xref: /netbsd-src/sys/kern/vfs_syscalls.c (revision 76dfffe33547c37f8bdd446e3e4ab0f3c16cea4b)
1 /*	$NetBSD: vfs_syscalls.c,v 1.73 1996/10/23 23:07:08 cgd 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 | MNT_NOCOREDUMP);
236 	mp->mnt_flag |= SCARG(uap, flags) & (MNT_NOSUID | MNT_NOEXEC |
237 	    MNT_NODEV | MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC |
238 	    MNT_NOCOREDUMP);
239 	/*
240 	 * Mount the filesystem.
241 	 */
242 	error = VFS_MOUNT(mp, SCARG(uap, path), SCARG(uap, data), &nd, p);
243 	if (mp->mnt_flag & MNT_UPDATE) {
244 		vrele(vp);
245 		if (mp->mnt_flag & MNT_WANTRDWR)
246 			mp->mnt_flag &= ~MNT_RDONLY;
247 		mp->mnt_flag &=~
248 		    (MNT_UPDATE | MNT_RELOAD | MNT_FORCE | MNT_WANTRDWR);
249 		if (error)
250 			mp->mnt_flag = flag;
251 		return (error);
252 	}
253 	/*
254 	 * Put the new filesystem on the mount list after root.
255 	 */
256 	cache_purge(vp);
257 	if (!error) {
258 		CIRCLEQ_INSERT_TAIL(&mountlist, mp, mnt_list);
259 		checkdirs(vp);
260 		VOP_UNLOCK(vp);
261 		vfs_unlock(mp);
262 		(void) VFS_STATFS(mp, &mp->mnt_stat, p);
263 		error = VFS_START(mp, 0, p);
264 	} else {
265 		mp->mnt_vnodecovered->v_mountedhere = (struct mount *)0;
266 		vfssw[fsindex]->vfs_refcount--;
267 		vfs_unlock(mp);
268 		free((caddr_t)mp, M_MOUNT);
269 		vput(vp);
270 	}
271 	return (error);
272 }
273 
274 /*
275  * Scan all active processes to see if any of them have a current
276  * or root directory onto which the new filesystem has just been
277  * mounted. If so, replace them with the new mount point.
278  */
279 void
280 checkdirs(olddp)
281 	struct vnode *olddp;
282 {
283 	struct filedesc *fdp;
284 	struct vnode *newdp;
285 	struct proc *p;
286 
287 	if (olddp->v_usecount == 1)
288 		return;
289 	if (VFS_ROOT(olddp->v_mountedhere, &newdp))
290 		panic("mount: lost mount");
291 	for (p = allproc.lh_first; p != 0; p = p->p_list.le_next) {
292 		fdp = p->p_fd;
293 		if (fdp->fd_cdir == olddp) {
294 			vrele(fdp->fd_cdir);
295 			VREF(newdp);
296 			fdp->fd_cdir = newdp;
297 		}
298 		if (fdp->fd_rdir == olddp) {
299 			vrele(fdp->fd_rdir);
300 			VREF(newdp);
301 			fdp->fd_rdir = newdp;
302 		}
303 	}
304 	if (rootvnode == olddp) {
305 		vrele(rootvnode);
306 		VREF(newdp);
307 		rootvnode = newdp;
308 	}
309 	vput(newdp);
310 }
311 
312 /*
313  * Unmount a file system.
314  *
315  * Note: unmount takes a path to the vnode mounted on as argument,
316  * not special file (as before).
317  */
318 /* ARGSUSED */
319 int
320 sys_unmount(p, v, retval)
321 	struct proc *p;
322 	void *v;
323 	register_t *retval;
324 {
325 	register struct sys_unmount_args /* {
326 		syscallarg(char *) path;
327 		syscallarg(int) flags;
328 	} */ *uap = v;
329 	register struct vnode *vp;
330 	struct mount *mp;
331 	int error;
332 	struct nameidata nd;
333 
334 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
335 	    SCARG(uap, path), p);
336 	if ((error = namei(&nd)) != 0)
337 		return (error);
338 	vp = nd.ni_vp;
339 	mp = vp->v_mount;
340 
341 	/*
342 	 * Only root, or the user that did the original mount is
343 	 * permitted to unmount this filesystem.
344 	 */
345 	if ((mp->mnt_stat.f_owner != p->p_ucred->cr_uid) &&
346 	    (error = suser(p->p_ucred, &p->p_acflag))) {
347 		vput(vp);
348 		return (error);
349 	}
350 
351 	/*
352 	 * Don't allow unmounting the root file system.
353 	 */
354 	if (mp->mnt_flag & MNT_ROOTFS) {
355 		vput(vp);
356 		return (EINVAL);
357 	}
358 
359 	/*
360 	 * Must be the root of the filesystem
361 	 */
362 	if ((vp->v_flag & VROOT) == 0) {
363 		vput(vp);
364 		return (EINVAL);
365 	}
366 	vput(vp);
367 	return (dounmount(mp, SCARG(uap, flags), p));
368 }
369 
370 /*
371  * Do the actual file system unmount.
372  */
373 int
374 dounmount(mp, flags, p)
375 	register struct mount *mp;
376 	int flags;
377 	struct proc *p;
378 {
379 	struct vnode *coveredvp;
380 	int error;
381 
382 	coveredvp = mp->mnt_vnodecovered;
383 	if (vfs_busy(mp))
384 		return (EBUSY);
385 	mp->mnt_flag |= MNT_UNMOUNT;
386 	if ((error = vfs_lock(mp)) != 0)
387 		return (error);
388 
389 	mp->mnt_flag &=~ MNT_ASYNC;
390 	vnode_pager_umount(mp);	/* release cached vnodes */
391 	cache_purgevfs(mp);	/* remove cache entries for this file sys */
392 	if ((error = VFS_SYNC(mp, MNT_WAIT, p->p_ucred, p)) == 0 ||
393 	    (flags & MNT_FORCE))
394 		error = VFS_UNMOUNT(mp, flags, p);
395 	mp->mnt_flag &= ~MNT_UNMOUNT;
396 	vfs_unbusy(mp);
397 	if (error) {
398 		vfs_unlock(mp);
399 	} else {
400 		CIRCLEQ_REMOVE(&mountlist, mp, mnt_list);
401 		if (coveredvp != NULLVP) {
402 			vrele(coveredvp);
403 			coveredvp->v_mountedhere = (struct mount *)0;
404 		}
405 		mp->mnt_op->vfs_refcount--;
406 		vfs_unlock(mp);
407 		if (mp->mnt_vnodelist.lh_first != NULL)
408 			panic("unmount: dangling vnode");
409 		free((caddr_t)mp, M_MOUNT);
410 	}
411 	return (error);
412 }
413 
414 /*
415  * Sync each mounted filesystem.
416  */
417 #ifdef DEBUG
418 int syncprt = 0;
419 struct ctldebug debug0 = { "syncprt", &syncprt };
420 #endif
421 
422 /* ARGSUSED */
423 int
424 sys_sync(p, v, retval)
425 	struct proc *p;
426 	void *v;
427 	register_t *retval;
428 {
429 	register struct mount *mp, *nmp;
430 	int asyncflag;
431 
432 	for (mp = mountlist.cqh_first; mp != (void *)&mountlist; mp = nmp) {
433 		/*
434 		 * Get the next pointer in case we hang on vfs_busy
435 		 * while we are being unmounted.
436 		 */
437 		nmp = mp->mnt_list.cqe_next;
438 		/*
439 		 * The lock check below is to avoid races with mount
440 		 * and unmount.
