1 2 Overview of the Linux Virtual File System 3 4 Original author: Richard Gooch <rgooch@atnf.csiro.au> 5 6 Last updated on June 24, 2007. 7 8 Copyright (C) 1999 Richard Gooch 9 Copyright (C) 2005 Pekka Enberg 10 11 This file is released under the GPLv2. 12 13 14Introduction 15============ 16 17The Virtual File System (also known as the Virtual Filesystem Switch) 18is the software layer in the kernel that provides the filesystem 19interface to userspace programs. It also provides an abstraction 20within the kernel which allows different filesystem implementations to 21coexist. 22 23VFS system calls open(2), stat(2), read(2), write(2), chmod(2) and so 24on are called from a process context. Filesystem locking is described 25in the document Documentation/filesystems/Locking. 26 27 28Directory Entry Cache (dcache) 29------------------------------ 30 31The VFS implements the open(2), stat(2), chmod(2), and similar system 32calls. The pathname argument that is passed to them is used by the VFS 33to search through the directory entry cache (also known as the dentry 34cache or dcache). This provides a very fast look-up mechanism to 35translate a pathname (filename) into a specific dentry. Dentries live 36in RAM and are never saved to disc: they exist only for performance. 37 38The dentry cache is meant to be a view into your entire filespace. As 39most computers cannot fit all dentries in the RAM at the same time, 40some bits of the cache are missing. In order to resolve your pathname 41into a dentry, the VFS may have to resort to creating dentries along 42the way, and then loading the inode. This is done by looking up the 43inode. 44 45 46The Inode Object 47---------------- 48 49An individual dentry usually has a pointer to an inode. Inodes are 50filesystem objects such as regular files, directories, FIFOs and other 51beasts. They live either on the disc (for block device filesystems) 52or in the memory (for pseudo filesystems). Inodes that live on the 53disc are copied into the memory when required and changes to the inode 54are written back to disc. A single inode can be pointed to by multiple 55dentries (hard links, for example, do this). 56 57To look up an inode requires that the VFS calls the lookup() method of 58the parent directory inode. This method is installed by the specific 59filesystem implementation that the inode lives in. Once the VFS has 60the required dentry (and hence the inode), we can do all those boring 61things like open(2) the file, or stat(2) it to peek at the inode 62data. The stat(2) operation is fairly simple: once the VFS has the 63dentry, it peeks at the inode data and passes some of it back to 64userspace. 65 66 67The File Object 68--------------- 69 70Opening a file requires another operation: allocation of a file 71structure (this is the kernel-side implementation of file 72descriptors). The freshly allocated file structure is initialized with 73a pointer to the dentry and a set of file operation member functions. 74These are taken from the inode data. The open() file method is then 75called so the specific filesystem implementation can do its work. You 76can see that this is another switch performed by the VFS. The file 77structure is placed into the file descriptor table for the process. 78 79Reading, writing and closing files (and other assorted VFS operations) 80is done by using the userspace file descriptor to grab the appropriate 81file structure, and then calling the required file structure method to 82do whatever is required. For as long as the file is open, it keeps the 83dentry in use, which in turn means that the VFS inode is still in use. 84 85 86Registering and Mounting a Filesystem 87===================================== 88 89To register and unregister a filesystem, use the following API 90functions: 91 92 #include <linux/fs.h> 93 94 extern int register_filesystem(struct file_system_type *); 95 extern int unregister_filesystem(struct file_system_type *); 96 97The passed struct file_system_type describes your filesystem. When a 98request is made to mount a filesystem onto a directory in your namespace, 99the VFS will call the appropriate mount() method for the specific 100filesystem. New vfsmount referring to the tree returned by ->mount() 101will be attached to the mountpoint, so that when pathname resolution 102reaches the mountpoint it will jump into the root of that vfsmount. 103 104You can see all filesystems that are registered to the kernel in the 105file /proc/filesystems. 106 107 108struct file_system_type 109----------------------- 110 111This describes the filesystem. As of kernel 2.6.39, the following 112members are defined: 113 114struct file_system_type { 115 const char *name; 116 int fs_flags; 117 struct dentry *(*mount) (struct file_system_type *, int, 118 const char *, void *); 119 void (*kill_sb) (struct super_block *); 120 struct module *owner; 121 struct file_system_type * next; 122 struct list_head fs_supers; 123 struct lock_class_key s_lock_key; 124 struct lock_class_key s_umount_key; 125}; 126 127 name: the name of the filesystem type, such as "ext2", "iso9660", 128 "msdos" and so on 129 130 fs_flags: various flags (i.e. FS_REQUIRES_DEV, FS_NO_DCACHE, etc.) 131 132 mount: the method to call when a new instance of this 133 filesystem should be mounted 134 135 kill_sb: the method to call when an instance of this filesystem 136 should be shut down 137 138 owner: for internal VFS use: you should initialize this to THIS_MODULE in 139 most cases. 140 141 next: for internal VFS use: you should initialize this to NULL 142 143 s_lock_key, s_umount_key: lockdep-specific 144 145The mount() method has the following arguments: 146 147 struct file_system_type *fs_type: describes the filesystem, partly initialized 148 by the specific filesystem code 149 150 int flags: mount flags 151 152 const char *dev_name: the device name we are mounting. 153 154 void *data: arbitrary mount options, usually comes as an ASCII 155 string (see "Mount Options" section) 156 157The mount() method must return the root dentry of the tree requested by 158caller. An active reference to its superblock must be grabbed and the 159superblock must be locked. On failure it should return ERR_PTR(error). 160 161The arguments match those of mount(2) and their interpretation 162depends on filesystem type. E.g. for block filesystems, dev_name is 163interpreted as block device name, that device is opened and if it 164contains a suitable filesystem image the method creates and initializes 165struct super_block accordingly, returning its root dentry to caller. 166 167->mount() may choose to return a subtree of existing filesystem - it 168doesn't have to create a new one. The main result from the caller's 169point of view is a reference to dentry at the root of (sub)tree to 170be attached; creation of new superblock is a common side effect. 