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 $
45 #include <sys/param.h>
46 #include <sys/systm.h>
47 #include <sys/kernel.h>
49 #include <sys/vnode.h>
50 #include <sys/mount.h>
51 #include <sys/namei.h>
52 #include <sys/malloc.h>
53 #include <sys/filedesc.h>
54 #include <miscfs/union/union.h>
55 #include <vm/vm_zone.h>
57 static MALLOC_DEFINE(M_UNIONFSMNT, "UNION mount", "UNION mount structure");
59 extern int union_init __P((struct vfsconf *));
60 static int union_mount __P((struct mount *mp, char *path, caddr_t data,
61 struct nameidata *ndp, struct proc *p));
62 static int union_root __P((struct mount *mp, struct vnode **vpp));
63 static int union_statfs __P((struct mount *mp, struct statfs *sbp,
65 static int union_unmount __P((struct mount *mp, int mntflags,
69 * Mount union filesystem
72 union_mount(mp, path, data, ndp, p)
76 struct nameidata *ndp;
80 struct union_args args;
81 struct vnode *lowerrootvp = NULLVP;
82 struct vnode *upperrootvp = NULLVP;
83 struct union_mount *um = 0;
84 struct ucred *cred = 0;
89 UDEBUG(("union_mount(mp = %p)\n", (void *)mp));
92 * Disable clustered write, otherwise system becomes unstable.
94 mp->mnt_flag |= MNT_NOCLUSTERW;
99 if (mp->mnt_flag & MNT_UPDATE) {
102 * 1. a way to convert between rdonly and rdwr mounts.
103 * 2. support for nfs exports.
112 error = copyin(data, (caddr_t)&args, sizeof(struct union_args));
117 * Obtain lower vnode. Vnode is stored in mp->mnt_vnodecovered.
118 * We need to reference it but not lock it.
121 lowerrootvp = mp->mnt_vnodecovered;
126 * Unlock lower node to avoid deadlock.
128 if (lowerrootvp->v_op == union_vnodeop_p)
129 VOP_UNLOCK(lowerrootvp, 0, p);
133 * Obtain upper vnode by calling namei() on the path. The
134 * upperrootvp will be turned referenced but not locked.
136 NDINIT(ndp, LOOKUP, FOLLOW|WANTPARENT,
137 UIO_USERSPACE, args.target, p);
142 if (lowerrootvp->v_op == union_vnodeop_p)
143 vn_lock(lowerrootvp, LK_EXCLUSIVE | LK_RETRY, p);
148 NDFREE(ndp, NDF_ONLY_PNBUF);
149 upperrootvp = ndp->ni_vp;
153 UDEBUG(("mount_root UPPERVP %p locked = %d\n", upperrootvp,
154 VOP_ISLOCKED(upperrootvp, NULL)));
157 * Check multi union mount to avoid `lock myself again' panic.
158 * Also require that it be a directory.
160 if (upperrootvp == VTOUNION(lowerrootvp)->un_uppervp) {
162 printf("union_mount: multi union mount?\n");
168 if (upperrootvp->v_type != VDIR) {
174 * Allocate our union_mount structure and populate the fields.
175 * The vnode references are stored in the union_mount as held,
176 * unlocked references. Depending on the _BELOW flag, the
177 * filesystems are viewed in a different order. In effect this
178 * is the same as providing a mount-under option to the mount
182 um = (struct union_mount *) malloc(sizeof(struct union_mount),
183 M_UNIONFSMNT, M_WAITOK);
185 bzero(um, sizeof(struct union_mount));
187 um->um_op = args.mntflags & UNMNT_OPMASK;
191 um->um_lowervp = lowerrootvp;
192 um->um_uppervp = upperrootvp;
198 um->um_lowervp = upperrootvp;
199 um->um_uppervp = lowerrootvp;
207 um->um_uppervp = upperrootvp;
208 um->um_lowervp = lowerrootvp;
218 * Unless the mount is readonly, ensure that the top layer
219 * supports whiteout operations
221 if ((mp->mnt_flag & MNT_RDONLY) == 0) {
222 error = VOP_WHITEOUT(um->um_uppervp, NULL, LOOKUP);
227 um->um_cred = p->p_ucred;
229 um->um_cmode = UN_DIRMODE &~ p->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 (void)union_statfs(mp, &mp->mnt_stat, p);
284 UDEBUG(("union_mount: from %s, on %s\n",
285 mp->mnt_stat.f_mntfromname, mp->mnt_stat.f_mntonname));
291 vrele(um->um_uppervp);
293 vrele(um->um_lowervp);
294 /* XXX other fields */
295 free(um, M_UNIONFSMNT);
307 * Free reference to union layer
310 union_unmount(mp, mntflags, p)
315 struct union_mount *um = MOUNTTOUNIONMOUNT(mp);
320 UDEBUG(("union_unmount(mp = %p)\n", (void *)mp));
322 if (mntflags & MNT_FORCE)
326 * Keep flushing vnodes from the mount list.