441 		 */
442 		if ((mp->mnt_flag & (MNT_MLOCK|MNT_RDONLY|MNT_MPBUSY)) == 0 &&
443 		    !vfs_busy(mp)) {
444 			asyncflag = mp->mnt_flag & MNT_ASYNC;
445 			mp->mnt_flag &= ~MNT_ASYNC;
446 			VFS_SYNC(mp, MNT_NOWAIT, p->p_ucred, p);
447 			if (asyncflag)
448 				mp->mnt_flag |= MNT_ASYNC;
449 			/*
450 			 * Get the next pointer again, as the next filesystem
451 			 * might have been unmounted while we were sync'ing.
452 			 */
453 			nmp = mp->mnt_list.cqe_next;
454 			vfs_unbusy(mp);
455 		}
456 	}
457 #ifdef DEBUG
458 	if (syncprt)
459 		vfs_bufstats();
460 #endif /* DEBUG */
461 	return (0);
462 }
463 
464 /*
465  * Change filesystem quotas.
466  */
467 /* ARGSUSED */
468 int
469 sys_quotactl(p, v, retval)
470 	struct proc *p;
471 	void *v;
472 	register_t *retval;
473 {
474 	register struct sys_quotactl_args /* {
475 		syscallarg(char *) path;
476 		syscallarg(int) cmd;
477 		syscallarg(int) uid;
478 		syscallarg(caddr_t) arg;
479 	} */ *uap = v;
480 	register struct mount *mp;
481 	int error;
482 	struct nameidata nd;
483 
484 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
485 	if ((error = namei(&nd)) != 0)
486 		return (error);
487 	mp = nd.ni_vp->v_mount;
488 	vrele(nd.ni_vp);
489 	return (VFS_QUOTACTL(mp, SCARG(uap, cmd), SCARG(uap, uid),
490 	    SCARG(uap, arg), p));
491 }
492 
493 /*
494  * Get filesystem statistics.
495  */
496 /* ARGSUSED */
497 int
498 sys_statfs(p, v, retval)
499 	struct proc *p;
500 	void *v;
501 	register_t *retval;
502 {
503 	register struct sys_statfs_args /* {
504 		syscallarg(char *) path;
505 		syscallarg(struct statfs *) buf;
506 	} */ *uap = v;
507 	register struct mount *mp;
508 	register struct statfs *sp;
509 	int error;
510 	struct nameidata nd;
511 
512 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
513 	if ((error = namei(&nd)) != 0)
514 		return (error);
515 	mp = nd.ni_vp->v_mount;
516 	sp = &mp->mnt_stat;
517 	vrele(nd.ni_vp);
518 	if ((error = VFS_STATFS(mp, sp, p)) != 0)
519 		return (error);
520 	sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
521 	return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
522 }
523 
524 /*
525  * Get filesystem statistics.
526  */
527 /* ARGSUSED */
528 int
529 sys_fstatfs(p, v, retval)
530 	struct proc *p;
531 	void *v;
532 	register_t *retval;
533 {
534 	register struct sys_fstatfs_args /* {
535 		syscallarg(int) fd;
536 		syscallarg(struct statfs *) buf;
537 	} */ *uap = v;
538 	struct file *fp;
539 	struct mount *mp;
540 	register struct statfs *sp;
541 	int error;
542 
543 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
544 		return (error);
545 	mp = ((struct vnode *)fp->f_data)->v_mount;
546 	sp = &mp->mnt_stat;
547 	if ((error = VFS_STATFS(mp, sp, p)) != 0)
548 		return (error);
549 	sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
550 	return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
551 }
552 
553 /*
554  * Get statistics on all filesystems.
555  */
556 int
557 sys_getfsstat(p, v, retval)
558 	struct proc *p;
559 	void *v;
560 	register_t *retval;
561 {
562 	register struct sys_getfsstat_args /* {
563 		syscallarg(struct statfs *) buf;
564 		syscallarg(long) bufsize;
565 		syscallarg(int) flags;
566 	} */ *uap = v;
567 	register struct mount *mp, *nmp;
568 	register struct statfs *sp;
569 	caddr_t sfsp;
570 	long count, maxcount, error;
571 
572 	maxcount = SCARG(uap, bufsize) / sizeof(struct statfs);
573 	sfsp = (caddr_t)SCARG(uap, buf);
574 	for (count = 0,
575 	     mp = mountlist.cqh_first; mp != (void *)&mountlist; mp = nmp) {
576 		nmp = mp->mnt_list.cqe_next;
577 		if (sfsp && count < maxcount &&
578 		    ((mp->mnt_flag & MNT_MLOCK) == 0)) {
579 			sp = &mp->mnt_stat;
580 			/*
581 			 * If MNT_NOWAIT is specified, do not refresh the
582 			 * fsstat cache. MNT_WAIT overrides MNT_NOWAIT.
583 			 */
584 			if (((SCARG(uap, flags) & MNT_NOWAIT) == 0 ||
585 			    (SCARG(uap, flags) & MNT_WAIT)) &&
586 			    (error = VFS_STATFS(mp, sp, p)))
587 				continue;
588 			sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
589 			error = copyout((caddr_t)sp, sfsp, sizeof(*sp));
590 			if (error)
591 				return (error);
592 			sfsp += sizeof(*sp);
593 		}
594 		count++;
595 	}
596 	if (sfsp && count > maxcount)
597 		*retval = maxcount;
598 	else
599 		*retval = count;
600 	return (0);
601 }
602 
603 /*
604  * Change current working directory to a given file descriptor.
605  */
606 /* ARGSUSED */
607 int
608 sys_fchdir(p, v, retval)
609 	struct proc *p;
610 	void *v;
611 	register_t *retval;
612 {
613 	struct sys_fchdir_args /* {
614 		syscallarg(int) fd;
615 	} */ *uap = v;
616 	register struct filedesc *fdp = p->p_fd;
617 	struct vnode *vp, *tdp;
618 	struct mount *mp;
619 	struct file *fp;
620 	int error;
621 
622 	if ((error = getvnode(fdp, SCARG(uap, fd), &fp)) != 0)
623 		return (error);
624 	vp = (struct vnode *)fp->f_data;
625 	VREF(vp);
626 	VOP_LOCK(vp);
627 	if (vp->v_type != VDIR)
628 		error = ENOTDIR;
629 	else
630 		error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
631 	while (!error && (mp = vp->v_mountedhere) != NULL) {
632 		if (mp->mnt_flag & MNT_MLOCK) {
633 			mp->mnt_flag |= MNT_MWAIT;
634 			sleep((caddr_t)mp, PVFS);
635 			continue;
636 		}
637 		if ((error = VFS_ROOT(mp, &tdp)) != 0)
638 			break;
639 		vput(vp);
640 		vp = tdp;
641 	}
642 	VOP_UNLOCK(vp);
643 	if (error) {
644 		vrele(vp);
645 		return (error);
646 	}
647 	vrele(fdp->fd_cdir);
648 	fdp->fd_cdir = vp;
649 	return (0);
650 }
651 
652 /*
653  * Change current working directory (``.'').
654  */
655 /* ARGSUSED */
656 int
657 sys_chdir(p, v, retval)
658 	struct proc *p;
659 	void *v;
660 	register_t *retval;
661 {
662 	struct sys_chdir_args /* {
663 		syscallarg(char *) path;
664 	} */ *uap = v;
665 	register struct filedesc *fdp = p->p_fd;
666 	int error;
667 	struct nameidata nd;
668 
669 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
670 	    SCARG(uap, path), p);
671 	if ((error = change_dir(&nd, p)) != 0)
672 		return (error);
673 	vrele(fdp->fd_cdir);
674 	fdp->fd_cdir = nd.ni_vp;
675 	return (0);
676 }
677 
678 /*
679  * Change notion of root (``/'') directory.
