X-Git-Url: https://git.subgeniuskitty.com/unix-history/.git/blobdiff_plain/12561664c1c9bff8b92347f86d3a60cb82a0d8b9..13f48dee70ec0b79753383e9734854a57464d8fa:/usr/src/sys/kern/vfs_syscalls.c

diff --git a/usr/src/sys/kern/vfs_syscalls.c b/usr/src/sys/kern/vfs_syscalls.c
index f0f0b8adb4..4063304af6 100644
--- a/usr/src/sys/kern/vfs_syscalls.c
+++ b/usr/src/sys/kern/vfs_syscalls.c
@@ -9,7 +9,7 @@
  *
  * %sccs.include.redist.c%
  *
- *	@(#)vfs_syscalls.c	8.12 (Berkeley) %G%
+ *	@(#)vfs_syscalls.c	8.28 (Berkeley) %G%
  */
 
 #include <sys/param.h>
@@ -63,13 +63,9 @@ mount(p, uap, retval)
 	register struct vnode *vp;
 	register struct mount *mp;
 	int error, flag;
+	struct vattr va;
 	struct nameidata nd;
 
-	/*
-	 * Must be super user
-	 */
-	if (error = suser(p->p_ucred, &p->p_acflag))
-		return (error);
 	/*
 	 * Get vnode to be covered
 	 */
@@ -95,9 +91,50 @@ mount(p, uap, retval)
 		}
 		mp->mnt_flag |=
 		    uap->flags & (MNT_RELOAD | MNT_FORCE | MNT_UPDATE);
+		/*
+		 * Only root, or the user that did the original mount is
+		 * permitted to update it.
+		 */
+		if (mp->mnt_stat.f_owner != p->p_ucred->cr_uid &&
+		    (error = suser(p->p_ucred, &p->p_acflag))) {
+			vput(vp);
+			return (error);
+		}
+		/*
+		 * Do not allow NFS export by non-root users. Silently
+		 * enforce MNT_NOSUID and MNT_NODEV for non-root users.
+		 */
+		if (p->p_ucred->cr_uid != 0) {
+			if (uap->flags & MNT_EXPORTED) {
+				vput(vp);
+				return (EPERM);
+			}
+			uap->flags |= MNT_NOSUID | MNT_NODEV;
+		}
 		VOP_UNLOCK(vp);
 		goto update;
 	}
+	/*
+	 * If the user is not root, ensure that they own the directory
+	 * onto which we are attempting to mount.
+	 */
+	if ((error = VOP_GETATTR(vp, &va, p->p_ucred, p)) ||
+	    (va.va_uid != p->p_ucred->cr_uid &&
+	     (error = suser(p->p_ucred, &p->p_acflag)))) {
+		vput(vp);
+		return (error);
+	}
+	/*
+	 * Do not allow NFS export by non-root users. Silently
+	 * enforce MNT_NOSUID and MNT_NODEV for non-root users.
+	 */
+	if (p->p_ucred->cr_uid != 0) {
+		if (uap->flags & MNT_EXPORTED) {
+			vput(vp);
+			return (EPERM);
+		}
+		uap->flags |= MNT_NOSUID | MNT_NODEV;
+	}
 	if (error = vinvalbuf(vp, V_SAVE, p->p_ucred, p, 0, 0))
 		return (error);
 	if (vp->v_type != VDIR) {
@@ -108,6 +145,10 @@ mount(p, uap, retval)
 		vput(vp);
 		return (ENODEV);
 	}
+	if (vp->v_mountedhere != NULL) {
+		vput(vp);
+		return (EBUSY);
+	}
 
