1af96c1e3STobin C. Harding.. SPDX-License-Identifier: GPL-2.0 2af96c1e3STobin C. Harding 3af96c1e3STobin C. Harding========================================= 4af96c1e3STobin C. HardingOverview of the Linux Virtual File System 5af96c1e3STobin C. Harding========================================= 6af96c1e3STobin C. Harding 7af96c1e3STobin C. HardingOriginal author: Richard Gooch <rgooch@atnf.csiro.au> 8af96c1e3STobin C. Harding 9af96c1e3STobin C. Harding- Copyright (C) 1999 Richard Gooch 10af96c1e3STobin C. Harding- Copyright (C) 2005 Pekka Enberg 11af96c1e3STobin C. Harding 12af96c1e3STobin C. Harding 13af96c1e3STobin C. HardingIntroduction 14af96c1e3STobin C. Harding============ 15af96c1e3STobin C. Harding 16af96c1e3STobin C. HardingThe Virtual File System (also known as the Virtual Filesystem Switch) is 17af96c1e3STobin C. Hardingthe software layer in the kernel that provides the filesystem interface 18af96c1e3STobin C. Hardingto userspace programs. It also provides an abstraction within the 19af96c1e3STobin C. Hardingkernel which allows different filesystem implementations to coexist. 20af96c1e3STobin C. Harding 21af96c1e3STobin C. HardingVFS system calls open(2), stat(2), read(2), write(2), chmod(2) and so on 22af96c1e3STobin C. Hardingare called from a process context. Filesystem locking is described in 23ec23eb54SMauro Carvalho Chehabthe document Documentation/filesystems/locking.rst. 24af96c1e3STobin C. Harding 25af96c1e3STobin C. Harding 26af96c1e3STobin C. HardingDirectory Entry Cache (dcache) 27af96c1e3STobin C. Harding------------------------------ 28af96c1e3STobin C. Harding 29af96c1e3STobin C. HardingThe VFS implements the open(2), stat(2), chmod(2), and similar system 30af96c1e3STobin C. Hardingcalls. The pathname argument that is passed to them is used by the VFS 31af96c1e3STobin C. Hardingto search through the directory entry cache (also known as the dentry 32af96c1e3STobin C. Hardingcache or dcache). This provides a very fast look-up mechanism to 33af96c1e3STobin C. Hardingtranslate a pathname (filename) into a specific dentry. Dentries live 34af96c1e3STobin C. Hardingin RAM and are never saved to disc: they exist only for performance. 35af96c1e3STobin C. Harding 36af96c1e3STobin C. HardingThe dentry cache is meant to be a view into your entire filespace. As 37af96c1e3STobin C. Hardingmost computers cannot fit all dentries in the RAM at the same time, some 38af96c1e3STobin C. Hardingbits of the cache are missing. In order to resolve your pathname into a 39af96c1e3STobin C. Hardingdentry, the VFS may have to resort to creating dentries along the way, 40af96c1e3STobin C. Hardingand then loading the inode. This is done by looking up the inode. 41af96c1e3STobin C. Harding 42af96c1e3STobin C. Harding 43af96c1e3STobin C. HardingThe Inode Object 44af96c1e3STobin C. Harding---------------- 45af96c1e3STobin C. Harding 46af96c1e3STobin C. HardingAn individual dentry usually has a pointer to an inode. Inodes are 47af96c1e3STobin C. Hardingfilesystem objects such as regular files, directories, FIFOs and other 48af96c1e3STobin C. Hardingbeasts. They live either on the disc (for block device filesystems) or 49af96c1e3STobin C. Hardingin the memory (for pseudo filesystems). Inodes that live on the disc 50af96c1e3STobin C. Hardingare copied into the memory when required and changes to the inode are 51af96c1e3STobin C. Hardingwritten back to disc. A single inode can be pointed to by multiple 52af96c1e3STobin C. Hardingdentries (hard links, for example, do this). 53af96c1e3STobin C. Harding 54af96c1e3STobin C. HardingTo look up an inode requires that the VFS calls the lookup() method of 55af96c1e3STobin C. Hardingthe parent directory inode. This method is installed by the specific 56af96c1e3STobin C. Hardingfilesystem implementation that the inode lives in. Once the VFS has the 57af96c1e3STobin C. Hardingrequired dentry (and hence the inode), we can do all those boring things 58af96c1e3STobin C. Hardinglike open(2) the file, or stat(2) it to peek at the inode data. The 59af96c1e3STobin C. Hardingstat(2) operation is fairly simple: once the VFS has the dentry, it 60af96c1e3STobin C. Hardingpeeks at the inode data and passes some of it back to userspace. 61af96c1e3STobin C. Harding 62af96c1e3STobin C. Harding 63af96c1e3STobin C. HardingThe File Object 64af96c1e3STobin C. Harding--------------- 65af96c1e3STobin C. Harding 66af96c1e3STobin C. HardingOpening a file requires another operation: allocation of a file 67af96c1e3STobin C. Hardingstructure (this is the kernel-side implementation of file descriptors). 68af96c1e3STobin C. HardingThe freshly allocated file structure is initialized with a pointer to 69af96c1e3STobin C. Hardingthe dentry and a set of file operation member functions. These are 70af96c1e3STobin C. Hardingtaken from the inode data. The open() file method is then called so the 71af96c1e3STobin C. Hardingspecific filesystem implementation can do its work. You can see that 72af96c1e3STobin C. Hardingthis is another switch performed by the VFS. The file structure is 73af96c1e3STobin C. Hardingplaced into the file descriptor table for the process. 74af96c1e3STobin C. Harding 75af96c1e3STobin C. HardingReading, writing and closing files (and other assorted VFS operations) 76af96c1e3STobin C. Hardingis done by using the userspace file descriptor to grab the appropriate 77af96c1e3STobin C. Hardingfile structure, and then calling the required file structure method to 78af96c1e3STobin C. Hardingdo whatever is required. For as long as the file is open, it keeps the 79af96c1e3STobin C. Hardingdentry in use, which in turn means that the VFS inode is still in use. 80af96c1e3STobin C. Harding 81af96c1e3STobin C. Harding 82af96c1e3STobin C. HardingRegistering and Mounting a Filesystem 83af96c1e3STobin C. Harding===================================== 84af96c1e3STobin C. Harding 85af96c1e3STobin C. HardingTo register and unregister a filesystem, use the following API 86af96c1e3STobin C. Hardingfunctions: 87af96c1e3STobin C. Harding 88af96c1e3STobin C. Harding.. code-block:: c 89af96c1e3STobin C. Harding 90af96c1e3STobin C. Harding #include <linux/fs.h> 91af96c1e3STobin C. Harding 92af96c1e3STobin C. Harding extern int register_filesystem(struct file_system_type *); 93af96c1e3STobin C. Harding extern int unregister_filesystem(struct file_system_type *); 94af96c1e3STobin C. Harding 95af96c1e3STobin C. HardingThe passed struct file_system_type describes your filesystem. When a 96af96c1e3STobin C. Hardingrequest is made to mount a filesystem onto a directory in your 97af96c1e3STobin C. Hardingnamespace, the VFS will call the appropriate mount() method for the 98af96c1e3STobin C. Hardingspecific filesystem. New vfsmount referring to the tree returned by 99af96c1e3STobin C. Harding->mount() will be attached to the mountpoint, so that when pathname 100af96c1e3STobin C. Hardingresolution reaches the mountpoint it will jump into the root of that 101af96c1e3STobin C. Hardingvfsmount. 102af96c1e3STobin C. Harding 103af96c1e3STobin C. HardingYou can see all filesystems that are registered to the kernel in the 104af96c1e3STobin C. Hardingfile /proc/filesystems. 105af96c1e3STobin C. Harding 106af96c1e3STobin C. Harding 107af96c1e3STobin C. Hardingstruct file_system_type 108af96c1e3STobin C. Harding----------------------- 109af96c1e3STobin C. Harding 110af96c1e3STobin C. HardingThis describes the filesystem. As of kernel 2.6.39, the following 111af96c1e3STobin C. Hardingmembers are defined: 112af96c1e3STobin C. Harding 113af96c1e3STobin C. Harding.. code-block:: c 114af96c1e3STobin C. Harding 1156a2195a1SLiao Pingfang struct file_system_type { 116af96c1e3STobin C. Harding const char *name; 117af96c1e3STobin C. Harding int fs_flags; 118af96c1e3STobin C. Harding struct dentry *(*mount) (struct file_system_type *, int, 119af96c1e3STobin C. Harding const char *, void *); 120af96c1e3STobin C. Harding void (*kill_sb) (struct super_block *); 121af96c1e3STobin C. Harding struct module *owner; 122af96c1e3STobin C. Harding struct file_system_type * next; 123af96c1e3STobin C. Harding struct list_head fs_supers; 124af96c1e3STobin C. Harding struct lock_class_key s_lock_key; 125af96c1e3STobin C. Harding struct lock_class_key s_umount_key; 126af96c1e3STobin C. Harding }; 127af96c1e3STobin C. Harding 128ee5dc049STobin C. Harding``name`` 129ee5dc049STobin C. Harding the name of the filesystem type, such as "ext2", "iso9660", 130af96c1e3STobin C. Harding "msdos" and so on 131af96c1e3STobin C. Harding 132ee5dc049STobin C. Harding``fs_flags`` 133ee5dc049STobin C. Harding various flags (i.e. FS_REQUIRES_DEV, FS_NO_DCACHE, etc.) 134af96c1e3STobin C. Harding 135ee5dc049STobin C. Harding``mount`` 136ee5dc049STobin C. Harding the method to call when a new instance of this filesystem should 137af96c1e3STobin C. Harding be mounted 138af96c1e3STobin C. Harding 139ee5dc049STobin C. Harding``kill_sb`` 140ee5dc049STobin C. Harding the method to call when an instance of this filesystem should be 141ee5dc049STobin C. Harding shut down 142af96c1e3STobin C. Harding 143af96c1e3STobin C. Harding 144ee5dc049STobin C. Harding``owner`` 145ee5dc049STobin C. Harding for internal VFS use: you should initialize this to THIS_MODULE 146ee5dc049STobin C. Harding in most cases. 147ee5dc049STobin C. Harding 148ee5dc049STobin C. Harding``next`` 149ee5dc049STobin C. Harding for internal VFS use: you should initialize this to NULL 150af96c1e3STobin C. Harding 151af96c1e3STobin C. Harding s_lock_key, s_umount_key: lockdep-specific 152af96c1e3STobin C. Harding 153af96c1e3STobin C. HardingThe mount() method has the following arguments: 154af96c1e3STobin C. Harding 155ee5dc049STobin C. Harding``struct file_system_type *fs_type`` 156ee5dc049STobin C. Harding describes the filesystem, partly initialized by the specific 157ee5dc049STobin C. Harding filesystem code 158af96c1e3STobin C. Harding 159ee5dc049STobin C. Harding``int flags`` 160ee5dc049STobin C. Harding mount flags 161af96c1e3STobin C. Harding 162ee5dc049STobin C. Harding``const char *dev_name`` 163ee5dc049STobin C. Harding the device name we are mounting. 164af96c1e3STobin C. Harding 165ee5dc049STobin C. Harding``void *data`` 166ee5dc049STobin C. Harding arbitrary mount options, usually comes as an ASCII string (see 167ee5dc049STobin C. Harding "Mount Options" section) 168af96c1e3STobin C. Harding 169af96c1e3STobin C. HardingThe mount() method must return the root dentry of the tree requested by 170af96c1e3STobin C. Hardingcaller. An active reference to its superblock must be grabbed and the 171af96c1e3STobin C. Hardingsuperblock must be locked. On failure it should return ERR_PTR(error). 172af96c1e3STobin C. Harding 173af96c1e3STobin C. HardingThe arguments match those of mount(2) and their interpretation depends 174af96c1e3STobin C. Hardingon filesystem type. E.g. for block filesystems, dev_name is interpreted 175af96c1e3STobin C. Hardingas block device name, that device is opened and if it contains a 176af96c1e3STobin C. Hardingsuitable filesystem image the method creates and initializes struct 177af96c1e3STobin C. Hardingsuper_block accordingly, returning its root dentry to caller. 178af96c1e3STobin C. Harding 179af96c1e3STobin C. Harding->mount() may choose to return a subtree of existing filesystem - it 180af96c1e3STobin C. Hardingdoesn't have to create a new one. The main result from the caller's 181af96c1e3STobin C. Hardingpoint of view is a reference to dentry at the root of (sub)tree to be 182af96c1e3STobin C. Hardingattached; creation of new superblock is a common side effect. 183af96c1e3STobin C. Harding 184af96c1e3STobin C. HardingThe most interesting member of the superblock structure that the mount() 185af96c1e3STobin C. Hardingmethod fills in is the "s_op" field. This is a pointer to a "struct 186af96c1e3STobin C. Hardingsuper_operations" which describes the next level of the filesystem 187af96c1e3STobin C. Hardingimplementation. 188af96c1e3STobin C. Harding 189af96c1e3STobin C. HardingUsually, a filesystem uses one of the generic mount() implementations 190af96c1e3STobin C. Hardingand provides a fill_super() callback instead. The generic variants are: 191af96c1e3STobin C. Harding 192ee5dc049STobin C. Harding``mount_bdev`` 193ee5dc049STobin C. Harding mount a filesystem residing on a block device 194af96c1e3STobin C. Harding 195ee5dc049STobin C. Harding``mount_nodev`` 196ee5dc049STobin C. Harding mount a filesystem that is not backed by a device 197af96c1e3STobin C. Harding 198ee5dc049STobin C. Harding``mount_single`` 199ee5dc049STobin C. Harding mount a filesystem which shares the instance between all mounts 200af96c1e3STobin C. Harding 201af96c1e3STobin C. HardingA fill_super() callback implementation has the following arguments: 202af96c1e3STobin C. Harding 203ee5dc049STobin C. Harding``struct super_block *sb`` 204ee5dc049STobin C. Harding the superblock structure. The callback must initialize this 205ee5dc049STobin C. Harding properly. 