680  */
681 /* ARGSUSED */
682 int
683 sys_chroot(p, v, retval)
684 	struct proc *p;
685 	void *v;
686 	register_t *retval;
687 {
688 	struct sys_chroot_args /* {
689 		syscallarg(char *) path;
690 	} */ *uap = v;
691 	register struct filedesc *fdp = p->p_fd;
692 	int error;
693 	struct nameidata nd;
694 
695 	if ((error = suser(p->p_ucred, &p->p_acflag)) != 0)
696 		return (error);
697 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
698 	    SCARG(uap, path), p);
699 	if ((error = change_dir(&nd, p)) != 0)
700 		return (error);
701 	if (fdp->fd_rdir != NULL)
702 		vrele(fdp->fd_rdir);
703 	fdp->fd_rdir = nd.ni_vp;
704 	return (0);
705 }
706 
707 /*
708  * Common routine for chroot and chdir.
709  */
710 static int
711 change_dir(ndp, p)
712 	register struct nameidata *ndp;
713 	struct proc *p;
714 {
715 	struct vnode *vp;
716 	int error;
717 
718 	if ((error = namei(ndp)) != 0)
719 		return (error);
720 	vp = ndp->ni_vp;
721 	if (vp->v_type != VDIR)
722 		error = ENOTDIR;
723 	else
724 		error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
725 	VOP_UNLOCK(vp);
726 	if (error)
727 		vrele(vp);
728 	return (error);
729 }
730 
731 /*
732  * Check permissions, allocate an open file structure,
733  * and call the device open routine if any.
734  */
735 int
736 sys_open(p, v, retval)
737 	struct proc *p;
738 	void *v;
739 	register_t *retval;
740 {
741 	register struct sys_open_args /* {
742 		syscallarg(char *) path;
743 		syscallarg(int) flags;
744 		syscallarg(int) mode;
745 	} */ *uap = v;
746 	register struct filedesc *fdp = p->p_fd;
747 	register struct file *fp;
748 	register struct vnode *vp;
749 	int flags, cmode;
750 	struct file *nfp;
751 	int type, indx, error;
752 	struct flock lf;
753 	struct nameidata nd;
754 	extern struct fileops vnops;
755 
756 	if ((error = falloc(p, &nfp, &indx)) != 0)
757 		return (error);
758 	fp = nfp;
759 	flags = FFLAGS(SCARG(uap, flags));
760 	cmode = ((SCARG(uap, mode) &~ fdp->fd_cmask) & ALLPERMS) &~ S_ISTXT;
761 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
762 	p->p_dupfd = -indx - 1;			/* XXX check for fdopen */
763 	if ((error = vn_open(&nd, flags, cmode)) != 0) {
764 		ffree(fp);
765 		if ((error == ENODEV || error == ENXIO) &&
766 		    p->p_dupfd >= 0 &&			/* XXX from fdopen */
767 		    (error =
768 			dupfdopen(fdp, indx, p->p_dupfd, flags, error)) == 0) {
769 			*retval = indx;
770 			return (0);
771 		}
772 		if (error == ERESTART)
773 			error = EINTR;
774 		fdp->fd_ofiles[indx] = NULL;
775 		return (error);
776 	}
777 	p->p_dupfd = 0;
778 	vp = nd.ni_vp;
779 	fp->f_flag = flags & FMASK;
780 	fp->f_type = DTYPE_VNODE;
781 	fp->f_ops = &vnops;
782 	fp->f_data = (caddr_t)vp;
783 	if (flags & (O_EXLOCK | O_SHLOCK)) {
784 		lf.l_whence = SEEK_SET;
785 		lf.l_start = 0;
786 		lf.l_len = 0;
787 		if (flags & O_EXLOCK)
788 			lf.l_type = F_WRLCK;
789 		else
790 			lf.l_type = F_RDLCK;
791 		type = F_FLOCK;
792 		if ((flags & FNONBLOCK) == 0)
793 			type |= F_WAIT;
794 		VOP_UNLOCK(vp);
795 		error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type);
796 		if (error) {
797 			(void) vn_close(vp, fp->f_flag, fp->f_cred, p);
798 			ffree(fp);
799 			fdp->fd_ofiles[indx] = NULL;
800 			return (error);
801 		}
802 		VOP_LOCK(vp);
803 		fp->f_flag |= FHASLOCK;
804 	}
805 	VOP_UNLOCK(vp);
806 	*retval = indx;
807 	return (0);
808 }
809 
810 /*
811  * Create a special file.
812  */
813 /* ARGSUSED */
814 int
815 sys_mknod(p, v, retval)
816 	struct proc *p;
817 	void *v;
818 	register_t *retval;
819 {
820 	register struct sys_mknod_args /* {
821 		syscallarg(char *) path;
822 		syscallarg(int) mode;
823 		syscallarg(int) dev;
824 	} */ *uap = v;
825 	register struct vnode *vp;
826 	struct vattr vattr;
827 	int error;
828 	int whiteout = 0;
829 	struct nameidata nd;
830 
831 	if ((error = suser(p->p_ucred, &p->p_acflag)) != 0)
832 		return (error);
833 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
834 	if ((error = namei(&nd)) != 0)
835 		return (error);
836 	vp = nd.ni_vp;
837 	if (vp != NULL)
838 		error = EEXIST;
839 	else {
840 		VATTR_NULL(&vattr);
841 		vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
842 		vattr.va_rdev = SCARG(uap, dev);
843 		whiteout = 0;
844 
845 		switch (SCARG(uap, mode) & S_IFMT) {
846 		case S_IFMT:	/* used by badsect to flag bad sectors */
847 			vattr.va_type = VBAD;
848 			break;
849 		case S_IFCHR:
850 			vattr.va_type = VCHR;
851 			break;
852 		case S_IFBLK:
853 			vattr.va_type = VBLK;
854 			break;
855 		case S_IFWHT:
856 			whiteout = 1;
857 			break;
858 		default:
859 			error = EINVAL;
860 			break;
861 		}
862 	}
863 	if (!error) {
864 		VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
865 		if (whiteout) {
866 			error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, CREATE);
867 			if (error)
868 				VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
869 			vput(nd.ni_dvp);
870 		} else {
871 			error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp,
872 						&nd.ni_cnd, &vattr);
873 		}
874 	} else {
875 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
876 		if (nd.ni_dvp == vp)
877 			vrele(nd.ni_dvp);
878 		else
879 			vput(nd.ni_dvp);
880 		if (vp)
881 			vrele(vp);
882 	}
883 	return (error);
884 }
885 
886 /*
887  * Create a named pipe.
888  */
889 /* ARGSUSED */
890 int
891 sys_mkfifo(p, v, retval)
892 	struct proc *p;
893 	void *v;
894 	register_t *retval;
895 {
896 #ifndef FIFO
897 	return (EOPNOTSUPP);
898 #else
899 	register struct sys_mkfifo_args /* {
900 		syscallarg(char *) path;
901 		syscallarg(int) mode;
902 	} */ *uap = v;
903 	struct vattr vattr;
904 	int error;
905 	struct nameidata nd;
906 
907 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
908 	if ((error = namei(&nd)) != 0)
909 		return (error);
910 	if (nd.ni_vp != NULL) {
911 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
912 		if (nd.ni_dvp == nd.ni_vp)
913 			vrele(nd.ni_dvp);
914 		else
915 			vput(nd.ni_dvp);
916 		vrele(nd.ni_vp);
917 		return (EEXIST);
918 	}
919 	VATTR_NULL(&vattr);
920 	vattr.va_type = VFIFO;
921 	vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
922 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
923 	return (VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr));
924 #endif /* FIFO */
925 }
926 
927 /*
928  * Make a hard file link.
