xref: /dflybsd-src/sys/kern/vfs_init.c (revision 10f84ad9e1cb939878f6c1ec7e73c30dd133a0f2)
1 /*
2  * Copyright (c) 2003,2004 The DragonFly Project.  All rights reserved.
3  *
4  * This code is derived from software contributed to The DragonFly Project
5  * by Matthew Dillon <dillon@backplane.com>
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  *
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in
15  *    the documentation and/or other materials provided with the
16  *    distribution.
17  * 3. Neither the name of The DragonFly Project nor the names of its
18  *    contributors may be used to endorse or promote products derived
19  *    from this software without specific, prior written permission.
20  *
21  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
22  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
23  * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
24  * FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE
25  * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
26  * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
27  * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
28  * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
29  * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
30  * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
31  * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32  * SUCH DAMAGE.
33  *
34  *
35  * Copyright (c) 1989, 1993
36  *	The Regents of the University of California.  All rights reserved.
37  *
38  * This code is derived from software contributed
39  * to Berkeley by John Heidemann of the UCLA Ficus project.
40  *
41  * Source: * @(#)i405_init.c 2.10 92/04/27 UCLA Ficus project
42  *
43  * Redistribution and use in source and binary forms, with or without
44  * modification, are permitted provided that the following conditions
45  * are met:
46  * 1. Redistributions of source code must retain the above copyright
47  *    notice, this list of conditions and the following disclaimer.
48  * 2. Redistributions in binary form must reproduce the above copyright
49  *    notice, this list of conditions and the following disclaimer in the
50  *    documentation and/or other materials provided with the distribution.
51  * 3. All advertising materials mentioning features or use of this software
52  *    must display the following acknowledgement:
53  *	This product includes software developed by the University of
54  *	California, Berkeley and its contributors.
55  * 4. Neither the name of the University nor the names of its contributors
56  *    may be used to endorse or promote products derived from this software
57  *    without specific prior written permission.
58  *
59  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
60  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
61  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
62  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
63  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
64  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
65  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
66  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
67  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
68  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
69  * SUCH DAMAGE.
70  *
71  *	@(#)vfs_init.c	8.3 (Berkeley) 1/4/94
72  * $FreeBSD: src/sys/kern/vfs_init.c,v 1.59 2002/04/30 18:44:32 dillon Exp $
73  * $DragonFly: src/sys/kern/vfs_init.c,v 1.6 2004/08/17 18:57:32 dillon Exp $
74  */
75 /*
76  * Manage vnode VOP operations vectors
77  */
78 #include <sys/param.h>
79 #include <sys/systm.h>
80 #include <sys/kernel.h>
81 #include <sys/mount.h>
82 #include <sys/sysctl.h>
83 #include <sys/vnode.h>
84 #include <sys/malloc.h>
85 #include <vm/vm_zone.h>
86 
87 MALLOC_DEFINE(M_VNODE, "vnodes", "Dynamically allocated vnodes");
88 
89 /*
90  * Zone for namei
91  */
92 struct vm_zone *namei_zone;
93 
94 /*
95  * vfs_init() will set maxvfsconf
96  * to the highest defined type number.
97  */
98 int maxvfsconf;
99 struct vfsconf *vfsconf;
100 
101 static TAILQ_HEAD(, vnodeopv_node) vnodeopv_list;
102 static void vfs_recalc_vnodeops(void);
103 
104 /*
105  * Add a vnode operations (vnops) vector to the global list.
106  */
107 void
108 vfs_add_vnodeops_sysinit(const void *data)
109 {
110 	const struct vnodeopv_desc *vdesc = data;
111 
112 	vfs_add_vnodeops(vdesc->opv_desc_vector, vdesc->opv_desc_ops);
113 }
114 
115 /*
116  * Unlink previously added vnode operations vector.