171 172The most interesting member of the superblock structure that the 173mount() method fills in is the "s_op" field. This is a pointer to 174a "struct super_operations" which describes the next level of the 175filesystem implementation. 176 177Usually, a filesystem uses one of the generic mount() implementations 178and provides a fill_super() callback instead. The generic variants are: 179 180 mount_bdev: mount a filesystem residing on a block device 181 182 mount_nodev: mount a filesystem that is not backed by a device 183 184 mount_single: mount a filesystem which shares the instance between 185 all mounts 186 187A fill_super() callback implementation has the following arguments: 188 189 struct super_block *sb: the superblock structure. The callback 190 must initialize this properly. 191 192 void *data: arbitrary mount options, usually comes as an ASCII 193 string (see "Mount Options" section) 194 195 int silent: whether or not to be silent on error 196 197 198The Superblock Object 199===================== 200 201A superblock object represents a mounted filesystem. 202 203 204struct super_operations 205----------------------- 206 207This describes how the VFS can manipulate the superblock of your 208filesystem. As of kernel 2.6.22, the following members are defined: 209 210struct super_operations { 211 struct inode *(*alloc_inode)(struct super_block *sb); 212 void (*destroy_inode)(struct inode *); 213 214 void (*dirty_inode) (struct inode *, int flags); 215 int (*write_inode) (struct inode *, int); 216 void (*drop_inode) (struct inode *); 217 void (*delete_inode) (struct inode *); 218 void (*put_super) (struct super_block *); 219 int (*sync_fs)(struct super_block *sb, int wait); 220 int (*freeze_fs) (struct super_block *); 221 int (*unfreeze_fs) (struct super_block *); 222 int (*statfs) (struct dentry *, struct kstatfs *); 223 int (*remount_fs) (struct super_block *, int *, char *); 224 void (*clear_inode) (struct inode *); 225 void (*umount_begin) (struct super_block *); 226 227 int (*show_options)(struct seq_file *, struct dentry *); 228 229 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t); 230 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t); 231 int (*nr_cached_objects)(struct super_block *); 232 void (*free_cached_objects)(struct super_block *, int); 233}; 234 235All methods are called without any locks being held, unless otherwise 236noted. This means that most methods can block safely. All methods are 237only called from a process context (i.e. not from an interrupt handler 238or bottom half). 239 240 alloc_inode: this method is called by alloc_inode() to allocate memory 241 for struct inode and initialize it. If this function is not 242 defined, a simple 'struct inode' is allocated. Normally 243 alloc_inode will be used to allocate a larger structure which 244 contains a 'struct inode' embedded within it. 245 246 destroy_inode: this method is called by destroy_inode() to release 247 resources allocated for struct inode. It is only required if 248 ->alloc_inode was defined and simply undoes anything done by 249 ->alloc_inode. 250 251 dirty_inode: this method is called by the VFS to mark an inode dirty. 252 253 write_inode: this method is called when the VFS needs to write an 254 inode to disc. The second parameter indicates whether the write 255 should be synchronous or not, not all filesystems check this flag. 256 257 drop_inode: called when the last access to the inode is dropped, 258 with the inode->i_lock spinlock held. 259 260 This method should be either NULL (normal UNIX filesystem 261 semantics) or "generic_delete_inode" (for filesystems that do not 262 want to cache inodes - causing "delete_inode" to always be 263 called regardless of the value of i_nlink) 264 265 The "generic_delete_inode()" behavior is equivalent to the 266 old practice of using "force_delete" in the put_inode() case, 267 but does not have the races that the "force_delete()" approach 268 had. 269 270 delete_inode: called when the VFS wants to delete an inode 271 272 put_super: called when the VFS wishes to free the superblock 273 (i.e. unmount). This is called with the superblock lock held 274 275 sync_fs: called when VFS is writing out all dirty data associated with 276 a superblock. The second parameter indicates whether the method 277 should wait until the write out has been completed. Optional. 278 279 freeze_fs: called when VFS is locking a filesystem and 280 forcing it into a consistent state. This method is currently 281 used by the Logical Volume Manager (LVM). 282 283 unfreeze_fs: called when VFS is unlocking a filesystem and making it writable 284 again. 285 286 statfs: called when the VFS needs to get filesystem statistics. 287 288 remount_fs: called when the filesystem is remounted. This is called 289 with the kernel lock held 290 291 clear_inode: called then the VFS clears the inode. Optional 292 293 umount_begin: called when the VFS is unmounting a filesystem. 294 295 show_options: called by the VFS to show mount options for 296 /proc/<pid>/mounts. (see "Mount Options" section) 297 298 quota_read: called by the VFS to read from filesystem quota file. 299 300 quota_write: called by the VFS to write to filesystem quota file. 301 302 nr_cached_objects: called by the sb cache shrinking function for the 303 filesystem to return the number of freeable cached objects it contains. 304 Optional. 305 306 free_cache_objects: called by the sb cache shrinking function for the 307 filesystem to scan the number of objects indicated to try to free them. 308 Optional, but any filesystem implementing this method needs to also 309 implement ->nr_cached_objects for it to be called correctly. 310 311 We can't do anything with any errors that the filesystem might 312 encountered, hence the void return type. This will never be called if 313 the VM is trying to reclaim under GFP_NOFS conditions, hence this 314 method does not need to handle that situation itself. 315 316 Implementations must include conditional reschedule calls inside any 317 scanning loop that is done. This allows the VFS to determine 318 appropriate scan batch sizes without having to worry about whether 319 implementations will cause holdoff problems due to large scan batch 320 sizes. 321 322Whoever sets up the inode is responsible for filling in the "i_op" field. This 323is a pointer to a "struct inode_operations" which describes the methods that 324can be performed on individual inodes. 325 326 327The Inode Object 328================ 329 330An inode object represents an object within the filesystem. 