock() spin_lock(&files_lock); 667 #define file_list_unlock() spin_unlock(&files_lock); 668 669 #define get_file(x) atomic_inc(&(x)->f_count) 670 #define file_count(x) atomic_read(&(x)->f_count) 671 672 #define MAX_NON_LFS ((1UL<<31) - 1) 673 674 /* Page cache limit. The filesystems should put that into their s_maxbytes 675 limits, otherwise bad things can happen in VM. */ 676 #if BITS_PER_LONG==32 677 #define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1) 678 #elif BITS_PER_LONG==64 679 #define MAX_LFS_FILESIZE 0x7fffffffffffffffUL 680 #endif 681 682 #define FL_POSIX 1 683 #define FL_FLOCK 2 684 #define FL_ACCESS 8 /* not trying to lock, just looking */ 685 #define FL_LEASE 32 /* lease held on this file */ 686 #define FL_CLOSE 64 /* unlock on close */ 687 #define FL_SLEEP 128 /* A blocking lock */ 688 689 /* 690 * The POSIX file lock owner is determined by 691 * the "struct files_struct" in the thread group 692 * (or NULL for no owner - BSD locks). 693 * 694 * Lockd stuffs a "host" pointer into this. 695 */ 696 typedef struct files_struct *fl_owner_t; 697 698 struct file_lock_operations { 699 void (*fl_insert)(struct file_lock *); /* lock insertion callback */ 700 void (*fl_remove)(struct file_lock *); /* lock removal callback */ 701 void (*fl_copy_lock)(struct file_lock *, struct file_lock *); 702 void (*fl_release_private)(struct file_lock *); 703 }; 704 705 struct lock_manager_operations { 706 int (*fl_compare_owner)(struct file_lock *, struct file_lock *); 707 void (*fl_notify)(struct file_lock *); /* unblock callback */ 708 void (*fl_copy_lock)(struct file_lock *, struct file_lock *); 709 void (*fl_release_private)(struct file_lock *); 710 void (*fl_break)(struct file_lock *); 711 int (*fl_mylease)(struct file_lock *, struct file_lock *); 712 int (*fl_change)(struct file_lock **, int); 713 }; 714 715 /* that will die - we need it for nfs_lock_info */ 716 #include <linux/nfs_fs_i.h> 717 718 struct file_lock { 719 struct file_lock *fl_next; /* singly linked list for this inode */ 720 struct list_head fl_link; /* doubly linked list of all locks */ 721 struct list_head fl_block; /* circular list of blocked processes */ 722 fl_owner_t fl_owner; 723 unsigned int fl_pid; 724 wait_queue_head_t fl_wait; 725 struct file *fl_file; 726 unsigned char fl_flags; 727 unsigned char fl_type; 728 loff_t fl_start; 729 loff_t fl_end; 730 731 struct fasync_struct * fl_fasync; /* for lease break notifications */ 732 unsigned long fl_break_time; /* for nonblocking lease breaks */ 733 734 struct file_lock_operations *fl_ops; /* Callbacks for filesystems */ 735 struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */ 736 union { 737 struct nfs_lock_info nfs_fl; 738 struct nfs4_lock_info nfs4_fl; 739 } fl_u; 740 }; 741 742 /* The following constant reflects the upper bound of the file/locking space */ 743 #ifndef OFFSET_MAX 744 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1))) 745 #define OFFSET_MAX INT_LIMIT(loff_t) 746 #define OFFT_OFFSET_MAX INT_LIMIT(off_t) 747 #endif 748 749 #include <linux/fcntl.h> 750 751 extern int fcntl_getlk(struct file *, struct flock __user *); 752 extern int fcntl_setlk(unsigned int, struct file *, unsigned int, 753 struct flock __user *); 754 755 #if BITS_PER_LONG == 32 756 extern int fcntl_getlk64(struct file *, struct flock64 __user *); 757 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int, 758 struct flock64 __user *); 759 #endif 760 761 extern void send_sigio(struct fown_struct *fown, int fd, int band); 762 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg); 763 extern int fcntl_getlease(struct file *filp); 764 765 /* fs/sync.c */ 766 extern int do_sync_file_range(struct file *file, loff_t offset, loff_t endbyte, 767 unsigned int flags); 768 769 /* fs/locks.c */ 770 extern void locks_init_lock(struct file_lock *); 771 extern void locks_copy_lock(struct file_lock *, struct file_lock *); 772 extern void locks_remove_posix(struct file *, fl_owner_t); 773 extern void locks_remove_flock(struct file *); 774 extern int posix_test_lock(struct file *, struct file_lock *, struct file_lock *); 775 extern int posix_lock_file_conf(struct file *, struct file_lock *, struct file_lock *); 776 extern int posix_lock_file(struct file *, struct file_lock *); 777 extern int posix_lock_file_wait(struct file *, struct file_lock *); 778 extern int posix_unblock_lock(struct file *, struct file_lock *); 779 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl); 780 extern int __break_lease(struct inode *inode, unsigned int flags); 781 extern void lease_get_mtime(struct inode *, struct timespec *time); 782 extern int setlease(struct file *, long, struct file_lock **); 783 extern int lease_modify(struct file_lock **, int); 784 extern int lock_may_read(struct inode *, loff_t start, unsigned long count); 785 extern int lock_may_write(struct inode *, loff_t start, unsigned long count); 786 787 struct fasync_struct { 788 int magic; 789 int fa_fd; 790 struct fasync_struct *fa_next; /* singly linked list */ 791 struct file *fa_file; 792 }; 793 794 #define FASYNC_MAGIC 0x4601 795 796 /* SMP safe fasync helpers: */ 797 extern int fasync_helper(int, struct file *, int, struct fasync_struct **); 798 /* can be called from interrupts */ 799 extern void kill_fasync(struct fasync_struct **, int, int); 800 /* only for net: no internal synchronization */ 801 extern void __kill_fasync(struct fasync_struct *, int, int); 802 803 extern int f_setown(struct file *filp, unsigned long arg, int force); 804 extern void f_delown(struct file *filp); 805 extern int send_sigurg(struct fown_struct *fown); 806 807 /* 808 * Umount options 809 */ 810 811 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */ 812 #define MNT_DETACH 0x00000002 /* Just detach from the tree */ 813 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */ 814 815 extern struct list_head super_blocks; 816 extern spinlock_t sb_lock; 817 818 #define sb_entry(list) list_entry((list), struct super_block, s_list) 819 #define S_BIAS (1<<30) 820 struct super_block { 821 struct list_head s_list; /* Keep this first */ 822 dev_t s_dev; /* search index; _not_ kdev_t */ 823 unsigned long s_blocksize; 824 unsigned char s_blocksize_bits; 825 unsigned char s_dirt; 826 unsigned long long s_maxbytes; /* Max file size */ 827 struct file_system_type *s_type; 828 struct