206af96c1e3STobin C. Harding 207ee5dc049STobin C. Harding``void *data`` 208ee5dc049STobin C. Harding arbitrary mount options, usually comes as an ASCII string (see 209ee5dc049STobin C. Harding "Mount Options" section) 210af96c1e3STobin C. Harding 211ee5dc049STobin C. Harding``int silent`` 212ee5dc049STobin C. Harding whether or not to be silent on error 213af96c1e3STobin C. Harding 214af96c1e3STobin C. Harding 215af96c1e3STobin C. HardingThe Superblock Object 216af96c1e3STobin C. Harding===================== 217af96c1e3STobin C. Harding 218af96c1e3STobin C. HardingA superblock object represents a mounted filesystem. 219af96c1e3STobin C. Harding 220af96c1e3STobin C. Harding 221af96c1e3STobin C. Hardingstruct super_operations 222af96c1e3STobin C. Harding----------------------- 223af96c1e3STobin C. Harding 224af96c1e3STobin C. HardingThis describes how the VFS can manipulate the superblock of your 225af96c1e3STobin C. Hardingfilesystem. As of kernel 2.6.22, the following members are defined: 226af96c1e3STobin C. Harding 227af96c1e3STobin C. Harding.. code-block:: c 228af96c1e3STobin C. Harding 229af96c1e3STobin C. Harding struct super_operations { 230af96c1e3STobin C. Harding struct inode *(*alloc_inode)(struct super_block *sb); 231af96c1e3STobin C. Harding void (*destroy_inode)(struct inode *); 232af96c1e3STobin C. Harding 233af96c1e3STobin C. Harding void (*dirty_inode) (struct inode *, int flags); 234af96c1e3STobin C. Harding int (*write_inode) (struct inode *, int); 235af96c1e3STobin C. Harding void (*drop_inode) (struct inode *); 236af96c1e3STobin C. Harding void (*delete_inode) (struct inode *); 237af96c1e3STobin C. Harding void (*put_super) (struct super_block *); 238af96c1e3STobin C. Harding int (*sync_fs)(struct super_block *sb, int wait); 239af96c1e3STobin C. Harding int (*freeze_fs) (struct super_block *); 240af96c1e3STobin C. Harding int (*unfreeze_fs) (struct super_block *); 241af96c1e3STobin C. Harding int (*statfs) (struct dentry *, struct kstatfs *); 242af96c1e3STobin C. Harding int (*remount_fs) (struct super_block *, int *, char *); 243af96c1e3STobin C. Harding void (*clear_inode) (struct inode *); 244af96c1e3STobin C. Harding void (*umount_begin) (struct super_block *); 245af96c1e3STobin C. Harding 246af96c1e3STobin C. Harding int (*show_options)(struct seq_file *, struct dentry *); 247af96c1e3STobin C. Harding 248af96c1e3STobin C. Harding ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t); 249af96c1e3STobin C. Harding ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t); 250af96c1e3STobin C. Harding int (*nr_cached_objects)(struct super_block *); 251af96c1e3STobin C. Harding void (*free_cached_objects)(struct super_block *, int); 252af96c1e3STobin C. Harding }; 253af96c1e3STobin C. Harding 254af96c1e3STobin C. HardingAll methods are called without any locks being held, unless otherwise 255af96c1e3STobin C. Hardingnoted. This means that most methods can block safely. All methods are 256af96c1e3STobin C. Hardingonly called from a process context (i.e. not from an interrupt handler 257af96c1e3STobin C. Hardingor bottom half). 258af96c1e3STobin C. Harding 259ee5dc049STobin C. Harding``alloc_inode`` 260ee5dc049STobin C. Harding this method is called by alloc_inode() to allocate memory for 261ee5dc049STobin C. Harding struct inode and initialize it. If this function is not 262af96c1e3STobin C. Harding defined, a simple 'struct inode' is allocated. Normally 263af96c1e3STobin C. Harding alloc_inode will be used to allocate a larger structure which 264af96c1e3STobin C. Harding contains a 'struct inode' embedded within it. 265af96c1e3STobin C. Harding 266ee5dc049STobin C. Harding``destroy_inode`` 267ee5dc049STobin C. Harding this method is called by destroy_inode() to release resources 268ee5dc049STobin C. Harding allocated for struct inode. It is only required if 269af96c1e3STobin C. Harding ->alloc_inode was defined and simply undoes anything done by 270af96c1e3STobin C. Harding ->alloc_inode. 271af96c1e3STobin C. Harding 272ee5dc049STobin C. Harding``dirty_inode`` 273a38ed483SEric Biggers this method is called by the VFS when an inode is marked dirty. 274a38ed483SEric Biggers This is specifically for the inode itself being marked dirty, 275a38ed483SEric Biggers not its data. If the update needs to be persisted by fdatasync(), 276a38ed483SEric Biggers then I_DIRTY_DATASYNC will be set in the flags argument. 277cbfecb92SLukas Czerner I_DIRTY_TIME will be set in the flags in case lazytime is enabled 278cbfecb92SLukas Czerner and struct inode has times updated since the last ->dirty_inode 279cbfecb92SLukas Czerner call. 280af96c1e3STobin C. Harding 281ee5dc049STobin C. Harding``write_inode`` 282ee5dc049STobin C. Harding this method is called when the VFS needs to write an inode to 283ee5dc049STobin C. Harding disc. The second parameter indicates whether the write should 284ee5dc049STobin C. Harding be synchronous or not, not all filesystems check this flag. 285af96c1e3STobin C. Harding 286ee5dc049STobin C. Harding``drop_inode`` 287ee5dc049STobin C. Harding called when the last access to the inode is dropped, with the 288ee5dc049STobin C. Harding inode->i_lock spinlock held. 289af96c1e3STobin C. Harding 290af96c1e3STobin C. Harding This method should be either NULL (normal UNIX filesystem 291ee5dc049STobin C. Harding semantics) or "generic_delete_inode" (for filesystems that do 292ee5dc049STobin C. Harding not want to cache inodes - causing "delete_inode" to always be 293af96c1e3STobin C. Harding called regardless of the value of i_nlink) 294af96c1e3STobin C. Harding 295ee5dc049STobin C. Harding The "generic_delete_inode()" behavior is equivalent to the old 296ee5dc049STobin C. Harding practice of using "force_delete" in the put_inode() case, but 297ee5dc049STobin C. Harding does not have the races that the "force_delete()" approach had. 298af96c1e3STobin C. Harding 299ee5dc049STobin C. Harding``delete_inode`` 300ee5dc049STobin C. Harding called when the VFS wants to delete an inode 301af96c1e3STobin C. Harding 302ee5dc049STobin C. Harding``put_super`` 303ee5dc049STobin C. Harding called when the VFS wishes to free the superblock 304af96c1e3STobin C. Harding (i.e. unmount). This is called with the superblock lock held 305af96c1e3STobin C. Harding 306ee5dc049STobin C. Harding``sync_fs`` 307ee5dc049STobin C. Harding called when VFS is writing out all dirty data associated with a 308ee5dc049STobin C. Harding superblock. The second parameter indicates whether the method 309af96c1e3STobin C. Harding should wait until the write out has been completed. Optional. 310af96c1e3STobin C. Harding 311ee5dc049STobin C. Harding``freeze_fs`` 312ee5dc049STobin C. Harding called when VFS is locking a filesystem and forcing it into a 313ee5dc049STobin C. Harding consistent state. This method is currently used by the Logical 314ee5dc049STobin C. Harding Volume Manager (LVM). 315af96c1e3STobin C. Harding 316ee5dc049STobin C. Harding``unfreeze_fs`` 317ee5dc049STobin C. Harding called when VFS is unlocking a filesystem and making it writable 318af96c1e3STobin C. Harding again. 319af96c1e3STobin C. Harding 320ee5dc049STobin C. Harding``statfs`` 321ee5dc049STobin C. Harding called when the VFS needs to get filesystem statistics. 322af96c1e3STobin C. Harding 323ee5dc049STobin C. Harding``remount_fs`` 324ee5dc049STobin C. Harding called when the filesystem is remounted. This is called with 325ee5dc049STobin C. Harding the kernel lock held 326af96c1e3STobin C. Harding 327ee5dc049STobin C. Harding``clear_inode`` 328ee5dc049STobin C. Harding called then the VFS clears the inode. Optional 329af96c1e3STobin C. Harding 330ee5dc049STobin C. Harding``umount_begin`` 331ee5dc049STobin C. Harding called when the VFS is unmounting a filesystem. 332af96c1e3STobin C. Harding 333ee5dc049STobin C. Harding``show_options`` 334ee5dc049STobin C. Harding called by the VFS to show mount options for /proc/<pid>/mounts. 335ee5dc049STobin C. Harding (see "Mount Options" section) 336af96c1e3STobin C. Harding 337ee5dc049STobin C. Harding``quota_read`` 338ee5dc049STobin C. Harding called by the VFS to read from filesystem quota file. 339af96c1e3STobin C. Harding 340ee5dc049STobin C. Harding``quota_write`` 341ee5dc049STobin C. Harding called by the VFS to write to filesystem quota file. 342af96c1e3STobin C. Harding 343ee5dc049STobin C. Harding``nr_cached_objects`` 344ee5dc049STobin C. Harding called by the sb cache shrinking function for the filesystem to 345ee5dc049STobin C. Harding return the number of freeable cached objects it contains. 346af96c1e3STobin C. Harding Optional. 347af96c1e3STobin C. Harding 348ee5dc049STobin C. Harding``free_cache_objects`` 349ee5dc049STobin C. Harding called by the sb cache shrinking function for the filesystem to 350ee5dc049STobin C. Harding scan the number of objects indicated to try to free them. 351ee5dc049STobin C. Harding Optional, but any filesystem implementing this method needs to 352ee5dc049STobin C. Harding also implement ->nr_cached_objects for it to be called 353ee5dc049STobin C. Harding correctly. 354af96c1e3STobin C. Harding 355af96c1e3STobin C. Harding We can't do anything with any errors that the filesystem might 356ee5dc049STobin C. Harding encountered, hence the void return type. This will never be 357ee5dc049STobin C. Harding called if the VM is trying to reclaim under GFP_NOFS conditions, 358ee5dc049STobin C. Harding hence this method does not need to handle that situation itself. 359af96c1e3STobin C. Harding 360ee5dc049STobin C. Harding Implementations must include conditional reschedule calls inside 361ee5dc049STobin C. Harding any scanning loop that is done. This allows the VFS to 362ee5dc049STobin C. Harding determine appropriate scan batch sizes without having to worry 363ee5dc049STobin C. Harding about whether implementations will cause holdoff problems due to 364ee5dc049STobin C. Harding large scan batch sizes. 365af96c1e3STobin C. Harding 366af96c1e3STobin C. HardingWhoever sets up the inode is responsible for filling in the "i_op" 367af96c1e3STobin C. Hardingfield. This is a pointer to a "struct inode_operations" which describes 368af96c1e3STobin C. Hardingthe methods that can be performed on individual inodes. 369af96c1e3STobin C. Harding 370af96c1e3STobin C. Harding 371af96c1e3STobin C. Hardingstruct xattr_handlers 372af96c1e3STobin C. Harding--------------------- 373af96c1e3STobin C. Harding 374af96c1e3STobin C. HardingOn filesystems that support extended attributes (xattrs), the s_xattr 375af96c1e3STobin C. Hardingsuperblock field points to a NULL-terminated array of xattr handlers. 376af96c1e3STobin C. HardingExtended attributes are name:value pairs. 377af96c1e3STobin C. Harding 378ee5dc049STobin C. Harding``name`` 379ee5dc049STobin C. Harding Indicates that the handler matches attributes with the specified 380ee5dc049STobin C. Harding name (such as "system.posix_acl_access"); the prefix field must 381ee5dc049STobin C. Harding be NULL. 382af96c1e3STobin C. Harding 383ee5dc049STobin C. Harding``prefix`` 384ee5dc049STobin C. Harding Indicates that the handler matches all attributes with the 385ee5dc049STobin C. Harding specified name prefix (such as "user."); the name field must be 386ee5dc049STobin C. Harding NULL. 387af96c1e3STobin C. Harding 388ee5dc049STobin C. Harding``list`` 389ee5dc049STobin C. Harding Determine if attributes matching this xattr handler should be 390ee5dc049STobin C. Harding listed for a particular dentry. Used by some listxattr 391ee5dc049STobin C. Harding implementations like generic_listxattr. 392af96c1e3STobin C. Harding 393ee5dc049STobin C. Harding``get`` 394ee5dc049STobin C. Harding Called by the VFS to get the value of a particular extended 395ee5dc049STobin C. Harding attribute. This method is called by the getxattr(2) system 396ee5dc049STobin C. Harding call. 397af96c1e3STobin C. Harding 398ee5dc049STobin C. Harding``set`` 399ee5dc049STobin C. Harding Called by the VFS to set the value of a particular extended 400ee5dc049STobin C. Harding attribute. When the new value is NULL, called to remove a 4018286de7cSRandy Dunlap particular extended attribute. This method is called by the 402ee5dc049STobin C. Harding setxattr(2) and removexattr(2) system calls. 