.\" $NetBSD: vfsops.9,v 1.3 2001/12/27 13:20:53 wiz Exp $ .\" .\" Copyright (c) 2001 The NetBSD Foundation, Inc. .\" All rights reserved. .\" .\" This code is derived from software contributed to The NetBSD Foundation .\" by Gregory McGarry. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice, this list of conditions and the following disclaimer in the .\" documentation and/or other materials provided with the distribution. .\" 3. All advertising materials mentioning features or use of this software .\" must display the following acknowledgement: .\" This product includes software developed by the NetBSD .\" Foundation, Inc. and its contributors. .\" 4. Neither the name of The NetBSD Foundation nor the names of its .\" contributors may be used to endorse or promote products derived .\" from this software without specific prior written permission. .\" .\" THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS .\" ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED .\" TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR .\" PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS .\" BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR .\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF .\" SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS .\" INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN .\" CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) .\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE .\" POSSIBILITY OF SUCH DAMAGE. .\" .Dd October 6, 2001 .Dt VFSOPS 9 .Os .Sh NAME .Nm vfsops , .Nm vfs_rootmountalloc , .Nm vfs_export_lookup , .Nm vfs_getnewfsid , .Nm VFS_MOUNT , .Nm VFS_START , .Nm VFS_UNMOUNT , .Nm VFS_ROOT , .Nm VFS_QUOTACTL , .Nm VFS_STATFS , .Nm VFS_STATFS , .Nm VFS_SYNC , .Nm VFS_VGET , .Nm VFS_FHTOVP , .Nm VFS_VPTOFH , .Nm VFS_CHECKEXP .Nd kernel file system interface .Sh SYNOPSIS .Fd #include .Fd #include .Fd #include .Ft int .Fn vfs_rootmountalloc "char *fstypename" "char *devname" \ "struct mount **mpp" .Ft int .Fn vfs_export_lookup "struct mount *mp" "struct netexport *nep" \ "struct export_args *argp" .Ft void .Fn vfs_getnewfsid "struct mount *mp" .Ft int .Fn VFS_MOUNT "struct mount *mp" "const char *path" "void *data" \ "struct nameidata *ndp" "struct proc *p" .Ft int .Fn VFS_START "struct mount *mp" "int flags" "struct proc *p" .Ft int .Fn VFS_UNMOUNT "struct mount *mp" "int mntflags" "struct proc *p" .Ft int .Fn VFS_ROOT "struct mount *mp" "struct vnode **vpp" .Ft int .Fn VFS_QUOTACTL "struct mount *mp" "int cmds" "uid_t uid" \ "caddr_t arg" "struct proc *p" .Ft int .Fn VFS_STATFS "struct mount *mp" "struct statfs *sbp" "struct proc *p" .Ft int .Fn VFS_SYNC "struct mount *mp" "int waitfor" "struct ucred *cred" \ "struct proc *p" .Ft int .Fn VFS_VGET "struct mount *mp" "ino_t ino" "struct vnode **vpp" .Ft int .Fn VFS_FHTOVP "struct mount *mp" "struct fid *fhp" "struct vnode **vpp" .Ft int .Fn VFS_VPTOFH "struct vnode *vp" "struct fid *fhp" .Ft int .Fn VFS_CHECKEXP "struct mount *mp" "struct mbuf *nam" \ "int *extflagsp" "struct ucred **credanonp" .Sh DESCRIPTION In a similar fashion to the .Xr vnode 9 interface, all operations that are done on a file system are conducted through a single interface that allows the system to carry out operations on a file system without knowing its construction or type. .Pp All supported file systems in the kernel have an entry in the .Va vfs_list_initial table. This table is generated by .Xr config 8 and is a NULL-terminated list of .Em vfsops structures. The vfsops structure describes the operations that can be done to a specific file system type. The following table list the elements of the vfsops vector, the corresponding invocation macro, and a description of the element. .Pp .nf .ta \w'int (*vfs_mountroot)()'u+2n +\w'VFS_QUOTACTL'u+2n +\w'Get the file system root vnode'u \fIVector element\fP \fIMacro\fP \fIDescription\fP .ta \w'int (*vfs_mountroot)()'u+2n +\w'VFS_QUOTACTL'u+2n +\w'Get the file system root vnode'u+6nC .sp 5p int (*vfs_mount)() VFS_MOUNT Mount a file system int (*vfs_start)() VFS_START Make operational int (*vfs_unmount)() VFS_UMOUNT Unmount a file system int (*vfs_root)() VFS_ROOT Get the file system root vnode int (*vfs_quotactl)() VFS_QUOTACTL Query/modify space quotas int (*vfs_statfs)() VFS_STATFS Get file system statistics int (*vfs_sync)() VFS_SYNC Flush file system buffers int (*vfs_vget)() VFS_VGET Get vnode from file ID int (*vfs_fhtovp)() VFS_FHTOVP