331 332 333struct inode_operations 334----------------------- 335 336This describes how the VFS can manipulate an inode in your 337filesystem. As of kernel 2.6.22, the following members are defined: 338 339struct inode_operations { 340 int (*create) (struct inode *,struct dentry *, umode_t, bool); 341 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int); 342 int (*link) (struct dentry *,struct inode *,struct dentry *); 343 int (*unlink) (struct inode *,struct dentry *); 344 int (*symlink) (struct inode *,struct dentry *,const char *); 345 int (*mkdir) (struct inode *,struct dentry *,umode_t); 346 int (*rmdir) (struct inode *,struct dentry *); 347 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t); 348 int (*rename) (struct inode *, struct dentry *, 349 struct inode *, struct dentry *); 350 int (*rename2) (struct inode *, struct dentry *, 351 struct inode *, struct dentry *, unsigned int); 352 int (*readlink) (struct dentry *, char __user *,int); 353 void * (*follow_link) (struct dentry *, struct nameidata *); 354 void (*put_link) (struct dentry *, struct nameidata *, void *); 355 int (*permission) (struct inode *, int); 356 int (*get_acl)(struct inode *, int); 357 int (*setattr) (struct dentry *, struct iattr *); 358 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *); 359 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int); 360 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t); 361 ssize_t (*listxattr) (struct dentry *, char *, size_t); 362 int (*removexattr) (struct dentry *, const char *); 363 void (*update_time)(struct inode *, struct timespec *, int); 364 int (*atomic_open)(struct inode *, struct dentry *, struct file *, 365 unsigned open_flag, umode_t create_mode, int *opened); 366 int (*tmpfile) (struct inode *, struct dentry *, umode_t); 367 int (*dentry_open)(struct dentry *, struct file *, const struct cred *); 368}; 369 370Again, all methods are called without any locks being held, unless 371otherwise noted. 372 373 create: called by the open(2) and creat(2) system calls. Only 374 required if you want to support regular files. The dentry you 375 get should not have an inode (i.e. it should be a negative 376 dentry). Here you will probably call d_instantiate() with the 377 dentry and the newly created inode 378 379 lookup: called when the VFS needs to look up an inode in a parent 380 directory. The name to look for is found in the dentry. This 381 method must call d_add() to insert the found inode into the 382 dentry. The "i_count" field in the inode structure should be 383 incremented. If the named inode does not exist a NULL inode 384 should be inserted into the dentry (this is called a negative 385 dentry). Returning an error code from this routine must only 386 be done on a real error, otherwise creating inodes with system 387 calls like create(2), mknod(2), mkdir(2) and so on will fail. 388 If you wish to overload the dentry methods then you should 389 initialise the "d_dop" field in the dentry; this is a pointer 390 to a struct "dentry_operations". 391 This method is called with the directory inode semaphore held 392 393 link: called by the link(2) system call. Only required if you want 394 to support hard links. You will probably need to call 395 d_instantiate() just as you would in the create() method 396 397 unlink: called by the unlink(2) system call. Only required if you 398 want to support deleting inodes 399 400 symlink: called by the symlink(2) system call. Only required if you 401 want to support symlinks. You will probably need to call 402 d_instantiate() just as you would in the create() method 403 404 mkdir: called by the mkdir(2) system call. Only required if you want 405 to support creating subdirectories. You will probably need to 406 call d_instantiate() just as you would in the create() method 407 408 rmdir: called by the rmdir(2) system call. Only required if you want 409 to support deleting subdirectories 410 411 mknod: called by the mknod(2) system call to create a device (char, 412 block) inode or a named pipe (FIFO) or socket. Only required 413 if you want to support creating these types of inodes. You 414 will probably need to call d_instantiate() just as you would 415 in the create() method 416 417 rename: called by the rename(2) system call to rename the object to 418 have the parent and name given by the second inode and dentry. 419 420 rename2: this has an additional flags argument compared to rename. 421 If no flags are supported by the filesystem then this method 422 need not be implemented. If some flags are supported then the 423 filesystem must return -EINVAL for any unsupported or unknown 424 flags. Currently the following flags are implemented: 425 (1) RENAME_NOREPLACE: this flag indicates that if the target 426 of the rename exists the rename should fail with -EEXIST 427 instead of replacing the target. The VFS already checks for 428 existence, so for local filesystems the RENAME_NOREPLACE 429 implementation is equivalent to plain rename. 430 (2) RENAME_EXCHANGE: exchange source and target. Both must 431 exist; this is checked by the VFS. Unlike plain rename, 432 source and target may be of different type. 433 434 readlink: called by the readlink(2) system call. Only required if 435 you want to support reading symbolic links 436 437 follow_link: called by the VFS to follow a symbolic link to the 438 inode it points to. Only required if you want to support 439 symbolic links. This method returns a void pointer cookie 440 that is passed to put_link(). 441 442 put_link: called by the VFS to release resources allocated by 443 follow_link(). The cookie returned by follow_link() is passed 444 to this method as the last parameter. It is used by 445 filesystems such as NFS where page cache is not stable 446 (i.e. page that was installed when the symbolic link walk 447 started might not be in the page cache at the end of the 448 walk). 449 450 permission: called by the VFS to check for access rights on a POSIX-like 451 filesystem. 452 453 May be called in rcu-walk mode (mask & MAY_NOT_BLOCK). If in rcu-walk 454 mode, the filesystem must check the permission without blocking or 455 storing to the inode. 456 457 If a situation is encountered that rcu-walk cannot handle, return 458 -ECHILD and it will be called again in ref-walk mode. 459 460 setattr: called by the VFS to set attributes for a file. This method 461 is called by chmod(2) and related system calls. 462 463 getattr: called by the VFS to get attributes of a file. This method 464 is called by stat(2) and related system calls. 