2 * Copyright (c) 1994, 1995 The Regents of the University of California.
3 * Copyright (c) 1994, 1995 Jan-Simon Pendry.
6 * This code is derived from software donated to Berkeley by
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. All advertising materials mentioning features or use of this software
18 * must display the following acknowledgement:
19 * This product includes software developed by the University of
20 * California, Berkeley and its contributors.
21 * 4. Neither the name of the University nor the names of its contributors
22 * may be used to endorse or promote products derived from this software
23 * without specific prior written permission.
25 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
26 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
29 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
30 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
31 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
32 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
33 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
34 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
37 * @(#)union_vfsops.c 8.20 (Berkeley) 5/20/95
38 * $FreeBSD: src/sys/miscfs/union/union_vfsops.c,v 1.39.2.2 2001/10/25 19:18:53 dillon Exp $
39 * $DragonFly: src/sys/vfs/union/union_vfsops.c,v 1.14 2004/08/17 18:57:36 dillon Exp $
46 #include <sys/param.h>
47 #include <sys/systm.h>
48 #include <sys/kernel.h>
50 #include <sys/vnode.h>
51 #include <sys/mount.h>
52 #include <sys/namei.h>
53 #include <sys/malloc.h>
54 #include <sys/filedesc.h>
56 #include <vm/vm_zone.h>
58 extern struct vnodeopv_entry_desc union_vnodeop_entries[];
60 static MALLOC_DEFINE(M_UNIONFSMNT, "UNION mount", "UNION mount structure");
62 extern int union_init (struct vfsconf *);
63 static int union_mount (struct mount *mp, char *path, caddr_t data,
64 struct nameidata *ndp, struct thread *td);
65 static int union_root (struct mount *mp, struct vnode **vpp);
66 static int union_statfs (struct mount *mp, struct statfs *sbp,
68 static int union_unmount (struct mount *mp, int mntflags,
72 * Mount union filesystem
75 union_mount(struct mount *mp, char *path, caddr_t data, struct nameidata *ndp,
79 struct union_args args;
80 struct vnode *lowerrootvp = NULLVP;
81 struct vnode *upperrootvp = NULLVP;
82 struct union_mount *um = 0;
83 struct ucred *cred = 0;
88 UDEBUG(("union_mount(mp = %p)\n", (void *)mp));
90 KKASSERT(td->td_proc);
93 * Disable clustered write, otherwise system becomes unstable.
95 mp->mnt_flag |= MNT_NOCLUSTERW;
100 if (mp->mnt_flag & MNT_UPDATE) {
103 * 1. a way to convert between rdonly and rdwr mounts.
104 * 2. support for nfs exports.
113 error = copyin(data, (caddr_t)&args, sizeof(struct union_args));
118 * Obtain lower vnode. Vnode is stored in mp->mnt_vnodecovered.
119 * We need to reference it but not lock it.
122 lowerrootvp = mp->mnt_vnodecovered;
127 * Unlock lower node to avoid deadlock.
129 if (lowerrootvp->v_tag == VT_UNION)
130 VOP_UNLOCK(lowerrootvp, NULL, 0, td);
134 * Obtain upper vnode by calling namei() on the path. The
135 * upperrootvp will be turned referenced but not locked.
137 NDINIT(ndp, NAMEI_LOOKUP, CNP_FOLLOW | CNP_WANTPARENT,
138 UIO_USERSPACE, args.target, td);
143 if (lowerrootvp->v_tag == VT_UNION)
144 vn_lock(lowerrootvp, NULL, LK_EXCLUSIVE | LK_RETRY, td);
149 NDFREE(ndp, NDF_ONLY_PNBUF);
150 upperrootvp = ndp->ni_vp;
154 UDEBUG(("mount_root UPPERVP %p locked = %d\n", upperrootvp,
155 VOP_ISLOCKED(upperrootvp, NULL)));
158 * Check multi union mount to avoid `lock myself again' panic.
159 * Also require that it be a directory.
161 if (upperrootvp == VTOUNION(lowerrootvp)->un_uppervp) {
163 printf("union_mount: multi union mount?\n");
169 if (upperrootvp->v_type != VDIR) {
175 * Allocate our union_mount structure and populate the fields.