 	/*
 	 * Allocate and initialize the file system.
@@ -121,14 +162,9 @@ mount(p, uap, retval)
 		vput(vp);
 		return (error);
 	}
-	if (vp->v_mountedhere != NULL) {
-		vfs_unlock(mp);
-		free((caddr_t)mp, M_MOUNT);
-		vput(vp);
-		return (EBUSY);
-	}
 	vp->v_mountedhere = mp;
 	mp->mnt_vnodecovered = vp;
+	mp->mnt_stat.f_owner = p->p_ucred->cr_uid;
 update:
 	/*
 	 * Set the mount level flags.
@@ -161,6 +197,7 @@ update:
 	cache_purge(vp);
 	if (!error) {
 		TAILQ_INSERT_TAIL(&mountlist, mp, mnt_list);
+		checkdirs(vp);
 		VOP_UNLOCK(vp);
 		vfs_unlock(mp);
 		error = VFS_START(mp, 0, p);
@@ -173,6 +210,43 @@ update:
 	return (error);
 }
 
+/*
+ * Scan all active processes to see if any of them have a current
+ * or root directory onto which the new filesystem has just been
+ * mounted. If so, replace them with the new mount point.
+ */
+checkdirs(olddp)
+	struct vnode *olddp;
+{
+	struct filedesc *fdp;
+	struct vnode *newdp;
+	struct proc *p;
+
+	if (olddp->v_usecount == 1)
+		return;
+	if (VFS_ROOT(olddp->v_mountedhere, &newdp))
+		panic("mount: lost mount");
+	for (p = allproc.lh_first; p != 0; p = p->p_list.le_next) {
+		fdp = p->p_fd;
+		if (fdp->fd_cdir == olddp) {
+			vrele(fdp->fd_cdir);
+			VREF(newdp);
+			fdp->fd_cdir = newdp;
+		}
+		if (fdp->fd_rdir == olddp) {
+			vrele(fdp->fd_rdir);
+			VREF(newdp);
+			fdp->fd_rdir = newdp;
+		}
+	}
+	if (rootvnode == olddp) {
+		vrele(rootvnode);
+		VREF(newdp);
+		rootvnode = newdp;
+	}
+	vput(newdp);
+}
+
 /*
  * Unmount a file system.
  *
@@ -198,12 +272,13 @@ unmount(p, uap, retval)
 	if (error = namei(&nd))
 		return (error);
 	vp = nd.ni_vp;
+	mp = vp->v_mount;
 
 	/*
-	 * Unless this is a user mount, then must
-	 * have suser privilege.
+	 * Only root, or the user that did the original mount is
+	 * permitted to unmount this filesystem.
 	 */
-	if (((vp->v_mount->mnt_flag & MNT_USER) == 0) &&
+	if ((mp->mnt_stat.f_owner != p->p_ucred->cr_uid) &&
 	    (error = suser(p->p_ucred, &p->p_acflag))) {
 		vput(vp);
 		return (error);
@@ -216,7 +291,6 @@ unmount(p, uap, retval)
 		vput(vp);
 		return (EINVAL);
 	}
-	mp = vp->v_mount;
 	vput(vp);
 	return (dounmount(mp, uap->flags, p));
 }
@@ -264,7 +338,7 @@ dounmount(mp, flags, p)
 /*
  * Sync each mounted filesystem.
  */
-#ifdef DIAGNOSTIC
+#ifdef DEBUG
 int syncprt = 0;
 struct ctldebug debug0 = { "syncprt", &syncprt };
 #endif
@@ -282,6 +356,10 @@ sync(p, uap, retval)
 	int asyncflag;
 
 	for (mp = mountlist.tqh_first; mp != NULL; mp = nmp) {
+		/*
+		 * Get the next pointer in case we hang on vfs_busy
+		 * while we are being unmounted.
+		 */
 		nmp = mp->mnt_list.tqe_next;
 		/*
 		 * The lock check below is to avoid races with mount
@@ -294,6 +372,11 @@ sync(p, uap, retval)
 			VFS_SYNC(mp, MNT_NOWAIT, p->p_ucred, p);
 			if (asyncflag)
 				mp->mnt_flag |= MNT_ASYNC;
+			/*
+			 * Get the next pointer again, as the next filesystem
+			 * might have been unmounted while we were sync'ing.
+			 */
+			nmp = mp->mnt_list.tqe_next;
 			vfs_unbusy(mp);
 		}
 	}
@@ -449,22 +532,36 @@ fchdir(p, uap, retval)
 	int *retval;
 {
 	register struct filedesc *fdp = p->p_fd;
-	register struct vnode *vp;
+	struct vnode *vp, *tdp;
+	struct mount *mp;
 	struct file *fp;
 	int error;
 