super_operations *s_op; 829 struct dquot_operations *dq_op; 830 struct quotactl_ops *s_qcop; 831 struct export_operations *s_export_op; 832 unsigned long s_flags; 833 unsigned long s_magic; 834 struct dentry *s_root; 835 struct rw_semaphore s_umount; 836 struct mutex s_lock; 837 int s_count; 838 int s_syncing; 839 int s_need_sync_fs; 840 atomic_t s_active; 841 void *s_security; 842 struct xattr_handler **s_xattr; 843 844 struct list_head s_inodes; /* all inodes */ 845 struct list_head s_dirty; /* dirty inodes */ 846 struct list_head s_io; /* parked for writeback */ 847 struct hlist_head s_anon; /* anonymous dentries for (nfs) exporting */ 848 struct list_head s_files; 849 850 struct block_device *s_bdev; 851 struct list_head s_instances; 852 struct quota_info s_dquot; /* Diskquota specific options */ 853 854 int s_frozen; 855 wait_queue_head_t s_wait_unfrozen; 856 857 char s_id[32]; /* Informational name */ 858 859 void *s_fs_info; /* Filesystem private info */ 860 861 /* 862 * The next field is for VFS *only*. No filesystems have any business 863 * even looking at it. You had been warned. 864 */ 865 struct mutex s_vfs_rename_mutex; /* Kludge */ 866 867 /* Granularity of c/m/atime in ns. 868 Cannot be worse than a second */ 869 u32 s_time_gran; 870 }; 871 872 extern struct timespec current_fs_time(struct super_block *sb); 873 874 /* 875 * Snapshotting support. 876 */ 877 enum { 878 SB_UNFROZEN = 0, 879 SB_FREEZE_WRITE = 1, 880 SB_FREEZE_TRANS = 2, 881 }; 882 883 #define vfs_check_frozen(sb, level) \ 884 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level))) 885 886 static inline void get_fs_excl(void) 887 { 888 atomic_inc(¤t->fs_excl); 889 } 890 891 static inline void put_fs_excl(void) 892 { 893 atomic_dec(¤t->fs_excl); 894 } 895 896 static inline int has_fs_excl(void) 897 { 898 return atomic_read(¤t->fs_excl); 899 } 900 901 902 /* 903 * Superblock locking. 904 */ 905 static inline void lock_super(struct super_block * sb) 906 { 907 get_fs_excl(); 908 mutex_lock(&sb->s_lock); 909 } 910 911 static inline void unlock_super(struct super_block * sb) 912 { 913 put_fs_excl(); 914 mutex_unlock(&sb->s_lock); 915 } 916 917 /* 918 * VFS helper functions.. 919 */ 920 extern int vfs_permission(struct nameidata *, int); 921 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *); 922 extern int vfs_mkdir(struct inode *, struct dentry *, int); 923 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t); 924 extern int vfs_symlink(struct inode *, struct dentry *, const char *, int); 925 extern int vfs_link(struct dentry *, struct inode *, struct dentry *); 926 extern int vfs_rmdir(struct inode *, struct dentry *); 927 extern int vfs_unlink(struct inode *, struct dentry *); 928 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *); 929 930 /* 931 * VFS dentry helper functions. 932 */ 933 extern void dentry_unhash(struct dentry *dentry); 934 935 /* 936 * VFS file helper functions. 937 */ 938 extern int file_permission(struct file *, int); 939 940 /* 941 * File types 942 * 943 * NOTE! These match bits 12..15 of stat.st_mode 944 * (ie "(i_mode >> 12) & 15"). 945 */ 946 #define DT_UNKNOWN 0 947 #define DT_FIFO 1 948 #define DT_CHR 2 949 #define DT_DIR 4 950 #define DT_BLK 6 951 #define DT_REG 8 952 #define DT_LNK 10 953 #define DT_SOCK 12 954 #define DT_WHT 14 955 956 #define OSYNC_METADATA (1<<0) 957 #define OSYNC_DATA (1<<1) 958 #define OSYNC_INODE (1<<2) 959 int generic_osync_inode(struct inode *, struct address_space *, int); 960 961 /* 962 * This is the "filldir" function type, used by readdir() to let 963 * the kernel specify what kind of dirent layout it wants to have. 964 * This allows the kernel to read directories into kernel space or 965 * to have different dirent layouts depending on the binary type. 966 */ 967 typedef int (*filldir_t)(void *, const char *, int, loff_t, ino_t, unsigned); 968 969 struct block_device_operations { 970 int (*open) (struct inode *, struct file *); 971 int (*release) (struct inode *, struct file *); 972 int (*ioctl) (struct inode *, struct file *, unsigned, unsigned long); 973 long (*unlocked_ioctl) (struct file *, unsigned, unsigned long); 974 long (*compat_ioctl) (struct file *, unsigned, unsigned long); 975 int (*direct_access) (struct block_device *, sector_t, unsigned long *); 976 int (*media_changed) (struct gendisk *); 977 int (*revalidate_disk) (struct gendisk *); 978 int (*getgeo)(struct block_device *, struct hd_geometry *); 979 struct module *owner; 980 }; 981 982 /* 983 * "descriptor" for what we're up to with a read for sendfile(). 984 * This allows us to use the same read code yet 985 * have multiple different users of the data that 986 * we read from a file. 987 * 988 * The simplest case just copies the data to user 989 * mode. 990 */ 991 typedef struct { 992 size_t written; 993 size_t count; 994 union { 995 char __user * buf; 996 void *data; 997 } arg; 998 int error; 999 } read_descriptor_t; 1000 1001 typedef int (*read_actor_t)(read_descriptor_t *, struct page *, unsigned long, unsigned long); 1002 1003 /* These macros are for out of kernel modules to test that 1004 * the kernel supports the unlocked_ioctl and compat_ioctl 1005 * fields in struct file_operations. */ 1006 #define HAVE_COMPAT_IOCTL 1 1007 #define HAVE_UNLOCKED_IOCTL 1 1008 1009 /* 1010 * NOTE: 1011 * read, write, poll, fsync, readv, writev, unlocked_ioctl and compat_ioctl 1012 * can be called without the big kernel lock held in all filesystems. 