vfs_init.c revision 138509
1/*
2 * Copyright (c) 1989, 1993
3 *	The Regents of the University of California.  All rights reserved.
4 *
5 * This code is derived from software contributed
6 * to Berkeley by John Heidemann of the UCLA Ficus project.
7 *
8 * Source: * @(#)i405_init.c 2.10 92/04/27 UCLA Ficus project
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 * 1. Redistributions of source code must retain the above copyright
14 *    notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 *    notice, this list of conditions and the following disclaimer in the
17 *    documentation and/or other materials provided with the distribution.
18 * 4. Neither the name of the University nor the names of its contributors
19 *    may be used to endorse or promote products derived from this software
20 *    without specific prior written permission.
21 *
22 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32 * SUCH DAMAGE.
33 *
34 *	@(#)vfs_init.c	8.3 (Berkeley) 1/4/94
35 */
36
37#include <sys/cdefs.h>
38__FBSDID("$FreeBSD: head/sys/kern/vfs_init.c 138509 2004-12-07 08:15:41Z phk $");
39
40#include <sys/param.h>
41#include <sys/systm.h>
42#include <sys/kernel.h>
43#include <sys/linker.h>
44#include <sys/mount.h>
45#include <sys/proc.h>
46#include <sys/sysctl.h>
47#include <sys/vnode.h>
48#include <sys/malloc.h>
49
50
51MALLOC_DEFINE(M_VNODE, "vnodes", "Dynamically allocated vnodes");
52
53/*
54 * The highest defined VFS number.
55 */
56int maxvfsconf = VFS_GENERIC + 1;
57
58/*
59 * Single-linked list of configured VFSes.
60 * New entries are added/deleted by vfs_register()/vfs_unregister()
61 */
62struct vfsconfhead vfsconf = TAILQ_HEAD_INITIALIZER(vfsconf);
63
64/*
65 * A Zen vnode attribute structure.
66 *
67 * Initialized when the first filesystem registers by vfs_register().
68 */
69struct vattr va_null;
70
71/*
72 * vfs_init.c
73 *
74 * Allocate and fill in operations vectors.
75 *
76 * An undocumented feature of this approach to defining operations is that
77 * there can be multiple entries in vfs_opv_descs for the same operations
78 * vector. This allows third parties to extend the set of operations
79 * supported by another layer in a binary compatibile way. For example,
80 * assume that NFS needed to be modified to support Ficus. NFS has an entry
81 * (probably nfs_vnopdeop_decls) declaring all the operations NFS supports by
82 * default. Ficus could add another entry (ficus_nfs_vnodeop_decl_entensions)
83 * listing those new operations Ficus adds to NFS, all without modifying the
84 * NFS code. (Of couse, the OTW NFS protocol still needs to be munged, but
85 * that is a(whole)nother story.) This is a feature.
86 */
87
88/*
89 * Routines having to do with the management of the vnode table.
90 */
91
92/*
93 * XXX: hack alert
94 */
95int
96vcall(struct vnode *vp, u_int off, void *ap)
97{
98	struct vop_vector *vop = vp->v_op;
99	vop_bypass_t **bpt;
100	int rc;
101
102	for(;;) {
103		bpt = (void *)((u_char *)vop + off);
104		if (vop != NULL && *bpt == NULL && vop->vop_bypass == NULL) {
105			vop = vop->vop_default;
106			continue;
107		}
108		break;
109	}
110	KASSERT(vop != NULL, ("No VCALL(%p...)", vp));
111	if (*bpt != NULL)
112		rc = (*bpt)(ap);
113	else
114		rc = vop->vop_bypass(ap);
115	return (rc);
116}
117
118struct vfsconf *
119vfs_byname(const char *name)
120{
121	struct vfsconf *vfsp;
122
123	if (!strcmp(name, "ffs"))
124		name = "ufs";
125	TAILQ_FOREACH(vfsp, &vfsconf, vfc_list)
126		if (!strcmp(name, vfsp->vfc_name))
127			return (vfsp);
128	return (NULL);
129}
130
131struct vfsconf *
132vfs_byname_kld(const char *fstype, struct thread *td, int *error)
133{
134	struct vfsconf *vfsp;
135	linker_file_t lf;
136
137	vfsp = vfs_byname(fstype);
138	if (vfsp != NULL)
139		return (vfsp);
140
141	/* Only load modules for root (very important!). */
142	*error = suser(td);
143	if (*error)
144		return (NULL);
145	*error = securelevel_gt(td->td_ucred, 0);
146	if (*error)
147		return (NULL);
148	*error = linker_load_module(NULL, fstype, NULL, NULL, &lf);
149	if (lf == NULL)
150		*error = ENODEV;
151	if (*error)
152		return (NULL);
153	lf->userrefs++;
154	/* Look up again to see if the VFS was loaded. */
155	vfsp = vfs_byname(fstype);
156	if (vfsp == NULL) {
157		lf->userrefs--;
158		linker_file_unload(lf, LINKER_UNLOAD_FORCE);
159		*error = ENODEV;
160		return (NULL);
161	}
162	return (vfsp);
163}
164
165
166/* Register a new filesystem type in the global table */
167int
168vfs_register(struct vfsconf *vfc)
169{
170	struct sysctl_oid *oidp;
171	struct vfsops *vfsops;
172	static int once;
173
174	if (!once) {
175		vattr_null(&va_null);
176		once = 1;
177	}
178
179	if (vfc->vfc_version != VFS_VERSION) {
180		printf("ERROR: filesystem %s, unsupported ABI version %x\n",
181		    vfc->vfc_name, vfc->vfc_version);
182		return (EINVAL);
183	}
184	if (vfs_byname(vfc->vfc_name) != NULL)
185		return EEXIST;
186
187	vfc->vfc_typenum = maxvfsconf++;
188	TAILQ_INSERT_TAIL(&vfsconf, vfc, vfc_list);
189
190	/*
191	 * If this filesystem has a sysctl node under vfs
192	 * (i.e. vfs.xxfs), then change the oid number of that node to
193	 * match the filesystem's type number.  This allows user code
194	 * which uses the type number to read sysctl variables defined
195	 * by the filesystem to continue working. Since the oids are
196	 * in a sorted list, we need to make sure the order is
197	 * preserved by re-registering the oid after modifying its
198	 * number.