117  */
118 void
119 vfs_rm_vnodeops_sysinit(const void *data)
120 {
121 	const struct vnodeopv_desc *vdesc = data;
122 
123 	vfs_rm_vnodeops(vdesc->opv_desc_vector);
124 }
125 
126 void
127 vfs_add_vnodeops(struct vop_ops **vops_pp, struct vnodeopv_entry_desc *descs)
128 {
129 	struct vnodeopv_node *node;
130 	struct vop_ops *ops;
131 
132 	node = malloc(sizeof(*node), M_VNODE, M_ZERO|M_WAITOK);
133 	if ((ops = *vops_pp) == NULL) {
134 		ops = malloc(sizeof(struct vop_ops), M_VNODE, M_ZERO|M_WAITOK);
135 		*vops_pp = ops;
136 	}
137 	node->ops = ops;
138 	node->descs = descs;
139 	++ops->vv_refs;
140 	TAILQ_INSERT_TAIL(&vnodeopv_list, node, entry);
141 	vfs_recalc_vnodeops();
142 }
143 
144 void
145 vfs_rm_vnodeops(struct vop_ops **vops_pp)
146 {
147 	struct vop_ops *ops = *vops_pp;
148 	struct vnodeopv_node *node;
149 
150 	if (ops == NULL)
151 		return;
152 
153 	TAILQ_FOREACH(node, &vnodeopv_list, entry) {
154 		if (node->ops == ops)
155 			break;
156 	}
157 	if (node == NULL) {
158 		printf("vfs_rm_vnodeops: unable to find ops: %p\n", ops);
159 		return;
160 	}
161 	TAILQ_REMOVE(&vnodeopv_list, node, entry);
162 	free(node, M_VNODE);
163 	KKASSERT(ops != NULL && ops->vv_refs > 0);
164 	if (--ops->vv_refs == 0) {
165 		*vops_pp = NULL;
166 		free(ops, M_VNODE);
167 	}
168 	vfs_recalc_vnodeops();
169 }
170 
171 /*
172  * Recalculate VFS operations vectors
173  */
174 static void
175 vfs_recalc_vnodeops(void)
176 {
177 	struct vnodeopv_node *node;
178 	struct vnodeopv_entry_desc *desc;
179 	struct vop_ops *ops;
180 	struct vop_ops *vnew;
181 	int off;
182 
183 	/*
184 	 * Because vop_ops may be active we can't just blow them away, we
185 	 * have to generate new vop_ops and then copy them into the running
186 	 * vop_ops.  Any missing entries will be assigned to the default
187 	 * entry.  If the default entry itself is missing it will be assigned
188 	 * to vop_eopnotsupp.
189 	 */
190 	TAILQ_FOREACH(node, &vnodeopv_list, entry) {
191 		ops = node->ops;
192 		if ((vnew = ops->vv_new) == NULL) {
193 			vnew = malloc(sizeof(struct vop_ops),
194 					M_VNODE, M_ZERO|M_WAITOK);
195 			ops->vv_new = vnew;
196 			vnew->vop_default = vop_eopnotsupp;
197 		}
198 		for (desc = node->descs; desc->opve_op; ++desc) {
199 			off = desc->opve_op->vdesc_offset;
200 			*(void **)((char *)vnew + off) = desc->opve_func;
201 		}
202 		for (off = __offsetof(struct vop_ops, vop_ops_first_field);
203 		     off <= __offsetof(struct vop_ops, vop_ops_last_field);
204 		     off += sizeof(void **)
205 		) {
206 			if (*(void **)((char *)vnew + off) == NULL)
207 			    *(void **)((char *)vnew + off) = vnew->vop_default;
208 		}
209 	}
210 
211 	/*
212 	 * Copy the temporary ops into the running configuration and then
213 	 * delete them.
214 	 */
215 	TAILQ_FOREACH(node, &vnodeopv_list, entry) {
216 		ops = node->ops;
217 		if ((vnew = ops->vv_new) == NULL)
218 			continue;
219 		for (off = __offsetof(struct vop_ops, vop_ops_first_field);
220 		     off <= __offsetof(struct vop_ops, vop_ops_last_field);
221 		     off += sizeof(void **)
222 		) {
223 			*(void **)((char *)ops + off) =
224 				*(void **)((char *)vnew + off);
225 		}
226 		ops->vv_new = NULL;
227 		free(vnew, M_VNODE);
228 	}
229 }
230 
231 /*
232  * Routines having to do with the management of the vnode table.
233  */
234 struct vattr va_null;
235 
236 /*
237  * Initialize the vnode structures and initialize each file system type.
238  */
239 /* ARGSUSED*/
240 static void
241 vfsinit(void *dummy)
242 {
243 	TAILQ_INIT(&vnodeopv_list);
244 	namei_zone = zinit("NAMEI", MAXPATHLEN, 0, 0, 2);
245 
246 	/*
247 	 * Initialize the vnode table
248 	 */
249 	vntblinit();
250 	/*
251 	 * Initialize the vnode name cache
252 	 */
253 	nchinit();
254 	/*
255 	 * Initialize each file system type.
256 	 * Vfs type numbers must be distinct from VFS_GENERIC (and VFS_VFSCONF).