465 466 setxattr: called by the VFS to set an extended attribute for a file. 467 Extended attribute is a name:value pair associated with an 468 inode. This method is called by setxattr(2) system call. 469 470 getxattr: called by the VFS to retrieve the value of an extended 471 attribute name. This method is called by getxattr(2) function 472 call. 473 474 listxattr: called by the VFS to list all extended attributes for a 475 given file. This method is called by listxattr(2) system call. 476 477 removexattr: called by the VFS to remove an extended attribute from 478 a file. This method is called by removexattr(2) system call. 479 480 update_time: called by the VFS to update a specific time or the i_version of 481 an inode. If this is not defined the VFS will update the inode itself 482 and call mark_inode_dirty_sync. 483 484 atomic_open: called on the last component of an open. Using this optional 485 method the filesystem can look up, possibly create and open the file in 486 one atomic operation. If it cannot perform this (e.g. the file type 487 turned out to be wrong) it may signal this by returning 1 instead of 488 usual 0 or -ve . This method is only called if the last component is 489 negative or needs lookup. Cached positive dentries are still handled by 490 f_op->open(). If the file was created, the FILE_CREATED flag should be 491 set in "opened". In case of O_EXCL the method must only succeed if the 492 file didn't exist and hence FILE_CREATED shall always be set on success. 493 494 tmpfile: called in the end of O_TMPFILE open(). Optional, equivalent to 495 atomically creating, opening and unlinking a file in given directory. 496 497The Address Space Object 498======================== 499 500The address space object is used to group and manage pages in the page 501cache. It can be used to keep track of the pages in a file (or 502anything else) and also track the mapping of sections of the file into 503process address spaces. 504 505There are a number of distinct yet related services that an 506address-space can provide. These include communicating memory 507pressure, page lookup by address, and keeping track of pages tagged as 508Dirty or Writeback. 509 510The first can be used independently to the others. The VM can try to 511either write dirty pages in order to clean them, or release clean 512pages in order to reuse them. To do this it can call the ->writepage 513method on dirty pages, and ->releasepage on clean pages with 514PagePrivate set. Clean pages without PagePrivate and with no external 515references will be released without notice being given to the 516address_space. 517 518To achieve this functionality, pages need to be placed on an LRU with 519lru_cache_add and mark_page_active needs to be called whenever the 520page is used. 521 522Pages are normally kept in a radix tree index by ->index. This tree 523maintains information about the PG_Dirty and PG_Writeback status of 524each page, so that pages with either of these flags can be found 525quickly. 526 527The Dirty tag is primarily used by mpage_writepages - the default 528->writepages method. It uses the tag to find dirty pages to call 529->writepage on. If mpage_writepages is not used (i.e. the address 530provides its own ->writepages) , the PAGECACHE_TAG_DIRTY tag is 531almost unused. write_inode_now and sync_inode do use it (through 532__sync_single_inode) to check if ->writepages has been successful in 533writing out the whole address_space. 534 535The Writeback tag is used by filemap*wait* and sync_page* functions, 536via filemap_fdatawait_range, to wait for all writeback to 537complete. While waiting ->sync_page (if defined) will be called on 538each page that is found to require writeback. 539 540An address_space handler may attach extra information to a page, 541typically using the 'private' field in the 'struct page'. If such 542information is attached, the PG_Private flag should be set. This will 543cause various VM routines to make extra calls into the address_space 544handler to deal with that data. 545 546An address space acts as an intermediate between storage and 547application. Data is read into the address space a whole page at a 548time, and provided to the application either by copying of the page, 549or by memory-mapping the page. 550Data is written into the address space by the application, and then 551written-back to storage typically in whole pages, however the 552address_space has finer control of write sizes. 553 554The read process essentially only requires 'readpage'. The write 555process is more complicated and uses write_begin/write_end or 556set_page_dirty to write data into the address_space, and writepage, 557sync_page, and writepages to writeback data to storage. 558 559Adding and removing pages to/from an address_space is protected by the 560inode's i_mutex. 561 562When data is written to a page, the PG_Dirty flag should be set. It 563typically remains set until writepage asks for it to be written. This 564should clear PG_Dirty and set PG_Writeback. It can be actually 565written at any point after PG_Dirty is clear. Once it is known to be 566safe, PG_Writeback is cleared. 567 568Writeback makes use of a writeback_control structure... 569 570struct address_space_operations 571------------------------------- 572 573This describes how the VFS can manipulate mapping of a file to page cache in 574your filesystem. The following members are defined: 575 576struct address_space_operations { 577 int (*writepage)(struct page *page, struct writeback_control *wbc); 578 int (*readpage)(struct file *, struct page *); 579 int (*writepages)(struct address_space *, struct writeback_control *); 580 int (*set_page_dirty)(struct page *page); 581 int (*readpages)(struct file *filp, struct address_space *mapping, 582 struct list_head *pages, unsigned nr_pages); 583 int (*write_begin)(struct file *, struct address_space *mapping, 584 loff_t pos, unsigned len, unsigned flags, 585 struct page **pagep, void **fsdata); 586 int (*write_end)(struct file *, struct address_space *mapping, 587 loff_t pos, unsigned len, unsigned copied, 588 struct page *page, void *fsdata); 589 sector_t (*bmap)(struct address_space *, sector_t); 590 void (*invalidatepage) (struct page *, unsigned int, unsigned int); 591 int (*releasepage) (struct page *, int); 592 void (*freepage)(struct page *); 593 ssize_t (*direct_IO)(int, struct kiocb *, struct iov_iter *iter, loff_t offset); 594 struct page* (*get_xip_page)(struct address_space *, sector_t, 595 int); 596 /* migrate the contents of a page to the specified target */ 597 int (*migratepage) (struct page *, struct page *); 598 int (*launder_page) (struct page *); 599 int (*is_partially_uptodate) (struct page *, unsigned long, 600 unsigned long); 601 void (*is_dirty_writeback) (struct page *, bool *, bool *); 602 int (*error_remove_page) (struct mapping *mapping, struct page *page); 603 int (*swap_activate)(struct file *); 604 int (*swap_deactivate)(struct file *); 605}; 606 607 writepage: called by the VM to write a dirty page to backing store. 