xref: /netbsd-src/sys/kern/vfs_syscalls.c (revision 201fac6e3d8ac6a5503c4810b0378f73101d50c3)
1 /*	$NetBSD: vfs_syscalls.c,v 1.45 1994/12/14 19:36:15 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 *ndp, struct proc *p));
63 
64 /*
65  * Virtual File System System Calls
66  */
67 
68 /*
69  * Mount a file system.
70  */
71 /* ARGSUSED */
72 mount(p, uap, retval)
73 	struct proc *p;
74 	register struct mount_args /* {
75 		syscallarg(char *) type;
76 		syscallarg(char *) path;
77 		syscallarg(int) flags;
78 		syscallarg(caddr_t) data;
79 	} */ *uap;
80 	register_t *retval;
81 {
82 	register struct vnode *vp;
83 	register struct mount *mp;
84 	int error, flag;
85 	u_long fsindex;
86 	char fstypename[MFSNAMELEN];
87 	struct vattr va;
88 	struct nameidata nd;
89 
90 	/*
91 	 * Get vnode to be covered
92 	 */
93 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
94 	    SCARG(uap, path), p);
95 	if (error = namei(&nd))
96 		return (error);
97 	vp = nd.ni_vp;
98 	if (SCARG(uap, flags) & MNT_UPDATE) {
99 		if ((vp->v_flag & VROOT) == 0) {
100 			vput(vp);
101 			return (EINVAL);
102 		}
103 		mp = vp->v_mount;
104 		flag = mp->mnt_flag;
105 		/*
106 		 * We only allow the filesystem to be reloaded if it
107 		 * is currently mounted read-only.
108 		 */
109 		if ((SCARG(uap, flags) & MNT_RELOAD) &&
110 		    ((mp->mnt_flag & MNT_RDONLY) == 0)) {
111 			vput(vp);
112 			return (EOPNOTSUPP);	/* Needs translation */
113 		}
114 		mp->mnt_flag |=
115 		    SCARG(uap, flags) & (MNT_RELOAD | MNT_FORCE | MNT_UPDATE);
116 		/*
117 		 * Only root, or the user that did the original mount is
118 		 * permitted to update it.
119 		 */
120 		if (mp->mnt_stat.f_owner != p->p_ucred->cr_uid &&
121 		    (error = suser(p->p_ucred, &p->p_acflag))) {
122 			vput(vp);
123 			return (error);
124 		}
125 		/*
126 		 * Do not allow NFS export by non-root users. Silently
127 		 * enforce MNT_NOSUID and MNT_NODEV for non-root users.
128 		 */
129 		if (p->p_ucred->cr_uid != 0) {
130 			if (SCARG(uap, flags) & MNT_EXPORTED) {
131 				vput(vp);
132 				return (EPERM);
133 			}
134 			SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
135 		}
136 		VOP_UNLOCK(vp);
137 		goto update;
138 	}
139 	/*
140 	 * If the user is not root, ensure that they own the directory
141 	 * onto which we are attempting to mount.
142 	 */
143 	if ((error = VOP_GETATTR(vp, &va, p->p_ucred, p)) ||
144 	    (va.va_uid != p->p_ucred->cr_uid &&
145 	     (error = suser(p->p_ucred, &p->p_acflag)))) {
146 		vput(vp);
147 		return (error);
148 	}
149 	/*
150 	 * Do not allow NFS export by non-root users. Silently
151 	 * enforce MNT_NOSUID and MNT_NODEV for non-root users.
152 	 */
153 	if (p->p_ucred->cr_uid != 0) {
154 		if (SCARG(uap, flags) & MNT_EXPORTED) {
155 			vput(vp);
156 			return (EPERM);
157 		}
158 		SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
159 	}
160 	if (error = vinvalbuf(vp, V_SAVE, p->p_ucred, p, 0, 0))
161 		return (error);
162 	if (vp->v_type != VDIR) {
163 		vput(vp);
164 		return (ENOTDIR);
165 	}
166 	if (error = copyin(SCARG(uap, type), fstypename, MFSNAMELEN)) {
167 #ifdef COMPAT_09
168 		goto check_num;
169 #else
170 		vput(vp);
171 		return (error);
172 #endif
173 	}
174 	fstypename[MFSNAMELEN] = '\0';
175 	for (fsindex = 0; fsindex < nvfssw; fsindex++)
176 		if (vfssw[fsindex] != NULL &&
177 		    !strncmp(vfssw[fsindex]->vfs_name, fstypename, MFSNAMELEN))
178 			break;
179 	if (fsindex >= nvfssw) {
180 #ifdef COMPAT_09
181 check_num:
182 		fsindex = (u_long)SCARG(uap, type);
183 		if (fsindex >= nvfssw || vfssw[fsindex] == NULL) {
184 			vput(vp);
185 			return (ENODEV);
186 		}
187 #else
188 		vput(vp);
189 		return (ENODEV);
190 #endif
191 	}
192 	if (vp->v_mountedhere != NULL) {
193 		vput(vp);
194 		return (EBUSY);
195 	}
196 
197 	/*
198 	 * Allocate and initialize the file system.
