【正文】 ture on mount ? distinction: disk superblock vs memory superblock ? dirty bit (s_dirt), copied to disk frequently ? important fields ? s_dev, s_bdev – device, devicedriver ? s_blocksize, s_maxbytes, s_type ? s_flags, s_magic, s_count, s_root, s_dquot ? s_dirty – dirty inodes for this filesystem ? s_op – superblock operations ? u – filesystem specific data ? 參見(jiàn) include/linux/ VFS 超級(jí)塊包含的信息 ? Device 這是包含文件系統(tǒng)的塊設(shè)備的設(shè)備標(biāo)識(shí)符。例如 ,/dev/hda1,設(shè)備標(biāo)識(shí)符是 0x301 ? Inode pointers 其中的 mounted inode指針指向該文件系統(tǒng)的第一個(gè) inode。Covered inode指針指向文件系統(tǒng)安裝到的目錄的 inode 。 ? Blocksize 文件系統(tǒng)塊的字節(jié)大小。 ? Superblock operations 指向一組本文件系統(tǒng)超級(jí)塊例程的指針。除了其他類(lèi)型之外， VFS 使用這些例程讀寫(xiě) inode 和超級(jí)塊 ? File System type 指向這個(gè)安裝的文件系統(tǒng)的 file_system_type 數(shù)據(jù)結(jié)構(gòu)的一個(gè)指針 ? File System Specific 指向這個(gè)文件系統(tǒng)需要的信息的一個(gè)指針 超級(jí)塊的操作 ? filesystemspecific operations ? read/write/clear/delete inode ? write_super, put_super (release) ? no get_super()! that lives in file_system_type descriptor ? write_super_lockfs, unlockfs, statfs ? file_handle ops (NFSrelated) ? show_options Inode ? index node – unique file or directory descriptor ? metadata: permissions, owner, timestamps, size, link count ? data: pointers to disk blocks containing actual data ?data pointers are indices into file contents (hence inode) ? inode unique integer (permounted filesystem) ? what about names and paths? ? highlevel fluff on top of a flatfilesystem ? implemented by directory files (directories) ? directory contents: name + inode Inode The VFS Inode ? 同 EXT2 文件系統(tǒng)， VFS 中每一個(gè)文件、目錄等等都用一個(gè)且只用一個(gè) VFS inode 代表。 ? 每一個(gè) VFS inode 中的信息使用文件系統(tǒng)相關(guān)的例程從底層的文件系統(tǒng)中獲取。 ? VFS inode 只在核心的內(nèi)存中存在，只要對(duì)系統(tǒng)有用，就一直保存在 VFS inode cache 中。 ? 參見(jiàn) include/linux/ The VFS Inode 包含的信息 ? device 存放這個(gè)文件的設(shè)備的設(shè)備標(biāo)識(shí)符。 ? Inode nunber 這個(gè) inode 的編號(hào)。 ? Mode 象 EXT2 一樣，這個(gè)域描述這個(gè) VFS inode 代表的東西和對(duì)它的訪問(wèn)權(quán)限。 ? Inode operations 指向一組例程地址的指針。例程和文件系統(tǒng)相關(guān) ,執(zhí)行對(duì)于此 inode的操作 ? Count 系統(tǒng)組件當(dāng)前使用這個(gè) VFS inode 的數(shù)目。 Count 0 意味著這個(gè) inode 是空閑，可以廢棄或者重用。 ? Lock 這個(gè)域用于鎖定這個(gè) VFS inode 。例如當(dāng)從文件系統(tǒng)讀取它的時(shí)候。 ? Dirty 顯示這個(gè) VFS inode 是否被寫(xiě)過(guò)，如果這樣，底層的文件系統(tǒng)需要更新。 Inode Fields ? large struct (~50 fields) ? important fields ? i_sb, i_ino (number), i_nlink (link count) ? metadata: i_mode, i_uid, i_gid, i_size, i_times ? i_flock (lock list), i_wait (waitq – for blocking ops) ? linkage: i_hash, i_list, i_dentry (aliases) ? i_op (inode ops), i_fop (default file ops) ? u (filesystem specific data – includes block map!) Inode Operations ? create – new inode for regular file ? link/unlink/rename – ? add/remove/modify dir entry ? symlink, readlink, follow_link – soft link ops ? mkdir/rmdir – new inode for directory file ? mknod – new inode for device file ? truncate – modify file size ? permission – check access permissions (Open) File Object ? struct file (usual variable name filp) ? association between file and process ? no disk representation ? created for each open (multiple possible, even same file) ? most important info: file pointer ? file descriptor (small ints) ? index into array of pointers to open file objects ? file object states ? unused (memory cache + root reserve (10)) ? get_empty_filp() ? inuse (persuperblock lists) ? systemwide max on open file objects (~8K) ? /proc/sys/fs/filemax File Object Fields ? important fields ? f_dentry (directory file is contained in) ? f_vfsmnt (fs mount point) ? f_op (fsspecific functions – table of function pointers) ? f_count, f_flags, f_mode (r/w, permissions, etc.) ? f_pos (current position – file pointer) ? info for readahead (more later) ? f_uid, f_gid, f_owner ? f_version (for consistency maintenance) ? private_data (fsspecific data) File Object Operations ? f_op field – table of function pointers ? copied from inode (i_fop) initially (fsspecific) ? possible to change to customize (peropen) ? devicedrivers do some tricks like this sometimes ? important operations ? llseek(), read(), write(), readdir(), poll() ? ioctl() – wildcard function for perfs semantics ? mmap(), open(), flush(), release(), fsync() ? fasync() – turn on/off asynchronous i/o notifications ? lock() – filelocks (more later) ? readv(), writev() – scatter/gather i/o ? read/write with discontiguous buffers (. packets) ? sendpage() – pageoptimized socket transfer 文件連接 ? UNIX link semantics ? hard links硬連接 – multiple dir entries with same inode ? equal status。 first is not real entry ? file deleted when link count goes to 0 ? restrictions ? can39。t hard link to directories (avoids cycles) ? or across filesystems ? soft (symbolic) links軟（符號(hào)）連接 – little files with pathnames ? just aliases for another pathname ? no restrictions, cycles possible, dangling links possible 硬連接 ?ln /a/ /b/ 符號(hào)連接 ?ln –s /a/ /b/ Dentry ( 1/2 ) ? abstraction of directory entry ? ex: line from ls l ? either files (hard links) or soft links or subdirectories ? every dentry has a parent dentry (except root) ? sibling dentries – other entries in the same directory ? directory api: dentry iterators ? posix: opendir(), readdir(), scandir(), seekdir(), rewinddir() ? syscall: getdents() ? why an abstraction? ? UNIX: directories are really files with directory records ? MSDOS, etc.: directory is just a big table on disk (FAT) ? no such thing as subdirectories! ? just fields in table (fileparentdir), (dirparentdir) Dentry ( 2/2 ) ? notdisk based (no dirty bit) ? dentry_cache – slab cache ? consistency maintenance using version numbers (later) ? important fields ? d_name, d_count, d_flags ? d_inode – associated inode ? d_parent – parent dentry ? d_child – siblings list ? d_subdirs – my children (if i39。m a subdirectory) ? d_alias – other names (links) for the same object (inode)? ? d_lru – unused state linkage ? d_op – dentry operations (function pointer tabl