929  */
930 /* ARGSUSED */
931 int
932 sys_link(p, v, retval)
933 	struct proc *p;
934 	void *v;
935 	register_t *retval;
936 {
937 	register struct sys_link_args /* {
938 		syscallarg(char *) path;
939 		syscallarg(char *) link;
940 	} */ *uap = v;
941 	register struct vnode *vp;
942 	struct nameidata nd;
943 	int error;
944 
945 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
946 	if ((error = namei(&nd)) != 0)
947 		return (error);
948 	vp = nd.ni_vp;
949 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, link), p);
950 	if ((error = namei(&nd)) != 0)
951 		goto out;
952 	if (nd.ni_vp) {
953 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
954 		if (nd.ni_dvp == nd.ni_vp)
955 			vrele(nd.ni_dvp);
956 		else
957 			vput(nd.ni_dvp);
958 		vrele(nd.ni_vp);
959 		error = EEXIST;
960 		goto out;
961 	}
962 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
963 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
964 	error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd);
965 out:
966 	vrele(vp);
967 	return (error);
968 }
969 
970 /*
971  * Make a symbolic link.
972  */
973 /* ARGSUSED */
974 int
975 sys_symlink(p, v, retval)
976 	struct proc *p;
977 	void *v;
978 	register_t *retval;
979 {
980 	register struct sys_symlink_args /* {
981 		syscallarg(char *) path;
982 		syscallarg(char *) link;
983 	} */ *uap = v;
984 	struct vattr vattr;
985 	char *path;
986 	int error;
987 	struct nameidata nd;
988 
989 	MALLOC(path, char *, MAXPATHLEN, M_NAMEI, M_WAITOK);
990 	error = copyinstr(SCARG(uap, path), path, MAXPATHLEN, NULL);
991 	if (error)
992 		goto out;
993 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, link), p);
994 	if ((error = namei(&nd)) != 0)
995 		goto out;
996 	if (nd.ni_vp) {
997 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
998 		if (nd.ni_dvp == nd.ni_vp)
999 			vrele(nd.ni_dvp);
1000 		else
1001 			vput(nd.ni_dvp);
1002 		vrele(nd.ni_vp);
1003 		error = EEXIST;
1004 		goto out;
1005 	}
1006 	VATTR_NULL(&vattr);
1007 	vattr.va_mode = ACCESSPERMS &~ p->p_fd->fd_cmask;
1008 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1009 	error = VOP_SYMLINK(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr, path);
1010 out:
1011 	FREE(path, M_NAMEI);
1012 	return (error);
1013 }
1014 
1015 /*
1016  * Delete a whiteout from the filesystem.
1017  */
1018 /* ARGSUSED */
1019 int
1020 sys_undelete(p, v, retval)
1021 	struct proc *p;
1022 	void *v;
1023 	register_t *retval;
1024 {
1025 	register struct sys_undelete_args /* {
1026 		syscallarg(char *) path;
1027 	} */ *uap = v;
1028 	int error;
1029 	struct nameidata nd;
1030 
1031 	NDINIT(&nd, DELETE, LOCKPARENT|DOWHITEOUT, UIO_USERSPACE,
1032 	    SCARG(uap, path), p);
1033 	error = namei(&nd);
1034 	if (error)
1035 		return (error);
1036 
1037 	if (nd.ni_vp != NULLVP || !(nd.ni_cnd.cn_flags & ISWHITEOUT)) {
1038 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1039 		if (nd.ni_dvp == nd.ni_vp)
1040 			vrele(nd.ni_dvp);
1041 		else
1042 			vput(nd.ni_dvp);
1043 		if (nd.ni_vp)
1044 			vrele(nd.ni_vp);
1045 		return (EEXIST);
1046 	}
1047 
1048 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1049 	if ((error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, DELETE)) != 0)
1050 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1051 	vput(nd.ni_dvp);
1052 	return (error);
1053 }
1054 
1055 /*
1056  * Delete a name from the filesystem.
1057  */
1058 /* ARGSUSED */
1059 int
1060 sys_unlink(p, v, retval)
1061 	struct proc *p;
1062 	void *v;
1063 	register_t *retval;
1064 {
1065 	struct sys_unlink_args /* {
1066 		syscallarg(char *) path;
1067 	} */ *uap = v;
1068 	register struct vnode *vp;
1069 	int error;
1070 	struct nameidata nd;
1071 
1072 	NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
1073 	    SCARG(uap, path), p);
1074 	if ((error = namei(&nd)) != 0)
1075 		return (error);
1076 	vp = nd.ni_vp;
1077 
1078 	/*
1079 	 * The root of a mounted filesystem cannot be deleted.
1080 	 */
1081 	if (vp->v_flag & VROOT) {
1082 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1083 		if (nd.ni_dvp == vp)
1084 			vrele(nd.ni_dvp);
1085 		else
1086 			vput(nd.ni_dvp);
1087 		vput(vp);
1088 		error = EBUSY;
1089 		goto out;
1090 	}
1091 
1092 	(void)vnode_pager_uncache(vp);
1093 
1094 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1095 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1096 	error = VOP_REMOVE(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
1097 out:
1098 	return (error);
1099 }
1100 
1101 /*
1102  * Reposition read/write file offset.
1103  */
1104 int
1105 sys_lseek(p, v, retval)
1106 	struct proc *p;
1107 	void *v;
1108 	register_t *retval;
1109 {
1110 	register struct sys_lseek_args /* {
1111 		syscallarg(int) fd;
1112 		syscallarg(int) pad;
1113 		syscallarg(off_t) offset;
1114 		syscallarg(int) whence;
1115 	} */ *uap = v;
1116 	struct ucred *cred = p->p_ucred;
1117 	register struct filedesc *fdp = p->p_fd;
1118 	register struct file *fp;
1119 	struct vattr vattr;
1120 	int error;
1121 
1122 	if ((u_int)SCARG(uap, fd) >= fdp->fd_nfiles ||
1123 	    (fp = fdp->fd_ofiles[SCARG(uap, fd)]) == NULL)
1124 		return (EBADF);
1125 	if (fp->f_type != DTYPE_VNODE
1126 #ifdef FIFO
1127 	    || ((struct vnode *)fp->f_data)->v_type == VFIFO
1128 #endif
1129 	)
1130 		return (ESPIPE);
1131 	switch (SCARG(uap, whence)) {
1132 	case L_INCR:
1133 		fp->f_offset += SCARG(uap, offset);
1134 		break;
1135 	case L_XTND:
1136 		error = VOP_GETATTR((struct vnode *)fp->f_data, &vattr,
1137 				    cred, p);
1138 		if (error)
1139 			return (error);
1140 		fp->f_offset = SCARG(uap, offset) + vattr.va_size;
1141 		break;
1142 	case L_SET:
1143 		fp->f_offset = SCARG(uap, offset);
1144 		break;
1145 	default:
1146 		return (EINVAL);
1147 	}
1148 	*(off_t *)retval = fp->f_offset;
1149 	return (0);
1150 }
1151 
1152 /*
1153  * Check access permissions.
1154  */
1155 int
1156 sys_access(p, v, retval)
1157 	struct proc *p;
1158 	void *v;
1159 	register_t *retval;
1160 {
1161 	register struct sys_access_args /* {
1162 		syscallarg(char *) path;
1163 		syscallarg(int) flags;
1164 	} */ *uap = v;
1165 	register struct ucred *cred = p->p_ucred;
1166 	register struct vnode *vp;
1167 	int error, flags, t_gid, t_uid;
1168 	struct nameidata nd;
1169 
1170 	t_uid = cred->cr_uid;
1171 	t_gid = cred->cr_gid;
1172 	cred->cr_uid = p->p_cred->p_ruid;
1173 	cred->cr_gid = p->p_cred->p_rgid;
1174 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1175 	    SCARG(uap, path), p);
1176 	if ((error = namei(&nd)) != 0)
1177 		goto out1;
1178 	vp = nd.ni_vp;
1179 
1180 	/* Flags == 0 means only check for existence. */
1181 	if (SCARG(uap, flags)) {
1182 		flags = 0;
1183 		if (SCARG(uap, flags) & R_OK)
1184 			flags |= VREAD;
1185 		if (SCARG(uap, flags) & W_OK)
1186 			flags |= VWRITE;
1187 		if (SCARG(uap, flags) & X_OK)
1188 			flags |= VEXEC;
1189 		if ((flags & VWRITE) == 0 || (error = vn_writechk(vp)) == 0)
1190 			error = VOP_ACCESS(vp, flags, cred, p);
1191 	}
1192 	vput(vp);
1193 out1:
1194 	cred->cr_uid = t_uid;
1195 	cred->cr_gid = t_gid;
1196 	return (error);
1197 }
1198 
1199 /*
1200  * Get file status; this version follows links.