403af96c1e3STobin C. Harding 404af96c1e3STobin C. HardingWhen none of the xattr handlers of a filesystem match the specified 405af96c1e3STobin C. Hardingattribute name or when a filesystem doesn't support extended attributes, 406af96c1e3STobin C. Hardingthe various ``*xattr(2)`` system calls return -EOPNOTSUPP. 407af96c1e3STobin C. Harding 408af96c1e3STobin C. Harding 409af96c1e3STobin C. HardingThe Inode Object 410af96c1e3STobin C. Harding================ 411af96c1e3STobin C. Harding 412af96c1e3STobin C. HardingAn inode object represents an object within the filesystem. 413af96c1e3STobin C. Harding 414af96c1e3STobin C. Harding 415af96c1e3STobin C. Hardingstruct inode_operations 416af96c1e3STobin C. Harding----------------------- 417af96c1e3STobin C. Harding 418af96c1e3STobin C. HardingThis describes how the VFS can manipulate an inode in your filesystem. 419af96c1e3STobin C. HardingAs of kernel 2.6.22, the following members are defined: 420af96c1e3STobin C. Harding 421af96c1e3STobin C. Harding.. code-block:: c 422af96c1e3STobin C. Harding 423af96c1e3STobin C. Harding struct inode_operations { 4246c960e68SChristian Brauner int (*create) (struct mnt_idmap *, struct inode *,struct dentry *, umode_t, bool); 425af96c1e3STobin C. Harding struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int); 426af96c1e3STobin C. Harding int (*link) (struct dentry *,struct inode *,struct dentry *); 427af96c1e3STobin C. Harding int (*unlink) (struct inode *,struct dentry *); 4287a77db95SChristian Brauner int (*symlink) (struct mnt_idmap *, struct inode *,struct dentry *,const char *); 429c54bd91eSChristian Brauner int (*mkdir) (struct mnt_idmap *, struct inode *,struct dentry *,umode_t); 430af96c1e3STobin C. Harding int (*rmdir) (struct inode *,struct dentry *); 4315ebb29beSChristian Brauner int (*mknod) (struct mnt_idmap *, struct inode *,struct dentry *,umode_t,dev_t); 432e18275aeSChristian Brauner int (*rename) (struct mnt_idmap *, struct inode *, struct dentry *, 433af96c1e3STobin C. Harding struct inode *, struct dentry *, unsigned int); 434af96c1e3STobin C. Harding int (*readlink) (struct dentry *, char __user *,int); 435af96c1e3STobin C. Harding const char *(*get_link) (struct dentry *, struct inode *, 436af96c1e3STobin C. Harding struct delayed_call *); 437549c7297SChristian Brauner int (*permission) (struct user_namespace *, struct inode *, int); 438cac2f8b8SChristian Brauner struct posix_acl * (*get_inode_acl)(struct inode *, int, bool); 439c1632a0fSChristian Brauner int (*setattr) (struct mnt_idmap *, struct dentry *, struct iattr *); 440b74d24f7SChristian Brauner int (*getattr) (struct mnt_idmap *, const struct path *, struct kstat *, u32, unsigned int); 441af96c1e3STobin C. Harding ssize_t (*listxattr) (struct dentry *, char *, size_t); 442af96c1e3STobin C. Harding void (*update_time)(struct inode *, struct timespec *, int); 443af96c1e3STobin C. Harding int (*atomic_open)(struct inode *, struct dentry *, struct file *, 444af96c1e3STobin C. Harding unsigned open_flag, umode_t create_mode); 445*011e2b71SChristian Brauner int (*tmpfile) (struct mnt_idmap *, struct inode *, struct file *, umode_t); 4467420332aSChristian Brauner struct posix_acl * (*get_acl)(struct user_namespace *, struct dentry *, int); 447138060baSChristian Brauner int (*set_acl)(struct user_namespace *, struct dentry *, struct posix_acl *, int); 4484c5b4799SMiklos Szeredi int (*fileattr_set)(struct user_namespace *mnt_userns, 4494c5b4799SMiklos Szeredi struct dentry *dentry, struct fileattr *fa); 4504c5b4799SMiklos Szeredi int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa); 451af96c1e3STobin C. Harding }; 452af96c1e3STobin C. Harding 453af96c1e3STobin C. HardingAgain, all methods are called without any locks being held, unless 454af96c1e3STobin C. Hardingotherwise noted. 455af96c1e3STobin C. Harding 456ee5dc049STobin C. Harding``create`` 457ee5dc049STobin C. Harding called by the open(2) and creat(2) system calls. Only required 458ee5dc049STobin C. Harding if you want to support regular files. The dentry you get should 459ee5dc049STobin C. Harding not have an inode (i.e. it should be a negative dentry). Here 460ee5dc049STobin C. Harding you will probably call d_instantiate() with the dentry and the 461ee5dc049STobin C. Harding newly created inode 462af96c1e3STobin C. Harding 463ee5dc049STobin C. Harding``lookup`` 464ee5dc049STobin C. Harding called when the VFS needs to look up an inode in a parent 465af96c1e3STobin C. Harding directory. The name to look for is found in the dentry. This 466af96c1e3STobin C. Harding method must call d_add() to insert the found inode into the 467af96c1e3STobin C. Harding dentry. The "i_count" field in the inode structure should be 468af96c1e3STobin C. Harding incremented. If the named inode does not exist a NULL inode 469af96c1e3STobin C. Harding should be inserted into the dentry (this is called a negative 470ee5dc049STobin C. Harding dentry). Returning an error code from this routine must only be 471ee5dc049STobin C. Harding done on a real error, otherwise creating inodes with system 472af96c1e3STobin C. Harding calls like create(2), mknod(2), mkdir(2) and so on will fail. 473af96c1e3STobin C. Harding If you wish to overload the dentry methods then you should 474ee5dc049STobin C. Harding initialise the "d_dop" field in the dentry; this is a pointer to 475ee5dc049STobin C. Harding a struct "dentry_operations". This method is called with the 476ee5dc049STobin C. Harding directory inode semaphore held 477af96c1e3STobin C. Harding 478ee5dc049STobin C. Harding``link`` 479ee5dc049STobin C. Harding called by the link(2) system call. Only required if you want to 480ee5dc049STobin C. Harding support hard links. You will probably need to call 481af96c1e3STobin C. Harding d_instantiate() just as you would in the create() method 482af96c1e3STobin C. Harding 483ee5dc049STobin C. Harding``unlink`` 484ee5dc049STobin C. Harding called by the unlink(2) system call. Only required if you want 485ee5dc049STobin C. Harding to support deleting inodes 486af96c1e3STobin C. Harding 487ee5dc049STobin C. Harding``symlink`` 488ee5dc049STobin C. Harding called by the symlink(2) system call. Only required if you want 489ee5dc049STobin C. Harding to support symlinks. You will probably need to call 490af96c1e3STobin C. Harding d_instantiate() just as you would in the create() method 491af96c1e3STobin C. Harding 492ee5dc049STobin C. Harding``mkdir`` 493ee5dc049STobin C. Harding called by the mkdir(2) system call. Only required if you want 494af96c1e3STobin C. Harding to support creating subdirectories. You will probably need to 495af96c1e3STobin C. Harding call d_instantiate() just as you would in the create() method 496af96c1e3STobin C. Harding 497ee5dc049STobin C. Harding``rmdir`` 498ee5dc049STobin C. Harding called by the rmdir(2) system call. Only required if you want 499af96c1e3STobin C. Harding to support deleting subdirectories 500af96c1e3STobin C. Harding 501ee5dc049STobin C. Harding``mknod`` 502ee5dc049STobin C. Harding called by the mknod(2) system call to create a device (char, 503ee5dc049STobin C. Harding block) inode or a named pipe (FIFO) or socket. Only required if 504ee5dc049STobin C. Harding you want to support creating these types of inodes. You will 505ee5dc049STobin C. Harding probably need to call d_instantiate() just as you would in the 506ee5dc049STobin C. Harding create() method 507af96c1e3STobin C. Harding 508ee5dc049STobin C. Harding``rename`` 509ee5dc049STobin C. Harding called by the rename(2) system call to rename the object to have 510ee5dc049STobin C. Harding the parent and name given by the second inode and dentry. 511af96c1e3STobin C. Harding 512af96c1e3STobin C. Harding The filesystem must return -EINVAL for any unsupported or 513af96c1e3STobin C. Harding unknown flags. Currently the following flags are implemented: 514ee5dc049STobin C. Harding (1) RENAME_NOREPLACE: this flag indicates that if the target of 515ee5dc049STobin C. Harding the rename exists the rename should fail with -EEXIST instead of 516ee5dc049STobin C. Harding replacing the target. The VFS already checks for existence, so 517ee5dc049STobin C. Harding for local filesystems the RENAME_NOREPLACE implementation is 518ee5dc049STobin C. Harding equivalent to plain rename. 519af96c1e3STobin C. Harding (2) RENAME_EXCHANGE: exchange source and target. Both must 520ee5dc049STobin C. Harding exist; this is checked by the VFS. Unlike plain rename, source 521ee5dc049STobin C. Harding and target may be of different type. 522af96c1e3STobin C. Harding 523ee5dc049STobin C. Harding``get_link`` 524ee5dc049STobin C. Harding called by the VFS to follow a symbolic link to the inode it 525ee5dc049STobin C. Harding points to. Only required if you want to support symbolic links. 526ee5dc049STobin C. Harding This method returns the symlink body to traverse (and possibly 527ee5dc049STobin C. Harding resets the current position with nd_jump_link()). If the body 528ee5dc049STobin C. Harding won't go away until the inode is gone, nothing else is needed; 529ee5dc049STobin C. Harding if it needs to be otherwise pinned, arrange for its release by 530ee5dc049STobin C. Harding having get_link(..., ..., done) do set_delayed_call(done, 531ee5dc049STobin C. Harding destructor, argument). In that case destructor(argument) will 532ee5dc049STobin C. Harding be called once VFS is done with the body you've returned. May 533ee5dc049STobin C. Harding be called in RCU mode; that is indicated by NULL dentry 534af96c1e3STobin C. Harding argument. If request can't be handled without leaving RCU mode, 535af96c1e3STobin C. Harding have it return ERR_PTR(-ECHILD). 536af96c1e3STobin C. Harding 537af96c1e3STobin C. Harding If the filesystem stores the symlink target in ->i_link, the 538af96c1e3STobin C. Harding VFS may use it directly without calling ->get_link(); however, 539af96c1e3STobin C. Harding ->get_link() must still be provided. ->i_link must not be 540af96c1e3STobin C. Harding freed until after an RCU grace period. Writing to ->i_link 541af96c1e3STobin C. Harding post-iget() time requires a 'release' memory barrier. 542af96c1e3STobin C. Harding 543ee5dc049STobin C. Harding``readlink`` 544ee5dc049STobin C. Harding this is now just an override for use by readlink(2) for the 545af96c1e3STobin C. Harding cases when ->get_link uses nd_jump_link() or object is not in 546af96c1e3STobin C. Harding fact a symlink. Normally filesystems should only implement 547af96c1e3STobin C. Harding ->get_link for symlinks and readlink(2) will automatically use 548af96c1e3STobin C. Harding that. 549af96c1e3STobin C. Harding 550ee5dc049STobin C. Harding``permission`` 551ee5dc049STobin C. Harding called by the VFS to check for access rights on a POSIX-like 552af96c1e3STobin C. Harding filesystem. 553af96c1e3STobin C. Harding 554ee5dc049STobin C. Harding May be called in rcu-walk mode (mask & MAY_NOT_BLOCK). If in 555ee5dc049STobin C. Harding rcu-walk mode, the filesystem must check the permission without 556ee5dc049STobin C. Harding blocking or storing to the inode. 557af96c1e3STobin C. Harding 558ee5dc049STobin C. Harding If a situation is encountered that rcu-walk cannot handle, 559ee5dc049STobin C. Harding return 560af96c1e3STobin C. Harding -ECHILD and it will be called again in ref-walk mode. 561af96c1e3STobin C. Harding 562ee5dc049STobin C. Harding``setattr`` 563ee5dc049STobin C. Harding called by the VFS to set attributes for a file. This method is 564ee5dc049STobin C. Harding called by chmod(2) and related system calls. 565af96c1e3STobin C. Harding 566ee5dc049STobin C. Harding``getattr`` 567ee5dc049STobin C. Harding called by the VFS to get attributes of a file. This method is 568ee5dc049STobin C. Harding called by stat(2) and related system calls. 