257 	 */
258 	vattr_null(&va_null);
259 	maxvfsconf = VFS_GENERIC + 1;
260 }
261 SYSINIT(vfs, SI_SUB_VFS, SI_ORDER_FIRST, vfsinit, NULL)
262 
263 /*
264  * Register a VFS.
265  *
266  * After doing general initialisation, this function will
267  * call the filesystem specific initialisation vector op,
268  * i.e. vfsops->vfs_init().
269  */
270 int
271 vfs_register(struct vfsconf *vfc)
272 {
273 	struct sysctl_oid *oidp;
274 	struct vfsconf *vfsp;
275 
276 	vfsp = NULL;
277 	if (vfsconf)
278 		for (vfsp = vfsconf; vfsp->vfc_next; vfsp = vfsp->vfc_next)
279 			if (strcmp(vfc->vfc_name, vfsp->vfc_name) == 0)
280 				return EEXIST;
281 
282 	vfc->vfc_typenum = maxvfsconf++;
283 	if (vfsp)
284 		vfsp->vfc_next = vfc;
285 	else
286 		vfsconf = vfc;
287 	vfc->vfc_next = NULL;
288 
289 	/*
290 	 * If this filesystem has a sysctl node under vfs
291 	 * (i.e. vfs.xxfs), then change the oid number of that node to
292 	 * match the filesystem's type number.  This allows user code
293 	 * which uses the type number to read sysctl variables defined
294 	 * by the filesystem to continue working. Since the oids are
295 	 * in a sorted list, we need to make sure the order is
296 	 * preserved by re-registering the oid after modifying its
297 	 * number.
298 	 */
299 	SLIST_FOREACH(oidp, &sysctl__vfs_children, oid_link)
300 		if (strcmp(oidp->oid_name, vfc->vfc_name) == 0) {
301 			sysctl_unregister_oid(oidp);
302 			oidp->oid_number = vfc->vfc_typenum;
303 			sysctl_register_oid(oidp);
304 		}
305 
306 	/*
307 	 * Call init function for this VFS...
308 	 */
309 	(*(vfc->vfc_vfsops->vfs_init))(vfc);
310 
311 	return 0;
312 }
313 
314 
315 /*
316  * Remove previously registered VFS.
317  *
318  * After doing general de-registration like removing sysctl
319  * nodes etc, it will call the filesystem specific vector
320  * op, i.e. vfsops->vfs_uninit().
321  *
322  */
323 int
324 vfs_unregister(struct vfsconf *vfc)
325 {
326 	struct vfsconf *vfsp, *prev_vfsp;
327 	int error, i, maxtypenum;
328 
329 	i = vfc->vfc_typenum;
330 
331 	prev_vfsp = NULL;
332 	for (vfsp = vfsconf; vfsp;
333 			prev_vfsp = vfsp, vfsp = vfsp->vfc_next) {
334 		if (!strcmp(vfc->vfc_name, vfsp->vfc_name))
335 			break;
336 	}
337 	if (vfsp == NULL)
338 		return EINVAL;
339 	if (vfsp->vfc_refcount)
340 		return EBUSY;
341 	if (vfc->vfc_vfsops->vfs_uninit != NULL) {
342 		error = (*vfc->vfc_vfsops->vfs_uninit)(vfsp);
343 		if (error)
344 			return (error);
345 	}
346 	if (prev_vfsp)
347 		prev_vfsp->vfc_next = vfsp->vfc_next;
348 	else
349 		vfsconf = vfsp->vfc_next;
350 	maxtypenum = VFS_GENERIC;
351 	for (vfsp = vfsconf; vfsp != NULL; vfsp = vfsp->vfc_next)
352 		if (maxtypenum < vfsp->vfc_typenum)
353 			maxtypenum = vfsp->vfc_typenum;
354 	maxvfsconf = maxtypenum + 1;
355 	return 0;
356 }
357 
358 int
359 vfs_modevent(module_t mod, int type, void *data)
360 {
361 	struct vfsconf *vfc;
362 	int error = 0;
363 
364 	vfc = (struct vfsconf *)data;
365 
366 	switch (type) {
367 	case MOD_LOAD:
368 		if (vfc)
369 			error = vfs_register(vfc);
370 		break;
371 
372 	case MOD_UNLOAD:
373 		if (vfc)
374 			error = vfs_unregister(vfc);
375 		break;
376 	default:	/* including MOD_SHUTDOWN */
377 		break;
378 	}
379 	return (error);
380 }
381