NFS file handle to vnode lookup int (*vfs_vptofh)() VFS_VPTOFH Vnode to NFS file handle lookup void (*vfs_init)() - Initialise file system void (*vfs_reinit)() - Reinitialise file system void (*vfs_done)() - Cleanup unmounted file system int (*vfs_sysctl)() - Query/modify kernel state int (*vfs_mountroot)() - Mount the root file system int (*vfs_checkexp)() VFS_CHECKEXP Check if fs is exported .fi .Pp Some additional non-function members of the vfsops structure are the file system name .Ns Em vfs_name and a reference count .Ns Em vfs_refcount . It is not mandatory for a file system type to support a particular operation, but it must assign each member function pointer to a suitable function to do the minimum required of it. In most cases, such functions either do nothing or return an error value to the effect that it is not supported. .Pp At system boot, each file system with an entry in .Va vfs_list_initial is established and initialised. Each initialised file system is recorded by the kernel in the list .Va vfs_list and the file system specific initialisation function .Em vfs_init in its vfsops vector is invoked. When the file system is not longer needed .Em vfs_done is invoked to run file system specific cleanups and the file system is removed from the kernel list. .Pp At system boot, the root filesystem is mounted by invoking the file system type specific .Em vfs_mountroot function in the vfsops vector. All filesystems that can be mounted as a root file system must define this function. It is responsible for initialising to list of mount structures for all future mounted file systems. .Pp Kernel state which affects a specific file system type can be queried and modified using the .Xr sysctl 8 interface. The .Em vfs_sysctl member of the vfsops structure is invoked by file system independent code for sysctl operations which are file system specific. .Sh FUNCTIONS .Bl -tag -width compact .It Fn vfs_rootmountalloc "fstypename" "devname" "mpp" Lookup a file system type specied by the name .Fa fstypename and if found allocate and initialise a mount structure for it. The allocated mount structure is returned in the address specified by .Fa mpp . The device the root file system was mounted from is specified by the argument .Fa devname and is recorded in the new mount structure. .It Fn vfs_export_lookup "mp" "nep" "argp" Check client permission on the exportable file system specified by the mount structure .Fa mp . The argument .Fa nam is the address of the networked client. This function is used by file system type specific functions to verify that the client can access the file system. .It Fn vfs_getnewfsid "mp" Get a new unique file system id type for the file system specified by the mount structure .Fa mp . The file system id type is stored in .Em mp->mnt_stat.f_fsid . .It Fn VFS_MOUNT "mp" "path" "data" "ndp" "p" Mount a file system specified by the mount structure .Fa mp on the mount point described by .Fa path . The argument .Fa data contains file system type specific data and is read into the kernel using .Xr copyin 9 . The argument .Fa ndp contains the result of a .Xr namei 9 call on the pathname of the mount point and .Fa p is the calling process. .Fn VFS_MOUNT initialises the mount structure for the mounted file system. This structure records mount-specific information for the file system and records the list of vnodes associated with the file system. This function is invoked both to mount new file systems and to change the attributes of an existing file system. If the MNT_UPDATE flag is set in .Em mp->mnt_flag then the filesystem should update its internal state from the value of .Em mp->mnt_flag . This can be used, for instance, to convert a read-only filesystem to read-write. If the MNT_UPDATE flag is not specified, then this is a newly mounted filesystem. .It Fn VFS_START " mp" "flags" "p" Make the file system specified by the mount structure .Fa mp operational. The argument .Fa p is the calling processing. The argument .Fa flags is a set of flags for controlling the operation of .Fn VOP_START . This function is invoked after .Fn VFS_MOUNT and before the first acces to the file system. .It Fn VFS_UNMOUNT "mp" "mntflags" "p" Unmount a file system specified by the mount structure .Fa mp . The argument .Fa p is the calling process. .Fa VFS_UNMOUNT performs any file system type specific operations required before the file system is unmounted, such are flushing