FILE SYSTEM FRAMEWORK —— virtual file system framework

1. FILE SYSTEM FRAMEWORK—— virtual file system framework

2. Outline • Solaris File System Framework • The vnode • The vfs Object

3. SolarisFILE SYSTEM FRAMEWORK • Solaris virtual file system framework • the virtual file system framework implementes multiple file system types. • It allows Sun’s distributed computing file system (NFS) to coexist with the UFS file system in SunOS 2.0 • Solaris file systems can be categorized into the following types: • Storage based — Regular file systems . The Solaris UFS and PC/DOS file systems are examples. • Network file systems—for example, NFS • Pseudo file systems —The /proc pseudo file systemis example.

4. Unified File System Interface • The framework provides a single set of well-defined interfaces . • Two key objects represent these interfaces: • the virtual file, or vnode: The vnode interfaces implement file-related functions. • the virtual file system, or vfs objects:the vfs interfaces implement file system management functions.

5. the file system layers is shown below T O P E N D B O T T O M E N D Figure 13.1 shows the file system layers.

7. File System Framework Facilities • The vnode/vfs interfaces • The “top end” of the file system module implement vnode and vfs objects. • The “bottom end” of the file system uses other kernel interfaces to access, store, and cache the data they represent. • Disk-based file systems interface to device drivers to provide persistent storage of their data. • they interface to network file systems access the networking subsystem to transmit and receive data to remote systems. • Pseudo file systems typically access local kernel functions and structures to gather the information they represent.

8. File System Framework Facilities

9. File System Framework Facilities • Loadable file system modules are dynamically loaded at the time each file system type is first mounted. • The vnode/vfs framework implementes file functions and file system management functions. • File system caching implements caching interface with the HAT layer of the virtual memory system to map, unmap, and manage the memory used for caching. • Path-name managementconverts paths into vnode pointers. • Directory name cachingprovides a mechanism to cache pathname-to-vnode mappings.

10. The vnode • A vnode is a representation of a file in the Solaris kernel. • The vnode is said to be objectlike . • it is an encapsulation of a file’s state and the methods that can be used to perform operations on that file. • The vnode hides the implementation of the file system and exposes file system-independent data and methods for that file to the rest of the kernel.

11. A vnode object Figure 13.2A vnode object

12. A vnode object contains three important items • File-system-independent data • the type of vnode :file, directory, character device, Block device, Hard link,Named pipe, etc. • flags of vnode : state, pointers to the file system that contains the vnode, a reference count to the vnode. • Functions to implement file methods • A structure of pointers to file-system-dependent functions, to implement file’s open(),close(), read(), and write(). • File-system-specific data • Data that is used internally by each file system implementation; typically the in-memory inode . UFS uses an inode, NFS uses an rnode,and tmpfs uses a tmpnode.

13. A vnode object • For example, to read from a file without knowing that it resides on a UFS file system, the kernel would simply call the file-system-independent macro for read(), VOP_READ(), which would call the vop_read() method of the vnode, which in turn calls the UFS function, ufs_read().

14. The data structure of a vnode • typedef struct vnode { • kmutex_t v_lock; /* protects vnode fields */ushort_t v_flag; /* vnode flags (see below) */uint_t v_count; /* reference count */struct vfs *v_vfsmountedhere; /* ptr to vfs mounted here */struct vnodeops *v_op; /* vnode operations */struct vfs *v_vfsp; /* ptr to containing VFS */struct stdata *v_stream; /* associated stream */struct page *v_pages; /* vnode pages list */enum vtype v_type; /* vnode type */dev_t v_rdev; /* device (VCHR, VBLK) */caddr_t v_data; /* private data for fs */struct filock *v_filocks; /* ptr to filock list */struct shrlocklist *v_shrlocks; /* ptr to shrlock list */kcondvar_t v_cv; /* synchronize locking */} vnode_t;

15. Vnode Methods • The vnode interfaceprovides the set of file system object methods • The Vnode Methods perform all file-system-specific file operations. • The figure is shown below.

16. Vnode Methods

17. vnode Reference Count • A vnode is created by the file system at the time a file is first opened or created and stays active until the file system decides the vnode is no longer needed. • The vnode framework provides an infrastructure that keeps track of the number of references to a vnode. • It is important to distinquish a vnode reference from a lock: • A lock ensures exclusive access to the data, • the reference count ensures persistence of the object.

18. Interfaces for Paging vnode Cache • Solaris unifies file and memory management by using a vnode to represent the backing store for virtual memory. • A page of memory represents a particular vnode and offset. • The file system uses the memory relationship to implement caching for vnodes within a file system. • The virtual memory system provides a set of functions for cache management and I/O for vnodes.

19. Block I/O on vnode Pages • The block I/O subsystem provides Three functions for initiating I/O to and from vnode pages. • The table shows to initiate I/O between a physical page and a device:

20. The vfs Object • The vfs layer provides an administrative interface into the file system to support commands like mount and umount in a file-system-independent manner. • The interface achieves independence by means of a virtual file system (vfs) object. • The vfs object represents an encapsulation of a file system’s state and a set of methods for each of the file system administrative interfaces.

