1/*-
2 * Copyright (c) 1989, 1993
3 *	The Regents of the University of California.  All rights reserved.
4 * (c) UNIX System Laboratories, Inc.
5 * All or some portions of this file are derived from material licensed
6 * to the University of California by American Telephone and Telegraph
7 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
8 * the permission of UNIX System Laboratories, Inc.
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_subr.c	8.31 (Berkeley) 5/26/95
35 */
36
37#include <sys/cdefs.h>
38__FBSDID("$FreeBSD$");
39
40#include <sys/param.h>
41#include <sys/dirent.h>
42#include <sys/domain.h>
43#include <sys/jail.h>
44#include <sys/kernel.h>
45#include <sys/lock.h>
46#include <sys/malloc.h>
47#include <sys/mbuf.h>
48#include <sys/mount.h>
49#include <sys/mutex.h>
50#include <sys/rwlock.h>
51#include <sys/refcount.h>
52#include <sys/signalvar.h>
53#include <sys/socket.h>
54#include <sys/systm.h>
55#include <sys/vnode.h>
56
57#include <net/radix.h>
58
59static MALLOC_DEFINE(M_NETADDR, "export_host", "Export host address structure");
60
61static void	vfs_free_addrlist(struct netexport *nep);
62static int	vfs_free_netcred(struct radix_node *rn, void *w);
63static int	vfs_hang_addrlist(struct mount *mp, struct netexport *nep,
64		    struct export_args *argp);
65static struct netcred *vfs_export_lookup(struct mount *, struct sockaddr *);
66
67/*
68 * Network address lookup element
69 */
70struct netcred {
71	struct	radix_node netc_rnodes[2];
72	int	netc_exflags;
73	struct	ucred *netc_anon;
74	int	netc_numsecflavors;
75	int	netc_secflavors[MAXSECFLAVORS];
76};
77
78/*
79 * Network export information
80 */
81struct netexport {
82	struct	netcred ne_defexported;		      /* Default export */
83	struct	radix_node_head *ne_rtable[AF_MAX+1]; /* Individual exports */
84};
85
86/*
87 * Build hash lists of net addresses and hang them off the mount point.
88 * Called by vfs_export() to set up the lists of export addresses.
89 */
90static int
91vfs_hang_addrlist(struct mount *mp, struct netexport *nep,
92    struct export_args *argp)
93{
94	register struct netcred *np;
95	register struct radix_node_head *rnh;
96	register int i;
97	struct radix_node *rn;
98	struct sockaddr *saddr, *smask = 0;
99	struct domain *dom;
100	int error;
101
102	/*
103	 * XXX: This routine converts from a `struct xucred'
104	 * (argp->ex_anon) to a `struct ucred' (np->netc_anon).  This
105	 * operation is questionable; for example, what should be done
106	 * with fields like cr_uidinfo and cr_prison?  Currently, this
107	 * routine does not touch them (leaves them as NULL).
108	 */
109	if (argp->ex_anon.cr_version != XUCRED_VERSION) {
110		vfs_mount_error(mp, "ex_anon.cr_version: %d != %d",
111		    argp->ex_anon.cr_version, XUCRED_VERSION);
112		return (EINVAL);
113	}
114
115	if (argp->ex_addrlen == 0) {
116		if (mp->mnt_flag & MNT_DEFEXPORTED) {
117			vfs_mount_error(mp,
118			    "MNT_DEFEXPORTED already set for mount %p", mp);
119			return (EPERM);
120		}
121		np = &nep->ne_defexported;
122		np->netc_exflags = argp->ex_flags;
123		np->netc_anon = crget();
124		np->netc_anon->cr_uid = argp->ex_anon.cr_uid;
125		crsetgroups(np->netc_anon, argp->ex_anon.cr_ngroups,
126		    argp->ex_anon.cr_groups);
127		np->netc_anon->cr_prison = &prison0;
128		prison_hold(np->netc_anon->cr_prison);
129		np->netc_numsecflavors = argp->ex_numsecflavors;
130		bcopy(argp->ex_secflavors, np->netc_secflavors,
131		    sizeof(np->netc_secflavors));
132		MNT_ILOCK(mp);