1013 */ 1014 struct file_operations { 1015 struct module *owner; 1016 loff_t (*llseek) (struct file *, loff_t, int); 1017 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *); 1018 ssize_t (*aio_read) (struct kiocb *, char __user *, size_t, loff_t); 1019 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *); 1020 ssize_t (*aio_write) (struct kiocb *, const char __user *, size_t, loff_t); 1021 int (*readdir) (struct file *, void *, filldir_t); 1022 unsigned int (*poll) (struct file *, struct poll_table_struct *); 1023 int (*ioctl) (struct inode *, struct file *, unsigned int, unsigned long); 1024 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long); 1025 long (*compat_ioctl) (struct file *, unsigned int, unsigned long); 1026 int (*mmap) (struct file *, struct vm_area_struct *); 1027 int (*open) (struct inode *, struct file *); 1028 int (*flush) (struct file *, fl_owner_t id); 1029 int (*release) (struct inode *, struct file *); 1030 int (*fsync) (struct file *, struct dentry *, int datasync); 1031 int (*aio_fsync) (struct kiocb *, int datasync); 1032 int (*fasync) (int, struct file *, int); 1033 int (*lock) (struct file *, int, struct file_lock *); 1034 ssize_t (*readv) (struct file *, const struct iovec *, unsigned long, loff_t *); 1035 ssize_t (*writev) (struct file *, const struct iovec *, unsigned long, loff_t *); 1036 ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t, void *); 1037 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int); 1038 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); 1039 int (*check_flags)(int); 1040 int (*dir_notify)(struct file *filp, unsigned long arg); 1041 int (*flock) (struct file *, int, struct file_lock *); 1042 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int); 1043 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int); 1044 }; 1045 1046 struct inode_operations { 1047 int (*create) (struct inode *,struct dentry *,int, struct nameidata *); 1048 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *); 1049 int (*link) (struct dentry *,struct inode *,struct dentry *); 1050 int (*unlink) (struct inode *,struct dentry *); 1051 int (*symlink) (struct inode *,struct dentry *,const char *); 1052 int (*mkdir) (struct inode *,struct dentry *,int); 1053 int (*rmdir) (struct inode *,struct dentry *); 1054 int (*mknod) (struct inode *,struct dentry *,int,dev_t); 1055 int (*rename) (struct inode *, struct dentry *, 1056 struct inode *, struct dentry *); 1057 int (*readlink) (struct dentry *, char __user *,int); 1058 void * (*follow_link) (struct dentry *, struct nameidata *); 1059 void (*put_link) (struct dentry *, struct nameidata *, void *); 1060 void (*truncate) (struct inode *); 1061 int (*permission) (struct inode *, int, struct nameidata *); 1062 int (*setattr) (struct dentry *, struct iattr *); 1063 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *); 1064 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int); 1065 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t); 1066 ssize_t (*listxattr) (struct dentry *, char *, size_t); 1067 int (*removexattr) (struct dentry *, const char *); 1068 void (*truncate_range)(struct inode *, loff_t, loff_t); 1069 }; 1070 1071 struct seq_file; 1072 1073 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *); 1074 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *); 1075 extern ssize_t vfs_readv(struct file *, const struct iovec __user *, 1076 unsigned long, loff_t *); 1077 extern ssize_t vfs_writev(struct file *, const struct iovec __user *, 1078 unsigned long, loff_t *); 1079 1080 /* 1081 * NOTE: write_inode, delete_inode, clear_inode, put_inode can be called 1082 * without the big kernel lock held in all filesystems. 1083 */ 1084 struct super_operations { 1085 struct inode *(*alloc_inode)(struct super_block *sb); 1086 void (*destroy_inode)(struct inode *); 1087 1088 void (*read_inode) (struct inode *); 1089 1090 void (*dirty_inode) (struct inode *); 1091 int (*write_inode) (struct inode *, int); 1092 void (*put_inode) (struct inode *); 1093 void (*drop_inode) (struct inode *); 1094 void (*delete_inode) (struct inode *); 1095 void (*put_super) (struct super_block *); 1096 void (*write_super) (struct super_block *); 1097 int (*sync_fs)(struct super_block *sb, int wait); 1098 void (*write_super_lockfs) (struct super_block *); 1099 void (*unlockfs) (struct super_block *); 1100 int (*statfs) (struct dentry *, struct kstatfs *); 1101 int (*remount_fs) (struct super_block *, int *, char *); 1102 void (*clear_inode) (struct inode *); 1103 void (*umount_begin) (struct vfsmount *, int); 1104 1105 int (*show_options)(struct seq_file *, struct vfsmount *); 1106 int (*show_stats)(struct seq_file *, struct vfsmount *); 1107 1108 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t); 1109 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t); 1110 }; 1111 1112 /* Inode state bits. Protected by inode_lock. */ 1113 #define I_DIRTY_SYNC 1 /* Not dirty enough for O_DATASYNC */ 1114 #define I_DIRTY_DATASYNC 2 /* Data-related inode changes pending */ 1115 #define I_DIRTY_PAGES 4 /* Data-related inode changes pending */ 1116 #define __I_LOCK 3 1117 #define I_LOCK (1 << __I_LOCK) 1118 #define I_FREEING 16 1119 #define I_CLEAR 32 1120 #define I_NEW 64 1121 #define I_WILL_FREE 128 1122 1123 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES) 1124 1125 extern void __mark_inode_dirty(struct inode *, int); 1126 static inline void mark_inode_dirty(struct inode *inode) 1127 { 1128 __mark_inode_dirty(inode, I_DIRTY); 1129 } 1130 1131 static inline void mark_inode_dirty_sync(struct inode *inode) 1132 { 1133 __mark_inode_dirty(inode, I_DIRTY_SYNC); 1134 } 1135 1136 static inline void inode_inc_link_count(struct inode *inode) 1137 { 1138 inode->i_nlink++; 1139 mark_inode_dirty(inode); 1140 } 1141 1142 static inline void inode_dec_link_count(struct inode *inode) 1143 { 1144 inode->i_nlink--; 1145 mark_inode_dirty(inode); 1146 } 1147 1148 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry); 1149 static inline void file_accessed(struct file *file) 1150 { 1151 if (!(file->f_flags & O_NOATIME)) 1152 touch_atime(file->f_vfsmnt, file->f_dentry); 1153 } 1154 1155 int sync_inode(struct inode *inode, struct writeback_control *wbc); 1156 1157 /** 1158 * struct export_operations - for nfsd to communicate with file systems 1159 * @decode_fh: decode a file handle fragment and return a &struct dentry 1160 * @encode_fh: encode a file handle fragment from a dentry 1161 * @get_name: find the name for a given inode in a given directory 1162 * @get_parent: find the parent of a given directory 1163 * @get_dentry: find a dentry for the inode given a file handle sub-fragment 1164 * @find_exported_dentry: 1165 * set by the exporting module to a standard helper function. 1166 * 1167 * Description: 1168 * The export_operations structure provides a means for nfsd to communicate 1169 * with a particular exported file system - particularly enabling nfsd and 1170 * the filesystem to co-operate when dealing with file handles. 1171 * 1172 * export_operations contains two basic operation for dealing with file 1173 * handles, decode_fh() and encode_fh(), and allows for some other 1174 * operations to be defined which standard helper routines use to get 1175 * specific information from the filesystem. 