1/*-
2 * Copyright (c) 1994, 1995 The Regents of the University of California.
3 * Copyright (c) 1994, 1995 Jan-Simon Pendry.
4 * Copyright (c) 2005, 2006, 2012 Masanori Ozawa <ozawa@ongs.co.jp>, ONGS Inc.
5 * Copyright (c) 2006, 2012 Daichi Goto <daichi@freebsd.org>
6 * All rights reserved.
7 *
8 * This code is derived from software donated to Berkeley by
9 * Jan-Simon Pendry.
10 *
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
13 * are met:
14 * 1. Redistributions of source code must retain the above copyright
15 *    notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 *    notice, this list of conditions and the following disclaimer in the
18 *    documentation and/or other materials provided with the distribution.
19 * 4. Neither the name of the University nor the names of its contributors
20 *    may be used to endorse or promote products derived from this software
21 *    without specific prior written permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * SUCH DAMAGE.
34 *
35 *	@(#)union_vfsops.c	8.20 (Berkeley) 5/20/95
36 * $FreeBSD$
37 */
38
39#include <sys/param.h>
40#include <sys/systm.h>
41#include <sys/kdb.h>
42#include <sys/fcntl.h>
43#include <sys/kernel.h>
44#include <sys/lock.h>
45#include <sys/malloc.h>
46#include <sys/mount.h>
47#include <sys/namei.h>
48#include <sys/proc.h>
49#include <sys/vnode.h>
50#include <sys/stat.h>
51
52#include <fs/unionfs/union.h>
53
54static MALLOC_DEFINE(M_UNIONFSMNT, "UNIONFS mount", "UNIONFS mount structure");
55
56static vfs_fhtovp_t	unionfs_fhtovp;
57static vfs_checkexp_t	unionfs_checkexp;
58static vfs_mount_t	unionfs_domount;
59static vfs_quotactl_t	unionfs_quotactl;
60static vfs_root_t	unionfs_root;
61static vfs_sync_t	unionfs_sync;
62static vfs_statfs_t	unionfs_statfs;
63static vfs_unmount_t	unionfs_unmount;
64static vfs_vget_t	unionfs_vget;
65static vfs_extattrctl_t	unionfs_extattrctl;
66
67static struct vfsops unionfs_vfsops;
68
69/*
70 * Mount unionfs layer.
71 */
72static int
73unionfs_domount(struct mount *mp)
74{
75	int		error;
76	struct vnode   *lowerrootvp;
77	struct vnode   *upperrootvp;
78	struct unionfs_mount *ump;
79	struct thread *td;
80	char           *target;
81	char           *tmp;
82	char           *ep;
83	int		len;
84	size_t		done;
85	int		below;
86	uid_t		uid;
87	gid_t		gid;
88	u_short		udir;
89	u_short		ufile;
90	unionfs_copymode copymode;
91	unionfs_whitemode whitemode;
92	struct nameidata nd, *ndp;
93	struct vattr	va;
94
95	UNIONFSDEBUG("unionfs_mount(mp = %p)\n", (void *)mp);
96
97	error = 0;
98	below = 0;
99	uid = 0;
100	gid = 0;
101	udir = 0;
102	ufile = 0;
103	copymode = UNIONFS_TRANSPARENT;	/* default */
104	whitemode = UNIONFS_WHITE_ALWAYS;
105	ndp = &nd;
106	td = curthread;
107
108	if (mp->mnt_flag & MNT_ROOTFS) {
109		vfs_mount_error(mp, "Cannot union mount root filesystem");
110		return (EOPNOTSUPP);
111	}
112
113	/*
114	 * Update is a no operation.