569af96c1e3STobin C. Harding 570ee5dc049STobin C. Harding``listxattr`` 571ee5dc049STobin C. Harding called by the VFS to list all extended attributes for a given 572ee5dc049STobin C. Harding file. This method is called by the listxattr(2) system call. 573af96c1e3STobin C. Harding 574ee5dc049STobin C. Harding``update_time`` 575ee5dc049STobin C. Harding called by the VFS to update a specific time or the i_version of 576ee5dc049STobin C. Harding an inode. If this is not defined the VFS will update the inode 577ee5dc049STobin C. Harding itself and call mark_inode_dirty_sync. 578af96c1e3STobin C. Harding 579ee5dc049STobin C. Harding``atomic_open`` 580ee5dc049STobin C. Harding called on the last component of an open. Using this optional 581ee5dc049STobin C. Harding method the filesystem can look up, possibly create and open the 582ee5dc049STobin C. Harding file in one atomic operation. If it wants to leave actual 583ee5dc049STobin C. Harding opening to the caller (e.g. if the file turned out to be a 584ee5dc049STobin C. Harding symlink, device, or just something filesystem won't do atomic 585ee5dc049STobin C. Harding open for), it may signal this by returning finish_no_open(file, 586ee5dc049STobin C. Harding dentry). This method is only called if the last component is 587ee5dc049STobin C. Harding negative or needs lookup. Cached positive dentries are still 588ee5dc049STobin C. Harding handled by f_op->open(). If the file was created, FMODE_CREATED 589ee5dc049STobin C. Harding flag should be set in file->f_mode. In case of O_EXCL the 590ee5dc049STobin C. Harding method must only succeed if the file didn't exist and hence 591ee5dc049STobin C. Harding FMODE_CREATED shall always be set on success. 592af96c1e3STobin C. Harding 593ee5dc049STobin C. Harding``tmpfile`` 594ee5dc049STobin C. Harding called in the end of O_TMPFILE open(). Optional, equivalent to 595ee5dc049STobin C. Harding atomically creating, opening and unlinking a file in given 596863f144fSMiklos Szeredi directory. On success needs to return with the file already 597863f144fSMiklos Szeredi open; this can be done by calling finish_open_simple() right at 598863f144fSMiklos Szeredi the end. 599af96c1e3STobin C. Harding 6004c5b4799SMiklos Szeredi``fileattr_get`` 6014c5b4799SMiklos Szeredi called on ioctl(FS_IOC_GETFLAGS) and ioctl(FS_IOC_FSGETXATTR) to 6024c5b4799SMiklos Szeredi retrieve miscellaneous file flags and attributes. Also called 6034c5b4799SMiklos Szeredi before the relevant SET operation to check what is being changed 6044c5b4799SMiklos Szeredi (in this case with i_rwsem locked exclusive). If unset, then 6054c5b4799SMiklos Szeredi fall back to f_op->ioctl(). 6064c5b4799SMiklos Szeredi 6074c5b4799SMiklos Szeredi``fileattr_set`` 6084c5b4799SMiklos Szeredi called on ioctl(FS_IOC_SETFLAGS) and ioctl(FS_IOC_FSSETXATTR) to 6094c5b4799SMiklos Szeredi change miscellaneous file flags and attributes. Callers hold 6104c5b4799SMiklos Szeredi i_rwsem exclusive. If unset, then fall back to f_op->ioctl(). 6114c5b4799SMiklos Szeredi 612af96c1e3STobin C. Harding 613af96c1e3STobin C. HardingThe Address Space Object 614af96c1e3STobin C. Harding======================== 615af96c1e3STobin C. Harding 616af96c1e3STobin C. HardingThe address space object is used to group and manage pages in the page 617af96c1e3STobin C. Hardingcache. It can be used to keep track of the pages in a file (or anything 618af96c1e3STobin C. Hardingelse) and also track the mapping of sections of the file into process 619af96c1e3STobin C. Hardingaddress spaces. 620af96c1e3STobin C. Harding 621af96c1e3STobin C. HardingThere are a number of distinct yet related services that an 622af96c1e3STobin C. Hardingaddress-space can provide. These include communicating memory pressure, 623af96c1e3STobin C. Hardingpage lookup by address, and keeping track of pages tagged as Dirty or 624af96c1e3STobin C. HardingWriteback. 625af96c1e3STobin C. Harding 626af96c1e3STobin C. HardingThe first can be used independently to the others. The VM can try to 627af96c1e3STobin C. Hardingeither write dirty pages in order to clean them, or release clean pages 628af96c1e3STobin C. Hardingin order to reuse them. To do this it can call the ->writepage method 629fa29000bSMatthew Wilcox (Oracle)on dirty pages, and ->release_folio on clean folios with the private 630fa29000bSMatthew Wilcox (Oracle)flag set. Clean pages without PagePrivate and with no external references 631fa29000bSMatthew Wilcox (Oracle)will be released without notice being given to the address_space. 632af96c1e3STobin C. Harding 633af96c1e3STobin C. HardingTo achieve this functionality, pages need to be placed on an LRU with 634af96c1e3STobin C. Hardinglru_cache_add and mark_page_active needs to be called whenever the page 635af96c1e3STobin C. Hardingis used. 636af96c1e3STobin C. Harding 637af96c1e3STobin C. HardingPages are normally kept in a radix tree index by ->index. This tree 638af96c1e3STobin C. Hardingmaintains information about the PG_Dirty and PG_Writeback status of each 639af96c1e3STobin C. Hardingpage, so that pages with either of these flags can be found quickly. 640af96c1e3STobin C. Harding 641af96c1e3STobin C. HardingThe Dirty tag is primarily used by mpage_writepages - the default 642af96c1e3STobin C. Harding->writepages method. It uses the tag to find dirty pages to call 643af96c1e3STobin C. Harding->writepage on. If mpage_writepages is not used (i.e. the address 644af96c1e3STobin C. Hardingprovides its own ->writepages) , the PAGECACHE_TAG_DIRTY tag is almost 645af96c1e3STobin C. Hardingunused. write_inode_now and sync_inode do use it (through 646af96c1e3STobin C. Harding__sync_single_inode) to check if ->writepages has been successful in 647af96c1e3STobin C. Hardingwriting out the whole address_space. 648af96c1e3STobin C. Harding 649af96c1e3STobin C. HardingThe Writeback tag is used by filemap*wait* and sync_page* functions, via 650af96c1e3STobin C. Hardingfilemap_fdatawait_range, to wait for all writeback to complete. 651af96c1e3STobin C. Harding 652af96c1e3STobin C. HardingAn address_space handler may attach extra information to a page, 653af96c1e3STobin C. Hardingtypically using the 'private' field in the 'struct page'. If such 654af96c1e3STobin C. Hardinginformation is attached, the PG_Private flag should be set. This will 655af96c1e3STobin C. Hardingcause various VM routines to make extra calls into the address_space 656af96c1e3STobin C. Hardinghandler to deal with that data. 657af96c1e3STobin C. Harding 658af96c1e3STobin C. HardingAn address space acts as an intermediate between storage and 659af96c1e3STobin C. Hardingapplication. Data is read into the address space a whole page at a 660af96c1e3STobin C. Hardingtime, and provided to the application either by copying of the page, or 661af96c1e3STobin C. Hardingby memory-mapping the page. Data is written into the address space by 662af96c1e3STobin C. Hardingthe application, and then written-back to storage typically in whole 663af96c1e3STobin C. Hardingpages, however the address_space has finer control of write sizes. 664af96c1e3STobin C. Harding 66508830c8bSMatthew Wilcox (Oracle)The read process essentially only requires 'read_folio'. The write 666af96c1e3STobin C. Hardingprocess is more complicated and uses write_begin/write_end or 6676f31a5a2SMatthew Wilcox (Oracle)dirty_folio to write data into the address_space, and writepage and 668af96c1e3STobin C. Hardingwritepages to writeback data to storage. 669af96c1e3STobin C. Harding 670af96c1e3STobin C. HardingAdding and removing pages to/from an address_space is protected by the 671af96c1e3STobin C. Hardinginode's i_mutex. 672af96c1e3STobin C. Harding 673af96c1e3STobin C. HardingWhen data is written to a page, the PG_Dirty flag should be set. It 674af96c1e3STobin C. Hardingtypically remains set until writepage asks for it to be written. This 675af96c1e3STobin C. Hardingshould clear PG_Dirty and set PG_Writeback. It can be actually written 676af96c1e3STobin C. Hardingat any point after PG_Dirty is clear. Once it is known to be safe, 677af96c1e3STobin C. HardingPG_Writeback is cleared. 678af96c1e3STobin C. Harding 679af96c1e3STobin C. HardingWriteback makes use of a writeback_control structure to direct the 6808286de7cSRandy Dunlapoperations. This gives the writepage and writepages operations some 681af96c1e3STobin C. Hardinginformation about the nature of and reason for the writeback request, 682af96c1e3STobin C. Hardingand the constraints under which it is being done. It is also used to 683af96c1e3STobin C. Hardingreturn information back to the caller about the result of a writepage or 684af96c1e3STobin C. Hardingwritepages request. 685af96c1e3STobin C. Harding 686af96c1e3STobin C. Harding 687af96c1e3STobin C. HardingHandling errors during writeback 688af96c1e3STobin C. Harding-------------------------------- 689af96c1e3STobin C. Harding 690af96c1e3STobin C. HardingMost applications that do buffered I/O will periodically call a file 691af96c1e3STobin C. Hardingsynchronization call (fsync, fdatasync, msync or sync_file_range) to 692af96c1e3STobin C. Hardingensure that data written has made it to the backing store. When there 693af96c1e3STobin C. Hardingis an error during writeback, they expect that error to be reported when 694af96c1e3STobin C. Hardinga file sync request is made. After an error has been reported on one 695af96c1e3STobin C. Hardingrequest, subsequent requests on the same file descriptor should return 696af96c1e3STobin C. Harding0, unless further writeback errors have occurred since the previous file 697af96c1e3STobin C. Hardingsyncronization. 698af96c1e3STobin C. Harding 699af96c1e3STobin C. HardingIdeally, the kernel would report errors only on file descriptions on 700af96c1e3STobin C. Hardingwhich writes were done that subsequently failed to be written back. The 701af96c1e3STobin C. Hardinggeneric pagecache infrastructure does not track the file descriptions 702af96c1e3STobin C. Hardingthat have dirtied each individual page however, so determining which 703af96c1e3STobin C. Hardingfile descriptors should get back an error is not possible. 704af96c1e3STobin C. Harding 705af96c1e3STobin C. HardingInstead, the generic writeback error tracking infrastructure in the 706af96c1e3STobin C. Hardingkernel settles for reporting errors to fsync on all file descriptions 707af96c1e3STobin C. Hardingthat were open at the time that the error occurred. In a situation with 708af96c1e3STobin C. Hardingmultiple writers, all of them will get back an error on a subsequent 709af96c1e3STobin C. Hardingfsync, even if all of the writes done through that particular file 710af96c1e3STobin C. Hardingdescriptor succeeded (or even if there were no writes on that file 711af96c1e3STobin C. Hardingdescriptor at all). 712af96c1e3STobin C. Harding 713af96c1e3STobin C. HardingFilesystems that wish to use this infrastructure should call 714af96c1e3STobin C. Hardingmapping_set_error to record the error in the address_space when it 715af96c1e3STobin C. Hardingoccurs. Then, after writing back data from the pagecache in their 716af96c1e3STobin C. Hardingfile->fsync operation, they should call file_check_and_advance_wb_err to 717af96c1e3STobin C. Hardingensure that the struct file's error cursor has advanced to the correct 718af96c1e3STobin C. Hardingpoint in the stream of errors emitted by the backing device(s). 719af96c1e3STobin C. Harding 720af96c1e3STobin C. Harding 721af96c1e3STobin C. Hardingstruct address_space_operations 722af96c1e3STobin C. Harding------------------------------- 723af96c1e3STobin C. Harding 724af96c1e3STobin C. HardingThis describes how the VFS can manipulate mapping of a file to page 725af96c1e3STobin C. Hardingcache in your filesystem. The following members are defined: 726af96c1e3STobin C. Harding 727af96c1e3STobin C. Harding.. code-block:: c 728af96c1e3STobin C. Harding 729af96c1e3STobin C. Harding struct address_space_operations { 730af96c1e3STobin C. Harding int (*writepage)(struct page *page, struct writeback_control *wbc); 73108830c8bSMatthew Wilcox (Oracle) int (*read_folio)(struct file *, struct folio *); 732af96c1e3STobin C. Harding int (*writepages)(struct address_space *, struct writeback_control *); 7336f31a5a2SMatthew Wilcox (Oracle) bool (*dirty_folio)(struct address_space *, struct folio *); 7348151b4c8SMatthew Wilcox (Oracle) void (*readahead)(struct readahead_control *); 735af96c1e3STobin C. Harding int (*write_begin)(struct file *, struct address_space *mapping, 7369d6b0cd7SMatthew Wilcox (Oracle) loff_t pos, unsigned len, 737af96c1e3STobin C. Harding struct page **pagep, void **fsdata); 738af96c1e3STobin C. Harding int (*write_end)(struct file *, struct address_space *mapping, 739af96c1e3STobin C. Harding loff_t pos, unsigned len, unsigned copied, 740af96c1e3STobin C. Harding struct page *page, void *fsdata); 741af96c1e3STobin C. Harding sector_t (*bmap)(struct address_space *, sector_t); 742128d1f82SMatthew Wilcox (Oracle) void (*invalidate_folio) (struct folio *, size_t start, size_t len); 743fa29000bSMatthew Wilcox (Oracle) bool (*release_folio)(struct folio *, gfp_t); 744d2329aa0SMatthew Wilcox (Oracle) void (*free_folio)(struct folio *); 745af96c1e3STobin C. Harding ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter); 7465490da4fSMatthew Wilcox (Oracle) int (*migrate_folio)(struct mapping *, struct folio *dst, 7475490da4fSMatthew Wilcox (Oracle) struct folio *src, enum migrate_mode); 748affa80e8SMatthew Wilcox (Oracle) int (*launder_folio) (struct folio *); 749af96c1e3STobin C. Harding 7502e7e80f7SMatthew Wilcox (Oracle) bool (*is_partially_uptodate) (struct folio *, size_t from, 7512e7e80f7SMatthew Wilcox (Oracle) size_t count); 752520f301cSMatthew Wilcox (Oracle) void (*is_dirty_writeback)(struct folio *, bool *, bool *); 753af96c1e3STobin C. Harding int (*error_remove_page) (struct mapping *mapping, struct page *page); 754cba738f6SNeilBrown int (*swap_activate)(struct swap_info_struct *sis, struct file *f, sector_t *span) 755af96c1e3STobin C. Harding int (*swap_deactivate)(struct file *); 756cba738f6SNeilBrown int (*swap_rw)(struct kiocb *iocb, struct iov_iter *iter); 757af96c1e3STobin C. Harding }; 758af96c1e3STobin C. Harding 759ee5dc049STobin C. Harding``writepage`` 760ee5dc049STobin C. Harding called by the VM to write a dirty page to backing store. This 761ee5dc049STobin C. Harding may happen for data integrity reasons (i.e. 