1176 * 1177 * nfsd encodes information use to determine which filesystem a filehandle 1178 * applies to in the initial part of the file handle. The remainder, termed 1179 * a file handle fragment, is controlled completely by the filesystem. The 1180 * standard helper routines assume that this fragment will contain one or 1181 * two sub-fragments, one which identifies the file, and one which may be 1182 * used to identify the (a) directory containing the file. 1183 * 1184 * In some situations, nfsd needs to get a dentry which is connected into a 1185 * specific part of the file tree. To allow for this, it passes the 1186 * function acceptable() together with a @context which can be used to see 1187 * if the dentry is acceptable. As there can be multiple dentrys for a 1188 * given file, the filesystem should check each one for acceptability before 1189 * looking for the next. As soon as an acceptable one is found, it should 1190 * be returned. 1191 * 1192 * decode_fh: 1193 * @decode_fh is given a &struct super_block (@sb), a file handle fragment 1194 * (@fh, @fh_len) and an acceptability testing function (@acceptable, 1195 * @context). It should return a &struct dentry which refers to the same 1196 * file that the file handle fragment refers to, and which passes the 1197 * acceptability test. If it cannot, it should return a %NULL pointer if 1198 * the file was found but no acceptable &dentries were available, or a 1199 * %ERR_PTR error code indicating why it couldn't be found (e.g. %ENOENT or 1200 * %ENOMEM). 1201 * 1202 * encode_fh: 1203 * @encode_fh should store in the file handle fragment @fh (using at most 1204 * @max_len bytes) information that can be used by @decode_fh to recover the 1205 * file refered to by the &struct dentry @de. If the @connectable flag is 1206 * set, the encode_fh() should store sufficient information so that a good 1207 * attempt can be made to find not only the file but also it's place in the 1208 * filesystem. This typically means storing a reference to de->d_parent in 1209 * the filehandle fragment. encode_fh() should return the number of bytes 1210 * stored or a negative error code such as %-ENOSPC 1211 * 1212 * get_name: 1213 * @get_name should find a name for the given @child in the given @parent 1214 * directory. The name should be stored in the @name (with the 1215 * understanding that it is already pointing to a a %NAME_MAX+1 sized 1216 * buffer. get_name() should return %0 on success, a negative error code 1217 * or error. @get_name will be called without @parent->i_mutex held. 1218 * 1219 * get_parent: 1220 * @get_parent should find the parent directory for the given @child which 1221 * is also a directory. In the event that it cannot be found, or storage 1222 * space cannot be allocated, a %ERR_PTR should be returned. 1223 * 1224 * get_dentry: 1225 * Given a &super_block (@sb) and a pointer to a file-system specific inode 1226 * identifier, possibly an inode number, (@inump) get_dentry() should find 1227 * the identified inode and return a dentry for that inode. Any suitable 1228 * dentry can be returned including, if necessary, a new dentry created with 1229 * d_alloc_root. The caller can then find any other extant dentrys by 1230 * following the d_alias links. If a new dentry was created using 1231 * d_alloc_root, DCACHE_NFSD_DISCONNECTED should be set, and the dentry 1232 * should be d_rehash()ed. 1233 * 1234 * If the inode cannot be found, either a %NULL pointer or an %ERR_PTR code 1235 * can be returned. The @inump will be whatever was passed to 1236 * nfsd_find_fh_dentry() in either the @obj or @parent parameters. 1237 * 1238 * Locking rules: 1239 * get_parent is called with child->d_inode->i_mutex down 1240 * get_name is not (which is possibly inconsistent) 1241 */ 1242 1243 struct export_operations { 1244 struct dentry *(*decode_fh)(struct super_block *sb, __u32 *fh, int fh_len, int fh_type, 1245 int (*acceptable)(void *context, struct dentry *de), 1246 void *context); 1247 int (*encode_fh)(struct dentry *de, __u32 *fh, int *max_len, 1248 int connectable); 1249 1250 /* the following are only called from the filesystem itself */ 1251 int (*get_name)(struct dentry *parent, char *name, 1252 struct dentry *child); 1253 struct dentry * (*get_parent)(struct dentry *child); 1254 struct dentry * (*get_dentry)(struct super_block *sb, void *inump); 1255 1256 /* This is set by the exporting module to a standard helper */ 1257 struct dentry * (*find_exported_dentry)( 1258 struct super_block *sb, void *obj, void *parent, 1259 int (*acceptable)(void *context, struct dentry *de), 1260 void *context); 1261 1262 1263 }; 1264 1265 extern struct dentry * 1266 find_exported_dentry(struct super_block *sb, void *obj, void *parent, 1267 int (*acceptable)(void *context, struct dentry *de), 1268 void *context); 1269 1270 struct file_system_type { 1271 const char *name; 1272 int fs_flags; 1273 int (*get_sb) (struct file_system_type *, int, 1274 const char *, void *, struct vfsmount *); 1275 void (*kill_sb) (struct super_block *); 1276 struct module *owner; 1277 struct file_system_type * next; 1278 struct list_head fs_supers; 1279 }; 1280 1281 extern int get_sb_bdev(struct file_system_type *fs_type, 1282 int flags, const char *dev_name, void *data, 1283 int (*fill_super)(struct super_block *, void *, int), 1284 struct vfsmount *mnt); 1285 extern int get_sb_single(struct file_system_type *fs_type, 1286 int flags, void *data, 1287 int (*fill_super)(struct super_block *, void *, int), 1288 struct vfsmount *mnt); 1289 extern int get_sb_nodev(struct file_system_type *fs_type, 1290 int flags, void *data, 1291 int (*fill_super)(struct super_block *, void *, int), 1292 struct vfsmount *mnt); 1293 void generic_shutdown_super(struct super_block *sb); 1294 void kill_block_super(struct super_block *sb); 1295 void kill_anon_super(struct super_block *sb); 1296 void kill_litter_super(struct super_block *sb); 1297 void deactivate_super(struct super_block *sb); 1298 int set_anon_super(struct super_block *s, void *data); 1299 struct super_block *sget(struct file_system_type *type, 1300 int (*test)(struct super_block *,void *), 1301 int (*set)(struct super_block *,void *), 1302 void *data); 1303 extern int