xref: /netbsd-src/sys/kern/vfs_syscalls.c (revision 81b108b45f75f89f1e3ffad9fb6f074e771c0935)
1 /*	$NetBSD: vfs_syscalls.c,v 1.71 1996/04/23 10:29:02 mycroft Exp $	*/
2 
3 /*
4  * Copyright (c) 1989, 1993
5  *	The Regents of the University of California.  All rights reserved.
6  * (c) UNIX System Laboratories, Inc.
7  * All or some portions of this file are derived from material licensed
8  * to the University of California by American Telephone and Telegraph
9  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
10  * the permission of UNIX System Laboratories, Inc.
11  *
12  * Redistribution and use in source and binary forms, with or without
13  * modification, are permitted provided that the following conditions
14  * are met:
15  * 1. Redistributions of source code must retain the above copyright
16  *    notice, this list of conditions and the following disclaimer.
17  * 2. Redistributions in binary form must reproduce the above copyright
18  *    notice, this list of conditions and the following disclaimer in the
19  *    documentation and/or other materials provided with the distribution.
20  * 3. All advertising materials mentioning features or use of this software
21  *    must display the following acknowledgement:
22  *	This product includes software developed by the University of
23  *	California, Berkeley and its contributors.
24  * 4. Neither the name of the University nor the names of its contributors
25  *    may be used to endorse or promote products derived from this software
26  *    without specific prior written permission.
27  *
28  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
29  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
30  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
31  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
32  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
33  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
34  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
35  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
36  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
37  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
38  * SUCH DAMAGE.
39  *
40  *	@(#)vfs_syscalls.c	8.28 (Berkeley) 12/10/94
41  */
42 
43 #include <sys/param.h>
44 #include <sys/systm.h>
45 #include <sys/namei.h>
46 #include <sys/filedesc.h>
47 #include <sys/kernel.h>
48 #include <sys/file.h>
49 #include <sys/stat.h>
50 #include <sys/vnode.h>
51 #include <sys/mount.h>
52 #include <sys/proc.h>
53 #include <sys/uio.h>
54 #include <sys/malloc.h>
55 #include <sys/dirent.h>
56 
57 #include <sys/syscallargs.h>
58 
59 #include <vm/vm.h>
60 #include <sys/sysctl.h>
61 
62 static int change_dir __P((struct nameidata *, struct proc *));
63 
64 void checkdirs __P((struct vnode *));
65 int dounmount __P((struct mount *, int, struct proc *));
66 
67 /*
68  * Virtual File System System Calls
69  */
70 
71 /*
72  * Mount a file system.
73  */
74 /* ARGSUSED */
75 int
76 sys_mount(p, v, retval)
77 	struct proc *p;
78 	void *v;
79 	register_t *retval;
80 {
81 	register struct sys_mount_args /* {
82 		syscallarg(char *) type;
83 		syscallarg(char *) path;
84 		syscallarg(int) flags;
85 		syscallarg(caddr_t) data;
86 	} */ *uap = v;
87 	register struct vnode *vp;
88 	register struct mount *mp;
89 	int error, flag = 0;
90 	u_long fsindex = 0;
91 	char fstypename[MFSNAMELEN];
92 	struct vattr va;
93 	struct nameidata nd;
94 
95 	/*
96 	 * Get vnode to be covered
97 	 */
98 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
99 	    SCARG(uap, path), p);
100 	if ((error = namei(&nd)) != 0)
101 		return (error);
102 	vp = nd.ni_vp;
103 	if (SCARG(uap, flags) & MNT_UPDATE) {
104 		if ((vp->v_flag & VROOT) == 0) {
105 			vput(vp);
106 			return (EINVAL);
107 		}
108 		mp = vp->v_mount;
109 		flag = mp->mnt_flag;
110 		/*
111 		 * We only allow the filesystem to be reloaded if it
112 		 * is currently mounted read-only.
113 		 */
114 		if ((SCARG(uap, flags) & MNT_RELOAD) &&
115 		    ((mp->mnt_flag & MNT_RDONLY) == 0)) {
116 			vput(vp);
117 			return (EOPNOTSUPP);	/* Needs translation */
118 		}
119 		mp->mnt_flag |=
120 		    SCARG(uap, flags) & (MNT_RELOAD | MNT_FORCE | MNT_UPDATE);
121 		/*
122 		 * Only root, or the user that did the original mount is
123 		 * permitted to update it.
124 		 */
125 		if (mp->mnt_stat.f_owner != p->p_ucred->cr_uid &&
126 		    (error = suser(p->p_ucred, &p->p_acflag))) {
127 			vput(vp);
128 			return (error);
129 		}
130 		/*
131 		 * Do not allow NFS export by non-root users. Silently
132 		 * enforce MNT_NOSUID and MNT_NODEV for non-root users.
133 		 */
134 		if (p->p_ucred->cr_uid != 0) {
135 			if (SCARG(uap, flags) & MNT_EXPORTED) {
136 				vput(vp);
137 				return (EPERM);
138 			}
139 			SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
140 		}
141 		VOP_UNLOCK(vp);
142 		goto update;
143 	}
144 	/*
145 	 * If the user is not root, ensure that they own the directory
146 	 * onto which we are attempting to mount.
147 	 */
148 	if ((error = VOP_GETATTR(vp, &va, p->p_ucred, p)) ||
149 	    (va.va_uid != p->p_ucred->cr_uid &&
150 	     (error = suser(p->p_ucred, &p->p_acflag)))) {
151 		vput(vp);
152 		return (error);
153 	}
154 	/*
155 	 * Do not allow NFS export by non-root users. Silently
156 	 * enforce MNT_NOSUID and MNT_NODEV for non-root users.
157 	 */
158 	if (p->p_ucred->cr_uid != 0) {
159 		if (SCARG(uap, flags) & MNT_EXPORTED) {
160 			vput(vp);
161 			return (EPERM);
162 		}
163 		SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
164 	}
165 	if ((error = vinvalbuf(vp, V_SAVE, p->p_ucred, p, 0, 0)) != 0)
166 		return (error);
167 	if (vp->v_type != VDIR) {
168 		vput(vp);
169 		return (ENOTDIR);
170 	}
171 	error = copyinstr(SCARG(uap, type), fstypename, MFSNAMELEN, NULL);
172 	if (error) {
173 #if defined(COMPAT_09) || defined(COMPAT_43)
174 		/*
175 		 * Historically filesystem types were identified by number.
176 		 * If we get an integer for the filesystem type instead of a
177 		 * string, we check to see if it matches one of the historic
178 		 * filesystem types.
179 		 */
180 		fsindex = (u_long)SCARG(uap, type);
181 		if (fsindex >= nvfssw || vfssw[fsindex] == NULL) {
182 			vput(vp);
183 			return (ENODEV);
184 		}
185 		strncpy(fstypename, vfssw[fsindex]->vfs_name, MFSNAMELEN);
186 #else
187 		vput(vp);
188 		return (error);
189 #endif
190 	}
191 #ifdef	COMPAT_10
192 	/* Accept `ufs' as an alias for `ffs'. */
193 	if (!strncmp(fstypename, "ufs", MFSNAMELEN))
194 		strncpy(fstypename, "ffs", MFSNAMELEN);
195 #endif
196 	for (fsindex = 0; fsindex < nvfssw; fsindex++)
197 		if (vfssw[fsindex] != NULL &&
198 		    !strncmp(vfssw[fsindex]->vfs_name, fstypename, MFSNAMELEN))
199 			break;
200 	if (fsindex >= nvfssw) {
201 		vput(vp);
202 		return (ENODEV);
203 	}
204 	if (vp->v_mountedhere != NULL) {
205 		vput(vp);
206 		return (EBUSY);
207 	}
208 
209 	/*
210 	 * Allocate and initialize the file system.
211 	 */
212 	mp = (struct mount *)malloc((u_long)sizeof(struct mount),
213 		M_MOUNT, M_WAITOK);
214 	bzero((char *)mp, (u_long)sizeof(struct mount));
215 	mp->mnt_op = vfssw[fsindex];
216 	if ((error = vfs_lock(mp)) != 0) {
217 		free((caddr_t)mp, M_MOUNT);
218 		vput(vp);
219 		return (error);
220 	}
221 	/* Do this early in case we block later. */
222 	vfssw[fsindex]->vfs_refcount++;
223 	vp->v_mountedhere = mp;
224 	mp->mnt_vnodecovered = vp;
225 	mp->mnt_stat.f_owner = p->p_ucred->cr_uid;
226 update:
227 	/*
228 	 * Set the mount level flags.