 	if (error = getvnode(fdp, uap->fd, &fp))
 		return (error);
 	vp = (struct vnode *)fp->f_data;
+	VREF(vp);
 	VOP_LOCK(vp);
 	if (vp->v_type != VDIR)
 		error = ENOTDIR;
 	else
 		error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
+	while (!error && (mp = vp->v_mountedhere) != NULL) {
+		if (mp->mnt_flag & MNT_MLOCK) {
+			mp->mnt_flag |= MNT_MWAIT;
+			sleep((caddr_t)mp, PVFS);
+			continue;
+		}
+		if (error = VFS_ROOT(mp, &tdp))
+			break;
+		vput(vp);
+		vp = tdp;
+	}
 	VOP_UNLOCK(vp);
-	if (error)
+	if (error) {
+		vrele(vp);
 		return (error);
-	VREF(vp);
+	}
 	vrele(fdp->fd_cdir);
 	fdp->fd_cdir = vp;
 	return (0);
@@ -661,6 +758,7 @@ mknod(p, uap, retval)
 	register struct vnode *vp;
 	struct vattr vattr;
 	int error;
+	int whiteout;
 	struct nameidata nd;
 
 	CHECKPOINTREF;
@@ -676,6 +774,7 @@ mknod(p, uap, retval)
 		VATTR_NULL(&vattr);
 		vattr.va_mode = (uap->mode & ALLPERMS) &~ p->p_fd->fd_cmask;
 		vattr.va_rdev = uap->dev;
+		whiteout = 0;
 
 		switch (uap->mode & S_IFMT) {
 		case S_IFMT:	/* used by badsect to flag bad sectors */
@@ -687,14 +786,25 @@ mknod(p, uap, retval)
 		case S_IFBLK:
 			vattr.va_type = VBLK;
 			break;
+		case S_IFWHT:
+			whiteout = 1;
+			break;
 		default:
 			error = EINVAL;
 			break;
 		}
 	}
 	if (!error) {
-		LEASE_CHECK(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
-		error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
+		VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+		if (whiteout) {
+			error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, CREATE);
+			if (error)
+				VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+			vput(nd.ni_dvp);
+		} else {
+			error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp,
+						&nd.ni_cnd, &vattr);
+		}
 	} else {
 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
 		if (nd.ni_dvp == vp)
@@ -743,7 +853,7 @@ mkfifo(p, uap, retval)
 	VATTR_NULL(&vattr);
 	vattr.va_type = VFIFO;
 	vattr.va_mode = (uap->mode & ALLPERMS) &~ p->p_fd->fd_cmask;
-	LEASE_CHECK(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
 	return (VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr));
 #endif /* FIFO */
 }
@@ -779,10 +889,9 @@ link(p, uap, retval)
 			if (nd.ni_vp != NULL)
 				error = EEXIST;
 			if (!error) {
-				LEASE_CHECK(nd.ni_dvp,
-				    p, p->p_ucred, LEASE_WRITE);
-				LEASE_CHECK(vp,
-				    p, p->p_ucred, LEASE_WRITE);
+				VOP_LEASE(nd.ni_dvp, p, p->p_ucred,
+				    LEASE_WRITE);
+				VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
 				error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd);
 			} else {
 				VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
@@ -837,7 +946,7 @@ symlink(p, uap, retval)
 	}
 	VATTR_NULL(&vattr);
 	vattr.va_mode = ACCESSPERMS &~ p->p_fd->fd_cmask;
-	LEASE_CHECK(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
 	error = VOP_SYMLINK(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr, path);
 out:
 	FREE(path, M_NAMEI);
@@ -845,6 +954,44 @@ out:
 	return (error);
 }
 