176 * The vnode references are stored in the union_mount as held,
177 * unlocked references. Depending on the _BELOW flag, the
178 * filesystems are viewed in a different order. In effect this
179 * is the same as providing a mount-under option to the mount
183 um = (struct union_mount *) malloc(sizeof(struct union_mount),
184 M_UNIONFSMNT, M_WAITOK);
186 bzero(um, sizeof(struct union_mount));
188 um->um_op = args.mntflags & UNMNT_OPMASK;
192 um->um_lowervp = lowerrootvp;
193 um->um_uppervp = upperrootvp;
199 um->um_lowervp = upperrootvp;
200 um->um_uppervp = lowerrootvp;
208 um->um_uppervp = upperrootvp;
209 um->um_lowervp = lowerrootvp;
219 * Unless the mount is readonly, ensure that the top layer
220 * supports whiteout operations
222 if ((mp->mnt_flag & MNT_RDONLY) == 0) {
223 error = VOP_WHITEOUT(um->um_uppervp, NCPNULL, NULL, NAMEI_LOOKUP);
228 um->um_cred = crhold(td->td_proc->p_ucred);
229 um->um_cmode = UN_DIRMODE & ~td->td_proc->p_fd->fd_cmask;
232 * Depending on what you think the MNT_LOCAL flag might mean,
233 * you may want the && to be || on the conditional below.
234 * At the moment it has been defined that the filesystem is
235 * only local if it is all local, ie the MNT_LOCAL flag implies
236 * that the entire namespace is local. If you think the MNT_LOCAL
237 * flag implies that some of the files might be stored locally
238 * then you will want to change the conditional.
240 if (um->um_op == UNMNT_ABOVE) {
241 if (((um->um_lowervp == NULLVP) ||
242 (um->um_lowervp->v_mount->mnt_flag & MNT_LOCAL)) &&
243 (um->um_uppervp->v_mount->mnt_flag & MNT_LOCAL))
244 mp->mnt_flag |= MNT_LOCAL;
248 * Copy in the upper layer's RDONLY flag. This is for the benefit
249 * of lookup() which explicitly checks the flag, rather than asking
250 * the filesystem for its own opinion. This means, that an update
251 * mount of the underlying filesystem to go from rdonly to rdwr
252 * will leave the unioned view as read-only.
254 mp->mnt_flag |= (um->um_uppervp->v_mount->mnt_flag & MNT_RDONLY);
256 mp->mnt_data = (qaddr_t) um;
259 (void) copyinstr(path, mp->mnt_stat.f_mntonname, MNAMELEN - 1, &size);
260 bzero(mp->mnt_stat.f_mntonname + size, MNAMELEN - size);
274 bcopy(cp, mp->mnt_stat.f_mntfromname, len);
276 cp = mp->mnt_stat.f_mntfromname + len;
277 len = MNAMELEN - len;
279 (void) copyinstr(args.target, cp, len - 1, &size);
280 bzero(cp + size, len - size);
282 vfs_add_vnodeops(&mp->mnt_vn_ops, union_vnodeop_entries);
284 (void)union_statfs(mp, &mp->mnt_stat, td);
286 UDEBUG(("union_mount: from %s, on %s\n",
287 mp->mnt_stat.f_mntfromname, mp->mnt_stat.f_mntonname));
293 vrele(um->um_uppervp);
295 vrele(um->um_lowervp);
296 /* XXX other fields */
297 free(um, M_UNIONFSMNT);
309 * Free reference to union layer
312 union_unmount(struct mount *mp, int mntflags, struct thread *td)
314 struct union_mount *um = MOUNTTOUNIONMOUNT(mp);
319 UDEBUG(("union_unmount(mp = %p)\n", (void *)mp));
321 if (mntflags & MNT_FORCE)
325 * Keep flushing vnodes from the mount list.
326 * This is needed because of the un_pvp held
327 * reference to the parent vnode.
328 * If more vnodes have been freed on a given pass,
329 * the try again. The loop will iterate at most
330 * (d) times, where (d) is the maximum tree depth
333 for (freeing = 0; (error = vflush(mp, 0, flags)) != 0;) {
334 int n = mp->mnt_nvnodelistsize;
336 /* if this is unchanged then stop */
340 /* otherwise try once more time */
344 /* If the most recent vflush failed, the filesystem is still busy. */
349 * Discard references to upper and lower target vnodes.