get_sb_pseudo(struct file_system_type *, char *, 1304 struct super_operations *ops, unsigned long, 1305 struct vfsmount *mnt); 1306 extern int simple_set_mnt(struct vfsmount *mnt, struct super_block *sb); 1307 int __put_super(struct super_block *sb); 1308 int __put_super_and_need_restart(struct super_block *sb); 1309 void unnamed_dev_init(void); 1310 1311 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */ 1312 #define fops_get(fops) \ 1313 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL)) 1314 #define fops_put(fops) \ 1315 do { if (fops) module_put((fops)->owner); } while(0) 1316 1317 extern int register_filesystem(struct file_system_type *); 1318 extern int unregister_filesystem(struct file_system_type *); 1319 extern struct vfsmount *kern_mount(struct file_system_type *); 1320 extern int may_umount_tree(struct vfsmount *); 1321 extern int may_umount(struct vfsmount *); 1322 extern void umount_tree(struct vfsmount *, int, struct list_head *); 1323 extern void release_mounts(struct list_head *); 1324 extern long do_mount(char *, char *, char *, unsigned long, void *); 1325 extern struct vfsmount *copy_tree(struct vfsmount *, struct dentry *, int); 1326 extern void mnt_set_mountpoint(struct vfsmount *, struct dentry *, 1327 struct vfsmount *); 1328 1329 extern int vfs_statfs(struct dentry *, struct kstatfs *); 1330 1331 /* /sys/fs */ 1332 extern struct subsystem fs_subsys; 1333 1334 #define FLOCK_VERIFY_READ 1 1335 #define FLOCK_VERIFY_WRITE 2 1336 1337 extern int locks_mandatory_locked(struct inode *); 1338 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t); 1339 1340 /* 1341 * Candidates for mandatory locking have the setgid bit set 1342 * but no group execute bit - an otherwise meaningless combination. 1343 */ 1344 #define MANDATORY_LOCK(inode) \ 1345 (IS_MANDLOCK(inode) && ((inode)->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID) 1346 1347 static inline int locks_verify_locked(struct inode *inode) 1348 { 1349 if (MANDATORY_LOCK(inode)) 1350 return locks_mandatory_locked(inode); 1351 return 0; 1352 } 1353 1354 extern int rw_verify_area(int, struct file *, loff_t *, size_t); 1355 1356 static inline int locks_verify_truncate(struct inode *inode, 1357 struct file *filp, 1358 loff_t size) 1359 { 1360 if (inode->i_flock && MANDATORY_LOCK(inode)) 1361 return locks_mandatory_area( 1362 FLOCK_VERIFY_WRITE, inode, filp, 1363 size < inode->i_size ? size : inode->i_size, 1364 (size < inode->i_size ? inode->i_size - size 1365 : size - inode->i_size) 1366 ); 1367 return 0; 1368 } 1369 1370 static inline int break_lease(struct inode *inode, unsigned int mode) 1371 { 1372 if (inode->i_flock) 1373 return __break_lease(inode, mode); 1374 return 0; 1375 } 1376 1377 /* fs/open.c */ 1378 1379 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs, 1380 struct file *filp); 1381 extern long do_sys_open(int fdf, const char __user *filename, int flags, 1382 int mode); 1383 extern struct file *filp_open(const char *, int, int); 1384 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int); 1385 extern int filp_close(struct file *, fl_owner_t id); 1386 extern char * getname(const char __user *); 1387 1388 /* fs/dcache.c */ 1389 extern void __init vfs_caches_init_early(void); 1390 extern void __init vfs_caches_init(unsigned long); 1391 1392 #define __getname() kmem_cache_alloc(names_cachep, SLAB_KERNEL) 1393 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name)) 1394 #ifndef CONFIG_AUDITSYSCALL 1395 #define putname(name) __putname(name) 1396 #else 1397 extern void putname(const char *name); 1398 #endif 1399 1400 extern int register_blkdev(unsigned int, const char *); 1401 extern int unregister_blkdev(unsigned int, const char *); 1402 extern struct block_device *bdget(dev_t); 1403 extern void bd_set_size(struct block_device *, loff_t size); 1404 extern void bd_forget(struct inode *inode); 1405 extern void bdput(struct block_device *); 1406 extern struct block_device *open_by_devnum(dev_t, unsigned); 1407 extern const struct file_operations def_blk_fops; 1408 extern const struct address_space_operations def_blk_aops; 1409 extern const struct file_operations def_chr_fops; 1410 extern const struct file_operations bad_sock_fops; 1411 extern const struct file_operations def_fifo_fops; 1412 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long); 1413 extern int blkdev_ioctl(struct inode *, struct file *, unsigned, unsigned long); 1414 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long); 1415 extern int blkdev_get(struct block_device *, mode_t, unsigned); 1416 extern int blkdev_put(struct block_device *); 1417 extern int bd_claim(struct block_device *, void *); 1418 extern void bd_release(struct block_device *); 1419 #ifdef CONFIG_SYSFS 1420 extern int bd_claim_by_disk(struct block_device *, void *, struct gendisk *); 1421 extern void bd_release_from_disk(struct block_device *, struct gendisk *); 1422 #else 1423 #define bd_claim_by_disk(bdev, holder, disk) bd_claim(bdev, holder) 1424 #define bd_release_from_disk(bdev, disk) bd_release(bdev) 1425 #endif 1426 1427 /* fs/char_dev.c */ 1428 #define CHRDEV_MAJOR_HASH_SIZE 255 1429 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *); 1430 extern int register_chrdev_region(dev_t, unsigned, const char *); 1431 extern int register_chrdev(unsigned int, const char *, 1432 const struct file_operations *); 1433 extern int unregister_chrdev(unsigned int, const char *); 1434 extern void unregister_chrdev_region(dev_t, unsigned); 1435 extern int chrdev_open(struct inode *, struct file *); 1436 extern void chrdev_show(struct seq_file *,off_t); 1437 1438 /* fs/block_dev.c */ 1439 #define BLKDEV_MAJOR_HASH_SIZE 255 1440 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */ 1441 extern const char *__bdevname(dev_t, char *buffer); 1442 extern const char *bdevname(struct block_device *bdev, char *buffer); 1443 extern struct block_device *lookup_bdev(const char *); 1444 extern struct block_device *open_bdev_excl(const char *, int, void *); 1445 extern void close_bdev_excl(struct block_device *); 1446 extern void blkdev_show(struct seq_file *,off_t); 1447 1448 extern void init_special_inode(struct inode *, umode_t, dev_t); 1449 1450 /* Invalid inode operations -- fs/bad_inode.c */ 1451 extern void make_bad_inode(struct inode *); 1452 extern int is_bad_inode(struct inode *); 1453 