+/*
+ * Delete a whiteout from the filesystem.
+ */
+struct undelete_args {
+	char	*path;
+};
+/* ARGSUSED */
+undelete(p, uap, retval)
+	struct proc *p;
+	struct undelete_args *uap;
+	int *retval;
+{
+	int error;
+	struct nameidata nd;
+
+	NDINIT(&nd, DELETE, LOCKPARENT|DOWHITEOUT, UIO_USERSPACE, uap->path, p);
+	error = namei(&nd);
+	if (error)
+		return (error);
+
+	if (nd.ni_vp != NULLVP || !(nd.ni_cnd.cn_flags & ISWHITEOUT)) {
+		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+		if (nd.ni_dvp == nd.ni_vp)
+			vrele(nd.ni_dvp);
+		else
+			vput(nd.ni_dvp);
+		if (nd.ni_vp)
+			vrele(nd.ni_vp);
+		return (EEXIST);
+	}
+
+	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+	if (error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, DELETE))
+		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+	vput(nd.ni_dvp);
+	return (error);
+}
+
 /*
  * Delete a name from the filesystem.
  */
@@ -866,7 +1013,7 @@ unlink(p, uap, retval)
 	if (error = namei(&nd))
 		return (error);
 	vp = nd.ni_vp;
-	LEASE_CHECK(vp, p, p->p_ucred, LEASE_WRITE);
+	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
 	VOP_LOCK(vp);
 
 	if (vp->v_type != VDIR ||
@@ -881,7 +1028,7 @@ unlink(p, uap, retval)
 	}
 
 	if (!error) {
-		LEASE_CHECK(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+		VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
 		error = VOP_REMOVE(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
 	} else {
 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
@@ -889,7 +1036,8 @@ unlink(p, uap, retval)
 			vrele(nd.ni_dvp);
 		else
 			vput(nd.ni_dvp);
-		vput(vp);
+		if (vp != NULLVP)
+			vput(vp);
 	}
 	CHECKREFS("unlink");
 	return (error);
@@ -1056,18 +1204,48 @@ olstat(p, uap, retval)
 	register struct olstat_args *uap;
 	int *retval;
 {
-	struct stat sb;
+	struct vnode *vp, *dvp;
+	struct stat sb, sb1;
 	struct ostat osb;
 	int error;
 	struct nameidata nd;
 