133		mp->mnt_flag |= MNT_DEFEXPORTED;
134		MNT_IUNLOCK(mp);
135		return (0);
136	}
137
138#if MSIZE <= 256
139	if (argp->ex_addrlen > MLEN) {
140		vfs_mount_error(mp, "ex_addrlen %d is greater than %d",
141		    argp->ex_addrlen, MLEN);
142		return (EINVAL);
143	}
144#endif
145
146	i = sizeof(struct netcred) + argp->ex_addrlen + argp->ex_masklen;
147	np = (struct netcred *) malloc(i, M_NETADDR, M_WAITOK | M_ZERO);
148	saddr = (struct sockaddr *) (np + 1);
149	if ((error = copyin(argp->ex_addr, saddr, argp->ex_addrlen)))
150		goto out;
151	if (saddr->sa_family == AF_UNSPEC || saddr->sa_family > AF_MAX) {
152		error = EINVAL;
153		vfs_mount_error(mp, "Invalid saddr->sa_family: %d");
154		goto out;
155	}
156	if (saddr->sa_len > argp->ex_addrlen)
157		saddr->sa_len = argp->ex_addrlen;
158	if (argp->ex_masklen) {
159		smask = (struct sockaddr *)((caddr_t)saddr + argp->ex_addrlen);
160		error = copyin(argp->ex_mask, smask, argp->ex_masklen);
161		if (error)
162			goto out;
163		if (smask->sa_len > argp->ex_masklen)
164			smask->sa_len = argp->ex_masklen;
165	}
166	i = saddr->sa_family;
167	if ((rnh = nep->ne_rtable[i]) == NULL) {
168		/*
169		 * Seems silly to initialize every AF when most are not used,
170		 * do so on demand here
171		 */
172		for (dom = domains; dom; dom = dom->dom_next) {
173			KASSERT(((i == AF_INET) || (i == AF_INET6)),
174			    ("unexpected protocol in vfs_hang_addrlist"));
175			if (dom->dom_family == i && dom->dom_rtattach) {
176				/*
177				 * XXX MRT
178				 * The INET and INET6 domains know the
179				 * offset already. We don't need to send it
180				 * So we just use it as a flag to say that
181				 * we are or are not setting up a real routing
182				 * table. Only IP and IPV6 need have this
183				 * be 0 so all other protocols can stay the
184				 * same (ABI compatible).
185				 */
186				dom->dom_rtattach(
187				    (void **) &nep->ne_rtable[i], 0);
188				break;
189			}
190		}
191		if ((rnh = nep->ne_rtable[i]) == NULL) {
192			error = ENOBUFS;
193			vfs_mount_error(mp, "%s %s %d",
194			    "Unable to initialize radix node head ",
195			    "for address family", i);
196			goto out;
197		}
198	}
199	RADIX_NODE_HEAD_LOCK(rnh);
200	rn = (*rnh->rnh_addaddr)(saddr, smask, rnh, np->netc_rnodes);
201	RADIX_NODE_HEAD_UNLOCK(rnh);
202	if (rn == NULL || np != (struct netcred *)rn) {	/* already exists */
203		error = EPERM;
204		vfs_mount_error(mp, "Invalid radix node head, rn: %p %p",
205		    rn, np);
206		goto out;
207	}
208	np->netc_exflags = argp->ex_flags;
209	np->netc_anon = crget();
210	np->netc_anon->cr_uid = argp->ex_anon.cr_uid;
211	crsetgroups(np->netc_anon, argp->ex_anon.cr_ngroups,
212	    argp->ex_anon.cr_groups);
213	np->netc_anon->cr_prison = &prison0;
214	prison_hold(np->netc_anon->cr_prison);
215	np->netc_numsecflavors = argp->ex_numsecflavors;
216	bcopy(argp->ex_secflavors, np->netc_secflavors,
217	    sizeof(np->netc_secflavors));
218	return (0);
219out:
220	free(np, M_NETADDR);
221	return (error);
222}
223
224/* Helper for vfs_free_addrlist. */
225/* ARGSUSED */
226static int
227vfs_free_netcred(struct radix_node *rn, void *w)
228{
229	struct radix_node_head *rnh = (struct radix_node_head *) w;
230	struct ucred *cred;
231
232	(*rnh->rnh_deladdr) (rn->rn_key, rn->rn_mask, rnh);
233	cred = ((struct netcred *)rn)->netc_anon;
234	if (cred != NULL)
235		crfree(cred);
236	free(rn, M_NETADDR);
237	return (0);
238}
239
240/*
241 * Free the net address hash lists that are hanging off the mount points.
