null_vfsops.c revision 247619
1/*-
2 * Copyright (c) 1992, 1993, 1995
3 *	The Regents of the University of California.  All rights reserved.
4 *
5 * This code is derived from software donated to Berkeley by
6 * Jan-Simon Pendry.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
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 the
15 *    documentation and/or other materials provided with the distribution.
16 * 4. Neither the name of the University nor the names of its contributors
17 *    may be used to endorse or promote products derived from this software
18 *    without specific prior written permission.
19 *
20 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * SUCH DAMAGE.
31 *
32 *	@(#)null_vfsops.c	8.2 (Berkeley) 1/21/94
33 *
34 * @(#)lofs_vfsops.c	1.2 (Berkeley) 6/18/92
35 * $FreeBSD: head/sys/fs/nullfs/null_vfsops.c 247619 2013-03-02 12:42:23Z jilles $
36 */
37
38/*
39 * Null Layer
40 * (See null_vnops.c for a description of what this does.)
41 */
42
43#include <sys/param.h>
44#include <sys/systm.h>
45#include <sys/fcntl.h>
46#include <sys/kernel.h>
47#include <sys/lock.h>
48#include <sys/malloc.h>
49#include <sys/mount.h>
50#include <sys/namei.h>
51#include <sys/proc.h>
52#include <sys/vnode.h>
53#include <sys/jail.h>
54
55#include <fs/nullfs/null.h>
56
57static MALLOC_DEFINE(M_NULLFSMNT, "nullfs_mount", "NULLFS mount structure");
58
59static vfs_fhtovp_t	nullfs_fhtovp;
60static vfs_mount_t	nullfs_mount;
61static vfs_quotactl_t	nullfs_quotactl;
62static vfs_root_t	nullfs_root;
63static vfs_sync_t	nullfs_sync;
64static vfs_statfs_t	nullfs_statfs;
65static vfs_unmount_t	nullfs_unmount;
66static vfs_vget_t	nullfs_vget;
67static vfs_extattrctl_t	nullfs_extattrctl;
68static vfs_reclaim_lowervp_t nullfs_reclaim_lowervp;
69
70/*
71 * Mount null layer
72 */
73static int
74nullfs_mount(struct mount *mp)
75{
76	int error = 0;
77	struct vnode *lowerrootvp, *vp;
78	struct vnode *nullm_rootvp;
79	struct null_mount *xmp;
80	struct thread *td = curthread;
81	char *target;
82	int isvnunlocked = 0, len;
83	struct nameidata nd, *ndp = &nd;
84
85	NULLFSDEBUG("nullfs_mount(mp = %p)\n", (void *)mp);
86
87	if (!prison_allow(td->td_ucred, PR_ALLOW_MOUNT_NULLFS))
88		return (EPERM);
89	if (mp->mnt_flag & MNT_ROOTFS)
90		return (EOPNOTSUPP);
91
92	/*
93	 * Update is a no-op
94	 */
95	if (mp->mnt_flag & MNT_UPDATE) {
96		/*
97		 * Only support update mounts for NFS export.
98		 */
99		if (vfs_flagopt(mp->mnt_optnew, "export", NULL, 0))
100			return (0);
101		else
102			return (EOPNOTSUPP);
103	}
104
105	/*
106	 * Get argument
107	 */
108	error = vfs_getopt(mp->mnt_optnew, "target", (void **)&target, &len);
109	if (error || target[len - 1] != '\0')
110		return (EINVAL);
111
112	/*
113	 * Unlock lower node to avoid possible deadlock.
114	 */
115	if ((mp->mnt_vnodecovered->v_op == &null_vnodeops) &&
116	    VOP_ISLOCKED(mp->mnt_vnodecovered) == LK_EXCLUSIVE) {
117		VOP_UNLOCK(mp->mnt_vnodecovered, 0);
118		isvnunlocked = 1;
119	}
120	/*
121	 * Find lower node
122	 */
123	NDINIT(ndp, LOOKUP, FOLLOW|LOCKLEAF, UIO_SYSSPACE, target, curthread);
124	error = namei(ndp);
125
126	/*
127	 * Re-lock vnode.
128	 * XXXKIB This is deadlock-prone as well.
129	 */
130	if (isvnunlocked)
131		vn_lock(mp->mnt_vnodecovered, LK_EXCLUSIVE | LK_RETRY);
132
133	if (error)
134		return (error);
135	NDFREE(ndp, NDF_ONLY_PNBUF);
136
137	/*
138	 * Sanity check on lower vnode
139	 */
140	lowerrootvp = ndp->ni_vp;
141
142	/*
143	 * Check multi null mount to avoid `lock against myself' panic.