229 	 */
230 	if (SCARG(uap, flags) & MNT_RDONLY)
231 		mp->mnt_flag |= MNT_RDONLY;
232 	else if (mp->mnt_flag & MNT_RDONLY)
233 		mp->mnt_flag |= MNT_WANTRDWR;
234 	mp->mnt_flag &=~ (MNT_NOSUID | MNT_NOEXEC | MNT_NODEV |
235 	    MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC);
236 	mp->mnt_flag |= SCARG(uap, flags) & (MNT_NOSUID | MNT_NOEXEC |
237 	    MNT_NODEV | MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC);
238 	/*
239 	 * Mount the filesystem.
240 	 */
241 	error = VFS_MOUNT(mp, SCARG(uap, path), SCARG(uap, data), &nd, p);
242 	if (mp->mnt_flag & MNT_UPDATE) {
243 		vrele(vp);
244 		if (mp->mnt_flag & MNT_WANTRDWR)
245 			mp->mnt_flag &= ~MNT_RDONLY;
246 		mp->mnt_flag &=~
247 		    (MNT_UPDATE | MNT_RELOAD | MNT_FORCE | MNT_WANTRDWR);
248 		if (error)
249 			mp->mnt_flag = flag;
250 		return (error);
251 	}
252 	/*
253 	 * Put the new filesystem on the mount list after root.
254 	 */
255 	cache_purge(vp);
256 	if (!error) {
257 		CIRCLEQ_INSERT_TAIL(&mountlist, mp, mnt_list);
258 		checkdirs(vp);
259 		VOP_UNLOCK(vp);
260 		vfs_unlock(mp);
261 		(void) VFS_STATFS(mp, &mp->mnt_stat, p);
262 		error = VFS_START(mp, 0, p);
263 	} else {
264 		mp->mnt_vnodecovered->v_mountedhere = (struct mount *)0;
265 		vfssw[fsindex]->vfs_refcount--;
266 		vfs_unlock(mp);
267 		free((caddr_t)mp, M_MOUNT);
268 		vput(vp);
269 	}
270 	return (error);
271 }
272 
273 /*
274  * Scan all active processes to see if any of them have a current
275  * or root directory onto which the new filesystem has just been
276  * mounted. If so, replace them with the new mount point.
277  */
278 void
279 checkdirs(olddp)
280 	struct vnode *olddp;
281 {
282 	struct filedesc *fdp;
283 	struct vnode *newdp;
284 	struct proc *p;
285 
286 	if (olddp->v_usecount == 1)
287 		return;
288 	if (VFS_ROOT(olddp->v_mountedhere, &newdp))
289 		panic("mount: lost mount");
290 	for (p = allproc.lh_first; p != 0; p = p->p_list.le_next) {
291 		fdp = p->p_fd;
292 		if (fdp->fd_cdir == olddp) {
293 			vrele(fdp->fd_cdir);
294 			VREF(newdp);
295 			fdp->fd_cdir = newdp;
296 		}
297 		if (fdp->fd_rdir == olddp) {
298 			vrele(fdp->fd_rdir);
299 			VREF(newdp);
300 			fdp->fd_rdir = newdp;
301 		}
302 	}
303 	if (rootvnode == olddp) {
304 		vrele(rootvnode);
305 		VREF(newdp);
306 		rootvnode = newdp;
307 	}
308 	vput(newdp);
309 }
310 
311 /*
312  * Unmount a file system.
313  *
314  * Note: unmount takes a path to the vnode mounted on as argument,
315  * not special file (as before).
316  */
317 /* ARGSUSED */
318 int
319 sys_unmount(p, v, retval)
320 	struct proc *p;
321 	void *v;
322 	register_t *retval;
323 {
324 	register struct sys_unmount_args /* {
325 		syscallarg(char *) path;
326 		syscallarg(int) flags;
327 	} */ *uap = v;
328 	register struct vnode *vp;
329 	struct mount *mp;
330 	int error;
331 	struct nameidata nd;
332 
333 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
334 	    SCARG(uap, path), p);
335 	if ((error = namei(&nd)) != 0)
336 		return (error);
337 	vp = nd.ni_vp;
338 	mp = vp->v_mount;
339 
340 	/*
341 	 * Only root, or the user that did the original mount is
342 	 * permitted to unmount this filesystem.
343 	 */
344 	if ((mp->mnt_stat.f_owner != p->p_ucred->cr_uid) &&
345 	    (error = suser(p->p_ucred, &p->p_acflag))) {
346 		vput(vp);
347 		return (error);
348 	}
349 
350 	/*
351 	 * Don't allow unmounting the root file system.
352 	 */
353 	if (mp->mnt_flag & MNT_ROOTFS) {
354 		vput(vp);
355 		return (EINVAL);
356 	}
357 
358 	/*
359 	 * Must be the root of the filesystem
360 	 */
361 	if ((vp->v_flag & VROOT) == 0) {
362 		vput(vp);
363 		return (EINVAL);
364 	}
365 	vput(vp);
366 	return (dounmount(mp, SCARG(uap, flags), p));
367 }
368 
369 /*
370  * Do the actual file system unmount.
371  */
372 int
373 dounmount(mp, flags, p)
374 	register struct mount *mp;
375 	int flags;
376 	struct proc *p;
377 {
378 	struct vnode *coveredvp;
379 	int error;
380 
381 	coveredvp = mp->mnt_vnodecovered;
382 	if (vfs_busy(mp))
383 		return (EBUSY);
384 	mp->mnt_flag |= MNT_UNMOUNT;
385 	if ((error = vfs_lock(mp)) != 0)
386 		return (error);
387 
388 	mp->mnt_flag &=~ MNT_ASYNC;
389 	vnode_pager_umount(mp);	/* release cached vnodes */
390 	cache_purgevfs(mp);	/* remove cache entries for this file sys */
391 	if ((error = VFS_SYNC(mp, MNT_WAIT, p->p_ucred, p)) == 0 ||
392 	    (flags & MNT_FORCE))
393 		error = VFS_UNMOUNT(mp, flags, p);
394 	mp->mnt_flag &= ~MNT_UNMOUNT;
395 	vfs_unbusy(mp);
396 	if (error) {
397 		vfs_unlock(mp);
398 	} else {
399 		CIRCLEQ_REMOVE(&mountlist, mp, mnt_list);
400 		if (coveredvp != NULLVP) {
401 			vrele(coveredvp);
402 			coveredvp->v_mountedhere = (struct mount *)0;
403 		}
404 		mp->mnt_op->vfs_refcount--;
405 		vfs_unlock(mp);
406 		if (mp->mnt_vnodelist.lh_first != NULL)
407 			panic("unmount: dangling vnode");
408 		free((caddr_t)mp, M_MOUNT);
409 	}
410 	return (error);
411 }
412 
413 /*
414  * Sync each mounted filesystem.
415  */
416 #ifdef DEBUG
417 int syncprt = 0;
418 struct ctldebug debug0 = { "syncprt", &syncprt };
419 #endif
420 
421 /* ARGSUSED */
422 int
423 sys_sync(p, v, retval)
424 	struct proc *p;
425 	void *v;
426 	register_t *retval;
427 {
428 	register struct mount *mp, *nmp;
429 	int asyncflag;
430 
431 	for (mp = mountlist.cqh_first; mp != (void *)&mountlist; mp = nmp) {
432 		/*
433 		 * Get the next pointer in case we hang on vfs_busy
434 		 * while we are being unmounted.
435 		 */
436 		nmp = mp->mnt_list.cqe_next;
437 		/*
438 		 * The lock check below is to avoid races with mount
439 		 * and unmount.
440 		 */
441 		if ((mp->mnt_flag & (MNT_MLOCK|MNT_RDONLY|MNT_MPBUSY)) == 0 &&
442 		    !vfs_busy(mp)) {
443 			asyncflag = mp->mnt_flag & MNT_ASYNC;
444 			mp->mnt_flag &= ~MNT_ASYNC;
445 			VFS_SYNC(mp, MNT_NOWAIT, p->p_ucred, p);
446 			if (asyncflag)
447 				mp->mnt_flag |= MNT_ASYNC;
448 			/*
449 			 * Get the next pointer again, as the next filesystem
450 			 * might have been unmounted while we were sync'ing.
451 			 */
452 			nmp = mp->mnt_list.cqe_next;
453 			vfs_unbusy(mp);
454 		}
455 	}
456 #ifdef DEBUG
457 	if (syncprt)
458 		vfs_bufstats();
459 #endif /* DEBUG */
460 	return (0);
461 }
462 
463 /*
464  * Change filesystem quotas.
465  */
466 /* ARGSUSED */
467 int
468 sys_quotactl(p, v, retval)
469 	struct proc *p;
470 	void *v;
471 	register_t *retval;
472 {
473 	register struct sys_quotactl_args /* {
474 		syscallarg(char *) path;
475 		syscallarg(int) cmd;
476 		syscallarg(int) uid;
477 		syscallarg(caddr_t) arg;
478 	} */ *uap = v;
479 	register struct mount *mp;
480 	int error;
481 	struct nameidata nd;
482 
483 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
484 	if ((error = namei(&nd)) != 0)
485 		return (error);
486 	mp = nd.ni_vp->v_mount;
487 	vrele(nd.ni_vp);
488 	return (VFS_QUOTACTL(mp, SCARG(uap, cmd), SCARG(uap, uid),
489 	    SCARG(uap, arg), p));
490 }
491 
492 /*
493  * Get filesystem statistics.