buffers. If MNT_FORCE is specified in the flags .Fa mntflags then open files are forcibly closed. The function also deallocates space associated with data structure that were allocated for the file system when it was mounted. .It Fn VFS_ROOT "mp" "vpp" Get the root vnode of the file system specified by the mount structure .Fa mp . The vnode is returned in the address given by .Fa vpp . This function is used by the pathname translation algorithms when a vnode that has been covered by a mounted file system is encountered. While resolving the pathname, the pathname translation algorithm will have to go through the directory tree in the file system associated with that mount point and therefore requires the root vnode of the file system. .It Fn VFS_QUOTACTL "mp" "cmds" "uid" "arg" "p" Query/modify user space quotas for the file system specified by the mount structure .Fa mp . The argument specifies the control command to perform. The userid is specified in .Fa id , the calling process is .Fa p and .Fa arg allows command-specific data to be returned to the system call interface. .Fn VFS_QUOTACTL is the file system type specific implementation of the .Xr quotactl 2 system call. .It Fn VFS_STATFS "mp" "sbp" "p" Get file system statistics for the file system specified by the mount structure .Fa mp . The argument .Fa p is the calling processing. A statfs structure filled with the statistics is returned in .Fa sbp . .Fn VFS_STATFS is the file system type specific implementation of the .Xr statfs 2 and .Xr fstatfs 2 system calls. .It Fn VFS_SYNC "mp" "waitfor" "cred" "p" Flush file system I/O buffers for the file system specified by the mount structure .Fa mp . The .Fa waitfor argument indicates whether a partial flush or complete flush should be performed. The arguments .Fa p and .Fa cred specific the calling process and its credentials respectively. .Fn VFS_SYNC does not provide any return value since the operation can never fail. .It Fn VFS_VGET "mp" "ino" "vpp" Get vnode for a file system type specific file id .Fa ino for the file system specified by the mount structure .Fa mp . The vnode is returned in the address specified .Fa vpp . The function is optional for file systems which have a unique id number for every file in the file system. It is used internally by the UFS file system and also by the NFSv3 server to implement the READDIRPLUS nfs call. If the file system does not support this invoked, it should return EOPNOTSUPP. .It Fn VFS_FHTOVP "mp" "fhp" "vpp" Get the vnode for the NFS file specified by the file handle .Fa fhp in the file system specified by the mount structure .Fa mp . The locked vnode is returned in .Fa vpp . .Pp A call to .Fa VFS_FHTOVP should generally be followed by a call to .Fn VFS_CHECKEXP to check if the file is accessable to the client. .It Fn VFS_VPTOFH "vp" "fhp" Get a unique NFS file handle for the file specified by the vnode .Fa vp . The file handle is returned in .Fa fhp . The contents of the file handle are defined by the file system and are not examined by an eother subsystems. It should contain enough information to uniquely identify a file within the file system as well as noticing when a file has been removed and the file system resources have been recycled for a new file. .It Fn VFS_CHECKEXP "mp" "nam" "extflagsp" "credanonp" Check if the file system specified by the mount structure .Fa mp is exported to a client with anonymous credentials .Fa credanonp . The argument .Fa nam is an mbuf containing the network address of the client. The return parameters for the export flags for the client are returned in the address specified by .Fa exflagsp . This function is used by the NFS server. It is generally invoked before .Fn VFS_FHTOVP to validate that client has access to the file system. The file system should call .Fn vfs_export_lookup with the address of an appropriate netexport structure and the address of the client to verify that the client can access this file system. .El .Sh CODE REFERENCES This section describes places within the .Nx source tree where actual code implementing or utilising the vfs operations can be found. All pathnames are relative to .Pa /usr/src . .Pp The vfs operations are implemented within the files .Pa sys/kern/vfs_subr.c and .Pa sys/kern/vfs_init.c . .Sh SEE ALSO .Xr intro 9 , .Xr namei 9 , .Xr vfs 9 , .Xr vnode 9 .Sh HISTORY The vfs operations vector, its functions and the corresponding macros appeared in .Bx 4.3 .