115	 */
116	if (mp->mnt_flag & MNT_UPDATE) {
117		vfs_mount_error(mp, "unionfs does not support mount update");
118		return (EOPNOTSUPP);
119	}
120
121	/*
122	 * Get argument
123	 */
124	error = vfs_getopt(mp->mnt_optnew, "target", (void **)&target, &len);
125	if (error)
126		error = vfs_getopt(mp->mnt_optnew, "from", (void **)&target,
127		    &len);
128	if (error || target[len - 1] != '\0') {
129		vfs_mount_error(mp, "Invalid target");
130		return (EINVAL);
131	}
132	if (vfs_getopt(mp->mnt_optnew, "below", NULL, NULL) == 0)
133		below = 1;
134	if (vfs_getopt(mp->mnt_optnew, "udir", (void **)&tmp, NULL) == 0) {
135		if (tmp != NULL)
136			udir = (mode_t)strtol(tmp, &ep, 8);
137		if (tmp == NULL || *ep) {
138			vfs_mount_error(mp, "Invalid udir");
139			return (EINVAL);
140		}
141		udir &= S_IRWXU | S_IRWXG | S_IRWXO;
142	}
143	if (vfs_getopt(mp->mnt_optnew, "ufile", (void **)&tmp, NULL) == 0) {
144		if (tmp != NULL)
145			ufile = (mode_t)strtol(tmp, &ep, 8);
146		if (tmp == NULL || *ep) {
147			vfs_mount_error(mp, "Invalid ufile");
148			return (EINVAL);
149		}
150		ufile &= S_IRWXU | S_IRWXG | S_IRWXO;
151	}
152	/* check umask, uid and gid */
153	if (udir == 0 && ufile != 0)
154		udir = ufile;
155	if (ufile == 0 && udir != 0)
156		ufile = udir;
157
158	vn_lock(mp->mnt_vnodecovered, LK_SHARED | LK_RETRY);
159	error = VOP_GETATTR(mp->mnt_vnodecovered, &va, mp->mnt_cred);
160	if (!error) {
161		if (udir == 0)
162			udir = va.va_mode;
163		if (ufile == 0)
164			ufile = va.va_mode;
165		uid = va.va_uid;
166		gid = va.va_gid;
167	}
168	VOP_UNLOCK(mp->mnt_vnodecovered, LK_RELEASE);
169	if (error)
170		return (error);
171
172	if (mp->mnt_cred->cr_ruid == 0) {	/* root only */
173		if (vfs_getopt(mp->mnt_optnew, "uid", (void **)&tmp,
174		    NULL) == 0) {
175			if (tmp != NULL)
176				uid = (uid_t)strtol(tmp, &ep, 10);
177			if (tmp == NULL || *ep) {
178				vfs_mount_error(mp, "Invalid uid");
179				return (EINVAL);
180			}
181		}
182		if (vfs_getopt(mp->mnt_optnew, "gid", (void **)&tmp,
183		    NULL) == 0) {
184			if (tmp != NULL)
185				gid = (gid_t)strtol(tmp, &ep, 10);
186			if (tmp == NULL || *ep) {
187				vfs_mount_error(mp, "Invalid gid");
188				return (EINVAL);
189			}
190		}
191		if (vfs_getopt(mp->mnt_optnew, "copymode", (void **)&tmp,
192		    NULL) == 0) {
193			if (tmp == NULL) {
194				vfs_mount_error(mp, "Invalid copymode");
195				return (EINVAL);
196			} else if (strcasecmp(tmp, "traditional") == 0)
197				copymode = UNIONFS_TRADITIONAL;
198			else if (strcasecmp(tmp, "transparent") == 0)
199				copymode = UNIONFS_TRANSPARENT;
200			else if (strcasecmp(tmp, "masquerade") == 0)
201				copymode = UNIONFS_MASQUERADE;
202			else {
203				vfs_mount_error(mp, "Invalid copymode");
204				return (EINVAL);
205			}
206		}
207		if (vfs_getopt(mp->mnt_optnew, "whiteout", (void **)&tmp,
208		    NULL) == 0) {
209			if (tmp == NULL) {
210				vfs_mount_error(mp, "Invalid whiteout mode");
211				return (EINVAL);
212			} else if (strcasecmp(tmp, "always") == 0)
213				whitemode = UNIONFS_WHITE_ALWAYS;
214			else if (strcasecmp(tmp, "whenneeded") == 0)
215				whitemode = UNIONFS_WHITE_WHENNEEDED;
216			else {
217				vfs_mount_error(mp, "Invalid whiteout mode");
218				return (EINVAL);
219			}
220		}
221	}
222	/* If copymode is UNIONFS_TRADITIONAL, uid/gid is mounted user. */
223	if (copymode == UNIONFS_TRADITIONAL) {
224		uid = mp->mnt_cred->cr_ruid;
225		gid = mp->mnt_cred->cr_rgid;
226	}
227
228	UNIONFSDEBUG("unionfs_mount: uid=%d, gid=%d\n", uid, gid);
229	UNIONFSDEBUG("unionfs_mount: udir=0%03o, ufile=0%03o\n", udir, ufile);
230	UNIONFSDEBUG("unionfs_mount: copymode=%d\n", copymode);
231
232	/*