1201  */
1202 /* ARGSUSED */
1203 int
1204 sys_stat(p, v, retval)
1205 	struct proc *p;
1206 	void *v;
1207 	register_t *retval;
1208 {
1209 	register struct sys_stat_args /* {
1210 		syscallarg(char *) path;
1211 		syscallarg(struct stat *) ub;
1212 	} */ *uap = v;
1213 	struct stat sb;
1214 	int error;
1215 	struct nameidata nd;
1216 
1217 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1218 	    SCARG(uap, path), p);
1219 	if ((error = namei(&nd)) != 0)
1220 		return (error);
1221 	error = vn_stat(nd.ni_vp, &sb, p);
1222 	vput(nd.ni_vp);
1223 	if (error)
1224 		return (error);
1225 	error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
1226 	return (error);
1227 }
1228 
1229 /*
1230  * Get file status; this version does not follow links.
1231  */
1232 /* ARGSUSED */
1233 int
1234 sys_lstat(p, v, retval)
1235 	struct proc *p;
1236 	void *v;
1237 	register_t *retval;
1238 {
1239 	register struct sys_lstat_args /* {
1240 		syscallarg(char *) path;
1241 		syscallarg(struct stat *) ub;
1242 	} */ *uap = v;
1243 	struct stat sb;
1244 	int error;
1245 	struct nameidata nd;
1246 
1247 	NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF, UIO_USERSPACE,
1248 	    SCARG(uap, path), p);
1249 	if ((error = namei(&nd)) != 0)
1250 		return (error);
1251 	error = vn_stat(nd.ni_vp, &sb, p);
1252 	vput(nd.ni_vp);
1253 	if (error)
1254 		return (error);
1255 	error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
1256 	return (error);
1257 }
1258 
1259 /*
1260  * Get configurable pathname variables.
1261  */
1262 /* ARGSUSED */
1263 int
1264 sys_pathconf(p, v, retval)
1265 	struct proc *p;
1266 	void *v;
1267 	register_t *retval;
1268 {
1269 	register struct sys_pathconf_args /* {
1270 		syscallarg(char *) path;
1271 		syscallarg(int) name;
1272 	} */ *uap = v;
1273 	int error;
1274 	struct nameidata nd;
1275 
1276 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1277 	    SCARG(uap, path), p);
1278 	if ((error = namei(&nd)) != 0)
1279 		return (error);
1280 	error = VOP_PATHCONF(nd.ni_vp, SCARG(uap, name), retval);
1281 	vput(nd.ni_vp);
1282 	return (error);
1283 }
1284 
1285 /*
1286  * Return target name of a symbolic link.
1287  */
1288 /* ARGSUSED */
1289 int
1290 sys_readlink(p, v, retval)
1291 	struct proc *p;
1292 	void *v;
1293 	register_t *retval;
1294 {
1295 	register struct sys_readlink_args /* {
1296 		syscallarg(char *) path;
1297 		syscallarg(char *) buf;
1298 		syscallarg(int) count;
1299 	} */ *uap = v;
1300 	register struct vnode *vp;
1301 	struct iovec aiov;
1302 	struct uio auio;
1303 	int error;
1304 	struct nameidata nd;
1305 
1306 	NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF, UIO_USERSPACE,
1307 	    SCARG(uap, path), p);
1308 	if ((error = namei(&nd)) != 0)
1309 		return (error);
1310 	vp = nd.ni_vp;
1311 	if (vp->v_type != VLNK)
1312 		error = EINVAL;
1313 	else {
1314 		aiov.iov_base = SCARG(uap, buf);
1315 		aiov.iov_len = SCARG(uap, count);
1316 		auio.uio_iov = &aiov;
1317 		auio.uio_iovcnt = 1;
1318 		auio.uio_offset = 0;
1319 		auio.uio_rw = UIO_READ;
1320 		auio.uio_segflg = UIO_USERSPACE;
1321 		auio.uio_procp = p;
1322 		auio.uio_resid = SCARG(uap, count);
1323 		error = VOP_READLINK(vp, &auio, p->p_ucred);
1324 	}
1325 	vput(vp);
1326 	*retval = SCARG(uap, count) - auio.uio_resid;
1327 	return (error);
1328 }
1329 
1330 /*
1331  * Change flags of a file given a path name.
1332  */
1333 /* ARGSUSED */
1334 int
1335 sys_chflags(p, v, retval)
1336 	struct proc *p;
1337 	void *v;
1338 	register_t *retval;
1339 {
1340 	register struct sys_chflags_args /* {
1341 		syscallarg(char *) path;
1342 		syscallarg(int) flags;
1343 	} */ *uap = v;
1344 	register struct vnode *vp;
1345 	struct vattr vattr;
1346 	int error;
1347 	struct nameidata nd;
1348 
1349 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1350 	if ((error = namei(&nd)) != 0)
1351 		return (error);
1352 	vp = nd.ni_vp;
1353 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1354 	VOP_LOCK(vp);
1355 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1356 		error = EROFS;
1357 	else {
1358 		VATTR_NULL(&vattr);
1359 		vattr.va_flags = SCARG(uap, flags);
1360 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1361 	}
1362 	vput(vp);
1363 	return (error);
1364 }
1365 
1366 /*
1367  * Change flags of a file given a file descriptor.
1368  */
1369 /* ARGSUSED */
1370 int
1371 sys_fchflags(p, v, retval)
1372 	struct proc *p;
1373 	void *v;
1374 	register_t *retval;
1375 {
1376 	register struct sys_fchflags_args /* {
1377 		syscallarg(int) fd;
1378 		syscallarg(int) flags;
1379 	} */ *uap = v;
1380 	struct vattr vattr;
1381 	struct vnode *vp;
1382 	struct file *fp;
1383 	int error;
1384 
1385 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
1386 		return (error);
1387 	vp = (struct vnode *)fp->f_data;
1388 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1389 	VOP_LOCK(vp);
1390 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1391 		error = EROFS;
1392 	else {
1393 		VATTR_NULL(&vattr);
1394 		vattr.va_flags = SCARG(uap, flags);
1395 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1396 	}
1397 	VOP_UNLOCK(vp);
1398 	return (error);
1399 }
1400 
1401 /*
1402  * Change mode of a file given path name.
1403  */
1404 /* ARGSUSED */
1405 int
1406 sys_chmod(p, v, retval)
1407 	struct proc *p;
1408 	void *v;
1409 	register_t *retval;
1410 {
1411 	register struct sys_chmod_args /* {
1412 		syscallarg(char *) path;
1413 		syscallarg(int) mode;
1414 	} */ *uap = v;
1415 	register struct vnode *vp;
1416 	struct vattr vattr;
1417 	int error;
1418 	struct nameidata nd;
1419 
1420 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1421 	if ((error = namei(&nd)) != 0)
1422 		return (error);
1423 	vp = nd.ni_vp;
1424 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1425 	VOP_LOCK(vp);
1426 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1427 		error = EROFS;
1428 	else {
1429 		VATTR_NULL(&vattr);
1430 		vattr.va_mode = SCARG(uap, mode) & ALLPERMS;
1431 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1432 	}
1433 	vput(vp);
1434 	return (error);
1435 }
1436 
1437 /*
1438  * Change mode of a file given a file descriptor.