'sync'), or to free 762ee5dc049STobin C. Harding up memory (flush). The difference can be seen in 763ee5dc049STobin C. Harding wbc->sync_mode. The PG_Dirty flag has been cleared and 764ee5dc049STobin C. Harding PageLocked is true. writepage should start writeout, should set 765ee5dc049STobin C. Harding PG_Writeback, and should make sure the page is unlocked, either 766ee5dc049STobin C. Harding synchronously or asynchronously when the write operation 767ee5dc049STobin C. Harding completes. 768af96c1e3STobin C. Harding 769af96c1e3STobin C. Harding If wbc->sync_mode is WB_SYNC_NONE, ->writepage doesn't have to 770af96c1e3STobin C. Harding try too hard if there are problems, and may choose to write out 771af96c1e3STobin C. Harding other pages from the mapping if that is easier (e.g. due to 772af96c1e3STobin C. Harding internal dependencies). If it chooses not to start writeout, it 773ee5dc049STobin C. Harding should return AOP_WRITEPAGE_ACTIVATE so that the VM will not 774ee5dc049STobin C. Harding keep calling ->writepage on that page. 775af96c1e3STobin C. Harding 776af96c1e3STobin C. Harding See the file "Locking" for more details. 777af96c1e3STobin C. Harding 77808830c8bSMatthew Wilcox (Oracle)``read_folio`` 77990c02eb9SMatthew Wilcox (Oracle) Called by the page cache to read a folio from the backing store. 78090c02eb9SMatthew Wilcox (Oracle) The 'file' argument supplies authentication information to network 78190c02eb9SMatthew Wilcox (Oracle) filesystems, and is generally not used by block based filesystems. 78290c02eb9SMatthew Wilcox (Oracle) It may be NULL if the caller does not have an open file (eg if 78390c02eb9SMatthew Wilcox (Oracle) the kernel is performing a read for itself rather than on behalf 78490c02eb9SMatthew Wilcox (Oracle) of a userspace process with an open file). 78590c02eb9SMatthew Wilcox (Oracle) 78690c02eb9SMatthew Wilcox (Oracle) If the mapping does not support large folios, the folio will 78790c02eb9SMatthew Wilcox (Oracle) contain a single page. The folio will be locked when read_folio 78890c02eb9SMatthew Wilcox (Oracle) is called. If the read completes successfully, the folio should 78990c02eb9SMatthew Wilcox (Oracle) be marked uptodate. The filesystem should unlock the folio 79090c02eb9SMatthew Wilcox (Oracle) once the read has completed, whether it was successful or not. 79190c02eb9SMatthew Wilcox (Oracle) The filesystem does not need to modify the refcount on the folio; 79290c02eb9SMatthew Wilcox (Oracle) the page cache holds a reference count and that will not be 79390c02eb9SMatthew Wilcox (Oracle) released until the folio is unlocked. 79490c02eb9SMatthew Wilcox (Oracle) 79590c02eb9SMatthew Wilcox (Oracle) Filesystems may implement ->read_folio() synchronously. 79690c02eb9SMatthew Wilcox (Oracle) In normal operation, folios are read through the ->readahead() 79790c02eb9SMatthew Wilcox (Oracle) method. Only if this fails, or if the caller needs to wait for 79890c02eb9SMatthew Wilcox (Oracle) the read to complete will the page cache call ->read_folio(). 79990c02eb9SMatthew Wilcox (Oracle) Filesystems should not attempt to perform their own readahead 80090c02eb9SMatthew Wilcox (Oracle) in the ->read_folio() operation. 80190c02eb9SMatthew Wilcox (Oracle) 80290c02eb9SMatthew Wilcox (Oracle) If the filesystem cannot perform the read at this time, it can 80390c02eb9SMatthew Wilcox (Oracle) unlock the folio, do whatever action it needs to ensure that the 80490c02eb9SMatthew Wilcox (Oracle) read will succeed in the future and return AOP_TRUNCATED_PAGE. 80590c02eb9SMatthew Wilcox (Oracle) In this case, the caller should look up the folio, lock it, 80690c02eb9SMatthew Wilcox (Oracle) and call ->read_folio again. 80790c02eb9SMatthew Wilcox (Oracle) 80890c02eb9SMatthew Wilcox (Oracle) Callers may invoke the ->read_folio() method directly, but using 80990c02eb9SMatthew Wilcox (Oracle) read_mapping_folio() will take care of locking, waiting for the 81090c02eb9SMatthew Wilcox (Oracle) read to complete and handle cases such as AOP_TRUNCATED_PAGE. 811af96c1e3STobin C. Harding 812ee5dc049STobin C. Harding``writepages`` 813ee5dc049STobin C. Harding called by the VM to write out pages associated with the 814e9b2f15bSJulia Lawall address_space object. If wbc->sync_mode is WB_SYNC_ALL, then 815af96c1e3STobin C. Harding the writeback_control will specify a range of pages that must be 816e9b2f15bSJulia Lawall written out. If it is WB_SYNC_NONE, then a nr_to_write is 817ee5dc049STobin C. Harding given and that many pages should be written if possible. If no 818ee5dc049STobin C. Harding ->writepages is given, then mpage_writepages is used instead. 819ee5dc049STobin C. Harding This will choose pages from the address space that are tagged as 820ee5dc049STobin C. Harding DIRTY and will pass them to ->writepage. 821af96c1e3STobin C. Harding 8226f31a5a2SMatthew Wilcox (Oracle)``dirty_folio`` 8236f31a5a2SMatthew Wilcox (Oracle) called by the VM to mark a folio as dirty. This is particularly 8246f31a5a2SMatthew Wilcox (Oracle) needed if an address space attaches private data to a folio, and 8256f31a5a2SMatthew Wilcox (Oracle) that data needs to be updated when a folio is dirtied. This is 826ee5dc049STobin C. Harding called, for example, when a memory mapped page gets modified. 8276f31a5a2SMatthew Wilcox (Oracle) If defined, it should set the folio dirty flag, and the 8286f31a5a2SMatthew Wilcox (Oracle) PAGECACHE_TAG_DIRTY search mark in i_pages. 829af96c1e3STobin C. Harding 8308151b4c8SMatthew Wilcox (Oracle)``readahead`` 8318151b4c8SMatthew Wilcox (Oracle) Called by the VM to read pages associated with the address_space 8328151b4c8SMatthew Wilcox (Oracle) object. The pages are consecutive in the page cache and are 8338151b4c8SMatthew Wilcox (Oracle) locked. The implementation should decrement the page refcount 8348151b4c8SMatthew Wilcox (Oracle) after starting I/O on each page. Usually the page will be 83584dacdbdSNeilBrown unlocked by the I/O completion handler. The set of pages are 83684dacdbdSNeilBrown divided into some sync pages followed by some async pages, 83784dacdbdSNeilBrown rac->ra->async_size gives the number of async pages. The 83884dacdbdSNeilBrown filesystem should attempt to read all sync pages but may decide 83984dacdbdSNeilBrown to stop once it reaches the async pages. If it does decide to 84084dacdbdSNeilBrown stop attempting I/O, it can simply return. The caller will 84184dacdbdSNeilBrown remove the remaining pages from the address space, unlock them 84284dacdbdSNeilBrown and decrement the page refcount. Set PageUptodate if the I/O 84384dacdbdSNeilBrown completes successfully. Setting PageError on any page will be 84484dacdbdSNeilBrown ignored; simply unlock the page if an I/O error occurs. 8458151b4c8SMatthew Wilcox (Oracle) 846ee5dc049STobin C. Harding``write_begin`` 847ee5dc049STobin C. Harding Called by the generic buffered write code to ask the filesystem 848ee5dc049STobin C. Harding to prepare to write len bytes at the given offset in the file. 849ee5dc049STobin C. Harding The address_space should check that the write will be able to 850ee5dc049STobin C. Harding complete, by allocating space if necessary and doing any other 851ee5dc049STobin C. Harding internal housekeeping. If the write will update parts of any 852ee5dc049STobin C. Harding basic-blocks on storage, then those blocks should be pre-read 853ee5dc049STobin C. Harding (if they haven't been read already) so that the updated blocks 854ee5dc049STobin C. Harding can be written out properly. 855af96c1e3STobin C. Harding 856ee5dc049STobin C. Harding The filesystem must return the locked pagecache page for the 857ee5dc049STobin C. Harding specified offset, in ``*pagep``, for the caller to write into. 858af96c1e3STobin C. Harding 859ee5dc049STobin C. Harding It must be able to cope with short writes (where the length 860ee5dc049STobin C. Harding passed to write_begin is greater than the number of bytes copied 861ee5dc049STobin C. Harding into the page). 862af96c1e3STobin C. Harding 863af96c1e3STobin C. Harding A void * may be returned in fsdata, which then gets passed into 864af96c1e3STobin C. Harding write_end. 865af96c1e3STobin C. Harding 866ee5dc049STobin C. Harding Returns 0 on success; < 0 on failure (which is the error code), 867ee5dc049STobin C. Harding in which case write_end is not called. 868af96c1e3STobin C. Harding 869ee5dc049STobin C. Harding``write_end`` 870ee5dc049STobin C. Harding After a successful write_begin, and data copy, write_end must be 871ee5dc049STobin C. Harding called. len is the original len passed to write_begin, and 872ee5dc049STobin C. Harding copied is the amount that was able to be copied. 873af96c1e3STobin C. Harding 874ee5dc049STobin C. Harding The filesystem must take care of unlocking the page and 875ee5dc049STobin C. Harding releasing it refcount, and updating i_size. 876af96c1e3STobin C. Harding 877ee5dc049STobin C. Harding Returns < 0 on failure, otherwise the number of bytes (<= 878ee5dc049STobin C. Harding 'copied') that were able to be copied into pagecache. 879af96c1e3STobin C. Harding 880ee5dc049STobin C. Harding``bmap`` 881ee5dc049STobin C. Harding called by the VFS to map a logical block offset within object to 882ee5dc049STobin C. Harding physical block number. This method is used by the FIBMAP ioctl 883ee5dc049STobin C. Harding and for working with swap-files. To be able to swap to a file, 884ee5dc049STobin C. Harding the file must have a stable mapping to a block device. The swap 885ee5dc049STobin C. Harding system does not go through the filesystem but instead uses bmap 886ee5dc049STobin C. Harding to find out where the blocks in the file are and uses those 887ee5dc049STobin C. Harding addresses directly. 888af96c1e3STobin C. Harding 889128d1f82SMatthew Wilcox (Oracle)``invalidate_folio`` 890128d1f82SMatthew Wilcox (Oracle) If a folio has private data, then invalidate_folio will be 891128d1f82SMatthew Wilcox (Oracle) called when part or all of the folio is to be removed from the 892ee5dc049STobin C. Harding address space. This generally corresponds to either a 893af96c1e3STobin C. Harding truncation, punch hole or a complete invalidation of the address 894af96c1e3STobin C. Harding space (in the latter case 'offset' will always be 0 and 'length' 895fa29000bSMatthew Wilcox (Oracle) will be folio_size()). Any private data associated with the folio 896ee5dc049STobin C. Harding should be updated to reflect this truncation. If offset is 0 897128d1f82SMatthew Wilcox (Oracle) and length is folio_size(), then the private data should be 898fa29000bSMatthew Wilcox (Oracle) released, because the folio must be able to be completely 899fa29000bSMatthew Wilcox (Oracle) discarded. This may be done by calling the ->release_folio 900ee5dc049STobin C. Harding function, but in this case the release MUST succeed. 901af96c1e3STobin C. Harding 902fa29000bSMatthew Wilcox (Oracle)``release_folio`` 903fa29000bSMatthew Wilcox (Oracle) release_folio is called on folios with private data to tell the 904fa29000bSMatthew Wilcox (Oracle) filesystem that the folio is about to be freed. ->release_folio 905fa29000bSMatthew Wilcox (Oracle) should remove any private data from the folio and clear the 906fa29000bSMatthew Wilcox (Oracle) private flag. If release_folio() fails, it should return false. 907fa29000bSMatthew Wilcox (Oracle) release_folio() is used in two distinct though related cases. 908fa29000bSMatthew Wilcox (Oracle) The first is when the VM wants to free a clean folio with no 909fa29000bSMatthew Wilcox (Oracle) active users. If ->release_folio succeeds, the folio will be 910fa29000bSMatthew Wilcox (Oracle) removed from the address_space and be freed. 