233	 * Find upper node
234	 */
235	NDINIT(ndp, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE, target, td);
236	if ((error = namei(ndp)))
237		return (error);
238
239	NDFREE(ndp, NDF_ONLY_PNBUF);
240
241	/* get root vnodes */
242	lowerrootvp = mp->mnt_vnodecovered;
243	upperrootvp = ndp->ni_vp;
244
245	/* create unionfs_mount */
246	ump = (struct unionfs_mount *)malloc(sizeof(struct unionfs_mount),
247	    M_UNIONFSMNT, M_WAITOK | M_ZERO);
248
249	/*
250	 * Save reference
251	 */
252	if (below) {
253		VOP_UNLOCK(upperrootvp, LK_RELEASE);
254		vn_lock(lowerrootvp, LK_EXCLUSIVE | LK_RETRY);
255		ump->um_lowervp = upperrootvp;
256		ump->um_uppervp = lowerrootvp;
257	} else {
258		ump->um_lowervp = lowerrootvp;
259		ump->um_uppervp = upperrootvp;
260	}
261	ump->um_rootvp = NULLVP;
262	ump->um_uid = uid;
263	ump->um_gid = gid;
264	ump->um_udir = udir;
265	ump->um_ufile = ufile;
266	ump->um_copymode = copymode;
267	ump->um_whitemode = whitemode;
268
269	MNT_ILOCK(mp);
270	if ((lowerrootvp->v_mount->mnt_kern_flag & MNTK_MPSAFE) &&
271	    (upperrootvp->v_mount->mnt_kern_flag & MNTK_MPSAFE))
272		mp->mnt_kern_flag |= MNTK_MPSAFE;
273	MNT_IUNLOCK(mp);
274	mp->mnt_data = ump;
275
276	/*
277	 * Copy upper layer's RDONLY flag.
278	 */
279	mp->mnt_flag |= ump->um_uppervp->v_mount->mnt_flag & MNT_RDONLY;
280
281	/*
282	 * Unlock the node
283	 */
284	VOP_UNLOCK(ump->um_uppervp, LK_RELEASE);
285
286	/*
287	 * Get the unionfs root vnode.
288	 */
289	error = unionfs_nodeget(mp, ump->um_uppervp, ump->um_lowervp,
290	    NULLVP, &(ump->um_rootvp), NULL, td);
291	vrele(upperrootvp);
292	if (error) {
293		free(ump, M_UNIONFSMNT);
294		mp->mnt_data = NULL;
295		return (error);
296	}
297
298	/*
299	 * Check mnt_flag
300	 */
301	if ((ump->um_lowervp->v_mount->mnt_flag & MNT_LOCAL) &&
302	    (ump->um_uppervp->v_mount->mnt_flag & MNT_LOCAL))
303		mp->mnt_flag |= MNT_LOCAL;
304
305	/*
306	 * Get new fsid
307	 */
308	vfs_getnewfsid(mp);
309
310	len = MNAMELEN - 1;
311	tmp = mp->mnt_stat.f_mntfromname;
312	copystr((below ? "<below>:" : "<above>:"), tmp, len, &done);
313	len -= done - 1;
314	tmp += done - 1;
315	copystr(target, tmp, len, NULL);
316
317	UNIONFSDEBUG("unionfs_mount: from %s, on %s\n",
318	    mp->mnt_stat.f_mntfromname, mp->mnt_stat.f_mntonname);
319
320	return (0);
321}
322
323/*
324 * Free reference to unionfs layer
325 */
326static int
327unionfs_unmount(struct mount *mp, int mntflags)
328{
329	struct unionfs_mount *ump;
330	int		error;
331	int		num;
332	int		freeing;
333	int		flags;
334
335	UNIONFSDEBUG("unionfs_unmount: mp = %p\n", (void *)mp);
336
337	ump = MOUNTTOUNIONFSMOUNT(mp);
338	flags = 0;
339
340	if (mntflags & MNT_FORCE)
341		flags |= FORCECLOSE;
342
343	/* vflush (no need to call vrele) */
344	for (freeing = 0; (error = vflush(mp, 1, flags, curthread)) != 0;) {
345		num = mp->mnt_nvnodelistsize;
346		if (num == freeing)
347			break;
348		freeing = num;
349	}
350
351	if (error)
352		return (error);
353
354	free(ump, M_UNIONFSMNT);
355	mp->mnt_data = 0;
356
357	return (0);
358}
359
360static int
361unionfs_root(struct mount *mp, int flags, struct vnode **vpp)
362{
363	struct unionfs_mount *ump;
364	struct vnode   *vp;
365
366	ump = MOUNTTOUNIONFSMOUNT(mp);
367	vp = ump->um_rootvp;
368
369	UNIONFSDEBUG("unionfs_root: rootvp=%p locked=%x\n",
370	    vp, VOP_ISLOCKED(vp));
371
372	vref(vp);
373	if (flags & LK_TYPE_MASK)
374		vn_lock(vp, flags);
375
376	*vpp = vp;
377
378	return (0);
379}
380
381static int
382unionfs_quotactl(struct mount *mp, int cmd, uid_t uid, void *arg)
383{
384	struct unionfs_mount *ump;
385
386	ump = MOUNTTOUNIONFSMOUNT(mp);
387
388	/*
389	 * Writing is always performed to upper vnode.