144	 */
145	if (lowerrootvp == VTONULL(mp->mnt_vnodecovered)->null_lowervp) {
146		NULLFSDEBUG("nullfs_mount: multi null mount?\n");
147		vput(lowerrootvp);
148		return (EDEADLK);
149	}
150
151	xmp = (struct null_mount *) malloc(sizeof(struct null_mount),
152	    M_NULLFSMNT, M_WAITOK | M_ZERO);
153
154	/*
155	 * Save reference to underlying FS
156	 */
157	xmp->nullm_vfs = lowerrootvp->v_mount;
158
159	/*
160	 * Save reference.  Each mount also holds
161	 * a reference on the root vnode.
162	 */
163	error = null_nodeget(mp, lowerrootvp, &vp);
164	/*
165	 * Make sure the node alias worked
166	 */
167	if (error) {
168		free(xmp, M_NULLFSMNT);
169		return (error);
170	}
171
172	/*
173	 * Keep a held reference to the root vnode.
174	 * It is vrele'd in nullfs_unmount.
175	 */
176	nullm_rootvp = vp;
177	nullm_rootvp->v_vflag |= VV_ROOT;
178	xmp->nullm_rootvp = nullm_rootvp;
179
180	/*
181	 * Unlock the node (either the lower or the alias)
182	 */
183	VOP_UNLOCK(vp, 0);
184
185	if (NULLVPTOLOWERVP(nullm_rootvp)->v_mount->mnt_flag & MNT_LOCAL) {
186		MNT_ILOCK(mp);
187		mp->mnt_flag |= MNT_LOCAL;
188		MNT_IUNLOCK(mp);
189	}
190
191	xmp->nullm_flags |= NULLM_CACHE;
192	if (vfs_getopt(mp->mnt_optnew, "nocache", NULL, NULL) == 0)
193		xmp->nullm_flags &= ~NULLM_CACHE;
194
195	MNT_ILOCK(mp);
196	if ((xmp->nullm_flags & NULLM_CACHE) != 0) {
197		mp->mnt_kern_flag |= lowerrootvp->v_mount->mnt_kern_flag &
198		    (MNTK_SHARED_WRITES | MNTK_LOOKUP_SHARED |
199		    MNTK_EXTENDED_SHARED);
200	}
201	mp->mnt_kern_flag |= MNTK_LOOKUP_EXCL_DOTDOT;
202	MNT_IUNLOCK(mp);
203	mp->mnt_data = xmp;
204	vfs_getnewfsid(mp);
205	if ((xmp->nullm_flags & NULLM_CACHE) != 0) {
206		MNT_ILOCK(xmp->nullm_vfs);
207		TAILQ_INSERT_TAIL(&xmp->nullm_vfs->mnt_uppers, mp,
208		    mnt_upper_link);
209		MNT_IUNLOCK(xmp->nullm_vfs);
210	}
211
212	vfs_mountedfrom(mp, target);
213
214	NULLFSDEBUG("nullfs_mount: lower %s, alias at %s\n",
215		mp->mnt_stat.f_mntfromname, mp->mnt_stat.f_mntonname);
216	return (0);
217}
218
219/*
220 * Free reference to null layer
221 */
222static int
223nullfs_unmount(mp, mntflags)
224	struct mount *mp;
225	int mntflags;
226{
227	struct null_mount *mntdata;
228	struct mount *ump;
229	int error, flags;
230
231	NULLFSDEBUG("nullfs_unmount: mp = %p\n", (void *)mp);
232
233	if (mntflags & MNT_FORCE)
234		flags = FORCECLOSE;
235	else
236		flags = 0;
237
238	/* There is 1 extra root vnode reference (nullm_rootvp). */
239	error = vflush(mp, 1, flags, curthread);
240	if (error)
241		return (error);
242
243	/*
244	 * Finally, throw away the null_mount structure
245	 */
246	mntdata = mp->mnt_data;
247	ump = mntdata->nullm_vfs;
248	if ((mntdata->nullm_flags & NULLM_CACHE) != 0) {
249		MNT_ILOCK(ump);
250		while ((ump->mnt_kern_flag & MNTK_VGONE_UPPER) != 0) {
251			ump->mnt_kern_flag |= MNTK_VGONE_WAITER;
252			msleep(&ump->mnt_uppers, &ump->mnt_mtx, 0, "vgnupw", 0);
253		}
254		TAILQ_REMOVE(&ump->mnt_uppers, mp, mnt_upper_link);
255		MNT_IUNLOCK(ump);
256	}
257	mp->mnt_data = NULL;
258	free(mntdata, M_NULLFSMNT);
259	return (0);
260}
261
262static int
263nullfs_root(mp, flags, vpp)
264	struct mount *mp;
265	int flags;
266	struct vnode **vpp;
267{
268	struct vnode *vp;
269
270	NULLFSDEBUG("nullfs_root(mp = %p, vp = %p->%p)\n", (void *)mp,
271	    (void *)MOUNTTONULLMOUNT(mp)->nullm_rootvp,
272	    (void *)NULLVPTOLOWERVP(MOUNTTONULLMOUNT(mp)->nullm_rootvp));
273
274	/*
275	 * Return locked reference to root.