242 */
243static void
244vfs_free_addrlist(struct netexport *nep)
245{
246	int i;
247	struct radix_node_head *rnh;
248	struct ucred *cred;
249
250	for (i = 0; i <= AF_MAX; i++) {
251		if ((rnh = nep->ne_rtable[i])) {
252			RADIX_NODE_HEAD_LOCK(rnh);
253			(*rnh->rnh_walktree) (rnh, vfs_free_netcred, rnh);
254			RADIX_NODE_HEAD_UNLOCK(rnh);
255			RADIX_NODE_HEAD_DESTROY(rnh);
256			free(rnh, M_RTABLE);
257			nep->ne_rtable[i] = NULL;	/* not SMP safe XXX */
258		}
259	}
260	cred = nep->ne_defexported.netc_anon;
261	if (cred != NULL)
262		crfree(cred);
263
264}
265
266/*
267 * High level function to manipulate export options on a mount point
268 * and the passed in netexport.
269 * Struct export_args *argp is the variable used to twiddle options,
270 * the structure is described in sys/mount.h
271 */
272int
273vfs_export(struct mount *mp, struct export_args *argp)
274{
275	struct netexport *nep;
276	int error;
277
278	if (argp->ex_numsecflavors < 0
279	    || argp->ex_numsecflavors >= MAXSECFLAVORS)
280		return (EINVAL);
281
282	error = 0;
283	lockmgr(&mp->mnt_explock, LK_EXCLUSIVE, NULL);
284	nep = mp->mnt_export;
285	if (argp->ex_flags & MNT_DELEXPORT) {
286		if (nep == NULL) {
287			error = ENOENT;
288			goto out;
289		}
290		if (mp->mnt_flag & MNT_EXPUBLIC) {
291			vfs_setpublicfs(NULL, NULL, NULL);
292			MNT_ILOCK(mp);
293			mp->mnt_flag &= ~MNT_EXPUBLIC;
294			MNT_IUNLOCK(mp);
295		}
296		vfs_free_addrlist(nep);
297		mp->mnt_export = NULL;
298		free(nep, M_MOUNT);
299		nep = NULL;
300		MNT_ILOCK(mp);
301		mp->mnt_flag &= ~(MNT_EXPORTED | MNT_DEFEXPORTED);
302		MNT_IUNLOCK(mp);
303	}
304	if (argp->ex_flags & MNT_EXPORTED) {
305		if (nep == NULL) {
306			nep = malloc(sizeof(struct netexport), M_MOUNT, M_WAITOK | M_ZERO);
307			mp->mnt_export = nep;
308		}
309		if (argp->ex_flags & MNT_EXPUBLIC) {
310			if ((error = vfs_setpublicfs(mp, nep, argp)) != 0)
311				goto out;
312			MNT_ILOCK(mp);
313			mp->mnt_flag |= MNT_EXPUBLIC;
314			MNT_IUNLOCK(mp);
315		}
316		if ((error = vfs_hang_addrlist(mp, nep, argp)))
317			goto out;
318		MNT_ILOCK(mp);
319		mp->mnt_flag |= MNT_EXPORTED;
320		MNT_IUNLOCK(mp);
321	}
322
323out:
324	lockmgr(&mp->mnt_explock, LK_RELEASE, NULL);
325	/*
326	 * Once we have executed the vfs_export() command, we do
327	 * not want to keep the "export" option around in the
328	 * options list, since that will cause subsequent MNT_UPDATE
329	 * calls to fail.  The export information is saved in
330	 * mp->mnt_export, so we can safely delete the "export" mount option
331	 * here.
332	 */
333	vfs_deleteopt(mp->mnt_optnew, "export");
334	vfs_deleteopt(mp->mnt_opt, "export");
335	return (error);
336}
337
338/*
339 * Set the publicly exported filesystem (WebNFS). Currently, only
340 * one public filesystem is possible in the spec (RFC 2054 and 2055)
341 */
342int
343vfs_setpublicfs(struct mount *mp, struct netexport *nep,
344    struct export_args *argp)
345{
346	int error;
347	struct vnode *rvp;
348	char *cp;
349
350	/*
351	 * mp == NULL -> invalidate the current info, the FS is
352	 * no longer exported. May be called from either vfs_export
353	 * or unmount, so check if it hasn't already been done.
354	 */
355	if (mp == NULL) {
356		if (nfs_pub.np_valid) {
357			nfs_pub.np_valid = 0;
358			if (nfs_pub.np_index != NULL) {
359				free(nfs_pub.np_index, M_TEMP);
360				nfs_pub.np_index = NULL;
361			}
362		}
363		return (0);
364	}
365
366	/*
367	 * Only one allowed at a time.