911af96c1e3STobin C. Harding 912af96c1e3STobin C. Harding The second case is when a request has been made to invalidate 913fa29000bSMatthew Wilcox (Oracle) some or all folios in an address_space. This can happen 914fa29000bSMatthew Wilcox (Oracle) through the fadvise(POSIX_FADV_DONTNEED) system call or by the 915fa29000bSMatthew Wilcox (Oracle) filesystem explicitly requesting it as nfs and 9p do (when they 916ee5dc049STobin C. Harding believe the cache may be out of date with storage) by calling 917ee5dc049STobin C. Harding invalidate_inode_pages2(). If the filesystem makes such a call, 918fa29000bSMatthew Wilcox (Oracle) and needs to be certain that all folios are invalidated, then 919fa29000bSMatthew Wilcox (Oracle) its release_folio will need to ensure this. Possibly it can 920fa29000bSMatthew Wilcox (Oracle) clear the uptodate flag if it cannot free private data yet. 921af96c1e3STobin C. Harding 922d2329aa0SMatthew Wilcox (Oracle)``free_folio`` 923d2329aa0SMatthew Wilcox (Oracle) free_folio is called once the folio is no longer visible in the 924ee5dc049STobin C. Harding page cache in order to allow the cleanup of any private data. 925ee5dc049STobin C. Harding Since it may be called by the memory reclaimer, it should not 926ee5dc049STobin C. Harding assume that the original address_space mapping still exists, and 927ee5dc049STobin C. Harding it should not block. 928af96c1e3STobin C. Harding 929ee5dc049STobin C. Harding``direct_IO`` 930ee5dc049STobin C. Harding called by the generic read/write routines to perform direct_IO - 931ee5dc049STobin C. Harding that is IO requests which bypass the page cache and transfer 932ee5dc049STobin C. Harding data directly between the storage and the application's address 933ee5dc049STobin C. Harding space. 934af96c1e3STobin C. Harding 9355490da4fSMatthew Wilcox (Oracle)``migrate_folio`` 936ee5dc049STobin C. Harding This is used to compact the physical memory usage. If the VM 9375490da4fSMatthew Wilcox (Oracle) wants to relocate a folio (maybe from a memory device that is 9385490da4fSMatthew Wilcox (Oracle) signalling imminent failure) it will pass a new folio and an old 9395490da4fSMatthew Wilcox (Oracle) folio to this function. migrate_folio should transfer any private 9405490da4fSMatthew Wilcox (Oracle) data across and update any references that it has to the folio. 941af96c1e3STobin C. Harding 942affa80e8SMatthew Wilcox (Oracle)``launder_folio`` 943affa80e8SMatthew Wilcox (Oracle) Called before freeing a folio - it writes back the dirty folio. 944affa80e8SMatthew Wilcox (Oracle) To prevent redirtying the folio, it is kept locked during the 945ee5dc049STobin C. Harding whole operation. 946af96c1e3STobin C. Harding 947ee5dc049STobin C. Harding``is_partially_uptodate`` 948ee5dc049STobin C. Harding Called by the VM when reading a file through the pagecache when 9492e7e80f7SMatthew Wilcox (Oracle) the underlying blocksize is smaller than the size of the folio. 9502e7e80f7SMatthew Wilcox (Oracle) If the required block is up to date then the read can complete 9512e7e80f7SMatthew Wilcox (Oracle) without needing I/O to bring the whole page up to date. 952af96c1e3STobin C. Harding 953ee5dc049STobin C. Harding``is_dirty_writeback`` 954520f301cSMatthew Wilcox (Oracle) Called by the VM when attempting to reclaim a folio. The VM uses 955ee5dc049STobin C. Harding dirty and writeback information to determine if it needs to 956ee5dc049STobin C. Harding stall to allow flushers a chance to complete some IO. 957520f301cSMatthew Wilcox (Oracle) Ordinarily it can use folio_test_dirty and folio_test_writeback but 958520f301cSMatthew Wilcox (Oracle) some filesystems have more complex state (unstable folios in NFS 959ee5dc049STobin C. Harding prevent reclaim) or do not set those flags due to locking 960ee5dc049STobin C. Harding problems. This callback allows a filesystem to indicate to the 961520f301cSMatthew Wilcox (Oracle) VM if a folio should be treated as dirty or writeback for the 962ee5dc049STobin C. Harding purposes of stalling. 963af96c1e3STobin C. Harding 964ee5dc049STobin C. Harding``error_remove_page`` 965ee5dc049STobin C. Harding normally set to generic_error_remove_page if truncation is ok 966ee5dc049STobin C. Harding for this address space. Used for memory failure handling. 967af96c1e3STobin C. Harding Setting this implies you deal with pages going away under you, 968af96c1e3STobin C. Harding unless you have them locked or reference counts increased. 969af96c1e3STobin C. Harding 970ee5dc049STobin C. Harding``swap_activate`` 971cba738f6SNeilBrown 972cba738f6SNeilBrown Called to prepare the given file for swap. It should perform 973cba738f6SNeilBrown any validation and preparation necessary to ensure that writes 974cba738f6SNeilBrown can be performed with minimal memory allocation. It should call 975cba738f6SNeilBrown add_swap_extent(), or the helper iomap_swapfile_activate(), and 976cba738f6SNeilBrown return the number of extents added. If IO should be submitted 977cba738f6SNeilBrown through ->swap_rw(), it should set SWP_FS_OPS, otherwise IO will 978cba738f6SNeilBrown be submitted directly to the block device ``sis->bdev``. 979af96c1e3STobin C. Harding 980ee5dc049STobin C. Harding``swap_deactivate`` 981ee5dc049STobin C. Harding Called during swapoff on files where swap_activate was 982ee5dc049STobin C. Harding successful. 983af96c1e3STobin C. Harding 984cba738f6SNeilBrown``swap_rw`` 985cba738f6SNeilBrown Called to read or write swap pages when SWP_FS_OPS is set. 986af96c1e3STobin C. Harding 987af96c1e3STobin C. HardingThe File Object 988af96c1e3STobin C. Harding=============== 989af96c1e3STobin C. Harding 990af96c1e3STobin C. HardingA file object represents a file opened by a process. This is also known 991af96c1e3STobin C. Hardingas an "open file description" in POSIX parlance. 992af96c1e3STobin C. Harding 993af96c1e3STobin C. Harding 994af96c1e3STobin C. Hardingstruct file_operations 995af96c1e3STobin C. Harding---------------------- 996af96c1e3STobin C. Harding 997af96c1e3STobin C. HardingThis describes how the VFS can manipulate an open file. As of kernel 998af96c1e3STobin C. Harding4.18, the following members are defined: 999af96c1e3STobin C. Harding 1000af96c1e3STobin C. Harding.. code-block:: c 1001af96c1e3STobin C. Harding 1002af96c1e3STobin C. Harding struct file_operations { 1003af96c1e3STobin C. Harding struct module *owner; 1004af96c1e3STobin C. Harding loff_t (*llseek) (struct file *, loff_t, int); 1005af96c1e3STobin C. Harding ssize_t (*read) (struct file *, char __user *, size_t, loff_t *); 1006af96c1e3STobin C. Harding ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *); 1007af96c1e3STobin C. Harding ssize_t (*read_iter) (struct kiocb *, struct iov_iter *); 1008af96c1e3STobin C. Harding ssize_t (*write_iter) (struct kiocb *, struct iov_iter *); 1009af96c1e3STobin C. Harding int (*iopoll)(struct kiocb *kiocb, bool spin); 1010af96c1e3STobin C. Harding int (*iterate) (struct file *, struct dir_context *); 1011af96c1e3STobin C. Harding int (*iterate_shared) (struct file *, struct dir_context *); 1012af96c1e3STobin C. Harding __poll_t (*poll) (struct file *, struct poll_table_struct *); 1013af96c1e3STobin C. Harding long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long); 1014af96c1e3STobin C. Harding long (*compat_ioctl) (struct file *, unsigned int, unsigned long); 1015af96c1e3STobin C. Harding int (*mmap) (struct file *, struct vm_area_struct *); 1016af96c1e3STobin C. Harding int (*open) (struct inode *, struct file *); 1017af96c1e3STobin C. Harding int (*flush) (struct file *, fl_owner_t id); 1018af96c1e3STobin C. Harding int (*release) (struct inode *, struct file *); 1019af96c1e3STobin C. Harding int (*fsync) (struct file *, loff_t, loff_t, int datasync); 1020af96c1e3STobin C. Harding int (*fasync) (int, struct file *, int); 1021af96c1e3STobin C. Harding int (*lock) (struct file *, int, struct file_lock *); 1022af96c1e3STobin C. Harding ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int); 1023af96c1e3STobin C. Harding unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); 1024af96c1e3STobin C. Harding int (*check_flags)(int); 1025af96c1e3STobin C. Harding int (*flock) (struct file *, int, struct file_lock *); 1026af96c1e3STobin C. Harding ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int); 1027af96c1e3STobin C. Harding ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int); 1028af96c1e3STobin C. Harding int (*setlease)(struct file *, long, struct file_lock **, void **); 1029af96c1e3STobin C. Harding long (*fallocate)(struct file *file, int mode, loff_t offset, 1030af96c1e3STobin C. Harding loff_t len); 1031af96c1e3STobin C. Harding void (*show_fdinfo)(struct seq_file *m, struct file *f); 1032af96c1e3STobin C. Harding #ifndef CONFIG_MMU 1033af96c1e3STobin C. Harding unsigned (*mmap_capabilities)(struct file *); 1034af96c1e3STobin C. Harding #endif 1035af96c1e3STobin C. Harding ssize_t (*copy_file_range)(struct file *, loff_t, struct file *, loff_t, size_t, unsigned int); 1036af96c1e3STobin C. Harding loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in, 1037af96c1e3STobin C. Harding struct file *file_out, loff_t pos_out, 1038af96c1e3STobin C. Harding loff_t len, unsigned int remap_flags); 1039af96c1e3STobin C. Harding int (*fadvise)(struct file *, loff_t, loff_t, int); 1040af96c1e3STobin C. Harding }; 1041af96c1e3STobin C. Harding 1042af96c1e3STobin C. HardingAgain, all methods are called without any locks being held, unless 1043af96c1e3STobin C. Hardingotherwise noted. 1044af96c1e3STobin C. Harding 1045ee5dc049STobin C. Harding``llseek`` 1046ee5dc049STobin C. Harding called when the VFS needs to move the file position index 1047af96c1e3STobin C. Harding 1048ee5dc049STobin C. Harding``read`` 1049ee5dc049STobin C. Harding called by read(2) and related system calls 1050af96c1e3STobin C. Harding 1051ee5dc049STobin C. Harding``read_iter`` 1052ee5dc049STobin C. Harding possibly asynchronous read with iov_iter as destination 1053af96c1e3STobin C. Harding 1054ee5dc049STobin C. Harding``write`` 1055ee5dc049STobin C. Harding called by write(2) and related system calls 1056af96c1e3STobin C. Harding 1057ee5dc049STobin C. Harding``write_iter`` 1058ee5dc049STobin C. Harding possibly asynchronous write with iov_iter as source 1059af96c1e3STobin C. Harding 1060ee5dc049STobin C. Harding``iopoll`` 1061ee5dc049STobin C. Harding called when aio wants to poll for completions on HIPRI iocbs 1062af96c1e3STobin C. Harding 1063ee5dc049STobin C. Harding``iterate`` 1064ee5dc049STobin C. Harding called when the VFS needs to read the directory contents 1065af96c1e3STobin C. Harding 1066ee5dc049STobin C. Harding``iterate_shared`` 1067ee5dc049STobin C. Harding called when the VFS needs to read the directory contents when 1068ee5dc049STobin C. Harding filesystem supports concurrent dir iterators 1069af96c1e3STobin C. Harding 1070ee5dc049STobin C. Harding``poll`` 1071ee5dc049STobin C. Harding called by the VFS when a process wants to check if there is 1072af96c1e3STobin C. Harding activity on this file and (optionally) go to sleep until there 1073af96c1e3STobin C. Harding is activity. Called by the select(2) and poll(2) system calls 1074af96c1e3STobin C. Harding 1075ee5dc049STobin C. Harding``unlocked_ioctl`` 1076ee5dc049STobin C. Harding called by the ioctl(2) system call. 1077af96c1e3STobin C. Harding 1078ee5dc049STobin C. Harding``compat_ioctl`` 1079ee5dc049STobin C. Harding called by the ioctl(2) system call when 32 bit system calls are 1080ee5dc049STobin C. Harding used on 64 bit kernels. 1081af96c1e3STobin C. Harding 1082ee5dc049STobin C. Harding``mmap`` 1083ee5dc049STobin C. Harding called by the mmap(2) system call 1084af96c1e3STobin C. Harding 1085ee5dc049STobin C. Harding``open`` 1086ee5dc049STobin C. Harding called by the VFS when an inode should be opened. When the VFS 1087af96c1e3STobin C. Harding opens a file, it creates a new "struct file". It then calls the 1088af96c1e3STobin C. Harding open method for the newly allocated file structure. You might 1089ee5dc049STobin C. Harding think that the open method really belongs in "struct 1090ee5dc049STobin C. Harding inode_operations", and you may be right. I think it's done the 1091ee5dc049STobin C. Harding way it is because it makes filesystems simpler to implement. 