276	 */
277	vp = MOUNTTONULLMOUNT(mp)->nullm_rootvp;
278	VREF(vp);
279
280	ASSERT_VOP_UNLOCKED(vp, "root vnode is locked");
281	vn_lock(vp, flags | LK_RETRY);
282	*vpp = vp;
283	return 0;
284}
285
286static int
287nullfs_quotactl(mp, cmd, uid, arg)
288	struct mount *mp;
289	int cmd;
290	uid_t uid;
291	void *arg;
292{
293	return VFS_QUOTACTL(MOUNTTONULLMOUNT(mp)->nullm_vfs, cmd, uid, arg);
294}
295
296static int
297nullfs_statfs(mp, sbp)
298	struct mount *mp;
299	struct statfs *sbp;
300{
301	int error;
302	struct statfs mstat;
303
304	NULLFSDEBUG("nullfs_statfs(mp = %p, vp = %p->%p)\n", (void *)mp,
305	    (void *)MOUNTTONULLMOUNT(mp)->nullm_rootvp,
306	    (void *)NULLVPTOLOWERVP(MOUNTTONULLMOUNT(mp)->nullm_rootvp));
307
308	bzero(&mstat, sizeof(mstat));
309
310	error = VFS_STATFS(MOUNTTONULLMOUNT(mp)->nullm_vfs, &mstat);
311	if (error)
312		return (error);
313
314	/* now copy across the "interesting" information and fake the rest */
315	sbp->f_type = mstat.f_type;
316	sbp->f_flags = (sbp->f_flags & (MNT_RDONLY | MNT_NOEXEC | MNT_NOSUID |
317	    MNT_UNION | MNT_NOSYMFOLLOW)) | (mstat.f_flags & ~MNT_ROOTFS);
318	sbp->f_bsize = mstat.f_bsize;
319	sbp->f_iosize = mstat.f_iosize;
320	sbp->f_blocks = mstat.f_blocks;
321	sbp->f_bfree = mstat.f_bfree;
322	sbp->f_bavail = mstat.f_bavail;
323	sbp->f_files = mstat.f_files;
324	sbp->f_ffree = mstat.f_ffree;
325	return (0);
326}
327
328static int
329nullfs_sync(mp, waitfor)
330	struct mount *mp;
331	int waitfor;
332{
333	/*
334	 * XXX - Assumes no data cached at null layer.
335	 */
336	return (0);
337}
338
339static int
340nullfs_vget(mp, ino, flags, vpp)
341	struct mount *mp;
342	ino_t ino;
343	int flags;
344	struct vnode **vpp;
345{
346	int error;
347
348	KASSERT((flags & LK_TYPE_MASK) != 0,
349	    ("nullfs_vget: no lock requested"));
350
351	error = VFS_VGET(MOUNTTONULLMOUNT(mp)->nullm_vfs, ino, flags, vpp);
352	if (error != 0)
353		return (error);
354	return (null_nodeget(mp, *vpp, vpp));
355}
356
357static int
358nullfs_fhtovp(mp, fidp, flags, vpp)
359	struct mount *mp;
360	struct fid *fidp;
361	int flags;
362	struct vnode **vpp;
363{
364	int error;
365
366	error = VFS_FHTOVP(MOUNTTONULLMOUNT(mp)->nullm_vfs, fidp, flags,
367	    vpp);
368	if (error != 0)
369		return (error);
370	return (null_nodeget(mp, *vpp, vpp));
371}
372
373static int
374nullfs_extattrctl(mp, cmd, filename_vp, namespace, attrname)
375	struct mount *mp;
376	int cmd;
377	struct vnode *filename_vp;
378	int namespace;
379	const char *attrname;
380{
381
382	return (VFS_EXTATTRCTL(MOUNTTONULLMOUNT(mp)->nullm_vfs, cmd,
383	    filename_vp, namespace, attrname));
384}
385
386static void
387nullfs_reclaim_lowervp(struct mount *mp, struct vnode *lowervp)
388{
389	struct vnode *vp;
390
391	vp = null_hashget(mp, lowervp);
392	if (vp == NULL)
393		return;
394	vgone(vp);
395	vn_lock(lowervp, LK_EXCLUSIVE | LK_RETRY);
396}
397
398static struct vfsops null_vfsops = {
399	.vfs_extattrctl =	nullfs_extattrctl,
400	.vfs_fhtovp =		nullfs_fhtovp,
401	.vfs_init =		nullfs_init,
402	.vfs_mount =		nullfs_mount,
403	.vfs_quotactl =		nullfs_quotactl,
404	.vfs_root =		nullfs_root,
405	.vfs_statfs =		nullfs_statfs,
406	.vfs_sync =		nullfs_sync,
407	.vfs_uninit =		nullfs_uninit,
408	.vfs_unmount =		nullfs_unmount,
409	.vfs_vget =		nullfs_vget,
410	.vfs_reclaim_lowervp =	nullfs_reclaim_lowervp,
411};
412
413VFS_SET(null_vfsops, nullfs, VFCF_LOOPBACK | VFCF_JAIL);
414