-	NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF, UIO_USERSPACE, uap->path, p);
+	NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | LOCKPARENT, UIO_USERSPACE,
+	    uap->path, p);
 	if (error = namei(&nd))
 		return (error);
-	error = vn_stat(nd.ni_vp, &sb, p);
-	vput(nd.ni_vp);
-	if (error)
-		return (error);
+	/*
+	 * For symbolic links, always return the attributes of its
+	 * containing directory, except for mode, size, and links.
+	 */
+	vp = nd.ni_vp;
+	dvp = nd.ni_dvp;
+	if (vp->v_type != VLNK) {
+		if (dvp == vp)
+			vrele(dvp);
+		else
+			vput(dvp);
+		error = vn_stat(vp, &sb, p);
+		vput(vp);
+		if (error)
+			return (error);
+	} else {
+		error = vn_stat(dvp, &sb, p);
+		vput(dvp);
+		if (error) {
+			vput(vp);
+			return (error);
+		}
+		error = vn_stat(vp, &sb1, p);
+		vput(vp);
+		if (error)
+			return (error);
+		sb.st_mode &= ~S_IFDIR;
+		sb.st_mode |= S_IFLNK;
+		sb.st_nlink = sb1.st_nlink;
+		sb.st_size = sb1.st_size;
+		sb.st_blocks = sb1.st_blocks;
+	}
 	cvtstat(&sb, &osb);
 	error = copyout((caddr_t)&osb, (caddr_t)uap->ub, sizeof (osb));
 	return (error);
@@ -1279,7 +1457,7 @@ chflags(p, uap, retval)
 	if (error = namei(&nd))
 		return (error);
 	vp = nd.ni_vp;
-	LEASE_CHECK(vp, p, p->p_ucred, LEASE_WRITE);
+	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
 	VOP_LOCK(vp);
 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
 		error = EROFS;
@@ -1313,7 +1491,7 @@ fchflags(p, uap, retval)
 	if (error = getvnode(p->p_fd, uap->fd, &fp))
 		return (error);
 	vp = (struct vnode *)fp->f_data;
-	LEASE_CHECK(vp, p, p->p_ucred, LEASE_WRITE);
+	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
 	VOP_LOCK(vp);
 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
 		error = EROFS;
@@ -1348,7 +1526,7 @@ chmod(p, uap, retval)
 	if (error = namei(&nd))
 		return (error);
 	vp = nd.ni_vp;
-	LEASE_CHECK(vp, p, p->p_ucred, LEASE_WRITE);
+	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
 	VOP_LOCK(vp);
 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
 		error = EROFS;
@@ -1382,7 +1560,7 @@ fchmod(p, uap, retval)
 	if (error = getvnode(p->p_fd, uap->fd, &fp))
 		return (error);
 	vp = (struct vnode *)fp->f_data;
-	LEASE_CHECK(vp, p, p->p_ucred, LEASE_WRITE);
+	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
 	VOP_LOCK(vp);
 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
 		error = EROFS;
@@ -1418,7 +1596,7 @@ chown(p, uap, retval)
 	if (error = namei(&nd))
 		return (error);
 	vp = nd.ni_vp;
-	LEASE_CHECK(vp, p, p->p_ucred, LEASE_WRITE);
+	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
 	VOP_LOCK(vp);
 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
 		error = EROFS;
@@ -1454,7 +1632,7 @@ fchown(p, uap, retval)
 	if (error = getvnode(p->p_fd, uap->fd, &fp))
 		return (error);
 	vp = (struct vnode *)fp->f_data;
-	LEASE_CHECK(vp, p, p->p_ucred, LEASE_WRITE);
+	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
 	VOP_LOCK(vp);
 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
 		error = EROFS;
@@ -1498,7 +1676,7 @@ utimes(p, uap, retval)
 	if (error = namei(&nd))
 		return (error);
 	vp = nd.ni_vp;
-	LEASE_CHECK(vp, p, p->p_ucred, LEASE_WRITE);
+	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
 	VOP_LOCK(vp);
 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
 		error = EROFS;
@@ -1536,7 +1714,7 @@ truncate(p, uap, retval)
 	if (error = namei(&nd))
 		return (error);
 	vp = nd.ni_vp;
-	LEASE_CHECK(vp, p, p->p_ucred, LEASE_WRITE);
+	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
 	VOP_LOCK(vp);
 	if (vp->v_type == VDIR)
 		error = EISDIR;
@@ -1574,7 +1752,7 @@ ftruncate(p, uap, retval)
 	if ((fp->f_flag & FWRITE) == 0)
 		return (EINVAL);
 	vp = (struct vnode *)fp->f_data;
-	LEASE_CHECK(vp, p, p->p_ucred, LEASE_WRITE);
+	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
 	VOP_LOCK(vp);
 	if (vp->v_type == VDIR)
 		error = EISDIR;
@@ -1711,11 +1889,11 @@ rename(p, uap, retval)
 		error = -1;
 out:
 	if (!error) {
-		LEASE_CHECK(tdvp, p, p->p_ucred, LEASE_WRITE);
+		VOP_LEASE(tdvp, p, p->p_ucred, LEASE_WRITE);
 		if (fromnd.ni_dvp != tdvp)
-			LEASE_CHECK(fromnd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+			VOP_LEASE(fromnd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
 		if (tvp)
-			LEASE_CHECK(tvp, p, p->p_ucred, LEASE_WRITE);
+			VOP_LEASE(tvp, p, p->p_ucred, LEASE_WRITE);
 		error = VOP_RENAME(fromnd.ni_dvp, fromnd.ni_vp, &fromnd.ni_cnd,
 				   tond.ni_dvp, tond.ni_vp, &tond.ni_cnd);
 	} else {
@@ -1735,7 +1913,8 @@ out:
 	FREE(tond.ni_cnd.cn_pnbuf, M_NAMEI);
 out1:
 	p->p_spare[1]--;
-	vrele(fromnd.ni_startdir);
+	if (fromnd.ni_startdir)
+		vrele(fromnd.ni_startdir);
 	FREE(fromnd.ni_cnd.cn_pnbuf, M_NAMEI);
 	CHECKREFS("rename");
 	if (error == -1)
@@ -1779,7 +1958,7 @@ mkdir(p, uap, retval)
 	VATTR_NULL(&vattr);
 	vattr.va_type = VDIR;
 	vattr.va_mode = (uap->mode & ACCESSPERMS) &~ p->p_fd->fd_cmask;
-	LEASE_CHECK(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
 	error = VOP_MKDIR(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
 	if (!error)
 		vput(nd.ni_vp);
@@ -1826,8 +2005,8 @@ rmdir(p, uap, retval)
 		error = EBUSY;
 out:
 	if (!error) {
-		LEASE_CHECK(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
-		LEASE_CHECK(vp, p, p->p_ucred, LEASE_WRITE);
+		VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+		VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
 		error = VOP_RMDIR(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
 	} else {
 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
@@ -1862,7 +2041,7 @@ ogetdirentries(p, uap, retval)
 	struct iovec aiov, kiov;
 	struct dirent *dp, *edp;
 	caddr_t dirbuf;
-	int error, readcnt;
+	int error, eofflag, readcnt;
 	long loff;
 