494  */
495 /* ARGSUSED */
496 int
497 sys_statfs(p, v, retval)
498 	struct proc *p;
499 	void *v;
500 	register_t *retval;
501 {
502 	register struct sys_statfs_args /* {
503 		syscallarg(char *) path;
504 		syscallarg(struct statfs *) buf;
505 	} */ *uap = v;
506 	register struct mount *mp;
507 	register struct statfs *sp;
508 	int error;
509 	struct nameidata nd;
510 
511 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
512 	if ((error = namei(&nd)) != 0)
513 		return (error);
514 	mp = nd.ni_vp->v_mount;
515 	sp = &mp->mnt_stat;
516 	vrele(nd.ni_vp);
517 	if ((error = VFS_STATFS(mp, sp, p)) != 0)
518 		return (error);
519 	sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
520 	return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
521 }
522 
523 /*
524  * Get filesystem statistics.
525  */
526 /* ARGSUSED */
527 int
528 sys_fstatfs(p, v, retval)
529 	struct proc *p;
530 	void *v;
531 	register_t *retval;
532 {
533 	register struct sys_fstatfs_args /* {
534 		syscallarg(int) fd;
535 		syscallarg(struct statfs *) buf;
536 	} */ *uap = v;
537 	struct file *fp;
538 	struct mount *mp;
539 	register struct statfs *sp;
540 	int error;
541 
542 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
543 		return (error);
544 	mp = ((struct vnode *)fp->f_data)->v_mount;
545 	sp = &mp->mnt_stat;
546 	if ((error = VFS_STATFS(mp, sp, p)) != 0)
547 		return (error);
548 	sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
549 	return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
550 }
551 
552 /*
553  * Get statistics on all filesystems.
554  */
555 int
556 sys_getfsstat(p, v, retval)
557 	struct proc *p;
558 	void *v;
559 	register_t *retval;
560 {
561 	register struct sys_getfsstat_args /* {
562 		syscallarg(struct statfs *) buf;
563 		syscallarg(long) bufsize;
564 		syscallarg(int) flags;
565 	} */ *uap = v;
566 	register struct mount *mp, *nmp;
567 	register struct statfs *sp;
568 	caddr_t sfsp;
569 	long count, maxcount, error;
570 
571 	maxcount = SCARG(uap, bufsize) / sizeof(struct statfs);
572 	sfsp = (caddr_t)SCARG(uap, buf);
573 	for (count = 0,
574 	     mp = mountlist.cqh_first; mp != (void *)&mountlist; mp = nmp) {
575 		nmp = mp->mnt_list.cqe_next;
576 		if (sfsp && count < maxcount &&
577 		    ((mp->mnt_flag & MNT_MLOCK) == 0)) {
578 			sp = &mp->mnt_stat;
579 			/*
580 			 * If MNT_NOWAIT is specified, do not refresh the
581 			 * fsstat cache. MNT_WAIT overrides MNT_NOWAIT.
582 			 */
583 			if (((SCARG(uap, flags) & MNT_NOWAIT) == 0 ||
584 			    (SCARG(uap, flags) & MNT_WAIT)) &&
585 			    (error = VFS_STATFS(mp, sp, p)))
586 				continue;
587 			sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
588 			error = copyout((caddr_t)sp, sfsp, sizeof(*sp));
589 			if (error)
590 				return (error);
591 			sfsp += sizeof(*sp);
592 		}
593 		count++;
594 	}
595 	if (sfsp && count > maxcount)
596 		*retval = maxcount;
597 	else
598 		*retval = count;
599 	return (0);
600 }
601 
602 /*
603  * Change current working directory to a given file descriptor.
604  */
605 /* ARGSUSED */
606 int
607 sys_fchdir(p, v, retval)
608 	struct proc *p;
609 	void *v;
610 	register_t *retval;
611 {
612 	struct sys_fchdir_args /* {
613 		syscallarg(int) fd;
614 	} */ *uap = v;
615 	register struct filedesc *fdp = p->p_fd;
616 	struct vnode *vp, *tdp;
617 	struct mount *mp;
618 	struct file *fp;
619 	int error;
620 
621 	if ((error = getvnode(fdp, SCARG(uap, fd), &fp)) != 0)
622 		return (error);
623 	vp = (struct vnode *)fp->f_data;
624 	VREF(vp);
625 	VOP_LOCK(vp);
626 	if (vp->v_type != VDIR)
627 		error = ENOTDIR;
628 	else
629 		error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
630 	while (!error && (mp = vp->v_mountedhere) != NULL) {
631 		if (mp->mnt_flag & MNT_MLOCK) {
632 			mp->mnt_flag |= MNT_MWAIT;
633 			sleep((caddr_t)mp, PVFS);
634 			continue;
635 		}
636 		if ((error = VFS_ROOT(mp, &tdp)) != 0)
637 			break;
638 		vput(vp);
639 		vp = tdp;
640 	}
641 	VOP_UNLOCK(vp);
642 	if (error) {
643 		vrele(vp);
644 		return (error);
645 	}
646 	vrele(fdp->fd_cdir);
647 	fdp->fd_cdir = vp;
648 	return (0);
649 }
650 
651 /*
652  * Change current working directory (``.'').
653  */
654 /* ARGSUSED */
655 int
656 sys_chdir(p, v, retval)
657 	struct proc *p;
658 	void *v;
659 	register_t *retval;
660 {
661 	struct sys_chdir_args /* {
662 		syscallarg(char *) path;
663 	} */ *uap = v;
664 	register struct filedesc *fdp = p->p_fd;
665 	int error;
666 	struct nameidata nd;
667 
668 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
669 	    SCARG(uap, path), p);
670 	if ((error = change_dir(&nd, p)) != 0)
671 		return (error);
672 	vrele(fdp->fd_cdir);
673 	fdp->fd_cdir = nd.ni_vp;
674 	return (0);
675 }
676 
677 /*
678  * Change notion of root (``/'') directory.
679  */
680 /* ARGSUSED */
681 int
682 sys_chroot(p, v, retval)
683 	struct proc *p;
684 	void *v;
685 	register_t *retval;
686 {
687 	struct sys_chroot_args /* {
688 		syscallarg(char *) path;
689 	} */ *uap = v;
690 	register struct filedesc *fdp = p->p_fd;
691 	int error;
692 	struct nameidata nd;
693 
694 	if ((error = suser(p->p_ucred, &p->p_acflag)) != 0)
695 		return (error);
696 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
697 	    SCARG(uap, path), p);
698 	if ((error = change_dir(&nd, p)) != 0)
699 		return (error);
700 	if (fdp->fd_rdir != NULL)
701 		vrele(fdp->fd_rdir);
702 	fdp->fd_rdir = nd.ni_vp;
703 	return (0);
704 }
705 
706 /*
707  * Common routine for chroot and chdir.
708  */
709 static int
710 change_dir(ndp, p)
711 	register struct nameidata *ndp;
712 	struct proc *p;
713 {
714 	struct vnode *vp;
715 	int error;
716 
717 	if ((error = namei(ndp)) != 0)
718 		return (error);
719 	vp = ndp->ni_vp;
720 	if (vp->v_type != VDIR)
721 		error = ENOTDIR;
722 	else
723 		error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
724 	VOP_UNLOCK(vp);
725 	if (error)
726 		vrele(vp);
727 	return (error);
728 }
729 
730 /*
731  * Check permissions, allocate an open file structure,
732  * and call the device open routine if any.