368	 */
369	if (nfs_pub.np_valid != 0 && mp != nfs_pub.np_mount)
370		return (EBUSY);
371
372	/*
373	 * Get real filehandle for root of exported FS.
374	 */
375	bzero(&nfs_pub.np_handle, sizeof(nfs_pub.np_handle));
376	nfs_pub.np_handle.fh_fsid = mp->mnt_stat.f_fsid;
377
378	if ((error = VFS_ROOT(mp, LK_EXCLUSIVE, &rvp)))
379		return (error);
380
381	if ((error = VOP_VPTOFH(rvp, &nfs_pub.np_handle.fh_fid)))
382		return (error);
383
384	vput(rvp);
385
386	/*
387	 * If an indexfile was specified, pull it in.
388	 */
389	if (argp->ex_indexfile != NULL) {
390		if (nfs_pub.np_index != NULL)
391			nfs_pub.np_index = malloc(MAXNAMLEN + 1, M_TEMP,
392			    M_WAITOK);
393		error = copyinstr(argp->ex_indexfile, nfs_pub.np_index,
394		    MAXNAMLEN, (size_t *)0);
395		if (!error) {
396			/*
397			 * Check for illegal filenames.
398			 */
399			for (cp = nfs_pub.np_index; *cp; cp++) {
400				if (*cp == '/') {
401					error = EINVAL;
402					break;
403				}
404			}
405		}
406		if (error) {
407			free(nfs_pub.np_index, M_TEMP);
408			nfs_pub.np_index = NULL;
409			return (error);
410		}
411	}
412
413	nfs_pub.np_mount = mp;
414	nfs_pub.np_valid = 1;
415	return (0);
416}
417
418/*
419 * Used by the filesystems to determine if a given network address
420 * (passed in 'nam') is present in their exports list, returns a pointer
421 * to struct netcred so that the filesystem can examine it for
422 * access rights (read/write/etc).
423 */
424static struct netcred *
425vfs_export_lookup(struct mount *mp, struct sockaddr *nam)
426{
427	struct netexport *nep;
428	register struct netcred *np;
429	register struct radix_node_head *rnh;
430	struct sockaddr *saddr;
431
432	nep = mp->mnt_export;
433	if (nep == NULL)
434		return (NULL);
435	np = NULL;
436	if (mp->mnt_flag & MNT_EXPORTED) {
437		/*
438		 * Lookup in the export list first.
439		 */
440		if (nam != NULL) {
441			saddr = nam;
442			rnh = nep->ne_rtable[saddr->sa_family];
443			if (rnh != NULL) {
444				RADIX_NODE_HEAD_RLOCK(rnh);
445				np = (struct netcred *)
446				    (*rnh->rnh_matchaddr)(saddr, rnh);
447				RADIX_NODE_HEAD_RUNLOCK(rnh);
448				if (np && np->netc_rnodes->rn_flags & RNF_ROOT)
449					np = NULL;
450			}
451		}
452		/*
453		 * If no address match, use the default if it exists.
454		 */
455		if (np == NULL && mp->mnt_flag & MNT_DEFEXPORTED)
456			np = &nep->ne_defexported;
457	}
458	return (np);
459}
460
461/*
462 * XXX: This comment comes from the deprecated ufs_check_export()
463 * XXX: and may not entirely apply, but lacking something better:
464 * This is the generic part of fhtovp called after the underlying
465 * filesystem has validated the file handle.
466 *
467 * Verify that a host should have access to a filesystem.
468 */
469
470int
471vfs_stdcheckexp(struct mount *mp, struct sockaddr *nam, int *extflagsp,
472    struct ucred **credanonp, int *numsecflavors, int **secflavors)
473{
474	struct netcred *np;
475
476	lockmgr(&mp->mnt_explock, LK_SHARED, NULL);
477	np = vfs_export_lookup(mp, nam);
478	if (np == NULL) {
479		lockmgr(&mp->mnt_explock, LK_RELEASE, NULL);
480		*credanonp = NULL;
481		return (EACCES);
482	}
483	*extflagsp = np->netc_exflags;
484	if ((*credanonp = np->netc_anon) != NULL)
485		crhold(*credanonp);
486	if (numsecflavors)
487		*numsecflavors = np->netc_numsecflavors;
488	if (secflavors)
489		*secflavors = np->netc_secflavors;
490	lockmgr(&mp->mnt_explock, LK_RELEASE, NULL);
491	return (0);
492}
493
494