21. Structure per mounted file system typedef struct vfs { struct vfs *vfs_next; /* next VFS in VFS list */ struct vfsops *vfs_op; /* operations on VFS */ struct vnode *vfs_vnodecovered; /* vnode mounted on */ uint_t vfs_flag; /* flags */ uint_t vfs_bsize; /* native block size */ int vfs_fstype; /* file system type index */ fsid_t vfs_fsid; /* file system id */ caddr_t vfs_data; /* private data */ dev_t vfs_dev; /* device of mounted VFS */ ulong_t vfs_bcount; /* I/O count (accounting) */ ushort_t vfs_nsubmounts; /* immediate sub-mount count */ struct vfs *vfs_list; /* sync list pointer */ struct vfs *vfs_hash; /* hash list pointer */ ksema_t vfs_reflock; } /* mount/unmount/sync lock */

22. Operations supported on virtual file system • typedef struct vfsops { • int (*vfs_mount)(struct vfs *, struct vnode *, struct mounta *, • struct cred *); • int (*vfs_unmount)(struct vfs *, struct cred *); • int (*vfs_root)(struct vfs *, struct vnode **); • int (*vfs_statvfs)(struct vfs *, struct statvfs64 *); • int (*vfs_sync)(struct vfs *, short, struct cred *); • int (*vfs_vget)(struct vfs *, struct vnode **, struct fid *); • int (*vfs_mountroot)(struct vfs *, enum whymountroot); • int (*vfs_swapvp)(struct vfs *, struct vnode **, char *); • }

25. 启动时UFS的挂载分析 • 在Solaris的整个文件系统体系结构中，所有对于具体文件系统的操作都是通过VFS来进行抽象的。为了分析UFS中的操作，就必须找到VFS与UFS的结合部 。

26. 挂载相关函数 • 在挂载UFS时，用到的最重要的两个函数是vfs_mountroot()和rootconf()。其中vfs_mountroot()的函数原型定义在src\uts\common\fs\vfs.c中

27. void vfs_mountroot(void)； • 函数功能：挂载根文件系统 • 函数参数：无 • 函数返回值：无 • rootconf()函数的原型也定义在src\uts\common\fs\vfs.c中： • int rootconf()； • 函数功能：选取一种文件系统，作为vfs的根文件系统 • 函数参数：无 • 函数返回值：返回错误号

28. 挂载过程分析 • 在系统启动时，uts\common\os的main.c中调用了vfs_mountroot函数，这个函数的作用就是挂载根文件系统。

29. vfs_mountroot • if (rootconf()) • cmn_err(CE_PANIC, "vfs_mountroot: cannot mount root"); • /* • * Get vnode for '/'. Set up rootdir, u.u_rdir and u.u_cdir • * to point to it. These are used by lookuppn() so that it • * knows where to start from ('/' or '.'). • */ • vfs_setmntpoint(rootvfs, "/"); • if (VFS_ROOT(rootvfs, &rootdir)) • cmn_err(CE_PANIC, "vfs_mountroot: no root vnode"); • u.u_cdir = rootdir; • VN_HOLD(u.u_cdir); • u.u_rdir = NULL;

30. rootconf函数选取了一种文件系统，作为vfs的根文件系统。

31. rootconf • 调用getrootfs（）；它判断启动的是否是一个无disk的客户端，如果不是，就返回”ufs”； • 调用modload（”fs”, fstype）；它完成ufs模块的装载，其中的参数fstype就是上面getrootfs返回的”ufs”； • 调用vsw = vfs_getvfsswbyname(fstyp)；

32. VSW • struct vfssw { • char *vsw_name; /* type name -- max len _ST_FSTYPSZ */ • int (*vsw_init) (int, char *); • /* init routine (for non-loadable fs only) */ • int vsw_flag; /* flags */ • mntopts_t vsw_optproto; /* mount options table prototype */ • uint_t vsw_count; /* count of references */ • kmutex_t vsw_lock; /* lock to protect vsw_count */ • vfsops_t vsw_vfsops; /* filesystem operations vector */ • }

33. vsw_vfsops包含了一系列对文件系统的基本操作，是一个由函数指针构成的结构体。vfs_getvfsswbyname(fstyp)函数根据传进来的文件系统的类型名字（这里就是fstype==”ufs”），来找到该类型文件系统的vsw结构体。其实在vfsconf.c中定义了一个表struct vfssw vfssw[]，里面的每一项都是对于一种文件系统的描述（用vfssw结构）。vfs_getvfsswbyname(fstyp)函数就是根据fstype来查找对应的vfssw结构，并返回其指针给vsw。

34. 调用VFS_INIT(rootvfs, &vsw->vsw_vfsops, 0)；这个宏完成的工作是将rootvfs这个vfs结构体中的相关函数指针与vsw->vsw_vfsops相关联。 • 调用error = VFS_MOUNTROOT(rootvfs, ROOT_INIT)；这里VFS_MOUNTROOT宏只是调用了rootvfs中的mountroot函数，由于刚才已经完成了rootvfs同UFS的关联，所以这里也就是调用了ufs_mountroot。

35. vfssw[] 0 1 2 3 4 5 6 7 …. struct vfssw struct vfssw struct vfssw vsw_name: proc vsw_name: nfs vsw_name: ufs vsw_vfsops vsw_vfsops vsw_vfsops struct vfsops struct vfsops struct vfsops vfs_mount vfs_mount vfs_mount vfs_umount vfs_umount vfs_umount …. …. …. struct vfs struct vfs struct vfs vfs_op vfs_op vfs_op vfs_next vfs_next vfs_next vfs_prev vfs_prev vfs_prev vfs_fstype: proc vfs_fstype: nfs vfs_fstype: ufs vfs_vnodecovered vfs_vnodecovered vfs_vnodecovered …. …. …. struct vnode struct vnode struct vnode v_vfsmountedhere v_vfsmountedhere v_vfsmountedhere …. …. ….