1439  */
1440 /* ARGSUSED */
1441 int
1442 sys_fchmod(p, v, retval)
1443 	struct proc *p;
1444 	void *v;
1445 	register_t *retval;
1446 {
1447 	register struct sys_fchmod_args /* {
1448 		syscallarg(int) fd;
1449 		syscallarg(int) mode;
1450 	} */ *uap = v;
1451 	struct vattr vattr;
1452 	struct vnode *vp;
1453 	struct file *fp;
1454 	int error;
1455 
1456 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
1457 		return (error);
1458 	vp = (struct vnode *)fp->f_data;
1459 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1460 	VOP_LOCK(vp);
1461 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1462 		error = EROFS;
1463 	else {
1464 		VATTR_NULL(&vattr);
1465 		vattr.va_mode = SCARG(uap, mode) & ALLPERMS;
1466 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1467 	}
1468 	VOP_UNLOCK(vp);
1469 	return (error);
1470 }
1471 
1472 /*
1473  * Set ownership given a path name.
1474  */
1475 /* ARGSUSED */
1476 int
1477 sys_chown(p, v, retval)
1478 	struct proc *p;
1479 	void *v;
1480 	register_t *retval;
1481 {
1482 	register struct sys_chown_args /* {
1483 		syscallarg(char *) path;
1484 		syscallarg(int) uid;
1485 		syscallarg(int) gid;
1486 	} */ *uap = v;
1487 	register struct vnode *vp;
1488 	struct vattr vattr;
1489 	int error;
1490 	struct nameidata nd;
1491 
1492 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1493 	if ((error = namei(&nd)) != 0)
1494 		return (error);
1495 	vp = nd.ni_vp;
1496 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1497 	VOP_LOCK(vp);
1498 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1499 		error = EROFS;
1500 	else {
1501 		VATTR_NULL(&vattr);
1502 		vattr.va_uid = SCARG(uap, uid);
1503 		vattr.va_gid = SCARG(uap, gid);
1504 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1505 	}
1506 	vput(vp);
1507 	return (error);
1508 }
1509 
1510 /*
1511  * Set ownership given a file descriptor.
1512  */
1513 /* ARGSUSED */
1514 int
1515 sys_fchown(p, v, retval)
1516 	struct proc *p;
1517 	void *v;
1518 	register_t *retval;
1519 {
1520 	register struct sys_fchown_args /* {
1521 		syscallarg(int) fd;
1522 		syscallarg(int) uid;
1523 		syscallarg(int) gid;
1524 	} */ *uap = v;
1525 	register struct vnode *vp;
1526 	struct vattr vattr;
1527 	int error;
1528 	struct file *fp;
1529 
1530 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
1531 		return (error);
1532 	vp = (struct vnode *)fp->f_data;
1533 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1534 	VOP_LOCK(vp);
1535 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1536 		error = EROFS;
1537 	else {
1538 		VATTR_NULL(&vattr);
1539 		vattr.va_uid = SCARG(uap, uid);
1540 		vattr.va_gid = SCARG(uap, gid);
1541 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1542 	}
1543 	VOP_UNLOCK(vp);
1544 	return (error);
1545 }
1546 
1547 /*
1548  * Set the access and modification times given a path name.
1549  */
1550 /* ARGSUSED */
1551 int
1552 sys_utimes(p, v, retval)
1553 	struct proc *p;
1554 	void *v;
1555 	register_t *retval;
1556 {
1557 	register struct sys_utimes_args /* {
1558 		syscallarg(char *) path;
1559 		syscallarg(struct timeval *) tptr;
1560 	} */ *uap = v;
1561 	register struct vnode *vp;
1562 	struct timeval tv[2];
1563 	struct vattr vattr;
1564 	int error;
1565 	struct nameidata nd;
1566 
1567 	VATTR_NULL(&vattr);
1568 	if (SCARG(uap, tptr) == NULL) {
1569 		microtime(&tv[0]);
1570 		tv[1] = tv[0];
1571 		vattr.va_vaflags |= VA_UTIMES_NULL;
1572 	} else {
1573 		error = copyin((caddr_t)SCARG(uap, tptr), (caddr_t)tv,
1574 			       sizeof (tv));
1575 		if (error)
1576 			return (error);
1577 	}
1578 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1579 	if ((error = namei(&nd)) != 0)
1580 		return (error);
1581 	vp = nd.ni_vp;
1582 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1583 	VOP_LOCK(vp);
1584 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1585 		error = EROFS;
1586 	else {
1587 		vattr.va_atime.tv_sec = tv[0].tv_sec;
1588 		vattr.va_atime.tv_nsec = tv[0].tv_usec * 1000;
1589 		vattr.va_mtime.tv_sec = tv[1].tv_sec;
1590 		vattr.va_mtime.tv_nsec = tv[1].tv_usec * 1000;
1591 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1592 	}
1593 	vput(vp);
1594 	return (error);
1595 }
1596 
1597 /*
1598  * Set the access and modification times given a file descriptor.
1599  */
1600 /* ARGSUSED */
1601 int
1602 sys_futimes(p, v, retval)
1603 	struct proc *p;
1604 	void *v;
1605 	register_t *retval;
1606 {
1607 	register struct sys_futimes_args /* {
1608 		syscallarg(int) fd;
1609 		syscallarg(struct timeval *) tptr;
1610 	} */ *uap = v;
1611 	register struct vnode *vp;
1612 	struct timeval tv[2];
1613 	struct vattr vattr;
1614 	int error;
1615 	struct file *fp;
1616 
1617 	VATTR_NULL(&vattr);
1618 	if (SCARG(uap, tptr) == NULL) {
1619 		microtime(&tv[0]);
1620 		tv[1] = tv[0];
1621 		vattr.va_vaflags |= VA_UTIMES_NULL;
1622 	} else {
1623 		error = copyin((caddr_t)SCARG(uap, tptr), (caddr_t)tv,
1624 			       sizeof (tv));
1625 		if (error)
1626 			return (error);
1627 	}
1628 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
1629 		return (error);
1630 	vp = (struct vnode *)fp->f_data;
1631 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1632 	VOP_LOCK(vp);
1633 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1634 		error = EROFS;
1635 	else {
1636 		vattr.va_atime.tv_sec = tv[0].tv_sec;
1637 		vattr.va_atime.tv_nsec = tv[0].tv_usec * 1000;
1638 		vattr.va_mtime.tv_sec = tv[1].tv_sec;
1639 		vattr.va_mtime.tv_nsec = tv[1].tv_usec * 1000;
1640 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1641 	}
1642 	VOP_UNLOCK(vp);
1643 	return (error);
1644 }
1645 
1646 /*
1647  * Truncate a file given its path name.
1648  */
1649 /* ARGSUSED */
1650 int
1651 sys_truncate(p, v, retval)
1652 	struct proc *p;
1653 	void *v;
1654 	register_t *retval;
1655 {
1656 	register struct sys_truncate_args /* {
1657 		syscallarg(char *) path;
1658 		syscallarg(int) pad;
1659 		syscallarg(off_t) length;
1660 	} */ *uap = v;
1661 	register struct vnode *vp;
1662 	struct vattr vattr;
1663 	int error;
1664 	struct nameidata nd;
1665 
1666 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1667 	if ((error = namei(&nd)) != 0)
1668 		return (error);
1669 	vp = nd.ni_vp;
1670 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1671 	VOP_LOCK(vp);
1672 	if (vp->v_type == VDIR)
1673 		error = EISDIR;
1674 	else if ((error = vn_writechk(vp)) == 0 &&
1675 	    (error = VOP_ACCESS(vp, VWRITE, p->p_ucred, p)) == 0) {
1676 		VATTR_NULL(&vattr);
1677 		vattr.va_size = SCARG(uap, length);
1678 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1679 	}
1680 	vput(vp);
1681 	return (error);
1682 }
1683 
1684 /*
1685  * Truncate a file given a file descriptor.