1454 extern const struct file_operations read_fifo_fops; 1455 extern const struct file_operations write_fifo_fops; 1456 extern const struct file_operations rdwr_fifo_fops; 1457 1458 extern int fs_may_remount_ro(struct super_block *); 1459 1460 /* 1461 * return READ, READA, or WRITE 1462 */ 1463 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK)) 1464 1465 /* 1466 * return data direction, READ or WRITE 1467 */ 1468 #define bio_data_dir(bio) ((bio)->bi_rw & 1) 1469 1470 extern int check_disk_change(struct block_device *); 1471 extern int invalidate_inodes(struct super_block *); 1472 extern int __invalidate_device(struct block_device *); 1473 extern int invalidate_partition(struct gendisk *, int); 1474 unsigned long invalidate_mapping_pages(struct address_space *mapping, 1475 pgoff_t start, pgoff_t end); 1476 unsigned long invalidate_inode_pages(struct address_space *mapping); 1477 static inline void invalidate_remote_inode(struct inode *inode) 1478 { 1479 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) || 1480 S_ISLNK(inode->i_mode)) 1481 invalidate_inode_pages(inode->i_mapping); 1482 } 1483 extern int invalidate_inode_pages2(struct address_space *mapping); 1484 extern int invalidate_inode_pages2_range(struct address_space *mapping, 1485 pgoff_t start, pgoff_t end); 1486 extern int write_inode_now(struct inode *, int); 1487 extern int filemap_fdatawrite(struct address_space *); 1488 extern int filemap_flush(struct address_space *); 1489 extern int filemap_fdatawait(struct address_space *); 1490 extern int filemap_write_and_wait(struct address_space *mapping); 1491 extern int filemap_write_and_wait_range(struct address_space *mapping, 1492 loff_t lstart, loff_t lend); 1493 extern int wait_on_page_writeback_range(struct address_space *mapping, 1494 pgoff_t start, pgoff_t end); 1495 extern int __filemap_fdatawrite_range(struct address_space *mapping, 1496 loff_t start, loff_t end, int sync_mode); 1497 1498 extern long do_fsync(struct file *file, int datasync); 1499 extern void sync_supers(void); 1500 extern void sync_filesystems(int wait); 1501 extern void emergency_sync(void); 1502 extern void emergency_remount(void); 1503 extern int do_remount_sb(struct super_block *sb, int flags, 1504 void *data, int force); 1505 extern sector_t bmap(struct inode *, sector_t); 1506 extern int notify_change(struct dentry *, struct iattr *); 1507 extern int permission(struct inode *, int, struct nameidata *); 1508 extern int generic_permission(struct inode *, int, 1509 int (*check_acl)(struct inode *, int)); 1510 1511 extern int get_write_access(struct inode *); 1512 extern int deny_write_access(struct file *); 1513 static inline void put_write_access(struct inode * inode) 1514 { 1515 atomic_dec(&inode->i_writecount); 1516 } 1517 static inline void allow_write_access(struct file *file) 1518 { 1519 if (file) 1520 atomic_inc(&file->f_dentry->d_inode->i_writecount); 1521 } 1522 extern int do_pipe(int *); 1523 1524 extern int open_namei(int dfd, const char *, int, int, struct nameidata *); 1525 extern int may_open(struct nameidata *, int, int); 1526 1527 extern int kernel_read(struct file *, unsigned long, char *, unsigned long); 1528 extern struct file * open_exec(const char *); 1529 1530 /* fs/dcache.c -- generic fs support functions */ 1531 extern int is_subdir(struct dentry *, struct dentry *); 1532 extern ino_t find_inode_number(struct dentry *, struct qstr *); 1533 1534 #include <linux/err.h> 1535 1536 /* needed for stackable file system support */ 1537 extern loff_t default_llseek(struct file *file, loff_t offset, int origin); 1538 1539 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin); 1540 1541 extern void inode_init_once(struct inode *); 1542 extern void iput(struct inode *); 1543 extern struct inode * igrab(struct inode *); 1544 extern ino_t iunique(struct super_block *, ino_t); 1545 extern int inode_needs_sync(struct inode *inode); 1546 extern void generic_delete_inode(struct inode *inode); 1547 extern void generic_drop_inode(struct inode *inode); 1548 1549 extern struct inode *ilookup5_nowait(struct super_block *sb, 1550 unsigned long hashval, int (*test)(struct inode *, void *), 1551 void *data); 1552 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval, 1553 int (*test)(struct inode *, void *), void *data); 1554 extern struct inode *ilookup(struct super_block *sb, unsigned long ino); 1555 1556 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *); 1557 extern struct inode * iget_locked(struct super_block *, unsigned long); 1558 extern void unlock_new_inode(struct inode *); 1559 1560 static inline struct inode *iget(struct super_block *sb, unsigned long ino) 1561 { 1562 struct inode *inode = iget_locked(sb, ino); 1563 1564 if (inode && (inode->i_state & I_NEW)) { 1565 sb->s_op->read_inode(inode); 1566 unlock_new_inode(inode); 1567 } 1568 1569 return inode; 1570 } 1571 1572 extern void __iget(struct inode * inode); 1573 extern void clear_inode(struct inode *); 1574 extern void destroy_inode(struct inode *); 1575 extern struct inode *new_inode(struct super_block *); 1576 extern int remove_suid(struct dentry *); 1577 extern void remove_dquot_ref(struct super_block *, int, struct list_head *); 1578 1579 extern void __insert_inode_hash(struct inode *, unsigned long hashval); 1580 extern void remove_inode_hash(struct inode *); 1581 static inline void insert_inode_hash(struct inode *inode) { 1582 __insert_inode_hash(inode, inode->i_ino); 1583 } 1584 1585 extern struct file * get_empty_filp(void); 1586 extern void file_move(struct file *f, struct list_head *list); 1587 extern void file_kill(struct file *f); 1588 struct bio; 1589 extern void submit_bio(int, struct bio *); 1590 extern int bdev_read_only(struct block_device *); 1591 extern int set_blocksize(struct block_device *, int); 1592 extern int sb_set_blocksize(struct super_block *, int); 1593 extern int sb_min_blocksize(struct super_block *, int); 1594 1595 extern int generic_file_mmap(struct file *, struct vm_area_struct *); 1596 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *); 1597 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size); 1598 extern int file_send_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size); 1599 extern ssize_t generic_file_read(struct file *, char __user *, size_t, loff_t *); 1600 