608 This may happen for data integrity reasons (i.e. 'sync'), or 609 to free up memory (flush). The difference can be seen in 610 wbc->sync_mode. 611 The PG_Dirty flag has been cleared and PageLocked is true. 612 writepage should start writeout, should set PG_Writeback, 613 and should make sure the page is unlocked, either synchronously 614 or asynchronously when the write operation completes. 615 616 If wbc->sync_mode is WB_SYNC_NONE, ->writepage doesn't have to 617 try too hard if there are problems, and may choose to write out 618 other pages from the mapping if that is easier (e.g. due to 619 internal dependencies). If it chooses not to start writeout, it 620 should return AOP_WRITEPAGE_ACTIVATE so that the VM will not keep 621 calling ->writepage on that page. 622 623 See the file "Locking" for more details. 624 625 readpage: called by the VM to read a page from backing store. 626 The page will be Locked when readpage is called, and should be 627 unlocked and marked uptodate once the read completes. 628 If ->readpage discovers that it needs to unlock the page for 629 some reason, it can do so, and then return AOP_TRUNCATED_PAGE. 630 In this case, the page will be relocated, relocked and if 631 that all succeeds, ->readpage will be called again. 632 633 writepages: called by the VM to write out pages associated with the 634 address_space object. If wbc->sync_mode is WBC_SYNC_ALL, then 635 the writeback_control will specify a range of pages that must be 636 written out. If it is WBC_SYNC_NONE, then a nr_to_write is given 637 and that many pages should be written if possible. 638 If no ->writepages is given, then mpage_writepages is used 639 instead. This will choose pages from the address space that are 640 tagged as DIRTY and will pass them to ->writepage. 641 642 set_page_dirty: called by the VM to set a page dirty. 643 This is particularly needed if an address space attaches 644 private data to a page, and that data needs to be updated when 645 a page is dirtied. This is called, for example, when a memory 646 mapped page gets modified. 647 If defined, it should set the PageDirty flag, and the 648 PAGECACHE_TAG_DIRTY tag in the radix tree. 649 650 readpages: called by the VM to read pages associated with the address_space 651 object. This is essentially just a vector version of 652 readpage. Instead of just one page, several pages are 653 requested. 654 readpages is only used for read-ahead, so read errors are 655 ignored. If anything goes wrong, feel free to give up. 656 657 write_begin: 658 Called by the generic buffered write code to ask the filesystem to 659 prepare to write len bytes at the given offset in the file. The 660 address_space should check that the write will be able to complete, 661 by allocating space if necessary and doing any other internal 662 housekeeping. If the write will update parts of any basic-blocks on 663 storage, then those blocks should be pre-read (if they haven't been 664 read already) so that the updated blocks can be written out properly. 665 666 The filesystem must return the locked pagecache page for the specified 667 offset, in *pagep, for the caller to write into. 668 669 It must be able to cope with short writes (where the length passed to 670 write_begin is greater than the number of bytes copied into the page). 671 672 flags is a field for AOP_FLAG_xxx flags, described in 673 include/linux/fs.h. 674 675 A void * may be returned in fsdata, which then gets passed into 676 write_end. 677 678 Returns 0 on success; < 0 on failure (which is the error code), in 679 which case write_end is not called. 680 681 write_end: After a successful write_begin, and data copy, write_end must 682 be called. len is the original len passed to write_begin, and copied 683 is the amount that was able to be copied (copied == len is always true 684 if write_begin was called with the AOP_FLAG_UNINTERRUPTIBLE flag). 685 686 The filesystem must take care of unlocking the page and releasing it 687 refcount, and updating i_size. 688 689 Returns < 0 on failure, otherwise the number of bytes (<= 'copied') 690 that were able to be copied into pagecache. 691 692 bmap: called by the VFS to map a logical block offset within object to 693 physical block number. This method is used by the FIBMAP 694 ioctl and for working with swap-files. To be able to swap to 695 a file, the file must have a stable mapping to a block 696 device. The swap system does not go through the filesystem 697 but instead uses bmap to find out where the blocks in the file 698 are and uses those addresses directly. 699 700 dentry_open: *WARNING: probably going away soon, do not use!* This is an 701 alternative to f_op->open(), the difference is that this method may open 702 a file not necessarily originating from the same filesystem as the one 703 i_op->open() was called on. It may be useful for stacking filesystems 704 which want to allow native I/O directly on underlying files. 705 706 707 invalidatepage: If a page has PagePrivate set, then invalidatepage 708 will be called when part or all of the page is to be removed 709 from the address space. This generally corresponds to either a 710 truncation, punch hole or a complete invalidation of the address 711 space (in the latter case 'offset' will always be 0 and 'length' 712 will be PAGE_CACHE_SIZE). Any private data associated with the page 713 should be updated to reflect this truncation. If offset is 0 and 714 length is PAGE_CACHE_SIZE, then the private data should be released, 715 because the page must be able to be completely discarded. This may 716 be done by calling the ->releasepage function, but in this case the 717 release MUST succeed. 718 719 releasepage: releasepage is called on PagePrivate pages to indicate 720 that the page should be freed if possible. ->releasepage 721 should remove any private data from the page and clear the 722 PagePrivate flag. If releasepage() fails for some reason, it must 723 indicate failure with a 0 return value. 724 releasepage() is used in two distinct though related cases. The 725 first is when the VM finds a clean page with no active users and 726 wants to make it a free page. If ->releasepage succeeds, the 727 page will be removed from the address_space and become free. 728 729 The second case is when a request has been made to invalidate 730 some or all pages in an address_space. This can happen 731 through the fadvice(POSIX_FADV_DONTNEED) system call or by the 732 filesystem explicitly requesting it as nfs and 9fs do (when 733 they believe the cache may be out of date with storage) by 734 calling invalidate_inode_pages2(). 