1686  */
1687 /* ARGSUSED */
1688 int
1689 sys_ftruncate(p, v, retval)
1690 	struct proc *p;
1691 	void *v;
1692 	register_t *retval;
1693 {
1694 	register struct sys_ftruncate_args /* {
1695 		syscallarg(int) fd;
1696 		syscallarg(int) pad;
1697 		syscallarg(off_t) length;
1698 	} */ *uap = v;
1699 	struct vattr vattr;
1700 	struct vnode *vp;
1701 	struct file *fp;
1702 	int error;
1703 
1704 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
1705 		return (error);
1706 	if ((fp->f_flag & FWRITE) == 0)
1707 		return (EINVAL);
1708 	vp = (struct vnode *)fp->f_data;
1709 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1710 	VOP_LOCK(vp);
1711 	if (vp->v_type == VDIR)
1712 		error = EISDIR;
1713 	else if ((error = vn_writechk(vp)) == 0) {
1714 		VATTR_NULL(&vattr);
1715 		vattr.va_size = SCARG(uap, length);
1716 		error = VOP_SETATTR(vp, &vattr, fp->f_cred, p);
1717 	}
1718 	VOP_UNLOCK(vp);
1719 	return (error);
1720 }
1721 
1722 /*
1723  * Sync an open file.
1724  */
1725 /* ARGSUSED */
1726 int
1727 sys_fsync(p, v, retval)
1728 	struct proc *p;
1729 	void *v;
1730 	register_t *retval;
1731 {
1732 	struct sys_fsync_args /* {
1733 		syscallarg(int) fd;
1734 	} */ *uap = v;
1735 	register struct vnode *vp;
1736 	struct file *fp;
1737 	int error;
1738 
1739 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
1740 		return (error);
1741 	vp = (struct vnode *)fp->f_data;
1742 	VOP_LOCK(vp);
1743 	error = VOP_FSYNC(vp, fp->f_cred, MNT_WAIT, p);
1744 	VOP_UNLOCK(vp);
1745 	return (error);
1746 }
1747 
1748 /*
1749  * Rename files.  Source and destination must either both be directories,
1750  * or both not be directories.  If target is a directory, it must be empty.
1751  */
1752 /* ARGSUSED */
1753 int
1754 sys_rename(p, v, retval)
1755 	struct proc *p;
1756 	void *v;
1757 	register_t *retval;
1758 {
1759 	register struct sys_rename_args /* {
1760 		syscallarg(char *) from;
1761 		syscallarg(char *) to;
1762 	} */ *uap = v;
1763 	register struct vnode *tvp, *fvp, *tdvp;
1764 	struct nameidata fromnd, tond;
1765 	int error;
1766 
1767 	NDINIT(&fromnd, DELETE, WANTPARENT | SAVESTART, UIO_USERSPACE,
1768 	    SCARG(uap, from), p);
1769 	if ((error = namei(&fromnd)) != 0)
1770 		return (error);
1771 	fvp = fromnd.ni_vp;
1772 	NDINIT(&tond, RENAME, LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART,
1773 	    UIO_USERSPACE, SCARG(uap, to), p);
1774 	if ((error = namei(&tond)) != 0) {
1775 		VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
1776 		vrele(fromnd.ni_dvp);
1777 		vrele(fvp);
1778 		goto out1;
1779 	}
1780 	tdvp = tond.ni_dvp;
1781 	tvp = tond.ni_vp;
1782 	if (tvp != NULL) {
1783 		if (fvp->v_type == VDIR && tvp->v_type != VDIR) {
1784 			error = ENOTDIR;
1785 			goto out;
1786 		} else if (fvp->v_type != VDIR && tvp->v_type == VDIR) {
1787 			error = EISDIR;
1788 			goto out;
1789 		}
1790 	}
1791 	if (fvp == tdvp)
1792 		error = EINVAL;
1793 	/*
1794 	 * If source is the same as the destination (that is the
1795 	 * same inode number with the same name in the same directory),
1796 	 * then there is nothing to do.
1797 	 */
1798 	if (fvp == tvp && fromnd.ni_dvp == tdvp &&
1799 	    fromnd.ni_cnd.cn_namelen == tond.ni_cnd.cn_namelen &&
1800 	    !bcmp(fromnd.ni_cnd.cn_nameptr, tond.ni_cnd.cn_nameptr,
1801 	      fromnd.ni_cnd.cn_namelen))
1802 		error = -1;
1803 out:
1804 	if (!error) {
1805 		VOP_LEASE(tdvp, p, p->p_ucred, LEASE_WRITE);
1806 		if (fromnd.ni_dvp != tdvp)
1807 			VOP_LEASE(fromnd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1808 		if (tvp)
1809 			VOP_LEASE(tvp, p, p->p_ucred, LEASE_WRITE);
1810 		error = VOP_RENAME(fromnd.ni_dvp, fromnd.ni_vp, &fromnd.ni_cnd,
1811 				   tond.ni_dvp, tond.ni_vp, &tond.ni_cnd);
1812 	} else {
1813 		VOP_ABORTOP(tond.ni_dvp, &tond.ni_cnd);
1814 		if (tdvp == tvp)
1815 			vrele(tdvp);
1816 		else
1817 			vput(tdvp);
1818 		if (tvp)
1819 			vput(tvp);
1820 		VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
1821 		vrele(fromnd.ni_dvp);
1822 		vrele(fvp);
1823 	}
1824 	vrele(tond.ni_startdir);
1825 	FREE(tond.ni_cnd.cn_pnbuf, M_NAMEI);
1826 out1:
1827 	if (fromnd.ni_startdir)
1828 		vrele(fromnd.ni_startdir);
1829 	FREE(fromnd.ni_cnd.cn_pnbuf, M_NAMEI);
1830 	if (error == -1)
1831 		return (0);
1832 	return (error);
1833 }
1834 
1835 /*
1836  * Make a directory file.
1837  */
1838 /* ARGSUSED */
1839 int
1840 sys_mkdir(p, v, retval)
1841 	struct proc *p;
1842 	void *v;
1843 	register_t *retval;
1844 {
1845 	register struct sys_mkdir_args /* {
1846 		syscallarg(char *) path;
1847 		syscallarg(int) mode;
1848 	} */ *uap = v;
1849 	register struct vnode *vp;
1850 	struct vattr vattr;
1851 	int error;
1852 	struct nameidata nd;
1853 
1854 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1855 	if ((error = namei(&nd)) != 0)
1856 		return (error);
1857 	vp = nd.ni_vp;
1858 	if (vp != NULL) {
1859 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1860 		if (nd.ni_dvp == vp)
1861 			vrele(nd.ni_dvp);
1862 		else
1863 			vput(nd.ni_dvp);
1864 		vrele(vp);
1865 		return (EEXIST);
1866 	}
1867 	VATTR_NULL(&vattr);
1868 	vattr.va_type = VDIR;
1869 	vattr.va_mode = (SCARG(uap, mode) & ACCESSPERMS) &~ p->p_fd->fd_cmask;
1870 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1871 	error = VOP_MKDIR(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
1872 	if (!error)
1873 		vput(nd.ni_vp);
1874 	return (error);
1875 }
1876 
1877 /*
1878  * Remove a directory file.