733  */
734 int
735 sys_open(p, v, retval)
736 	struct proc *p;
737 	void *v;
738 	register_t *retval;
739 {
740 	register struct sys_open_args /* {
741 		syscallarg(char *) path;
742 		syscallarg(int) flags;
743 		syscallarg(int) mode;
744 	} */ *uap = v;
745 	register struct filedesc *fdp = p->p_fd;
746 	register struct file *fp;
747 	register struct vnode *vp;
748 	int flags, cmode;
749 	struct file *nfp;
750 	int type, indx, error;
751 	struct flock lf;
752 	struct nameidata nd;
753 	extern struct fileops vnops;
754 
755 	if ((error = falloc(p, &nfp, &indx)) != 0)
756 		return (error);
757 	fp = nfp;
758 	flags = FFLAGS(SCARG(uap, flags));
759 	cmode = ((SCARG(uap, mode) &~ fdp->fd_cmask) & ALLPERMS) &~ S_ISTXT;
760 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
761 	p->p_dupfd = -indx - 1;			/* XXX check for fdopen */
762 	if ((error = vn_open(&nd, flags, cmode)) != 0) {
763 		ffree(fp);
764 		if ((error == ENODEV || error == ENXIO) &&
765 		    p->p_dupfd >= 0 &&			/* XXX from fdopen */
766 		    (error =
767 			dupfdopen(fdp, indx, p->p_dupfd, flags, error)) == 0) {
768 			*retval = indx;
769 			return (0);
770 		}
771 		if (error == ERESTART)
772 			error = EINTR;
773 		fdp->fd_ofiles[indx] = NULL;
774 		return (error);
775 	}
776 	p->p_dupfd = 0;
777 	vp = nd.ni_vp;
778 	fp->f_flag = flags & FMASK;
779 	fp->f_type = DTYPE_VNODE;
780 	fp->f_ops = &vnops;
781 	fp->f_data = (caddr_t)vp;
782 	if (flags & (O_EXLOCK | O_SHLOCK)) {
783 		lf.l_whence = SEEK_SET;
784 		lf.l_start = 0;
785 		lf.l_len = 0;
786 		if (flags & O_EXLOCK)
787 			lf.l_type = F_WRLCK;
788 		else
789 			lf.l_type = F_RDLCK;
790 		type = F_FLOCK;
791 		if ((flags & FNONBLOCK) == 0)
792 			type |= F_WAIT;
793 		VOP_UNLOCK(vp);
794 		error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type);
795 		if (error) {
796 			(void) vn_close(vp, fp->f_flag, fp->f_cred, p);
797 			ffree(fp);
798 			fdp->fd_ofiles[indx] = NULL;
799 			return (error);
800 		}
801 		VOP_LOCK(vp);
802 		fp->f_flag |= FHASLOCK;
803 	}
804 	VOP_UNLOCK(vp);
805 	*retval = indx;
806 	return (0);
807 }
808 
809 /*
810  * Create a special file.
811  */
812 /* ARGSUSED */
813 int
814 sys_mknod(p, v, retval)
815 	struct proc *p;
816 	void *v;
817 	register_t *retval;
818 {
819 	register struct sys_mknod_args /* {
820 		syscallarg(char *) path;
821 		syscallarg(int) mode;
822 		syscallarg(int) dev;
823 	} */ *uap = v;
824 	register struct vnode *vp;
825 	struct vattr vattr;
826 	int error;
827 	int whiteout = 0;
828 	struct nameidata nd;
829 
830 	if ((error = suser(p->p_ucred, &p->p_acflag)) != 0)
831 		return (error);
832 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
833 	if ((error = namei(&nd)) != 0)
834 		return (error);
835 	vp = nd.ni_vp;
836 	if (vp != NULL)
837 		error = EEXIST;
838 	else {
839 		VATTR_NULL(&vattr);
840 		vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
841 		vattr.va_rdev = SCARG(uap, dev);
842 		whiteout = 0;
843 
844 		switch (SCARG(uap, mode) & S_IFMT) {
845 		case S_IFMT:	/* used by badsect to flag bad sectors */
846 			vattr.va_type = VBAD;
847 			break;
848 		case S_IFCHR:
849 			vattr.va_type = VCHR;
850 			break;
851 		case S_IFBLK:
852 			vattr.va_type = VBLK;
853 			break;
854 		case S_IFWHT:
855 			whiteout = 1;
856 			break;
857 		default:
858 			error = EINVAL;
859 			break;
860 		}
861 	}
862 	if (!error) {
863 		VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
864 		if (whiteout) {
865 			error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, CREATE);
866 			if (error)
867 				VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
868 			vput(nd.ni_dvp);
869 		} else {
870 			error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp,
871 						&nd.ni_cnd, &vattr);
872 		}
873 	} else {
874 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
875 		if (nd.ni_dvp == vp)
876 			vrele(nd.ni_dvp);
877 		else
878 			vput(nd.ni_dvp);
879 		if (vp)
880 			vrele(vp);
881 	}
882 	return (error);
883 }
884 
885 /*
886  * Create a named pipe.
887  */
888 /* ARGSUSED */
889 int
890 sys_mkfifo(p, v, retval)
891 	struct proc *p;
892 	void *v;
893 	register_t *retval;
894 {
895 #ifndef FIFO
896 	return (EOPNOTSUPP);
897 #else
898 	register struct sys_mkfifo_args /* {
899 		syscallarg(char *) path;
900 		syscallarg(int) mode;
901 	} */ *uap = v;
902 	struct vattr vattr;
903 	int error;
904 	struct nameidata nd;
905 
906 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
907 	if ((error = namei(&nd)) != 0)
908 		return (error);
909 	if (nd.ni_vp != NULL) {
910 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
911 		if (nd.ni_dvp == nd.ni_vp)
912 			vrele(nd.ni_dvp);
913 		else
914 			vput(nd.ni_dvp);
915 		vrele(nd.ni_vp);
916 		return (EEXIST);
917 	}
918 	VATTR_NULL(&vattr);
919 	vattr.va_type = VFIFO;
920 	vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
921 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
922 	return (VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr));
923 #endif /* FIFO */
924 }
925 
926 /*
927  * Make a hard file link.
928  */
929 /* ARGSUSED */
930 int
931 sys_link(p, v, retval)
932 	struct proc *p;
933 	void *v;
934 	register_t *retval;
935 {
936 	register struct sys_link_args /* {
937 		syscallarg(char *) path;
938 		syscallarg(char *) link;
939 	} */ *uap = v;
940 	register struct vnode *vp;
941 	struct nameidata nd;
942 	int error;
943 
944 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
945 	if ((error = namei(&nd)) != 0)
946 		return (error);
947 	vp = nd.ni_vp;
948 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, link), p);
949 	if ((error = namei(&nd)) != 0)
950 		goto out;
951 	if (nd.ni_vp) {
952 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
953 		if (nd.ni_dvp == nd.ni_vp)
954 			vrele(nd.ni_dvp);
955 		else
956 			vput(nd.ni_dvp);
957 		vrele(nd.ni_vp);
958 		error = EEXIST;
959 		goto out;
960 	}
961 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
962 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
963 	error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd);
964 out:
965 	vrele(vp);
966 	return (error);
967 }
968 
969 /*
970  * Make a symbolic link.
971  */
972 /* ARGSUSED */
973 int
974 sys_symlink(p, v, retval)
975 	struct proc *p;
976 	void *v;
977 	register_t *retval;
978 {
979 	register struct sys_symlink_args /* {
980 		syscallarg(char *) path;
981 		syscallarg(char *) link;
982 	} */ *uap = v;
983 	struct vattr vattr;
984 	char *path;
985 	int error;
986 	struct nameidata nd;
987 
988 	MALLOC(path, char *, MAXPATHLEN, M_NAMEI, M_WAITOK);
989 	error = copyinstr(SCARG(uap, path), path, MAXPATHLEN, NULL);
990 	if (error)
991 		goto out;
992 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, link), p);
993 	if ((error = namei(&nd)) != 0)
994 		goto out;
995 	if (nd.ni_vp) {
996 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
997 		if (nd.ni_dvp == nd.ni_vp)
998 			vrele(nd.ni_dvp);
999 		else
1000 			vput(nd.ni_dvp);
1001 		vrele(nd.ni_vp);
1002 		error = EEXIST;
1003 		goto out;
1004 	}
1005 	VATTR_NULL(&vattr);
1006 	vattr.va_mode = ACCESSPERMS &~ p->p_fd->fd_cmask;
1007 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1008 	error = VOP_SYMLINK(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr, path);
1009 out:
1010 	FREE(path, M_NAMEI);
1011 	return (error);
1012 }
1013 
1014 /*
1015  * Delete a whiteout from the filesystem.
1016  */
1017 /* ARGSUSED */
1018 int
1019 sys_undelete(p, v, retval)
1020 	struct proc *p;
1021 	void *v;
1022 	register_t *retval;
1023 {
1024 	register struct sys_undelete_args /* {
1025 		syscallarg(char *) path;
1026 	} */ *uap = v;
1027 	int error;
1028 	struct nameidata nd;
1029 
1030 	NDINIT(&nd, DELETE, LOCKPARENT|DOWHITEOUT, UIO_USERSPACE,
1031 	    SCARG(uap, path), p);
1032 	error = namei(&nd);
1033 	if (error)
1034 		return (error);
1035 
1036 	if (nd.ni_vp != NULLVP || !(nd.ni_cnd.cn_flags & ISWHITEOUT)) {
1037 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1038 		if (nd.ni_dvp == nd.ni_vp)
1039 			vrele(nd.ni_dvp);
1040 		else
1041 			vput(nd.ni_dvp);
1042 		if (nd.ni_vp)
1043 			vrele(nd.ni_vp);
1044 		return (EEXIST);
1045 	}
1046 
1047 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1048 	if ((error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, DELETE)) != 0)
1049 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1050 	vput(nd.ni_dvp);
1051 	return (error);
1052 }
1053 
1054 /*
1055  * Delete a name from the filesystem.