1092ee5dc049STobin C. Harding The open() method is a good place to initialize the 1093af96c1e3STobin C. Harding "private_data" member in the file structure if you want to point 1094af96c1e3STobin C. Harding to a device structure 1095af96c1e3STobin C. Harding 1096ee5dc049STobin C. Harding``flush`` 1097ee5dc049STobin C. Harding called by the close(2) system call to flush a file 1098af96c1e3STobin C. Harding 1099ee5dc049STobin C. Harding``release`` 1100ee5dc049STobin C. Harding called when the last reference to an open file is closed 1101af96c1e3STobin C. Harding 1102ee5dc049STobin C. Harding``fsync`` 1103ee5dc049STobin C. Harding called by the fsync(2) system call. Also see the section above 1104af96c1e3STobin C. Harding entitled "Handling errors during writeback". 1105af96c1e3STobin C. Harding 1106ee5dc049STobin C. Harding``fasync`` 1107ee5dc049STobin C. Harding called by the fcntl(2) system call when asynchronous 1108af96c1e3STobin C. Harding (non-blocking) mode is enabled for a file 1109af96c1e3STobin C. Harding 1110ee5dc049STobin C. Harding``lock`` 1111ee5dc049STobin C. Harding called by the fcntl(2) system call for F_GETLK, F_SETLK, and 1112ee5dc049STobin C. Harding F_SETLKW commands 1113af96c1e3STobin C. Harding 1114ee5dc049STobin C. Harding``get_unmapped_area`` 1115ee5dc049STobin C. Harding called by the mmap(2) system call 1116af96c1e3STobin C. Harding 1117ee5dc049STobin C. Harding``check_flags`` 1118ee5dc049STobin C. Harding called by the fcntl(2) system call for F_SETFL command 1119af96c1e3STobin C. Harding 1120ee5dc049STobin C. Harding``flock`` 1121ee5dc049STobin C. Harding called by the flock(2) system call 1122af96c1e3STobin C. Harding 1123ee5dc049STobin C. Harding``splice_write`` 1124ee5dc049STobin C. Harding called by the VFS to splice data from a pipe to a file. This 1125af96c1e3STobin C. Harding method is used by the splice(2) system call 1126af96c1e3STobin C. Harding 1127ee5dc049STobin C. Harding``splice_read`` 1128ee5dc049STobin C. Harding called by the VFS to splice data from file to a pipe. This 1129af96c1e3STobin C. Harding method is used by the splice(2) system call 1130af96c1e3STobin C. Harding 1131ee5dc049STobin C. Harding``setlease`` 1132ee5dc049STobin C. Harding called by the VFS to set or release a file lock lease. setlease 1133af96c1e3STobin C. Harding implementations should call generic_setlease to record or remove 1134af96c1e3STobin C. Harding the lease in the inode after setting it. 1135af96c1e3STobin C. Harding 1136ee5dc049STobin C. Harding``fallocate`` 1137ee5dc049STobin C. Harding called by the VFS to preallocate blocks or punch a hole. 1138af96c1e3STobin C. Harding 1139ee5dc049STobin C. Harding``copy_file_range`` 1140ee5dc049STobin C. Harding called by the copy_file_range(2) system call. 1141af96c1e3STobin C. Harding 1142ee5dc049STobin C. Harding``remap_file_range`` 1143ee5dc049STobin C. Harding called by the ioctl(2) system call for FICLONERANGE and FICLONE 1144ee5dc049STobin C. Harding and FIDEDUPERANGE commands to remap file ranges. An 1145ee5dc049STobin C. Harding implementation should remap len bytes at pos_in of the source 1146ee5dc049STobin C. Harding file into the dest file at pos_out. Implementations must handle 1147ee5dc049STobin C. Harding callers passing in len == 0; this means "remap to the end of the 1148ee5dc049STobin C. Harding source file". The return value should the number of bytes 1149ee5dc049STobin C. Harding remapped, or the usual negative error code if errors occurred 1150ee5dc049STobin C. Harding before any bytes were remapped. The remap_flags parameter 1151ee5dc049STobin C. Harding accepts REMAP_FILE_* flags. If REMAP_FILE_DEDUP is set then the 1152ee5dc049STobin C. Harding implementation must only remap if the requested file ranges have 1153cb56ecaeSJulia Lawall identical contents. If REMAP_FILE_CAN_SHORTEN is set, the caller is 1154ee5dc049STobin C. Harding ok with the implementation shortening the request length to 1155ee5dc049STobin C. Harding satisfy alignment or EOF requirements (or any other reason). 1156af96c1e3STobin C. Harding 1157ee5dc049STobin C. Harding``fadvise`` 1158ee5dc049STobin C. Harding possibly called by the fadvise64() system call. 1159af96c1e3STobin C. Harding 1160af96c1e3STobin C. HardingNote that the file operations are implemented by the specific 1161af96c1e3STobin C. Hardingfilesystem in which the inode resides. When opening a device node 1162af96c1e3STobin C. Harding(character or block special) most filesystems will call special 1163af96c1e3STobin C. Hardingsupport routines in the VFS which will locate the required device 1164af96c1e3STobin C. Hardingdriver information. These support routines replace the filesystem file 1165af96c1e3STobin C. Hardingoperations with those for the device driver, and then proceed to call 1166af96c1e3STobin C. Hardingthe new open() method for the file. This is how opening a device file 1167af96c1e3STobin C. Hardingin the filesystem eventually ends up calling the device driver open() 1168af96c1e3STobin C. Hardingmethod. 1169af96c1e3STobin C. Harding 1170af96c1e3STobin C. Harding 1171af96c1e3STobin C. HardingDirectory Entry Cache (dcache) 1172af96c1e3STobin C. Harding============================== 1173af96c1e3STobin C. Harding 1174af96c1e3STobin C. Harding 1175af96c1e3STobin C. Hardingstruct dentry_operations 1176af96c1e3STobin C. Harding------------------------ 1177af96c1e3STobin C. Harding 1178af96c1e3STobin C. HardingThis describes how a filesystem can overload the standard dentry 1179af96c1e3STobin C. Hardingoperations. Dentries and the dcache are the domain of the VFS and the 1180af96c1e3STobin C. Hardingindividual filesystem implementations. Device drivers have no business 1181af96c1e3STobin C. Hardinghere. These methods may be set to NULL, as they are either optional or 1182af96c1e3STobin C. Hardingthe VFS uses a default. As of kernel 2.6.22, the following members are 1183af96c1e3STobin C. Hardingdefined: 1184af96c1e3STobin C. Harding 1185af96c1e3STobin C. Harding.. code-block:: c 1186af96c1e3STobin C. Harding 1187af96c1e3STobin C. Harding struct dentry_operations { 1188af96c1e3STobin C. Harding int (*d_revalidate)(struct dentry *, unsigned int); 1189af96c1e3STobin C. Harding int (*d_weak_revalidate)(struct dentry *, unsigned int); 1190af96c1e3STobin C. Harding int (*d_hash)(const struct dentry *, struct qstr *); 1191af96c1e3STobin C. Harding int (*d_compare)(const struct dentry *, 1192af96c1e3STobin C. Harding unsigned int, const char *, const struct qstr *); 1193af96c1e3STobin C. Harding int (*d_delete)(const struct dentry *); 1194af96c1e3STobin C. Harding int (*d_init)(struct dentry *); 1195af96c1e3STobin C. Harding void (*d_release)(struct dentry *); 1196af96c1e3STobin C. Harding void (*d_iput)(struct dentry *, struct inode *); 1197af96c1e3STobin C. Harding char *(*d_dname)(struct dentry *, char *, int); 1198af96c1e3STobin C. Harding struct vfsmount *(*d_automount)(struct path *); 1199af96c1e3STobin C. Harding int (*d_manage)(const struct path *, bool); 1200af96c1e3STobin C. Harding struct dentry *(*d_real)(struct dentry *, const struct inode *); 1201af96c1e3STobin C. Harding }; 1202af96c1e3STobin C. Harding 1203ee5dc049STobin C. Harding``d_revalidate`` 1204ee5dc049STobin C. Harding called when the VFS needs to revalidate a dentry. This is 1205ee5dc049STobin C. Harding called whenever a name look-up finds a dentry in the dcache. 1206ee5dc049STobin C. Harding Most local filesystems leave this as NULL, because all their 1207ee5dc049STobin C. Harding dentries in the dcache are valid. Network filesystems are 1208ee5dc049STobin C. Harding different since things can change on the server without the 1209ee5dc049STobin C. Harding client necessarily being aware of it. 1210af96c1e3STobin C. Harding 1211ee5dc049STobin C. Harding This function should return a positive value if the dentry is 1212ee5dc049STobin C. Harding still valid, and zero or a negative error code if it isn't. 1213af96c1e3STobin C. Harding 1214ee5dc049STobin C. Harding d_revalidate may be called in rcu-walk mode (flags & 1215ee5dc049STobin C. Harding LOOKUP_RCU). If in rcu-walk mode, the filesystem must 1216ee5dc049STobin C. Harding revalidate the dentry without blocking or storing to the dentry, 1217ee5dc049STobin C. Harding d_parent and d_inode should not be used without care (because 1218ee5dc049STobin C. Harding they can change and, in d_inode case, even become NULL under 1219ee5dc049STobin C. Harding us). 1220af96c1e3STobin C. Harding 1221ee5dc049STobin C. Harding If a situation is encountered that rcu-walk cannot handle, 1222ee5dc049STobin C. Harding return 1223af96c1e3STobin C. Harding -ECHILD and it will be called again in ref-walk mode. 1224af96c1e3STobin C. Harding 1225ee5dc049STobin C. Harding``_weak_revalidate`` 1226ee5dc049STobin C. Harding called when the VFS needs to revalidate a "jumped" dentry. This 1227ee5dc049STobin C. Harding is called when a path-walk ends at dentry that was not acquired 1228ee5dc049STobin C. Harding by doing a lookup in the parent directory. This includes "/", 1229ee5dc049STobin C. Harding "." and "..", as well as procfs-style symlinks and mountpoint 1230ee5dc049STobin C. Harding traversal. 1231af96c1e3STobin C. Harding 1232ee5dc049STobin C. Harding In this case, we are less concerned with whether the dentry is 1233ee5dc049STobin C. Harding still fully correct, but rather that the inode is still valid. 1234ee5dc049STobin C. Harding As with d_revalidate, most local filesystems will set this to 1235ee5dc049STobin C. Harding NULL since their dcache entries are always valid. 1236af96c1e3STobin C. Harding 1237ee5dc049STobin C. Harding This function has the same return code semantics as 1238ee5dc049STobin C. Harding d_revalidate. 1239af96c1e3STobin C. Harding 1240af96c1e3STobin C. Harding d_weak_revalidate is only called after leaving rcu-walk mode. 1241af96c1e3STobin C. Harding 1242ee5dc049STobin C. Harding``d_hash`` 1243ee5dc049STobin C. Harding called when the VFS adds a dentry to the hash table. The first 1244af96c1e3STobin C. Harding dentry passed to d_hash is the parent directory that the name is 1245af96c1e3STobin C. Harding to be hashed into. 1246af96c1e3STobin C. Harding 1247af96c1e3STobin C. Harding Same locking and synchronisation rules as d_compare regarding 1248af96c1e3STobin C. Harding what is safe to dereference etc. 1249af96c1e3STobin C. Harding 1250ee5dc049STobin C. Harding``d_compare`` 1251ee5dc049STobin C. Harding called to compare a dentry name with a given name. The first 1252af96c1e3STobin C. Harding dentry is the parent of the dentry to be compared, the second is 1253ee5dc049STobin C. Harding the child dentry. len and name string are properties of the 1254ee5dc049STobin C. Harding dentry to be compared. qstr is the name to compare it with. 1255af96c1e3STobin C. Harding 1256af96c1e3STobin C. Harding Must be constant and idempotent, and should not take locks if 1257ee5dc049STobin C. Harding possible, and should not or store into the dentry. Should not 1258ee5dc049STobin C. Harding dereference pointers outside the dentry without lots of care 1259ee5dc049STobin C. Harding (eg. d_parent, d_inode, d_name should not be used). 1260af96c1e3STobin C. Harding 1261ee5dc049STobin C. Harding However, our vfsmount is pinned, and RCU held, so the dentries 1262ee5dc049STobin C. Harding and inodes won't disappear, neither will our sb or filesystem 1263ee5dc049STobin C. Harding module. ->d_sb may be used. 1264af96c1e3STobin C. Harding 1265ee5dc049STobin C. Harding It is a tricky calling convention because it needs to be called 1266ee5dc049STobin C. Harding under "rcu-walk", ie. without any locks or references on things. 1267af96c1e3STobin C. Harding 1268ee5dc049STobin C. Harding``d_delete`` 1269ee5dc049STobin C. Harding called when the last reference to a dentry is dropped and the 1270ee5dc049STobin C. Harding dcache is deciding whether or not to cache it. Return 1 to 1271ee5dc049STobin C. Harding delete immediately, or 0 to cache the dentry. Default is NULL 1272ee5dc049STobin C. Harding which means to always cache a reachable dentry. d_delete must 1273ee5dc049STobin C. Harding be constant and idempotent. 1274af96c1e3STobin C. Harding 1275ee5dc049STobin C. Harding``d_init`` 1276ee5dc049STobin C. Harding called when a dentry is allocated 1277af96c1e3STobin C. Harding 1278ee5dc049STobin C. Harding``d_release`` 1279ee5dc049STobin C. Harding called when a dentry is really deallocated 1280af96c1e3STobin C. Harding 1281ee5dc049STobin C. Harding``d_iput`` 1282ee5dc049STobin C. Harding called when a dentry loses its inode (just prior to its being 1283ee5dc049STobin C. Harding deallocated). The default when this is NULL is that the VFS 1284ee5dc049STobin C. Harding calls iput(). If you define this method, you must call iput() 1285ee5dc049STobin C. Harding yourself 1286af96c1e3STobin C. Harding 1287ee5dc049STobin C. Harding``d_dname`` 1288ee5dc049STobin C. Harding called when the pathname of a dentry should be generated. 