735 If the filesystem makes such a call, and needs to be certain 736 that all pages are invalidated, then its releasepage will 737 need to ensure this. Possibly it can clear the PageUptodate 738 bit if it cannot free private data yet. 739 740 freepage: freepage is called once the page is no longer visible in 741 the page cache in order to allow the cleanup of any private 742 data. Since it may be called by the memory reclaimer, it 743 should not assume that the original address_space mapping still 744 exists, and it should not block. 745 746 direct_IO: called by the generic read/write routines to perform 747 direct_IO - that is IO requests which bypass the page cache 748 and transfer data directly between the storage and the 749 application's address space. 750 751 get_xip_page: called by the VM to translate a block number to a page. 752 The page is valid until the corresponding filesystem is unmounted. 753 Filesystems that want to use execute-in-place (XIP) need to implement 754 it. An example implementation can be found in fs/ext2/xip.c. 755 756 migrate_page: This is used to compact the physical memory usage. 757 If the VM wants to relocate a page (maybe off a memory card 758 that is signalling imminent failure) it will pass a new page 759 and an old page to this function. migrate_page should 760 transfer any private data across and update any references 761 that it has to the page. 762 763 launder_page: Called before freeing a page - it writes back the dirty page. To 764 prevent redirtying the page, it is kept locked during the whole 765 operation. 766 767 is_partially_uptodate: Called by the VM when reading a file through the 768 pagecache when the underlying blocksize != pagesize. If the required 769 block is up to date then the read can complete without needing the IO 770 to bring the whole page up to date. 771 772 is_dirty_writeback: Called by the VM when attempting to reclaim a page. 773 The VM uses dirty and writeback information to determine if it needs 774 to stall to allow flushers a chance to complete some IO. Ordinarily 775 it can use PageDirty and PageWriteback but some filesystems have 776 more complex state (unstable pages in NFS prevent reclaim) or 777 do not set those flags due to locking problems (jbd). This callback 778 allows a filesystem to indicate to the VM if a page should be 779 treated as dirty or writeback for the purposes of stalling. 780 781 error_remove_page: normally set to generic_error_remove_page if truncation 782 is ok for this address space. Used for memory failure handling. 783 Setting this implies you deal with pages going away under you, 784 unless you have them locked or reference counts increased. 785 786 swap_activate: Called when swapon is used on a file to allocate 787 space if necessary and pin the block lookup information in 788 memory. A return value of zero indicates success, 789 in which case this file can be used to back swapspace. The 790 swapspace operations will be proxied to this address space's 791 ->swap_{out,in} methods. 792 793 swap_deactivate: Called during swapoff on files where swap_activate 794 was successful. 795 796 797The File Object 798=============== 799 800A file object represents a file opened by a process. 801 802 803struct file_operations 804---------------------- 805 806This describes how the VFS can manipulate an open file. As of kernel 8073.12, the following members are defined: 808 809struct file_operations { 810 struct module *owner; 811 loff_t (*llseek) (struct file *, loff_t, int); 812 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *); 813 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *); 814 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t); 815 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t); 816 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *); 817 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *); 818 int (*iterate) (struct file *, struct dir_context *); 819 unsigned int (*poll) (struct file *, struct poll_table_struct *); 820 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long); 821 long (*compat_ioctl) (struct file *, unsigned int, unsigned long); 822 int (*mmap) (struct file *, struct vm_area_struct *); 823 int (*open) (struct inode *, struct file *); 824 int (*flush) (struct file *); 825 int (*release) (struct inode *, struct file *); 826 int (*fsync) (struct file *, loff_t, loff_t, int datasync); 827 int (*aio_fsync) (struct kiocb *, int datasync); 828 int (*fasync) (int, struct file *, int); 829 int (*lock) (struct file *, int, struct file_lock *); 830 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int); 831 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); 832 int (*check_flags)(int); 833 int (*flock) (struct file *, int, struct file_lock *); 834 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, size_t, unsigned int); 835 ssize_t (*splice_read)(struct file *, struct pipe_inode_info *, size_t, unsigned int); 836 int (*setlease)(struct file *, long arg, struct file_lock **, void **); 837 long (*fallocate)(struct file *, int mode, loff_t offset, loff_t len); 838 int (*show_fdinfo)(struct seq_file *m, struct file *f); 839}; 840 841Again, all methods are called without any locks being held, unless 842otherwise noted. 843 844 llseek: called when the VFS needs to move the file position index 845 846 read: called by read(2) and related system calls 847 848 aio_read: vectored, possibly asynchronous read 849 850 read_iter: possibly asynchronous read with iov_iter as destination 851 852 write: called by write(2) and related system calls 853 854 aio_write: vectored, possibly asynchronous write 855 856 write_iter: possibly asynchronous write with iov_iter as source 857 858 iterate: called when the VFS needs to read the directory contents 859 860 poll: called by the VFS when a process wants to check if there is 861 activity on this file and (optionally) go to sleep until there 862 is activity. Called by the select(2) and poll(2) system calls 863 864 unlocked_ioctl: called by the ioctl(2) system call. 865 866 compat_ioctl: called by the ioctl(2) system call when 32 bit system calls 867 are used on 64 bit kernels. 