1056  */
1057 /* ARGSUSED */
1058 int
1059 sys_unlink(p, v, retval)
1060 	struct proc *p;
1061 	void *v;
1062 	register_t *retval;
1063 {
1064 	struct sys_unlink_args /* {
1065 		syscallarg(char *) path;
1066 	} */ *uap = v;
1067 	register struct vnode *vp;
1068 	int error;
1069 	struct nameidata nd;
1070 
1071 	NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
1072 	    SCARG(uap, path), p);
1073 	if ((error = namei(&nd)) != 0)
1074 		return (error);
1075 	vp = nd.ni_vp;
1076 
1077 	/*
1078 	 * The root of a mounted filesystem cannot be deleted.
1079 	 */
1080 	if (vp->v_flag & VROOT) {
1081 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1082 		if (nd.ni_dvp == vp)
1083 			vrele(nd.ni_dvp);
1084 		else
1085 			vput(nd.ni_dvp);
1086 		vput(vp);
1087 		error = EBUSY;
1088 		goto out;
1089 	}
1090 
1091 	(void)vnode_pager_uncache(vp);
1092 
1093 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1094 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1095 	error = VOP_REMOVE(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
1096 out:
1097 	return (error);
1098 }
1099 
1100 /*
1101  * Reposition read/write file offset.
1102  */
1103 int
1104 sys_lseek(p, v, retval)
1105 	struct proc *p;
1106 	void *v;
1107 	register_t *retval;
1108 {
1109 	register struct sys_lseek_args /* {
1110 		syscallarg(int) fd;
1111 		syscallarg(int) pad;
1112 		syscallarg(off_t) offset;
1113 		syscallarg(int) whence;
1114 	} */ *uap = v;
1115 	struct ucred *cred = p->p_ucred;
1116 	register struct filedesc *fdp = p->p_fd;
1117 	register struct file *fp;
1118 	struct vattr vattr;
1119 	int error;
1120 
1121 	if ((u_int)SCARG(uap, fd) >= fdp->fd_nfiles ||
1122 	    (fp = fdp->fd_ofiles[SCARG(uap, fd)]) == NULL)
1123 		return (EBADF);
1124 	if (fp->f_type != DTYPE_VNODE)
1125 		return (ESPIPE);
1126 	switch (SCARG(uap, whence)) {
1127 	case L_INCR:
1128 		fp->f_offset += SCARG(uap, offset);
1129 		break;
1130 	case L_XTND:
1131 		error = VOP_GETATTR((struct vnode *)fp->f_data, &vattr,
1132 				    cred, p);
1133 		if (error)
1134 			return (error);
1135 		fp->f_offset = SCARG(uap, offset) + vattr.va_size;
1136 		break;
1137 	case L_SET:
1138 		fp->f_offset = SCARG(uap, offset);
1139 		break;
1140 	default:
1141 		return (EINVAL);
1142 	}
1143 	*(off_t *)retval = fp->f_offset;
1144 	return (0);
1145 }
1146 
1147 /*
1148  * Check access permissions.
1149  */
1150 int
1151 sys_access(p, v, retval)
1152 	struct proc *p;
1153 	void *v;
1154 	register_t *retval;
1155 {
1156 	register struct sys_access_args /* {
1157 		syscallarg(char *) path;
1158 		syscallarg(int) flags;
1159 	} */ *uap = v;
1160 	register struct ucred *cred = p->p_ucred;
1161 	register struct vnode *vp;
1162 	int error, flags, t_gid, t_uid;
1163 	struct nameidata nd;
1164 
1165 	t_uid = cred->cr_uid;
1166 	t_gid = cred->cr_gid;
1167 	cred->cr_uid = p->p_cred->p_ruid;
1168 	cred->cr_gid = p->p_cred->p_rgid;
1169 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1170 	    SCARG(uap, path), p);
1171 	if ((error = namei(&nd)) != 0)
1172 		goto out1;
1173 	vp = nd.ni_vp;
1174 
1175 	/* Flags == 0 means only check for existence. */
1176 	if (SCARG(uap, flags)) {
1177 		flags = 0;
1178 		if (SCARG(uap, flags) & R_OK)
1179 			flags |= VREAD;
1180 		if (SCARG(uap, flags) & W_OK)
1181 			flags |= VWRITE;
1182 		if (SCARG(uap, flags) & X_OK)
1183 			flags |= VEXEC;
1184 		if ((flags & VWRITE) == 0 || (error = vn_writechk(vp)) == 0)
1185 			error = VOP_ACCESS(vp, flags, cred, p);
1186 	}
1187 	vput(vp);
1188 out1:
1189 	cred->cr_uid = t_uid;
1190 	cred->cr_gid = t_gid;
1191 	return (error);
1192 }
1193 
1194 /*
1195  * Get file status; this version follows links.
1196  */
1197 /* ARGSUSED */
1198 int
1199 sys_stat(p, v, retval)
1200 	struct proc *p;
1201 	void *v;
1202 	register_t *retval;
1203 {
1204 	register struct sys_stat_args /* {
1205 		syscallarg(char *) path;
1206 		syscallarg(struct stat *) ub;
1207 	} */ *uap = v;
1208 	struct stat sb;
1209 	int error;
1210 	struct nameidata nd;
1211 
1212 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1213 	    SCARG(uap, path), p);
1214 	if ((error = namei(&nd)) != 0)
1215 		return (error);
1216 	error = vn_stat(nd.ni_vp, &sb, p);
1217 	vput(nd.ni_vp);
1218 	if (error)
1219 		return (error);
1220 	error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
1221 	return (error);
1222 }
1223 
1224 /*
1225  * Get file status; this version does not follow links.
1226  */
1227 /* ARGSUSED */
1228 int
1229 sys_lstat(p, v, retval)
1230 	struct proc *p;
1231 	void *v;
1232 	register_t *retval;
1233 {
1234 	register struct sys_lstat_args /* {
1235 		syscallarg(char *) path;
1236 		syscallarg(struct stat *) ub;
1237 	} */ *uap = v;
1238 	struct stat sb;
1239 	int error;
1240 	struct nameidata nd;
1241 
1242 	NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF, UIO_USERSPACE,
1243 	    SCARG(uap, path), p);
1244 	if ((error = namei(&nd)) != 0)
1245 		return (error);
1246 	error = vn_stat(nd.ni_vp, &sb, p);
1247 	vput(nd.ni_vp);
1248 	if (error)
1249 		return (error);
1250 	error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
1251 	return (error);
1252 }
1253 
1254 /*
1255  * Get configurable pathname variables.
1256  */
1257 /* ARGSUSED */
1258 int
1259 sys_pathconf(p, v, retval)
1260 	struct proc *p;
1261 	void *v;
1262 	register_t *retval;
1263 {
1264 	register struct sys_pathconf_args /* {
1265 		syscallarg(char *) path;
1266 		syscallarg(int) name;
1267 	} */ *uap = v;
1268 	int error;
1269 	struct nameidata nd;
1270 
1271 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1272 	    SCARG(uap, path), p);
1273 	if ((error = namei(&nd)) != 0)
1274 		return (error);
1275 	error = VOP_PATHCONF(nd.ni_vp, SCARG(uap, name), retval);
1276 	vput(nd.ni_vp);
1277 	return (error);
1278 }
1279 
1280 /*
1281  * Return target name of a symbolic link.
1282  */
1283 /* ARGSUSED */
1284 int
1285 sys_readlink(p, v, retval)
1286 	struct proc *p;
1287 	void *v;
1288 	register_t *retval;
1289 {
1290 	register struct sys_readlink_args /* {
1291 		syscallarg(char *) path;
1292 		syscallarg(char *) buf;
1293 		syscallarg(int) count;
1294 	} */ *uap = v;
1295 	register struct vnode *vp;
1296 	struct iovec aiov;
1297 	struct uio auio;
1298 	int error;
1299 	struct nameidata nd;
1300 
1301 	NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF, UIO_USERSPACE,
1302 	    SCARG(uap, path), p);
1303 	if ((error = namei(&nd)) != 0)
1304 		return (error);
1305 	vp = nd.ni_vp;
1306 	if (vp->v_type != VLNK)
1307 		error = EINVAL;
1308 	else {
1309 		aiov.iov_base = SCARG(uap, buf);
1310 		aiov.iov_len = SCARG(uap, count);
1311 		auio.uio_iov = &aiov;
1312 		auio.uio_iovcnt = 1;
1313 		auio.uio_offset = 0;
1314 		auio.uio_rw = UIO_READ;
1315 		auio.uio_segflg = UIO_USERSPACE;
1316 		auio.uio_procp = p;
1317 		auio.uio_resid = SCARG(uap, count);
1318 		error = VOP_READLINK(vp, &auio, p->p_ucred);
1319 	}
1320 	vput(vp);
1321 	*retval = SCARG(uap, count) - auio.uio_resid;
1322 	return (error);
1323 }
1324 
1325 /*
1326  * Change flags of a file given a path name.