352 vrele(um->um_lowervp);
353 vrele(um->um_uppervp);
356 * Finally, throw away the union_mount structure
358 free(mp->mnt_data, M_UNIONFSMNT); /* XXX */
364 union_root(struct mount *mp, struct vnode **vpp)
366 struct union_mount *um = MOUNTTOUNIONMOUNT(mp);
370 * Supply an unlocked reference to um_uppervp and to um_lowervp. It
371 * is possible for um_uppervp to be locked without the associated
372 * root union_node being locked. We let union_allocvp() deal with
375 UDEBUG(("union_root UPPERVP %p locked = %d\n", um->um_uppervp,
376 VOP_ISLOCKED(um->um_uppervp, NULL)));
378 vref(um->um_uppervp);
380 vref(um->um_lowervp);
382 error = union_allocvp(vpp, mp, NULLVP, NULLVP, NULL,
383 um->um_uppervp, um->um_lowervp, 1);
384 UDEBUG(("error %d\n", error));
385 UDEBUG(("union_root2 UPPERVP %p locked = %d\n", um->um_uppervp,
386 VOP_ISLOCKED(um->um_uppervp, NULL)));
392 union_statfs(struct mount *mp, struct statfs *sbp, struct thread *td)
395 struct union_mount *um = MOUNTTOUNIONMOUNT(mp);
399 UDEBUG(("union_statfs(mp = %p, lvp = %p, uvp = %p)\n",
400 (void *)mp, (void *)um->um_lowervp, (void *)um->um_uppervp));
402 bzero(&mstat, sizeof(mstat));
404 if (um->um_lowervp) {
405 error = VFS_STATFS(um->um_lowervp->v_mount, &mstat, td);
410 /* now copy across the "interesting" information and fake the rest */
412 sbp->f_type = mstat.f_type;
413 sbp->f_flags = mstat.f_flags;
414 sbp->f_bsize = mstat.f_bsize;
415 sbp->f_iosize = mstat.f_iosize;
417 lbsize = mstat.f_bsize;
418 sbp->f_blocks = mstat.f_blocks;
419 sbp->f_bfree = mstat.f_bfree;
420 sbp->f_bavail = mstat.f_bavail;
421 sbp->f_files = mstat.f_files;
422 sbp->f_ffree = mstat.f_ffree;
424 error = VFS_STATFS(um->um_uppervp->v_mount, &mstat, td);
428 sbp->f_flags = mstat.f_flags;
429 sbp->f_bsize = mstat.f_bsize;
430 sbp->f_iosize = mstat.f_iosize;
433 * if the lower and upper blocksizes differ, then frig the
434 * block counts so that the sizes reported by df make some
435 * kind of sense. none of this makes sense though.
438 if (mstat.f_bsize != lbsize)
439 sbp->f_blocks = ((off_t) sbp->f_blocks * lbsize) / mstat.f_bsize;
442 * The "total" fields count total resources in all layers,
443 * the "free" fields count only those resources which are
444 * free in the upper layer (since only the upper layer
447 sbp->f_blocks += mstat.f_blocks;
448 sbp->f_bfree = mstat.f_bfree;
449 sbp->f_bavail = mstat.f_bavail;
450 sbp->f_files += mstat.f_files;
451 sbp->f_ffree = mstat.f_ffree;
453 if (sbp != &mp->mnt_stat) {
454 sbp->f_type = mp->mnt_vfc->vfc_typenum;
455 bcopy(&mp->mnt_stat.f_fsid, &sbp->f_fsid, sizeof(sbp->f_fsid));
456 bcopy(mp->mnt_stat.f_mntonname, sbp->f_mntonname, MNAMELEN);
457 bcopy(mp->mnt_stat.f_mntfromname, sbp->f_mntfromname, MNAMELEN);
462 static struct vfsops union_vfsops = {
464 vfs_stdstart, /* underlying start already done */
469 vfs_stdsync, /* XXX assumes no cached data on union level */
479 VFS_SET(union_vfsops, union, VFCF_LOOPBACK);