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk); 1601 extern ssize_t generic_file_write(struct file *, const char __user *, size_t, loff_t *); 1602 extern ssize_t generic_file_aio_read(struct kiocb *, char __user *, size_t, loff_t); 1603 extern ssize_t __generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t *); 1604 extern ssize_t generic_file_aio_write(struct kiocb *, const char __user *, size_t, loff_t); 1605 extern ssize_t generic_file_aio_write_nolock(struct kiocb *, const struct iovec *, 1606 unsigned long, loff_t *); 1607 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *, 1608 unsigned long *, loff_t, loff_t *, size_t, size_t); 1609 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *, 1610 unsigned long, loff_t, loff_t *, size_t, ssize_t); 1611 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos); 1612 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos); 1613 ssize_t generic_file_write_nolock(struct file *file, const struct iovec *iov, 1614 unsigned long nr_segs, loff_t *ppos); 1615 extern ssize_t generic_file_sendfile(struct file *, loff_t *, size_t, read_actor_t, void *); 1616 extern void do_generic_mapping_read(struct address_space *mapping, 1617 struct file_ra_state *, struct file *, 1618 loff_t *, read_descriptor_t *, read_actor_t); 1619 1620 /* fs/splice.c */ 1621 extern ssize_t generic_file_splice_read(struct file *, loff_t *, 1622 struct pipe_inode_info *, size_t, unsigned int); 1623 extern ssize_t generic_file_splice_write(struct pipe_inode_info *, 1624 struct file *, loff_t *, size_t, unsigned int); 1625 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe, 1626 struct file *out, loff_t *, size_t len, unsigned int flags); 1627 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out, 1628 size_t len, unsigned int flags); 1629 1630 extern void 1631 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping); 1632 extern ssize_t generic_file_readv(struct file *filp, const struct iovec *iov, 1633 unsigned long nr_segs, loff_t *ppos); 1634 ssize_t generic_file_writev(struct file *filp, const struct iovec *iov, 1635 unsigned long nr_segs, loff_t *ppos); 1636 extern loff_t no_llseek(struct file *file, loff_t offset, int origin); 1637 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin); 1638 extern loff_t remote_llseek(struct file *file, loff_t offset, int origin); 1639 extern int generic_file_open(struct inode * inode, struct file * filp); 1640 extern int nonseekable_open(struct inode * inode, struct file * filp); 1641 1642 #ifdef CONFIG_FS_XIP 1643 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len, 1644 loff_t *ppos); 1645 extern ssize_t xip_file_sendfile(struct file *in_file, loff_t *ppos, 1646 size_t count, read_actor_t actor, 1647 void *target); 1648 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma); 1649 extern ssize_t xip_file_write(struct file *filp, const char __user *buf, 1650 size_t len, loff_t *ppos); 1651 extern int xip_truncate_page(struct address_space *mapping, loff_t from); 1652 #else 1653 static inline int xip_truncate_page(struct address_space *mapping, loff_t from) 1654 { 1655 return 0; 1656 } 1657 #endif 1658 1659 static inline void do_generic_file_read(struct file * filp, loff_t *ppos, 1660 read_descriptor_t * desc, 1661 read_actor_t actor) 1662 { 1663 do_generic_mapping_read(filp->f_mapping, 1664 &filp->f_ra, 1665 filp, 1666 ppos, 1667 desc, 1668 actor); 1669 } 1670 1671 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode, 1672 struct block_device *bdev, const struct iovec *iov, loff_t offset, 1673 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io, 1674 int lock_type); 1675 1676 enum { 1677 DIO_LOCKING = 1, /* need locking between buffered and direct access */ 1678 DIO_NO_LOCKING, /* bdev; no locking at all between buffered/direct */ 1679 DIO_OWN_LOCKING, /* filesystem locks buffered and direct internally */ 1680 }; 1681 1682 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb, 1683 struct inode *inode, struct block_device *bdev, const struct iovec *iov, 1684 loff_t offset, unsigned long nr_segs, get_block_t get_block, 1685 dio_iodone_t end_io) 1686 { 1687 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset, 1688 nr_segs, get_block, end_io, DIO_LOCKING); 1689 } 1690 1691 static inline ssize_t blockdev_direct_IO_no_locking(int rw, struct kiocb *iocb, 1692 struct inode *inode, struct block_device *bdev, const struct iovec *iov, 1693 loff_t offset, unsigned long nr_segs, get_block_t get_block, 1694 dio_iodone_t end_io) 1695 { 1696 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset, 1697 nr_segs, get_block, end_io, DIO_NO_LOCKING); 1698 } 1699 1700 static inline ssize_t blockdev_direct_IO_own_locking(int rw, struct kiocb *iocb, 1701 struct inode *inode, struct block_device *bdev, const struct iovec *iov, 1702 loff_t offset, unsigned long nr_segs, get_block_t get_block, 1703 dio_iodone_t end_io) 1704 { 1705 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset, 1706 nr_segs, get_block, end_io, DIO_OWN_LOCKING); 1707 } 1708 1709 extern const struct file_operations generic_ro_fops; 1710 1711 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m)) 1712 1713 extern int vfs_readlink(struct dentry *, char __user *, int, const char *); 1714 extern int vfs_follow_link(struct nameidata *, const char *); 1715 extern int page_readlink(struct dentry *, char __user *, int); 1716 extern void *page_follow_link_light(struct dentry *, struct nameidata *); 1717 extern void page_put_link(struct dentry *, struct nameidata *, void *); 1718 extern int __page_symlink(struct inode *inode, const char *symname, int len, 1719 gfp_t gfp_mask); 1720 extern int page_symlink(struct inode *inode, const char *symname, int len); 1721 extern struct inode_operations page_symlink_inode_operations; 1722 extern int generic_readlink(struct dentry *, char __user *, int); 1723 extern void generic_fillattr(struct inode *, struct kstat *); 1724 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *); 1725 void inode_add_bytes(struct inode *inode, loff_t bytes); 1726 void inode_sub_bytes(struct inode *inode, loff_t bytes); 1727 loff_t inode_get_bytes(struct inode *inode); 1728 void inode_set_bytes(struct inode *inode, loff_t bytes); 1729 1730 extern int vfs_readdir(struct file *, filldir_t, void *); 1731 1732 extern int vfs_stat(char __user *, struct kstat *); 1733 extern int