1879  */
1880 /* ARGSUSED */
1881 int
1882 sys_rmdir(p, v, retval)
1883 	struct proc *p;
1884 	void *v;
1885 	register_t *retval;
1886 {
1887 	struct sys_rmdir_args /* {
1888 		syscallarg(char *) path;
1889 	} */ *uap = v;
1890 	register struct vnode *vp;
1891 	int error;
1892 	struct nameidata nd;
1893 
1894 	NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
1895 	    SCARG(uap, path), p);
1896 	if ((error = namei(&nd)) != 0)
1897 		return (error);
1898 	vp = nd.ni_vp;
1899 	if (vp->v_type != VDIR) {
1900 		error = ENOTDIR;
1901 		goto out;
1902 	}
1903 	/*
1904 	 * No rmdir "." please.
1905 	 */
1906 	if (nd.ni_dvp == vp) {
1907 		error = EINVAL;
1908 		goto out;
1909 	}
1910 	/*
1911 	 * The root of a mounted filesystem cannot be deleted.
1912 	 */
1913 	if (vp->v_flag & VROOT)
1914 		error = EBUSY;
1915 out:
1916 	if (!error) {
1917 		VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1918 		VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1919 		error = VOP_RMDIR(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
1920 	} else {
1921 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1922 		if (nd.ni_dvp == vp)
1923 			vrele(nd.ni_dvp);
1924 		else
1925 			vput(nd.ni_dvp);
1926 		vput(vp);
1927 	}
1928 	return (error);
1929 }
1930 
1931 /*
1932  * Read a block of directory entries in a file system independent format.
1933  */
1934 int
1935 sys_getdirentries(p, v, retval)
1936 	struct proc *p;
1937 	void *v;
1938 	register_t *retval;
1939 {
1940 	register struct sys_getdirentries_args /* {
1941 		syscallarg(int) fd;
1942 		syscallarg(char *) buf;
1943 		syscallarg(u_int) count;
1944 		syscallarg(long *) basep;
1945 	} */ *uap = v;
1946 	register struct vnode *vp;
1947 	struct file *fp;
1948 	struct uio auio;
1949 	struct iovec aiov;
1950 	long loff;
1951 	int error, eofflag;
1952 
1953 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
1954 		return (error);
1955 	if ((fp->f_flag & FREAD) == 0)
1956 		return (EBADF);
1957 	vp = (struct vnode *)fp->f_data;
1958 unionread:
1959 	if (vp->v_type != VDIR)
1960 		return (EINVAL);
1961 	aiov.iov_base = SCARG(uap, buf);
1962 	aiov.iov_len = SCARG(uap, count);
1963 	auio.uio_iov = &aiov;
1964 	auio.uio_iovcnt = 1;
1965 	auio.uio_rw = UIO_READ;
1966 	auio.uio_segflg = UIO_USERSPACE;
1967 	auio.uio_procp = p;
1968 	auio.uio_resid = SCARG(uap, count);
1969 	VOP_LOCK(vp);
1970 	loff = auio.uio_offset = fp->f_offset;
1971 	error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, (u_long *)0, 0);
1972 	fp->f_offset = auio.uio_offset;
1973 	VOP_UNLOCK(vp);
1974 	if (error)
1975 		return (error);
1976 
1977 #ifdef UNION
1978 {
1979 	extern int (**union_vnodeop_p) __P((void *));
1980 	extern struct vnode *union_dircache __P((struct vnode *));
1981 
1982 	if ((SCARG(uap, count) == auio.uio_resid) &&
1983 	    (vp->v_op == union_vnodeop_p)) {
1984 		struct vnode *lvp;
1985 
1986 		lvp = union_dircache(vp);
1987 		if (lvp != NULLVP) {
1988 			struct vattr va;
1989 
1990 			/*
1991 			 * If the directory is opaque,
1992 			 * then don't show lower entries
1993 			 */
1994 			error = VOP_GETATTR(vp, &va, fp->f_cred, p);
1995 			if (va.va_flags & OPAQUE) {
1996 				vput(lvp);
1997 				lvp = NULL;
1998 			}
1999 		}
2000 
2001 		if (lvp != NULLVP) {
2002 			error = VOP_OPEN(lvp, FREAD, fp->f_cred, p);
2003 			VOP_UNLOCK(lvp);
2004 
2005 			if (error) {
2006 				vrele(lvp);
2007 				return (error);
2008 			}
2009 			fp->f_data = (caddr_t) lvp;
2010 			fp->f_offset = 0;
2011 			error = vn_close(vp, FREAD, fp->f_cred, p);
2012 			if (error)
2013 				return (error);
2014 			vp = lvp;
2015 			goto unionread;
2016 		}
2017 	}
2018 }
2019 #endif /* UNION */
2020 
2021 	if ((SCARG(uap, count) == auio.uio_resid) &&
2022 	    (vp->v_flag & VROOT) &&
2023 	    (vp->v_mount->mnt_flag & MNT_UNION)) {
2024 		struct vnode *tvp = vp;
2025 		vp = vp->v_mount->mnt_vnodecovered;
2026 		VREF(vp);
2027 		fp->f_data = (caddr_t) vp;
2028 		fp->f_offset = 0;
2029 		vrele(tvp);
2030 		goto unionread;
2031 	}
2032 	error = copyout((caddr_t)&loff, (caddr_t)SCARG(uap, basep),
2033 	    sizeof(long));
2034 	*retval = SCARG(uap, count) - auio.uio_resid;
2035 	return (error);
2036 }
2037 
2038 /*
2039  * Set the mode mask for creation of filesystem nodes.
2040  */
2041 int
2042 sys_umask(p, v, retval)
2043 	struct proc *p;
2044 	void *v;
2045 	register_t *retval;
2046 {
2047 	struct sys_umask_args /* {
2048 		syscallarg(int) newmask;
2049 	} */ *uap = v;
2050 	register struct filedesc *fdp;
2051 
2052 	fdp = p->p_fd;
2053 	*retval = fdp->fd_cmask;
2054 	fdp->fd_cmask = SCARG(uap, newmask) & ALLPERMS;
2055 	return (0);
2056 }
2057 
2058 /*
2059  * Void all references to file by ripping underlying filesystem
2060  * away from vnode.
2061  */
2062 /* ARGSUSED */
2063 int
2064 sys_revoke(p, v, retval)
2065 	struct proc *p;
2066 	void *v;
2067 	register_t *retval;
2068 {
2069 	register struct sys_revoke_args /* {
2070 		syscallarg(char *) path;
2071 	} */ *uap = v;
2072 	register struct vnode *vp;
2073 	struct vattr vattr;
2074 	int error;
2075 	struct nameidata nd;
2076 
2077 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2078 	if ((error = namei(&nd)) != 0)
2079 		return (error);
2080 	vp = nd.ni_vp;
2081 	if (vp->v_type != VCHR && vp->v_type != VBLK) {
2082 		error = EINVAL;
2083 		goto out;
2084 	}
2085 	if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)) != 0)
2086 		goto out;
2087 	if (p->p_ucred->cr_uid != vattr.va_uid &&
2088 	    (error = suser(p->p_ucred, &p->p_acflag)))
2089 		goto out;
2090 	if (vp->v_usecount > 1 || (vp->v_flag & VALIASED))
2091 		vgoneall(vp);
2092 out:
2093 	vrele(vp);
2094 	return (error);
2095 }
2096 
2097 /*
2098  * Convert a user file descriptor to a kernel file entry.
2099  */
2100 int
2101 getvnode(fdp, fd, fpp)
2102 	struct filedesc *fdp;
2103 	int fd;
2104 	struct file **fpp;
2105 {
2106 	struct vnode *vp;
2107 	struct file *fp;
2108 
2109 	if ((u_int)fd >= fdp->fd_nfiles ||
2110 	    (fp = fdp->fd_ofiles[fd]) == NULL)
2111 		return (EBADF);
2112 	if (fp->f_type != DTYPE_VNODE)
2113 		return (EINVAL);
2114 	vp = (struct vnode *)fp->f_data;
2115 	if (vp->v_type == VBAD)
2116 		return (EBADF);
2117 	*fpp = fp;
2118 	return (0);
2119 }
2120