 	if (error = getvnode(p->p_fd, uap->fd, &fp))
@@ -1870,6 +2049,7 @@ ogetdirentries(p, uap, retval)
 	if ((fp->f_flag & FREAD) == 0)
 		return (EBADF);
 	vp = (struct vnode *)fp->f_data;
+unionread:
 	if (vp->v_type != VDIR)
 		return (EINVAL);
 	aiov.iov_base = uap->buf;
@@ -1884,7 +2064,8 @@ ogetdirentries(p, uap, retval)
 	loff = auio.uio_offset = fp->f_offset;
 #	if (BYTE_ORDER != LITTLE_ENDIAN)
 		if (vp->v_mount->mnt_maxsymlinklen <= 0) {
-			error = VOP_READDIR(vp, &auio, fp->f_cred);
+			error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag,
+			    (u_long *)0, 0);
 			fp->f_offset = auio.uio_offset;
 		} else
 #	endif
@@ -1895,7 +2076,8 @@ ogetdirentries(p, uap, retval)
 		kiov.iov_len = uap->count;
 		MALLOC(dirbuf, caddr_t, uap->count, M_TEMP, M_WAITOK);
 		kiov.iov_base = dirbuf;
-		error = VOP_READDIR(vp, &kuio, fp->f_cred);
+		error = VOP_READDIR(vp, &kuio, fp->f_cred, &eofflag,
+			    (u_long *)0, 0);
 		fp->f_offset = kuio.uio_offset;
 		if (error == 0) {
 			readcnt = uap->count - kuio.uio_resid;
@@ -1934,11 +2116,67 @@ ogetdirentries(p, uap, retval)
 	VOP_UNLOCK(vp);
 	if (error)
 		return (error);
+
+#ifdef UNION
+{
+	extern int (**union_vnodeop_p)();
+	extern struct vnode *union_dircache __P((struct vnode *));
+
+	if ((uap->count == auio.uio_resid) &&
+	    (vp->v_op == union_vnodeop_p)) {
+		struct vnode *lvp;
+
+		lvp = union_dircache(vp);
+		if (lvp != NULLVP) {
+			struct vattr va;
+
+			/*
+			 * If the directory is opaque,
+			 * then don't show lower entries
+			 */
+			error = VOP_GETATTR(vp, &va, fp->f_cred, p);
+			if (va.va_flags & OPAQUE) {
+				vput(lvp);
+				lvp = NULL;
+			}
+		}
+		
+		if (lvp != NULLVP) {
+			error = VOP_OPEN(lvp, FREAD, fp->f_cred, p);
+			VOP_UNLOCK(lvp);
+
+			if (error) {
+				vrele(lvp);
+				return (error);
+			}
+			fp->f_data = (caddr_t) lvp;
+			fp->f_offset = 0;
+			error = vn_close(vp, FREAD, fp->f_cred, p);
+			if (error)
+				return (error);
+			vp = lvp;
+			goto unionread;
+		}
+	}
+}
+#endif /* UNION */
+
+	if ((uap->count == auio.uio_resid) &&
+	    (vp->v_flag & VROOT) &&
+	    (vp->v_mount->mnt_flag & MNT_UNION)) {
+		struct vnode *tvp = vp;
+		vp = vp->v_mount->mnt_vnodecovered;
+		VREF(vp);
+		fp->f_data = (caddr_t) vp;
+		fp->f_offset = 0;
+		vrele(tvp);
+		goto unionread;
+	}
 	error = copyout((caddr_t)&loff, (caddr_t)uap->basep, sizeof(long));
 	*retval = uap->count - auio.uio_resid;
 	return (error);
 }
-#endif
+#endif /* COMPAT_43 */
 