1289ee5dc049STobin C. Harding Useful for some pseudo filesystems (sockfs, pipefs, ...) to 1290ee5dc049STobin C. Harding delay pathname generation. (Instead of doing it when dentry is 1291ee5dc049STobin C. Harding created, it's done only when the path is needed.). Real 1292ee5dc049STobin C. Harding filesystems probably dont want to use it, because their dentries 1293ee5dc049STobin C. Harding are present in global dcache hash, so their hash should be an 1294ee5dc049STobin C. Harding invariant. As no lock is held, d_dname() should not try to 1295ee5dc049STobin C. Harding modify the dentry itself, unless appropriate SMP safety is used. 1296ee5dc049STobin C. Harding CAUTION : d_path() logic is quite tricky. The correct way to 1297ee5dc049STobin C. Harding return for example "Hello" is to put it at the end of the 1298ee5dc049STobin C. Harding buffer, and returns a pointer to the first char. 1299ee5dc049STobin C. Harding dynamic_dname() helper function is provided to take care of 1300ee5dc049STobin C. Harding this. 1301af96c1e3STobin C. Harding 1302af96c1e3STobin C. Harding Example : 1303af96c1e3STobin C. Harding 1304af96c1e3STobin C. Harding.. code-block:: c 1305af96c1e3STobin C. Harding 1306af96c1e3STobin C. Harding static char *pipefs_dname(struct dentry *dent, char *buffer, int buflen) 1307af96c1e3STobin C. Harding { 1308af96c1e3STobin C. Harding return dynamic_dname(dentry, buffer, buflen, "pipe:[%lu]", 1309af96c1e3STobin C. Harding dentry->d_inode->i_ino); 1310af96c1e3STobin C. Harding } 1311af96c1e3STobin C. Harding 1312ee5dc049STobin C. Harding``d_automount`` 1313ee5dc049STobin C. Harding called when an automount dentry is to be traversed (optional). 1314ee5dc049STobin C. Harding This should create a new VFS mount record and return the record 1315ee5dc049STobin C. Harding to the caller. The caller is supplied with a path parameter 1316ee5dc049STobin C. Harding giving the automount directory to describe the automount target 1317ee5dc049STobin C. Harding and the parent VFS mount record to provide inheritable mount 1318ee5dc049STobin C. Harding parameters. NULL should be returned if someone else managed to 1319ee5dc049STobin C. Harding make the automount first. If the vfsmount creation failed, then 1320ee5dc049STobin C. Harding an error code should be returned. If -EISDIR is returned, then 1321ee5dc049STobin C. Harding the directory will be treated as an ordinary directory and 1322ee5dc049STobin C. Harding returned to pathwalk to continue walking. 1323af96c1e3STobin C. Harding 1324ee5dc049STobin C. Harding If a vfsmount is returned, the caller will attempt to mount it 1325ee5dc049STobin C. Harding on the mountpoint and will remove the vfsmount from its 1326ee5dc049STobin C. Harding expiration list in the case of failure. The vfsmount should be 1327ee5dc049STobin C. Harding returned with 2 refs on it to prevent automatic expiration - the 1328ee5dc049STobin C. Harding caller will clean up the additional ref. 1329af96c1e3STobin C. Harding 1330ee5dc049STobin C. Harding This function is only used if DCACHE_NEED_AUTOMOUNT is set on 1331ee5dc049STobin C. Harding the dentry. This is set by __d_instantiate() if S_AUTOMOUNT is 1332ee5dc049STobin C. Harding set on the inode being added. 1333af96c1e3STobin C. Harding 1334ee5dc049STobin C. Harding``d_manage`` 1335ee5dc049STobin C. Harding called to allow the filesystem to manage the transition from a 1336ee5dc049STobin C. Harding dentry (optional). This allows autofs, for example, to hold up 1337ee5dc049STobin C. Harding clients waiting to explore behind a 'mountpoint' while letting 1338ee5dc049STobin C. Harding the daemon go past and construct the subtree there. 0 should be 1339ee5dc049STobin C. Harding returned to let the calling process continue. -EISDIR can be 1340ee5dc049STobin C. Harding returned to tell pathwalk to use this directory as an ordinary 1341ee5dc049STobin C. Harding directory and to ignore anything mounted on it and not to check 1342ee5dc049STobin C. Harding the automount flag. Any other error code will abort pathwalk 1343ee5dc049STobin C. Harding completely. 1344af96c1e3STobin C. Harding 1345af96c1e3STobin C. Harding If the 'rcu_walk' parameter is true, then the caller is doing a 1346ee5dc049STobin C. Harding pathwalk in RCU-walk mode. Sleeping is not permitted in this 1347ee5dc049STobin C. Harding mode, and the caller can be asked to leave it and call again by 1348ee5dc049STobin C. Harding returning -ECHILD. -EISDIR may also be returned to tell 1349ee5dc049STobin C. Harding pathwalk to ignore d_automount or any mounts. 1350af96c1e3STobin C. Harding 1351ee5dc049STobin C. Harding This function is only used if DCACHE_MANAGE_TRANSIT is set on 1352ee5dc049STobin C. Harding the dentry being transited from. 1353af96c1e3STobin C. Harding 1354ee5dc049STobin C. Harding``d_real`` 1355ee5dc049STobin C. Harding overlay/union type filesystems implement this method to return 1356ee5dc049STobin C. Harding one of the underlying dentries hidden by the overlay. It is 1357ee5dc049STobin C. Harding used in two different modes: 1358af96c1e3STobin C. Harding 1359ee5dc049STobin C. Harding Called from file_dentry() it returns the real dentry matching 1360ee5dc049STobin C. Harding the inode argument. The real dentry may be from a lower layer 1361ee5dc049STobin C. Harding already copied up, but still referenced from the file. This 1362ee5dc049STobin C. Harding mode is selected with a non-NULL inode argument. 1363af96c1e3STobin C. Harding 1364af96c1e3STobin C. Harding With NULL inode the topmost real underlying dentry is returned. 1365af96c1e3STobin C. Harding 1366af96c1e3STobin C. HardingEach dentry has a pointer to its parent dentry, as well as a hash list 1367af96c1e3STobin C. Hardingof child dentries. Child dentries are basically like files in a 1368af96c1e3STobin C. Hardingdirectory. 1369af96c1e3STobin C. Harding 1370af96c1e3STobin C. Harding 1371af96c1e3STobin C. HardingDirectory Entry Cache API 1372af96c1e3STobin C. Harding-------------------------- 1373af96c1e3STobin C. Harding 1374af96c1e3STobin C. HardingThere are a number of functions defined which permit a filesystem to 1375af96c1e3STobin C. Hardingmanipulate dentries: 1376af96c1e3STobin C. Harding 1377ee5dc049STobin C. Harding``dget`` 1378ee5dc049STobin C. Harding open a new handle for an existing dentry (this just increments 1379af96c1e3STobin C. Harding the usage count) 1380af96c1e3STobin C. Harding 1381ee5dc049STobin C. Harding``dput`` 1382ee5dc049STobin C. Harding close a handle for a dentry (decrements the usage count). If 1383af96c1e3STobin C. Harding the usage count drops to 0, and the dentry is still in its 1384af96c1e3STobin C. Harding parent's hash, the "d_delete" method is called to check whether 1385ee5dc049STobin C. Harding it should be cached. If it should not be cached, or if the 1386ee5dc049STobin C. Harding dentry is not hashed, it is deleted. Otherwise cached dentries 1387ee5dc049STobin C. Harding are put into an LRU list to be reclaimed on memory shortage. 1388af96c1e3STobin C. Harding 1389ee5dc049STobin C. Harding``d_drop`` 1390ee5dc049STobin C. Harding this unhashes a dentry from its parents hash list. A subsequent 1391ee5dc049STobin C. Harding call to dput() will deallocate the dentry if its usage count 1392ee5dc049STobin C. Harding drops to 0 1393af96c1e3STobin C. Harding 1394ee5dc049STobin C. Harding``d_delete`` 1395ee5dc049STobin C. Harding delete a dentry. If there are no other open references to the 1396ee5dc049STobin C. Harding dentry then the dentry is turned into a negative dentry (the 1397ee5dc049STobin C. Harding d_iput() method is called). If there are other references, then 1398ee5dc049STobin C. Harding d_drop() is called instead 1399af96c1e3STobin C. Harding 1400ee5dc049STobin C. Harding``d_add`` 1401ee5dc049STobin C. Harding add a dentry to its parents hash list and then calls 1402af96c1e3STobin C. Harding d_instantiate() 1403af96c1e3STobin C. Harding 1404ee5dc049STobin C. Harding``d_instantiate`` 1405ee5dc049STobin C. Harding add a dentry to the alias hash list for the inode and updates 1406ee5dc049STobin C. Harding the "d_inode" member. The "i_count" member in the inode 1407ee5dc049STobin C. Harding structure should be set/incremented. If the inode pointer is 1408ee5dc049STobin C. Harding NULL, the dentry is called a "negative dentry". This function 1409ee5dc049STobin C. Harding is commonly called when an inode is created for an existing 1410ee5dc049STobin C. Harding negative dentry 1411af96c1e3STobin C. Harding 1412ee5dc049STobin C. Harding``d_lookup`` 1413ee5dc049STobin C. Harding look up a dentry given its parent and path name component It 1414ee5dc049STobin C. Harding looks up the child of that given name from the dcache hash 1415ee5dc049STobin C. Harding table. If it is found, the reference count is incremented and 1416ee5dc049STobin C. Harding the dentry is returned. The caller must use dput() to free the 1417ee5dc049STobin C. Harding dentry when it finishes using it. 1418af96c1e3STobin C. Harding 1419af96c1e3STobin C. Harding 1420af96c1e3STobin C. HardingMount Options 1421af96c1e3STobin C. Harding============= 1422af96c1e3STobin C. Harding 1423af96c1e3STobin C. Harding 1424af96c1e3STobin C. HardingParsing options 1425af96c1e3STobin C. Harding--------------- 1426af96c1e3STobin C. Harding 1427af96c1e3STobin C. HardingOn mount and remount the filesystem is passed a string containing a 1428af96c1e3STobin C. Hardingcomma separated list of mount options. The options can have either of 1429af96c1e3STobin C. Hardingthese forms: 1430af96c1e3STobin C. Harding 1431af96c1e3STobin C. Harding option 1432af96c1e3STobin C. Harding option=value 1433af96c1e3STobin C. Harding 1434af96c1e3STobin C. HardingThe <linux/parser.h> header defines an API that helps parse these 1435af96c1e3STobin C. Hardingoptions. There are plenty of examples on how to use it in existing 1436af96c1e3STobin C. Hardingfilesystems. 1437af96c1e3STobin C. Harding 1438af96c1e3STobin C. Harding 1439af96c1e3STobin C. HardingShowing options 1440af96c1e3STobin C. Harding--------------- 1441af96c1e3STobin C. Harding 1442af96c1e3STobin C. HardingIf a filesystem accepts mount options, it must define show_options() to 1443af96c1e3STobin C. Hardingshow all the currently active options. The rules are: 1444af96c1e3STobin C. Harding 1445af96c1e3STobin C. Harding - options MUST be shown which are not default or their values differ 1446af96c1e3STobin C. Harding from the default 1447af96c1e3STobin C. Harding 1448af96c1e3STobin C. Harding - options MAY be shown which are enabled by default or have their 1449af96c1e3STobin C. Harding default value 1450af96c1e3STobin C. Harding 1451af96c1e3STobin C. HardingOptions used only internally between a mount helper and the kernel (such 1452af96c1e3STobin C. Hardingas file descriptors), or which only have an effect during the mounting 1453af96c1e3STobin C. Harding(such as ones controlling the creation of a journal) are exempt from the 1454af96c1e3STobin C. Hardingabove rules. 1455af96c1e3STobin C. Harding 1456af96c1e3STobin C. HardingThe underlying reason for the above rules is to make sure, that a mount 1457af96c1e3STobin C. Hardingcan be accurately replicated (e.g. umounting and mounting again) based 1458af96c1e3STobin C. Hardingon the information found in /proc/mounts. 1459af96c1e3STobin C. Harding 1460af96c1e3STobin C. Harding 1461af96c1e3STobin C. HardingResources 1462af96c1e3STobin C. Harding========= 1463af96c1e3STobin C. Harding 1464af96c1e3STobin C. Harding(Note some of these resources are not up-to-date with the latest kernel 1465af96c1e3STobin C. Harding version.) 1466af96c1e3STobin C. Harding 1467af96c1e3STobin C. HardingCreating Linux virtual filesystems. 2002 1468c69f22f2SAlexander A. Klimov <https://lwn.net/Articles/13325/> 1469af96c1e3STobin C. Harding 1470af96c1e3STobin C. HardingThe Linux Virtual File-system Layer by Neil Brown. 1999 1471af96c1e3STobin C. Harding <http://www.cse.unsw.edu.au/~neilb/oss/linux-commentary/vfs.html> 1472af96c1e3STobin C. Harding 1473af96c1e3STobin C. HardingA tour of the Linux VFS by Michael K. Johnson. 1996 1474c69f22f2SAlexander A. Klimov <https://www.tldp.org/LDP/khg/HyperNews/get/fs/vfstour.html> 1475af96c1e3STobin C. Harding 1476af96c1e3STobin C. HardingA small trail through the Linux kernel by Andries Brouwer. 2001 1477c69f22f2SAlexander A. Klimov <https://www.win.tue.nl/~aeb/linux/vfs/trail.html> 1478