390	 */
391	return (VFS_QUOTACTL(ump->um_uppervp->v_mount, cmd, uid, arg));
392}
393
394static int
395unionfs_statfs(struct mount *mp, struct statfs *sbp)
396{
397	struct unionfs_mount *ump;
398	int		error;
399	struct statfs	mstat;
400	uint64_t	lbsize;
401
402	ump = MOUNTTOUNIONFSMOUNT(mp);
403
404	UNIONFSDEBUG("unionfs_statfs(mp = %p, lvp = %p, uvp = %p)\n",
405	    (void *)mp, (void *)ump->um_lowervp, (void *)ump->um_uppervp);
406
407	bzero(&mstat, sizeof(mstat));
408
409	error = VFS_STATFS(ump->um_lowervp->v_mount, &mstat);
410	if (error)
411		return (error);
412
413	/* now copy across the "interesting" information and fake the rest */
414	sbp->f_blocks = mstat.f_blocks;
415	sbp->f_files = mstat.f_files;
416
417	lbsize = mstat.f_bsize;
418
419	error = VFS_STATFS(ump->um_uppervp->v_mount, &mstat);
420	if (error)
421		return (error);
422
423	/*
424	 * The FS type etc is copy from upper vfs.
425	 * (write able vfs have priority)
426	 */
427	sbp->f_type = mstat.f_type;
428	sbp->f_flags = mstat.f_flags;
429	sbp->f_bsize = mstat.f_bsize;
430	sbp->f_iosize = mstat.f_iosize;
431
432	if (mstat.f_bsize != lbsize)
433		sbp->f_blocks = ((off_t)sbp->f_blocks * lbsize) / mstat.f_bsize;
434
435	sbp->f_blocks += mstat.f_blocks;
436	sbp->f_bfree = mstat.f_bfree;
437	sbp->f_bavail = mstat.f_bavail;
438	sbp->f_files += mstat.f_files;
439	sbp->f_ffree = mstat.f_ffree;
440	return (0);
441}
442
443static int
444unionfs_sync(struct mount *mp, int waitfor)
445{
446	/* nothing to do */
447	return (0);
448}
449
450static int
451unionfs_vget(struct mount *mp, ino_t ino, int flags, struct vnode **vpp)
452{
453	return (EOPNOTSUPP);
454}
455
456static int
457unionfs_fhtovp(struct mount *mp, struct fid *fidp, int flags,
458    struct vnode **vpp)
459{
460	return (EOPNOTSUPP);
461}
462
463static int
464unionfs_checkexp(struct mount *mp, struct sockaddr *nam, int *extflagsp,
465    struct ucred **credanonp, int *numsecflavors, int **secflavors)
466{
467	return (EOPNOTSUPP);
468}
469
470static int
471unionfs_extattrctl(struct mount *mp, int cmd, struct vnode *filename_vp,
472    int namespace, const char *attrname)
473{
474	struct unionfs_mount *ump;
475	struct unionfs_node *unp;
476
477	ump = MOUNTTOUNIONFSMOUNT(mp);
478	unp = VTOUNIONFS(filename_vp);
479
480	if (unp->un_uppervp != NULLVP) {
481		return (VFS_EXTATTRCTL(ump->um_uppervp->v_mount, cmd,
482		    unp->un_uppervp, namespace, attrname));
483	} else {
484		return (VFS_EXTATTRCTL(ump->um_lowervp->v_mount, cmd,
485		    unp->un_lowervp, namespace, attrname));
486	}
487}
488
489static struct vfsops unionfs_vfsops = {
490	.vfs_checkexp =		unionfs_checkexp,
491	.vfs_extattrctl =	unionfs_extattrctl,
492	.vfs_fhtovp =		unionfs_fhtovp,
493	.vfs_init =		unionfs_init,
494	.vfs_mount =		unionfs_domount,
495	.vfs_quotactl =		unionfs_quotactl,
496	.vfs_root =		unionfs_root,
497	.vfs_statfs =		unionfs_statfs,
498	.vfs_sync =		unionfs_sync,
499	.vfs_uninit =		unionfs_uninit,
500	.vfs_unmount =		unionfs_unmount,
501	.vfs_vget =		unionfs_vget,
502};
503
504VFS_SET(unionfs_vfsops, unionfs, VFCF_LOOPBACK);
505