199	 */
200	SLIST_FOREACH(oidp, &sysctl__vfs_children, oid_link)
201		if (strcmp(oidp->oid_name, vfc->vfc_name) == 0) {
202			sysctl_unregister_oid(oidp);
203			oidp->oid_number = vfc->vfc_typenum;
204			sysctl_register_oid(oidp);
205		}
206
207	/*
208	 * Initialise unused ``struct vfsops'' fields, to use
209	 * the vfs_std*() functions.  Note, we need the mount
210	 * and unmount operations, at the least.  The check
211	 * for vfsops available is just a debugging aid.
212	 */
213	KASSERT(vfc->vfc_vfsops != NULL,
214	    ("Filesystem %s has no vfsops", vfc->vfc_name));
215	/*
216	 * Check the mount and unmount operations.
217	 */
218	vfsops = vfc->vfc_vfsops;
219	KASSERT(vfsops->vfs_mount != NULL,
220	    ("Filesystem %s has no mount op", vfc->vfc_name));
221	KASSERT(vfsops->vfs_unmount != NULL,
222	    ("Filesystem %s has no unmount op", vfc->vfc_name));
223
224	if (vfsops->vfs_start == NULL)
225		/* make a file system operational */
226		vfsops->vfs_start =	vfs_stdstart;
227	if (vfsops->vfs_root == NULL)
228		/* return file system's root vnode */
229		vfsops->vfs_root =	vfs_stdroot;
230	if (vfsops->vfs_quotactl == NULL)
231		/* quota control */
232		vfsops->vfs_quotactl =	vfs_stdquotactl;
233	if (vfsops->vfs_statfs == NULL)
234		/* return file system's status */
235		vfsops->vfs_statfs =	vfs_stdstatfs;
236	if (vfsops->vfs_sync == NULL)
237		/*
238		 * flush unwritten data (nosync)
239		 * file systems can use vfs_stdsync
240		 * explicitly by setting it in the
241		 * vfsop vector.
242		 */
243		vfsops->vfs_sync =	vfs_stdnosync;
244	if (vfsops->vfs_vget == NULL)
245		/* convert an inode number to a vnode */
246		vfsops->vfs_vget =	vfs_stdvget;
247	if (vfsops->vfs_fhtovp == NULL)
248		/* turn an NFS file handle into a vnode */
249		vfsops->vfs_fhtovp =	vfs_stdfhtovp;
250	if (vfsops->vfs_checkexp == NULL)
251		/* check if file system is exported */
252		vfsops->vfs_checkexp =	vfs_stdcheckexp;
253	if (vfsops->vfs_vptofh == NULL)
254		/* turn a vnode into an NFS file handle */
255		vfsops->vfs_vptofh =	vfs_stdvptofh;
256	if (vfsops->vfs_init == NULL)
257		/* file system specific initialisation */
258		vfsops->vfs_init =	vfs_stdinit;
259	if (vfsops->vfs_uninit == NULL)
260		/* file system specific uninitialisation */
261		vfsops->vfs_uninit =	vfs_stduninit;
262	if (vfsops->vfs_extattrctl == NULL)
263		/* extended attribute control */
264		vfsops->vfs_extattrctl = vfs_stdextattrctl;
265	if (vfsops->vfs_sysctl == NULL)
266		vfsops->vfs_sysctl = vfs_stdsysctl;
267
268	/*
269	 * Call init function for this VFS...
270	 */
271	(*(vfc->vfc_vfsops->vfs_init))(vfc);
272
273	return 0;
274}
275
276
277/* Remove registration of a filesystem type */
278int
279vfs_unregister(struct vfsconf *vfc)
280{
281	struct vfsconf *vfsp;
282	int error, i, maxtypenum;
283
284	i = vfc->vfc_typenum;
285
286	vfsp = vfs_byname(vfc->vfc_name);
287	if (vfsp == NULL)
288		return EINVAL;
289	if (vfsp->vfc_refcount)
290		return EBUSY;
291	if (vfc->vfc_vfsops->vfs_uninit != NULL) {
292		error = (*vfc->vfc_vfsops->vfs_uninit)(vfsp);
293		if (error)
294			return (error);
295	}
296	TAILQ_REMOVE(&vfsconf, vfsp, vfc_list);
297	maxtypenum = VFS_GENERIC;
298	TAILQ_FOREACH(vfsp, &vfsconf, vfc_list)
299		if (maxtypenum < vfsp->vfc_typenum)
300			maxtypenum = vfsp->vfc_typenum;
301	maxvfsconf = maxtypenum + 1;
302	return 0;
303}
304
305/*
306 * Standard kernel module handling code for filesystem modules.
307 * Referenced from VFS_SET().
308 */
309int
310vfs_modevent(module_t mod, int type, void *data)
311{
312	struct vfsconf *vfc;
313	int error = 0;
314
315	vfc = (struct vfsconf *)data;
316
317	switch (type) {
318	case MOD_LOAD:
319		if (vfc)
320			error = vfs_register(vfc);
321		break;
322
323	case MOD_UNLOAD:
324		if (vfc)
325			error = vfs_unregister(vfc);
326		break;
327	default:
328		error = EOPNOTSUPP;
329		break;
330	}
331	return (error);
332}
333