1327  */
1328 /* ARGSUSED */
1329 int
1330 sys_chflags(p, v, retval)
1331 	struct proc *p;
1332 	void *v;
1333 	register_t *retval;
1334 {
1335 	register struct sys_chflags_args /* {
1336 		syscallarg(char *) path;
1337 		syscallarg(int) flags;
1338 	} */ *uap = v;
1339 	register struct vnode *vp;
1340 	struct vattr vattr;
1341 	int error;
1342 	struct nameidata nd;
1343 
1344 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1345 	if ((error = namei(&nd)) != 0)
1346 		return (error);
1347 	vp = nd.ni_vp;
1348 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1349 	VOP_LOCK(vp);
1350 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1351 		error = EROFS;
1352 	else {
1353 		VATTR_NULL(&vattr);
1354 		vattr.va_flags = SCARG(uap, flags);
1355 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1356 	}
1357 	vput(vp);
1358 	return (error);
1359 }
1360 
1361 /*
1362  * Change flags of a file given a file descriptor.
1363  */
1364 /* ARGSUSED */
1365 int
1366 sys_fchflags(p, v, retval)
1367 	struct proc *p;
1368 	void *v;
1369 	register_t *retval;
1370 {
1371 	register struct sys_fchflags_args /* {
1372 		syscallarg(int) fd;
1373 		syscallarg(int) flags;
1374 	} */ *uap = v;
1375 	struct vattr vattr;
1376 	struct vnode *vp;
1377 	struct file *fp;
1378 	int error;
1379 
1380 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
1381 		return (error);
1382 	vp = (struct vnode *)fp->f_data;
1383 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1384 	VOP_LOCK(vp);
1385 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1386 		error = EROFS;
1387 	else {
1388 		VATTR_NULL(&vattr);
1389 		vattr.va_flags = SCARG(uap, flags);
1390 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1391 	}
1392 	VOP_UNLOCK(vp);
1393 	return (error);
1394 }
1395 
1396 /*
1397  * Change mode of a file given path name.
1398  */
1399 /* ARGSUSED */
1400 int
1401 sys_chmod(p, v, retval)
1402 	struct proc *p;
1403 	void *v;
1404 	register_t *retval;
1405 {
1406 	register struct sys_chmod_args /* {
1407 		syscallarg(char *) path;
1408 		syscallarg(int) mode;
1409 	} */ *uap = v;
1410 	register struct vnode *vp;
1411 	struct vattr vattr;
1412 	int error;
1413 	struct nameidata nd;
1414 
1415 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1416 	if ((error = namei(&nd)) != 0)
1417 		return (error);
1418 	vp = nd.ni_vp;
1419 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1420 	VOP_LOCK(vp);
1421 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1422 		error = EROFS;
1423 	else {
1424 		VATTR_NULL(&vattr);
1425 		vattr.va_mode = SCARG(uap, mode) & ALLPERMS;
1426 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1427 	}
1428 	vput(vp);
1429 	return (error);
1430 }
1431 
1432 /*
1433  * Change mode of a file given a file descriptor.
1434  */
1435 /* ARGSUSED */
1436 int
1437 sys_fchmod(p, v, retval)
1438 	struct proc *p;
1439 	void *v;
1440 	register_t *retval;
1441 {
1442 	register struct sys_fchmod_args /* {
1443 		syscallarg(int) fd;
1444 		syscallarg(int) mode;
1445 	} */ *uap = v;
1446 	struct vattr vattr;
1447 	struct vnode *vp;
1448 	struct file *fp;
1449 	int error;
1450 
1451 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
1452 		return (error);
1453 	vp = (struct vnode *)fp->f_data;
1454 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1455 	VOP_LOCK(vp);
1456 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1457 		error = EROFS;
1458 	else {
1459 		VATTR_NULL(&vattr);
1460 		vattr.va_mode = SCARG(uap, mode) & ALLPERMS;
1461 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1462 	}
1463 	VOP_UNLOCK(vp);
1464 	return (error);
1465 }
1466 
1467 /*
1468  * Set ownership given a path name.
1469  */
1470 /* ARGSUSED */
1471 int
1472 sys_chown(p, v, retval)
1473 	struct proc *p;
1474 	void *v;
1475 	register_t *retval;
1476 {
1477 	register struct sys_chown_args /* {
1478 		syscallarg(char *) path;
1479 		syscallarg(int) uid;
1480 		syscallarg(int) gid;
1481 	} */ *uap = v;
1482 	register struct vnode *vp;
1483 	struct vattr vattr;
1484 	int error;
1485 	struct nameidata nd;
1486 
1487 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1488 	if ((error = namei(&nd)) != 0)
1489 		return (error);
1490 	vp = nd.ni_vp;
1491 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1492 	VOP_LOCK(vp);
1493 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1494 		error = EROFS;
1495 	else {
1496 		VATTR_NULL(&vattr);
1497 		vattr.va_uid = SCARG(uap, uid);
1498 		vattr.va_gid = SCARG(uap, gid);
1499 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1500 	}
1501 	vput(vp);
1502 	return (error);
1503 }
1504 
1505 /*
1506  * Set ownership given a file descriptor.
1507  */
1508 /* ARGSUSED */
1509 int
1510 sys_fchown(p, v, retval)
1511 	struct proc *p;
1512 	void *v;
1513 	register_t *retval;
1514 {
1515 	register struct sys_fchown_args /* {
1516 		syscallarg(int) fd;
1517 		syscallarg(int) uid;
1518 		syscallarg(int) gid;
1519 	} */ *uap = v;
1520 	register struct vnode *vp;
1521 	struct vattr vattr;
1522 	int error;
1523 	struct file *fp;
1524 
1525 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
1526 		return (error);
1527 	vp = (struct vnode *)fp->f_data;
1528 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1529 	VOP_LOCK(vp);
1530 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1531 		error = EROFS;
1532 	else {
1533 		VATTR_NULL(&vattr);
1534 		vattr.va_uid = SCARG(uap, uid);
1535 		vattr.va_gid = SCARG(uap, gid);
1536 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1537 	}
1538 	VOP_UNLOCK(vp);
1539 	return (error);
1540 }
1541 
1542 /*
1543  * Set the access and modification times given a path name.
1544  */
1545 /* ARGSUSED */
1546 int
1547 sys_utimes(p, v, retval)
1548 	struct proc *p;
1549 	void *v;
1550 	register_t *retval;
1551 {
1552 	register struct sys_utimes_args /* {
1553 		syscallarg(char *) path;
1554 		syscallarg(struct timeval *) tptr;
1555 	} */ *uap = v;
1556 	register struct vnode *vp;
1557 	struct timeval tv[2];
1558 	struct vattr vattr;
1559 	int error;
1560 	struct nameidata nd;
1561 
1562 	VATTR_NULL(&vattr);
1563 	if (SCARG(uap, tptr) == NULL) {
1564 		microtime(&tv[0]);
1565 		tv[1] = tv[0];
1566 		vattr.va_vaflags |= VA_UTIMES_NULL;
1567 	} else {
1568 		error = copyin((caddr_t)SCARG(uap, tptr), (caddr_t)tv,
1569 			       sizeof (tv));
1570 		if (error)
1571 			return (error);
1572 	}
1573 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1574 	if ((error = namei(&nd)) != 0)
1575 		return (error);
1576 	vp = nd.ni_vp;
1577 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1578 	VOP_LOCK(vp);
1579 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1580 		error = EROFS;
1581 	else {
1582 		vattr.va_atime.tv_sec = tv[0].tv_sec;
1583 		vattr.va_atime.tv_nsec = tv[0].tv_usec * 1000;
1584 		vattr.va_mtime.tv_sec = tv[1].tv_sec;
1585 		vattr.va_mtime.tv_nsec = tv[1].tv_usec * 1000;
1586 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1587 	}
1588 	vput(vp);
1589 	return (error);
1590 }
1591 
1592 /*
1593  * Set the access and modification times given a file descriptor.