868 869 mmap: called by the mmap(2) system call 870 871 open: called by the VFS when an inode should be opened. When the VFS 872 opens a file, it creates a new "struct file". It then calls the 873 open method for the newly allocated file structure. You might 874 think that the open method really belongs in 875 "struct inode_operations", and you may be right. I think it's 876 done the way it is because it makes filesystems simpler to 877 implement. The open() method is a good place to initialize the 878 "private_data" member in the file structure if you want to point 879 to a device structure 880 881 flush: called by the close(2) system call to flush a file 882 883 release: called when the last reference to an open file is closed 884 885 fsync: called by the fsync(2) system call 886 887 fasync: called by the fcntl(2) system call when asynchronous 888 (non-blocking) mode is enabled for a file 889 890 lock: called by the fcntl(2) system call for F_GETLK, F_SETLK, and F_SETLKW 891 commands 892 893 get_unmapped_area: called by the mmap(2) system call 894 895 check_flags: called by the fcntl(2) system call for F_SETFL command 896 897 flock: called by the flock(2) system call 898 899 splice_write: called by the VFS to splice data from a pipe to a file. This 900 method is used by the splice(2) system call 901 902 splice_read: called by the VFS to splice data from file to a pipe. This 903 method is used by the splice(2) system call 904 905 setlease: called by the VFS to set or release a file lock lease. setlease 906 implementations should call generic_setlease to record or remove 907 the lease in the inode after setting it. 908 909 fallocate: called by the VFS to preallocate blocks or punch a hole. 910 911Note that the file operations are implemented by the specific 912filesystem in which the inode resides. When opening a device node 913(character or block special) most filesystems will call special 914support routines in the VFS which will locate the required device 915driver information. These support routines replace the filesystem file 916operations with those for the device driver, and then proceed to call 917the new open() method for the file. This is how opening a device file 918in the filesystem eventually ends up calling the device driver open() 919method. 920 921 922Directory Entry Cache (dcache) 923============================== 924 925 926struct dentry_operations 927------------------------ 928 929This describes how a filesystem can overload the standard dentry 930operations. Dentries and the dcache are the domain of the VFS and the 931individual filesystem implementations. Device drivers have no business 932here. These methods may be set to NULL, as they are either optional or 933the VFS uses a default. As of kernel 2.6.22, the following members are 934defined: 935 936struct dentry_operations { 937 int (*d_revalidate)(struct dentry *, unsigned int); 938 int (*d_weak_revalidate)(struct dentry *, unsigned int); 939 int (*d_hash)(const struct dentry *, struct qstr *); 940 int (*d_compare)(const struct dentry *, const struct dentry *, 941 unsigned int, const char *, const struct qstr *); 942 int (*d_delete)(const struct dentry *); 943 void (*d_release)(struct dentry *); 944 void (*d_iput)(struct dentry *, struct inode *); 945 char *(*d_dname)(struct dentry *, char *, int); 946 struct vfsmount *(*d_automount)(struct path *); 947 int (*d_manage)(struct dentry *, bool); 948}; 949 950 d_revalidate: called when the VFS needs to revalidate a dentry. This 951 is called whenever a name look-up finds a dentry in the 952 dcache. Most local filesystems leave this as NULL, because all their 953 dentries in the dcache are valid. Network filesystems are different 954 since things can change on the server without the client necessarily 955 being aware of it. 956 957 This function should return a positive value if the dentry is still 958 valid, and zero or a negative error code if it isn't. 959 960 d_revalidate may be called in rcu-walk mode (flags & LOOKUP_RCU). 961 If in rcu-walk mode, the filesystem must revalidate the dentry without 962 blocking or storing to the dentry, d_parent and d_inode should not be 963 used without care (because they can change and, in d_inode case, even 964 become NULL under us). 965 966 If a situation is encountered that rcu-walk cannot handle, return 967 -ECHILD and it will be called again in ref-walk mode. 968 969 d_weak_revalidate: called when the VFS needs to revalidate a "jumped" dentry. 970 This is called when a path-walk ends at dentry that was not acquired by 971 doing a lookup in the parent directory. This includes "/", "." and "..", 972 as well as procfs-style symlinks and mountpoint traversal. 973 974 In this case, we are less concerned with whether the dentry is still 975 fully correct, but rather that the inode is still valid. As with 976 d_revalidate, most local filesystems will set this to NULL since their 977 dcache entries are always valid. 978 979 This function has the same return code semantics as d_revalidate. 980 981 d_weak_revalidate is only called after leaving rcu-walk mode. 982 983 d_hash: called when the VFS adds a dentry to the hash table. The first 984 dentry passed to d_hash is the parent directory that the name is 985 to be hashed into. 986 987 Same locking and synchronisation rules as d_compare regarding 988 what is safe to dereference etc. 989 990 d_compare: called to compare a dentry name with a given name. The first 991 dentry is the parent of the dentry to be compared, the second is 992 the child dentry. len and name string are properties of the dentry 993 to be compared. qstr is the name to compare it with. 994 995 Must be constant and idempotent, and should not take locks if 996 possible, and should not or store into the dentry. 997 Should not dereference pointers outside the dentry without 998 lots of care (eg. d_parent, d_inode, d_name should not be used). 999 1000 However, our vfsmount is pinned, and RCU held, so the dentries and 1001 inodes won't disappear, neither will our sb or filesystem module. 1002 ->d_sb may be used. 1003 1004 It is a tricky calling convention because it needs to be called under 1005 "rcu-walk", ie. without any locks or references on things. 1006 1007 d_delete: called when the last reference to a dentry is dropped and the 1008 dcache is deciding whether or not to cache it. Return 1 to delete 1009 immediately, or 0 to cache the dentry. Default is NULL which means to 1010 always cache a reachable dentry. d_delete must be constant and 1011 idempotent. 1012 1013 d_release: called when a dentry is really deallocated 1014 1015 d_iput: called when a dentry loses its inode (just prior to its 1016 being deallocated). The default when this is NULL is that the 1017 VFS calls iput(). If you define this method, you must call 1018 iput() yourself 1019 1020 d_dname: called when the pathname of a dentry should be generated. 1021 Useful for some pseudo filesystems (sockfs, pipefs, ...) to delay 1022 pathname generation. (Instead of doing it when dentry is created, 1023 it's done only when the path is needed.). Real filesystems probably 1024 dont want to use it, because their dentries are present in global 1025 dcache hash, so their hash should be an invariant. As no lock is 1026 held, d_dname() should not try to modify the dentry itself, unless 1027 appropriate SMP safety is used. CAUTION : d_path() logic is quite 1028 tricky. The correct way to return for example "Hello" is to put it 1029 at the end of the buffer, and returns a pointer to the first char. 1030 dynamic_dname() helper function is provided to take care of this. 1031 1032 d_automount: called when an automount dentry is to be traversed (optional). 