 /*
  * Read a block of directory entries in a file system independent format.
@@ -1959,7 +2197,7 @@ getdirentries(p, uap, retval)
 	struct uio auio;
 	struct iovec aiov;
 	long loff;
-	int error;
+	int error, eofflag;
 
 	if (error = getvnode(p->p_fd, uap->fd, &fp))
 		return (error);
@@ -1979,7 +2217,7 @@ unionread:
 	auio.uio_resid = uap->count;
 	VOP_LOCK(vp);
 	loff = auio.uio_offset = fp->f_offset;
-	error = VOP_READDIR(vp, &auio, fp->f_cred);
+	error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, (u_long *)0, 0);
 	fp->f_offset = auio.uio_offset;
 	VOP_UNLOCK(vp);
 	if (error)
@@ -1988,27 +2226,41 @@ unionread:
 #ifdef UNION
 {
 	extern int (**union_vnodeop_p)();
-	extern struct vnode *union_lowervp __P((struct vnode *));
+	extern struct vnode *union_dircache __P((struct vnode *));
 
 	if ((uap->count == auio.uio_resid) &&
 	    (vp->v_op == union_vnodeop_p)) {
-		struct vnode *tvp = vp;
+		struct vnode *lvp;
 
-		vp = union_lowervp(vp);
-		if (vp != NULLVP) {
-			VOP_LOCK(vp);
-			error = VOP_OPEN(vp, FREAD);
-			VOP_UNLOCK(vp);
+		lvp = union_dircache(vp);
+		if (lvp != NULLVP) {
+			struct vattr va;
+
+			/*
+			 * If the directory is opaque,
+			 * then don't show lower entries
+			 */
+			error = VOP_GETATTR(vp, &va, fp->f_cred, p);
+			if (va.va_flags & OPAQUE) {
+				vput(lvp);
+				lvp = NULL;
+			}
+		}
+		
+		if (lvp != NULLVP) {
+			error = VOP_OPEN(lvp, FREAD, fp->f_cred, p);
+			VOP_UNLOCK(lvp);
 
 			if (error) {
-				vrele(vp);
+				vrele(lvp);
 				return (error);
 			}
-			fp->f_data = (caddr_t) vp;
+			fp->f_data = (caddr_t) lvp;
 			fp->f_offset = 0;
-			error = vn_close(tvp, FREAD, fp->f_cred, p);
+			error = vn_close(vp, FREAD, fp->f_cred, p);
 			if (error)
 				return (error);
+			vp = lvp;
 			goto unionread;
 		}
 	}