1594  */
1595 /* ARGSUSED */
1596 int
1597 sys_futimes(p, v, retval)
1598 	struct proc *p;
1599 	void *v;
1600 	register_t *retval;
1601 {
1602 	register struct sys_futimes_args /* {
1603 		syscallarg(int) fd;
1604 		syscallarg(struct timeval *) tptr;
1605 	} */ *uap = v;
1606 	register struct vnode *vp;
1607 	struct timeval tv[2];
1608 	struct vattr vattr;
1609 	int error;
1610 	struct file *fp;
1611 
1612 	VATTR_NULL(&vattr);
1613 	if (SCARG(uap, tptr) == NULL) {
1614 		microtime(&tv[0]);
1615 		tv[1] = tv[0];
1616 		vattr.va_vaflags |= VA_UTIMES_NULL;
1617 	} else {
1618 		error = copyin((caddr_t)SCARG(uap, tptr), (caddr_t)tv,
1619 			       sizeof (tv));
1620 		if (error)
1621 			return (error);
1622 	}
1623 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
1624 		return (error);
1625 	vp = (struct vnode *)fp->f_data;
1626 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1627 	VOP_LOCK(vp);
1628 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1629 		error = EROFS;
1630 	else {
1631 		vattr.va_atime.tv_sec = tv[0].tv_sec;
1632 		vattr.va_atime.tv_nsec = tv[0].tv_usec * 1000;
1633 		vattr.va_mtime.tv_sec = tv[1].tv_sec;
1634 		vattr.va_mtime.tv_nsec = tv[1].tv_usec * 1000;
1635 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1636 	}
1637 	VOP_UNLOCK(vp);
1638 	return (error);
1639 }
1640 
1641 /*
1642  * Truncate a file given its path name.
1643  */
1644 /* ARGSUSED */
1645 int
1646 sys_truncate(p, v, retval)
1647 	struct proc *p;
1648 	void *v;
1649 	register_t *retval;
1650 {
1651 	register struct sys_truncate_args /* {
1652 		syscallarg(char *) path;
1653 		syscallarg(int) pad;
1654 		syscallarg(off_t) length;
1655 	} */ *uap = v;
1656 	register struct vnode *vp;
1657 	struct vattr vattr;
1658 	int error;
1659 	struct nameidata nd;
1660 
1661 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1662 	if ((error = namei(&nd)) != 0)
1663 		return (error);
1664 	vp = nd.ni_vp;
1665 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1666 	VOP_LOCK(vp);
1667 	if (vp->v_type == VDIR)
1668 		error = EISDIR;
1669 	else if ((error = vn_writechk(vp)) == 0 &&
1670 	    (error = VOP_ACCESS(vp, VWRITE, p->p_ucred, p)) == 0) {
1671 		VATTR_NULL(&vattr);
1672 		vattr.va_size = SCARG(uap, length);
1673 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1674 	}
1675 	vput(vp);
1676 	return (error);
1677 }
1678 
1679 /*
1680  * Truncate a file given a file descriptor.
1681  */
1682 /* ARGSUSED */
1683 int
1684 sys_ftruncate(p, v, retval)
1685 	struct proc *p;
1686 	void *v;
1687 	register_t *retval;
1688 {
1689 	register struct sys_ftruncate_args /* {
1690 		syscallarg(int) fd;
1691 		syscallarg(int) pad;
1692 		syscallarg(off_t) length;
1693 	} */ *uap = v;
1694 	struct vattr vattr;
1695 	struct vnode *vp;
1696 	struct file *fp;
1697 	int error;
1698 
1699 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
1700 		return (error);
1701 	if ((fp->f_flag & FWRITE) == 0)
1702 		return (EINVAL);
1703 	vp = (struct vnode *)fp->f_data;
1704 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1705 	VOP_LOCK(vp);
1706 	if (vp->v_type == VDIR)
1707 		error = EISDIR;
1708 	else if ((error = vn_writechk(vp)) == 0) {
1709 		VATTR_NULL(&vattr);
1710 		vattr.va_size = SCARG(uap, length);
1711 		error = VOP_SETATTR(vp, &vattr, fp->f_cred, p);
1712 	}
1713 	VOP_UNLOCK(vp);
1714 	return (error);
1715 }
1716 
1717 /*
1718  * Sync an open file.
1719  */
1720 /* ARGSUSED */
1721 int
1722 sys_fsync(p, v, retval)
1723 	struct proc *p;
1724 	void *v;
1725 	register_t *retval;
1726 {
1727 	struct sys_fsync_args /* {
1728 		syscallarg(int) fd;
1729 	} */ *uap = v;
1730 	register struct vnode *vp;
1731 	struct file *fp;
1732 	int error;
1733 
1734 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
1735 		return (error);
1736 	vp = (struct vnode *)fp->f_data;
1737 	VOP_LOCK(vp);
1738 	error = VOP_FSYNC(vp, fp->f_cred, MNT_WAIT, p);
1739 	VOP_UNLOCK(vp);
1740 	return (error);
1741 }
1742 
1743 /*
1744  * Rename files.  Source and destination must either both be directories,
1745  * or both not be directories.  If target is a directory, it must be empty.
1746  */
1747 /* ARGSUSED */
1748 int
1749 sys_rename(p, v, retval)
1750 	struct proc *p;
1751 	void *v;
1752 	register_t *retval;
1753 {
1754 	register struct sys_rename_args /* {
1755 		syscallarg(char *) from;
1756 		syscallarg(char *) to;
1757 	} */ *uap = v;
1758 	register struct vnode *tvp, *fvp, *tdvp;
1759 	struct nameidata fromnd, tond;
1760 	int error;
1761 
1762 	NDINIT(&fromnd, DELETE, WANTPARENT | SAVESTART, UIO_USERSPACE,
1763 	    SCARG(uap, from), p);
1764 	if ((error = namei(&fromnd)) != 0)
1765 		return (error);
1766 	fvp = fromnd.ni_vp;
1767 	NDINIT(&tond, RENAME, LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART,
1768 	    UIO_USERSPACE, SCARG(uap, to), p);
1769 	if ((error = namei(&tond)) != 0) {
1770 		VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
1771 		vrele(fromnd.ni_dvp);
1772 		vrele(fvp);
1773 		goto out1;
1774 	}
1775 	tdvp = tond.ni_dvp;
1776 	tvp = tond.ni_vp;
1777 	if (tvp != NULL) {
1778 		if (fvp->v_type == VDIR && tvp->v_type != VDIR) {
1779 			error = ENOTDIR;
1780 			goto out;
1781 		} else if (fvp->v_type != VDIR && tvp->v_type == VDIR) {
1782 			error = EISDIR;
1783 			goto out;
1784 		}
1785 	}
1786 	if (fvp == tdvp)
1787 		error = EINVAL;
1788 	/*
1789 	 * If source is the same as the destination (that is the
1790 	 * same inode number with the same name in the same directory),
1791 	 * then there is nothing to do.
1792 	 */
1793 	if (fvp == tvp && fromnd.ni_dvp == tdvp &&
1794 	    fromnd.ni_cnd.cn_namelen == tond.ni_cnd.cn_namelen &&
1795 	    !bcmp(fromnd.ni_cnd.cn_nameptr, tond.ni_cnd.cn_nameptr,
1796 	      fromnd.ni_cnd.cn_namelen))
1797 		error = -1;
1798 out:
1799 	if (!error) {
1800 		VOP_LEASE(tdvp, p, p->p_ucred, LEASE_WRITE);
1801 		if (fromnd.ni_dvp != tdvp)
1802 			VOP_LEASE(fromnd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1803 		if (tvp)
1804 			VOP_LEASE(tvp, p, p->p_ucred, LEASE_WRITE);
1805 		error = VOP_RENAME(fromnd.ni_dvp, fromnd.ni_vp, &fromnd.ni_cnd,
1806 				   tond.ni_dvp, tond.ni_vp, &tond.ni_cnd);
1807 	} else {
1808 		VOP_ABORTOP(tond.ni_dvp, &tond.ni_cnd);
1809 		if (tdvp == tvp)
1810 			vrele(tdvp);
1811 		else
1812 			vput(tdvp);
1813 		if (tvp)
1814 			vput(tvp);
1815 		VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
1816 		vrele(fromnd.ni_dvp);
1817 		vrele(fvp);
1818 	}
1819 	vrele(tond.ni_startdir);
1820 	FREE(tond.ni_cnd.cn_pnbuf, M_NAMEI);
1821 out1:
1822 	if (fromnd.ni_startdir)
1823 		vrele(fromnd.ni_startdir);
1824 	FREE(fromnd.ni_cnd.cn_pnbuf, M_NAMEI);
1825 	if (error == -1)
1826 		return (0);
1827 	return (error);
1828 }
1829 
1830 /*
1831  * Make a directory file.
1832  */
1833 /* ARGSUSED */
1834 int
1835 sys_mkdir(p, v, retval)
1836 	struct proc *p;
1837 	void *v;
1838 	register_t *retval;
1839 {
1840 	register struct sys_mkdir_args /* {
1841 		syscallarg(char *) path;
1842 		syscallarg(int) mode;
1843 	} */ *uap = v;
1844 	register struct vnode *vp;
1845 	struct vattr vattr;
1846 	int error;
1847 	struct nameidata nd;
1848 
1849 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1850 	if ((error = namei(&nd)) != 0)
1851 		return (error);
1852 	vp = nd.ni_vp;
1853 	if (vp != NULL) {
1854 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1855 		if (nd.ni_dvp == vp)
1856 			vrele(nd.ni_dvp);
1857 		else
1858 			vput(nd.ni_dvp);
1859 		vrele(vp);
1860 		return (EEXIST);
1861 	}
1862 	VATTR_NULL(&vattr);
1863 	vattr.va_type = VDIR;
1864 	vattr.va_mode = (SCARG(uap, mode) & ACCESSPERMS) &~ p->p_fd->fd_cmask;
1865 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1866 	error = VOP_MKDIR(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
1867 	if (!error)
1868 		vput(nd.ni_vp);
1869 	return (error);
1870 }
1871 
1872 /*
1873  * Remove a directory file.