1033 This should create a new VFS mount record and return the record to the 1034 caller. The caller is supplied with a path parameter giving the 1035 automount directory to describe the automount target and the parent 1036 VFS mount record to provide inheritable mount parameters. NULL should 1037 be returned if someone else managed to make the automount first. If 1038 the vfsmount creation failed, then an error code should be returned. 1039 If -EISDIR is returned, then the directory will be treated as an 1040 ordinary directory and returned to pathwalk to continue walking. 1041 1042 If a vfsmount is returned, the caller will attempt to mount it on the 1043 mountpoint and will remove the vfsmount from its expiration list in 1044 the case of failure. The vfsmount should be returned with 2 refs on 1045 it to prevent automatic expiration - the caller will clean up the 1046 additional ref. 1047 1048 This function is only used if DCACHE_NEED_AUTOMOUNT is set on the 1049 dentry. This is set by __d_instantiate() if S_AUTOMOUNT is set on the 1050 inode being added. 1051 1052 d_manage: called to allow the filesystem to manage the transition from a 1053 dentry (optional). This allows autofs, for example, to hold up clients 1054 waiting to explore behind a 'mountpoint' whilst letting the daemon go 1055 past and construct the subtree there. 0 should be returned to let the 1056 calling process continue. -EISDIR can be returned to tell pathwalk to 1057 use this directory as an ordinary directory and to ignore anything 1058 mounted on it and not to check the automount flag. Any other error 1059 code will abort pathwalk completely. 1060 1061 If the 'rcu_walk' parameter is true, then the caller is doing a 1062 pathwalk in RCU-walk mode. Sleeping is not permitted in this mode, 1063 and the caller can be asked to leave it and call again by returning 1064 -ECHILD. -EISDIR may also be returned to tell pathwalk to 1065 ignore d_automount or any mounts. 1066 1067 This function is only used if DCACHE_MANAGE_TRANSIT is set on the 1068 dentry being transited from. 1069 1070Example : 1071 1072static char *pipefs_dname(struct dentry *dent, char *buffer, int buflen) 1073{ 1074 return dynamic_dname(dentry, buffer, buflen, "pipe:[%lu]", 1075 dentry->d_inode->i_ino); 1076} 1077 1078Each dentry has a pointer to its parent dentry, as well as a hash list 1079of child dentries. Child dentries are basically like files in a 1080directory. 1081 1082 1083Directory Entry Cache API 1084-------------------------- 1085 1086There are a number of functions defined which permit a filesystem to 1087manipulate dentries: 1088 1089 dget: open a new handle for an existing dentry (this just increments 1090 the usage count) 1091 1092 dput: close a handle for a dentry (decrements the usage count). If 1093 the usage count drops to 0, and the dentry is still in its 1094 parent's hash, the "d_delete" method is called to check whether 1095 it should be cached. If it should not be cached, or if the dentry 1096 is not hashed, it is deleted. Otherwise cached dentries are put 1097 into an LRU list to be reclaimed on memory shortage. 1098 1099 d_drop: this unhashes a dentry from its parents hash list. A 1100 subsequent call to dput() will deallocate the dentry if its 1101 usage count drops to 0 1102 1103 d_delete: delete a dentry. If there are no other open references to 1104 the dentry then the dentry is turned into a negative dentry 1105 (the d_iput() method is called). If there are other 1106 references, then d_drop() is called instead 1107 1108 d_add: add a dentry to its parents hash list and then calls 1109 d_instantiate() 1110 1111 d_instantiate: add a dentry to the alias hash list for the inode and 1112 updates the "d_inode" member. The "i_count" member in the 1113 inode structure should be set/incremented. If the inode 1114 pointer is NULL, the dentry is called a "negative 1115 dentry". This function is commonly called when an inode is 1116 created for an existing negative dentry 1117 1118 d_lookup: look up a dentry given its parent and path name component 1119 It looks up the child of that given name from the dcache 1120 hash table. If it is found, the reference count is incremented 1121 and the dentry is returned. The caller must use dput() 1122 to free the dentry when it finishes using it. 1123 1124Mount Options 1125============= 1126 1127Parsing options 1128--------------- 1129 1130On mount and remount the filesystem is passed a string containing a 1131comma separated list of mount options. The options can have either of 1132these forms: 1133 1134 option 1135 option=value 1136 1137The <linux/parser.h> header defines an API that helps parse these 1138options. There are plenty of examples on how to use it in existing 1139filesystems. 1140 1141Showing options 1142--------------- 1143 1144If a filesystem accepts mount options, it must define show_options() 1145to show all the currently active options. The rules are: 1146 1147 - options MUST be shown which are not default or their values differ 1148 from the default 1149 1150 - options MAY be shown which are enabled by default or have their 1151 default value 1152 1153Options used only internally between a mount helper and the kernel 1154(such as file descriptors), or which only have an effect during the 1155mounting (such as ones controlling the creation of a journal) are exempt 1156from the above rules. 1157 1158The underlying reason for the above rules is to make sure, that a 1159mount can be accurately replicated (e.g. umounting and mounting again) 1160based on the information found in /proc/mounts. 1161 1162A simple method of saving options at mount/remount time and showing 1163them is provided with the save_mount_options() and 1164generic_show_options() helper functions. Please note, that using 1165these may have drawbacks. For more info see header comments for these 1166functions in fs/namespace.c. 1167 1168Resources 1169========= 1170 1171(Note some of these resources are not up-to-date with the latest kernel 1172 version.) 1173 1174Creating Linux virtual filesystems. 2002 1175 <http://lwn.net/Articles/13325/> 1176 1177The Linux Virtual File-system Layer by Neil Brown. 1999 1178 <http://www.cse.unsw.edu.au/~neilb/oss/linux-commentary/vfs.html> 1179 1180A tour of the Linux VFS by Michael K. Johnson. 1996 1181 <http://www.tldp.org/LDP/khg/HyperNews/get/fs/vfstour.html> 1182 1183A small trail through the Linux kernel by Andries Brouwer. 2001 1184 <http://www.win.tue.nl/~aeb/linux/vfs/trail.html> 1185