vfs_lstat(char __user *, struct kstat *); 1734 extern int vfs_stat_fd(int dfd, char __user *, struct kstat *); 1735 extern int vfs_lstat_fd(int dfd, char __user *, struct kstat *); 1736 extern int vfs_fstat(unsigned int, struct kstat *); 1737 1738 extern int vfs_ioctl(struct file *, unsigned int, unsigned int, unsigned long); 1739 1740 extern struct file_system_type *get_fs_type(const char *name); 1741 extern struct super_block *get_super(struct block_device *); 1742 extern struct super_block *user_get_super(dev_t); 1743 extern void drop_super(struct super_block *sb); 1744 1745 extern int dcache_dir_open(struct inode *, struct file *); 1746 extern int dcache_dir_close(struct inode *, struct file *); 1747 extern loff_t dcache_dir_lseek(struct file *, loff_t, int); 1748 extern int dcache_readdir(struct file *, void *, filldir_t); 1749 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *); 1750 extern int simple_statfs(struct dentry *, struct kstatfs *); 1751 extern int simple_link(struct dentry *, struct inode *, struct dentry *); 1752 extern int simple_unlink(struct inode *, struct dentry *); 1753 extern int simple_rmdir(struct inode *, struct dentry *); 1754 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *); 1755 extern int simple_sync_file(struct file *, struct dentry *, int); 1756 extern int simple_empty(struct dentry *); 1757 extern int simple_readpage(struct file *file, struct page *page); 1758 extern int simple_prepare_write(struct file *file, struct page *page, 1759 unsigned offset, unsigned to); 1760 extern int simple_commit_write(struct file *file, struct page *page, 1761 unsigned offset, unsigned to); 1762 1763 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *); 1764 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *); 1765 extern const struct file_operations simple_dir_operations; 1766 extern struct inode_operations simple_dir_inode_operations; 1767 struct tree_descr { char *name; const struct file_operations *ops; int mode; }; 1768 struct dentry *d_alloc_name(struct dentry *, const char *); 1769 extern int simple_fill_super(struct super_block *, int, struct tree_descr *); 1770 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count); 1771 extern void simple_release_fs(struct vfsmount **mount, int *count); 1772 1773 extern ssize_t simple_read_from_buffer(void __user *, size_t, loff_t *, const void *, size_t); 1774 1775 #ifdef CONFIG_MIGRATION 1776 extern int buffer_migrate_page(struct address_space *, 1777 struct page *, struct page *); 1778 #else 1779 #define buffer_migrate_page NULL 1780 #endif 1781 1782 extern int inode_change_ok(struct inode *, struct iattr *); 1783 extern int __must_check inode_setattr(struct inode *, struct iattr *); 1784 1785 extern void file_update_time(struct file *file); 1786 1787 static inline ino_t parent_ino(struct dentry *dentry) 1788 { 1789 ino_t res; 1790 1791 spin_lock(&dentry->d_lock); 1792 res = dentry->d_parent->d_inode->i_ino; 1793 spin_unlock(&dentry->d_lock); 1794 return res; 1795 } 1796 1797 /* kernel/fork.c */ 1798 extern int unshare_files(void); 1799 1800 /* Transaction based IO helpers */ 1801 1802 /* 1803 * An argresp is stored in an allocated page and holds the 1804 * size of the argument or response, along with its content 1805 */ 1806 struct simple_transaction_argresp { 1807 ssize_t size; 1808 char data[0]; 1809 }; 1810 1811 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp)) 1812 1813 char *simple_transaction_get(struct file *file, const char __user *buf, 1814 size_t size); 1815 ssize_t simple_transaction_read(struct file *file, char __user *buf, 1816 size_t size, loff_t *pos); 1817 int simple_transaction_release(struct inode *inode, struct file *file); 1818 1819 static inline void simple_transaction_set(struct file *file, size_t n) 1820 { 1821 struct simple_transaction_argresp *ar = file->private_data; 1822 1823 BUG_ON(n > SIMPLE_TRANSACTION_LIMIT); 1824 1825 /* 1826 * The barrier ensures that ar->size will really remain zero until 1827 * ar->data is ready for reading. 1828 */ 1829 smp_mb(); 1830 ar->size = n; 1831 } 1832 1833 /* 1834 * simple attribute files 1835 * 1836 * These attributes behave similar to those in sysfs: 1837 * 1838 * Writing to an attribute immediately sets a value, an open file can be 1839 * written to multiple times. 1840 * 1841 * Reading from an attribute creates a buffer from the value that might get 1842 * read with multiple read calls. When the attribute has been read 1843 * completely, no further read calls are possible until the file is opened 1844 * again. 1845 * 1846 * All attributes contain a text representation of a numeric value 1847 * that are accessed with the get() and set() functions. 1848 */ 1849 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \ 1850 static int __fops ## _open(struct inode *inode, struct file *file) \ 1851 { \ 1852 __simple_attr_check_format(__fmt, 0ull); \ 1853 return simple_attr_open(inode, file, __get, __set, __fmt); \ 1854 } \ 1855 static struct file_operations __fops = { \ 1856 .owner = THIS_MODULE, \ 1857 .open = __fops ## _open, \ 1858 .release = simple_attr_close, \ 1859 .read = simple_attr_read, \ 1860 .write = simple_attr_write, \ 1861 }; 1862 1863 static inline void __attribute__((format(printf, 1, 2))) 1864 __simple_attr_check_format(const char *fmt, ...) 1865 { 1866 /* don't do anything, just let the compiler check the arguments; */ 1867 } 1868 1869 int simple_attr_open(struct inode *inode, struct file *file, 1870 u64 (*get)(void *), void (*set)(void *, u64), 1871 const char *fmt); 1872 int simple_attr_close(struct inode *inode, struct file *file); 1873 ssize_t simple_attr_read(struct file *file, char __user *buf, 1874 size_t len, loff_t *ppos); 1875 ssize_t simple_attr_write(struct file *file, const char __user *buf, 1876 size_t len, loff_t *ppos); 1877 1878 1879 #ifdef CONFIG_SECURITY 1880 static inline char *alloc_secdata(void) 1881 { 1882 return (char *)get_zeroed_page(GFP_KERNEL); 1883 } 1884 1885 static inline void free_secdata(void *secdata) 1886 { 1887 free_page((unsigned long)secdata); 1888 } 1889 #else 1890 static inline char *alloc_secdata(void) 1891 { 1892 return (char *)1; 1893 } 1894 1895 static inline void free_secdata(void *secdata) 1896 { } 1897 #endif /* CONFIG_SECURITY */ 1898 1899 #endif /* __KERNEL__ */ 1900 #endif /* _LINUX_FS_H */ 1901