1874  */
1875 /* ARGSUSED */
1876 int
1877 sys_rmdir(p, v, retval)
1878 	struct proc *p;
1879 	void *v;
1880 	register_t *retval;
1881 {
1882 	struct sys_rmdir_args /* {
1883 		syscallarg(char *) path;
1884 	} */ *uap = v;
1885 	register struct vnode *vp;
1886 	int error;
1887 	struct nameidata nd;
1888 
1889 	NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
1890 	    SCARG(uap, path), p);
1891 	if ((error = namei(&nd)) != 0)
1892 		return (error);
1893 	vp = nd.ni_vp;
1894 	if (vp->v_type != VDIR) {
1895 		error = ENOTDIR;
1896 		goto out;
1897 	}
1898 	/*
1899 	 * No rmdir "." please.
1900 	 */
1901 	if (nd.ni_dvp == vp) {
1902 		error = EINVAL;
1903 		goto out;
1904 	}
1905 	/*
1906 	 * The root of a mounted filesystem cannot be deleted.
1907 	 */
1908 	if (vp->v_flag & VROOT)
1909 		error = EBUSY;
1910 out:
1911 	if (!error) {
1912 		VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1913 		VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1914 		error = VOP_RMDIR(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
1915 	} else {
1916 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1917 		if (nd.ni_dvp == vp)
1918 			vrele(nd.ni_dvp);
1919 		else
1920 			vput(nd.ni_dvp);
1921 		vput(vp);
1922 	}
1923 	return (error);
1924 }
1925 
1926 /*
1927  * Read a block of directory entries in a file system independent format.
1928  */
1929 int
1930 sys_getdirentries(p, v, retval)
1931 	struct proc *p;
1932 	void *v;
1933 	register_t *retval;
1934 {
1935 	register struct sys_getdirentries_args /* {
1936 		syscallarg(int) fd;
1937 		syscallarg(char *) buf;
1938 		syscallarg(u_int) count;
1939 		syscallarg(long *) basep;
1940 	} */ *uap = v;
1941 	register struct vnode *vp;
1942 	struct file *fp;
1943 	struct uio auio;
1944 	struct iovec aiov;
1945 	long loff;
1946 	int error, eofflag;
1947 
1948 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
1949 		return (error);
1950 	if ((fp->f_flag & FREAD) == 0)
1951 		return (EBADF);
1952 	vp = (struct vnode *)fp->f_data;
1953 unionread:
1954 	if (vp->v_type != VDIR)
1955 		return (EINVAL);
1956 	aiov.iov_base = SCARG(uap, buf);
1957 	aiov.iov_len = SCARG(uap, count);
1958 	auio.uio_iov = &aiov;
1959 	auio.uio_iovcnt = 1;
1960 	auio.uio_rw = UIO_READ;
1961 	auio.uio_segflg = UIO_USERSPACE;
1962 	auio.uio_procp = p;
1963 	auio.uio_resid = SCARG(uap, count);
1964 	VOP_LOCK(vp);
1965 	loff = auio.uio_offset = fp->f_offset;
1966 	error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, (u_long *)0, 0);
1967 	fp->f_offset = auio.uio_offset;
1968 	VOP_UNLOCK(vp);
1969 	if (error)
1970 		return (error);
1971 
1972 #ifdef UNION
1973 {
1974 	extern int (**union_vnodeop_p) __P((void *));
1975 	extern struct vnode *union_dircache __P((struct vnode *));
1976 
1977 	if ((SCARG(uap, count) == auio.uio_resid) &&
1978 	    (vp->v_op == union_vnodeop_p)) {
1979 		struct vnode *lvp;
1980 
1981 		lvp = union_dircache(vp);
1982 		if (lvp != NULLVP) {
1983 			struct vattr va;
1984 
1985 			/*
1986 			 * If the directory is opaque,
1987 			 * then don't show lower entries
1988 			 */
1989 			error = VOP_GETATTR(vp, &va, fp->f_cred, p);
1990 			if (va.va_flags & OPAQUE) {
1991 				vput(lvp);
1992 				lvp = NULL;
1993 			}
1994 		}
1995 
1996 		if (lvp != NULLVP) {
1997 			error = VOP_OPEN(lvp, FREAD, fp->f_cred, p);
1998 			VOP_UNLOCK(lvp);
1999 
2000 			if (error) {
2001 				vrele(lvp);
2002 				return (error);
2003 			}
2004 			fp->f_data = (caddr_t) lvp;
2005 			fp->f_offset = 0;
2006 			error = vn_close(vp, FREAD, fp->f_cred, p);
2007 			if (error)
2008 				return (error);
2009 			vp = lvp;
2010 			goto unionread;
2011 		}
2012 	}
2013 }
2014 #endif /* UNION */
2015 
2016 	if ((SCARG(uap, count) == auio.uio_resid) &&
2017 	    (vp->v_flag & VROOT) &&
2018 	    (vp->v_mount->mnt_flag & MNT_UNION)) {
2019 		struct vnode *tvp = vp;
2020 		vp = vp->v_mount->mnt_vnodecovered;
2021 		VREF(vp);
2022 		fp->f_data = (caddr_t) vp;
2023 		fp->f_offset = 0;
2024 		vrele(tvp);
2025 		goto unionread;
2026 	}
2027 	error = copyout((caddr_t)&loff, (caddr_t)SCARG(uap, basep),
2028 	    sizeof(long));
2029 	*retval = SCARG(uap, count) - auio.uio_resid;
2030 	return (error);
2031 }
2032 
2033 /*
2034  * Set the mode mask for creation of filesystem nodes.
2035  */
2036 int
2037 sys_umask(p, v, retval)
2038 	struct proc *p;
2039 	void *v;
2040 	register_t *retval;
2041 {
2042 	struct sys_umask_args /* {
2043 		syscallarg(int) newmask;
2044 	} */ *uap = v;
2045 	register struct filedesc *fdp;
2046 
2047 	fdp = p->p_fd;
2048 	*retval = fdp->fd_cmask;
2049 	fdp->fd_cmask = SCARG(uap, newmask) & ALLPERMS;
2050 	return (0);
2051 }
2052 
2053 /*
2054  * Void all references to file by ripping underlying filesystem
2055  * away from vnode.
2056  */
2057 /* ARGSUSED */
2058 int
2059 sys_revoke(p, v, retval)
2060 	struct proc *p;
2061 	void *v;
2062 	register_t *retval;
2063 {
2064 	register struct sys_revoke_args /* {
2065 		syscallarg(char *) path;
2066 	} */ *uap = v;
2067 	register struct vnode *vp;
2068 	struct vattr vattr;
2069 	int error;
2070 	struct nameidata nd;
2071 
2072 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2073 	if ((error = namei(&nd)) != 0)
2074 		return (error);
2075 	vp = nd.ni_vp;
2076 	if (vp->v_type != VCHR && vp->v_type != VBLK) {
2077 		error = EINVAL;
2078 		goto out;
2079 	}
2080 	if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)) != 0)
2081 		goto out;
2082 	if (p->p_ucred->cr_uid != vattr.va_uid &&
2083 	    (error = suser(p->p_ucred, &p->p_acflag)))
2084 		goto out;
2085 	if (vp->v_usecount > 1 || (vp->v_flag & VALIASED))
2086 		vgoneall(vp);
2087 out:
2088 	vrele(vp);
2089 	return (error);
2090 }
2091 
2092 /*
2093  * Convert a user file descriptor to a kernel file entry.
2094  */
2095 int
2096 getvnode(fdp, fd, fpp)
2097 	struct filedesc *fdp;
2098 	int fd;
2099 	struct file **fpp;
2100 {
2101 	struct vnode *vp;
2102 	struct file *fp;
2103 
2104 	if ((u_int)fd >= fdp->fd_nfiles ||
2105 	    (fp = fdp->fd_ofiles[fd]) == NULL)
2106 		return (EBADF);
2107 	if (fp->f_type != DTYPE_VNODE)
2108 		return (EINVAL);
2109 	vp = (struct vnode *)fp->f_data;
2110 	if (vp->v_type == VBAD)
2111 		return (EBADF);
2112 	*fpp = fp;
2113 	return (0);
2114 }
2115