327 * This is needed because of the un_pvp held
328 * reference to the parent vnode.
329 * If more vnodes have been freed on a given pass,
330 * the try again. The loop will iterate at most
331 * (d) times, where (d) is the maximum tree depth
334 for (freeing = 0; (error = vflush(mp, 0, flags)) != 0;) {
338 /* count #vnodes held on mount list */
339 for (n = 0, vp = TAILQ_FIRST(&mp->mnt_nvnodelist);
341 vp = TAILQ_NEXT(vp, v_nmntvnodes)) {
345 /* if this is unchanged then stop */
349 /* otherwise try once more time */
353 /* If the most recent vflush failed, the filesystem is still busy. */
358 * Discard references to upper and lower target vnodes.
361 vrele(um->um_lowervp);
362 vrele(um->um_uppervp);
365 * Finally, throw away the union_mount structure
367 free(mp->mnt_data, M_UNIONFSMNT); /* XXX */
377 struct union_mount *um = MOUNTTOUNIONMOUNT(mp);
381 * Supply an unlocked reference to um_uppervp and to um_lowervp. It
382 * is possible for um_uppervp to be locked without the associated
383 * root union_node being locked. We let union_allocvp() deal with
386 UDEBUG(("union_root UPPERVP %p locked = %d\n", um->um_uppervp,
387 VOP_ISLOCKED(um->um_uppervp, NULL)));
389 VREF(um->um_uppervp);
391 VREF(um->um_lowervp);
393 error = union_allocvp(vpp, mp, NULLVP, NULLVP, NULL,
394 um->um_uppervp, um->um_lowervp, 1);
395 UDEBUG(("error %d\n", error));
396 UDEBUG(("union_root2 UPPERVP %p locked = %d\n", um->um_uppervp,
397 VOP_ISLOCKED(um->um_uppervp, NULL)));
403 union_statfs(mp, sbp, p)
409 struct union_mount *um = MOUNTTOUNIONMOUNT(mp);
413 UDEBUG(("union_statfs(mp = %p, lvp = %p, uvp = %p)\n",
414 (void *)mp, (void *)um->um_lowervp, (void *)um->um_uppervp));
416 bzero(&mstat, sizeof(mstat));
418 if (um->um_lowervp) {
419 error = VFS_STATFS(um->um_lowervp->v_mount, &mstat, p);
424 /* now copy across the "interesting" information and fake the rest */
426 sbp->f_type = mstat.f_type;
427 sbp->f_flags = mstat.f_flags;
428 sbp->f_bsize = mstat.f_bsize;
429 sbp->f_iosize = mstat.f_iosize;
431 lbsize = mstat.f_bsize;
432 sbp->f_blocks = mstat.f_blocks;
433 sbp->f_bfree = mstat.f_bfree;
434 sbp->f_bavail = mstat.f_bavail;
435 sbp->f_files = mstat.f_files;
436 sbp->f_ffree = mstat.f_ffree;
438 error = VFS_STATFS(um->um_uppervp->v_mount, &mstat, p);
442 sbp->f_flags = mstat.f_flags;
443 sbp->f_bsize = mstat.f_bsize;
444 sbp->f_iosize = mstat.f_iosize;
447 * if the lower and upper blocksizes differ, then frig the
448 * block counts so that the sizes reported by df make some
449 * kind of sense. none of this makes sense though.
452 if (mstat.f_bsize != lbsize)
453 sbp->f_blocks = ((off_t) sbp->f_blocks * lbsize) / mstat.f_bsize;
456 * The "total" fields count total resources in all layers,
457 * the "free" fields count only those resources which are
458 * free in the upper layer (since only the upper layer
461 sbp->f_blocks += mstat.f_blocks;
462 sbp->f_bfree = mstat.f_bfree;
463 sbp->f_bavail = mstat.f_bavail;
464 sbp->f_files += mstat.f_files;
465 sbp->f_ffree = mstat.f_ffree;
467 if (sbp != &mp->mnt_stat) {
468 sbp->f_type = mp->mnt_vfc->vfc_typenum;
469 bcopy(&mp->mnt_stat.f_fsid, &sbp->f_fsid, sizeof(sbp->f_fsid));
470 bcopy(mp->mnt_stat.f_mntonname, sbp->f_mntonname, MNAMELEN);
471 bcopy(mp->mnt_stat.f_mntfromname, sbp->f_mntfromname, MNAMELEN);
476 static struct vfsops union_vfsops = {
478 vfs_stdstart, /* underlying start already done */
483 vfs_stdsync, /* XXX assumes no cached data on union level */
493 VFS_SET(union_vfsops, union, VFCF_LOOPBACK);