199 	 */
200 	mp = (struct mount *)malloc((u_long)sizeof(struct mount),
201 		M_MOUNT, M_WAITOK);
202 	bzero((char *)mp, (u_long)sizeof(struct mount));
203 	mp->mnt_op = vfssw[fsindex];
204 	if (error = vfs_lock(mp)) {
205 		free((caddr_t)mp, M_MOUNT);
206 		vput(vp);
207 		return (error);
208 	}
209 	/* Do this early in case we block later. */
210 	vfssw[fsindex]->vfs_refcount++;
211 	vp->v_mountedhere = mp;
212 	mp->mnt_vnodecovered = vp;
213 	mp->mnt_stat.f_owner = p->p_ucred->cr_uid;
214 update:
215 	/*
216 	 * Set the mount level flags.
217 	 */
218 	if (SCARG(uap, flags) & MNT_RDONLY)
219 		mp->mnt_flag |= MNT_RDONLY;
220 	else if (mp->mnt_flag & MNT_RDONLY)
221 		mp->mnt_flag |= MNT_WANTRDWR;
222 	mp->mnt_flag &=~ (MNT_NOSUID | MNT_NOEXEC | MNT_NODEV |
223 	    MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC);
224 	mp->mnt_flag |= SCARG(uap, flags) & (MNT_NOSUID | MNT_NOEXEC |
225 	    MNT_NODEV | MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC);
226 	/*
227 	 * Mount the filesystem.
228 	 */
229 	error = VFS_MOUNT(mp, SCARG(uap, path), SCARG(uap, data), &nd, p);
230 	if (mp->mnt_flag & MNT_UPDATE) {
231 		vrele(vp);
232 		if (mp->mnt_flag & MNT_WANTRDWR)
233 			mp->mnt_flag &= ~MNT_RDONLY;
234 		mp->mnt_flag &=~
235 		    (MNT_UPDATE | MNT_RELOAD | MNT_FORCE | MNT_WANTRDWR);
236 		if (error)
237 			mp->mnt_flag = flag;
238 		return (error);
239 	}
240 	/*
241 	 * Put the new filesystem on the mount list after root.
242 	 */
243 	cache_purge(vp);
244 	if (!error) {
245 		TAILQ_INSERT_TAIL(&mountlist, mp, mnt_list);
246 		checkdirs(vp);
247 		VOP_UNLOCK(vp);
248 		vfs_unlock(mp);
249 		error = VFS_START(mp, 0, p);
250 	} else {
251 		mp->mnt_vnodecovered->v_mountedhere = (struct mount *)0;
252 		vfssw[fsindex]->vfs_refcount--;
253 		vfs_unlock(mp);
254 		free((caddr_t)mp, M_MOUNT);
255 		vput(vp);
256 	}
257 	return (error);
258 }
259 
260 /*
261  * Scan all active processes to see if any of them have a current
262  * or root directory onto which the new filesystem has just been
263  * mounted. If so, replace them with the new mount point.
264  */
265 checkdirs(olddp)
266 	struct vnode *olddp;
267 {
268 	struct filedesc *fdp;
269 	struct vnode *newdp;
270 	struct proc *p;
271 
272 	if (olddp->v_usecount == 1)
273 		return;
274 	if (VFS_ROOT(olddp->v_mountedhere, &newdp))
275 		panic("mount: lost mount");
276 	for (p = allproc.lh_first; p != 0; p = p->p_list.le_next) {
277 		fdp = p->p_fd;
278 		if (fdp->fd_cdir == olddp) {
279 			vrele(fdp->fd_cdir);
280 			VREF(newdp);
281 			fdp->fd_cdir = newdp;
282 		}
283 		if (fdp->fd_rdir == olddp) {
284 			vrele(fdp->fd_rdir);
285 			VREF(newdp);
286 			fdp->fd_rdir = newdp;
287 		}
288 	}
289 	if (rootvnode == olddp) {
290 		vrele(rootvnode);
291 		VREF(newdp);
292 		rootvnode = newdp;
293 	}
294 	vput(newdp);
295 }
296 
297 /*
298  * Unmount a file system.
299  *
300  * Note: unmount takes a path to the vnode mounted on as argument,
301  * not special file (as before).
302  */
303 /* ARGSUSED */
304 unmount(p, uap, retval)
305 	struct proc *p;
306 	register struct unmount_args /* {
307 		syscallarg(char *) path;
308 		syscallarg(int) flags;
309 	} */ *uap;
310 	register_t *retval;
311 {
312 	register struct vnode *vp;
313 	struct mount *mp;
314 	int error;
315 	struct nameidata nd;
316 
317 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
318 	    SCARG(uap, path), p);
319 	if (error = namei(&nd))
320 		return (error);
321 	vp = nd.ni_vp;
322 	mp = vp->v_mount;
323 
324 	/*
325 	 * Only root, or the user that did the original mount is
326 	 * permitted to unmount this filesystem.
327 	 */
328 	if ((mp->mnt_stat.f_owner != p->p_ucred->cr_uid) &&
329 	    (error = suser(p->p_ucred, &p->p_acflag))) {
330 		vput(vp);
331 		return (error);
332 	}
333 
334 	/*
335 	 * Must be the root of the filesystem
336 	 */
337 	if ((vp->v_flag & VROOT) == 0) {
338 		vput(vp);
339 		return (EINVAL);
340 	}
341 	vput(vp);
342 	return (dounmount(mp, SCARG(uap, flags), p));
343 }
344 
345 /*
346  * Do the actual file system unmount.
347  */
348 dounmount(mp, flags, p)
349 	register struct mount *mp;
350 	int flags;
351 	struct proc *p;
352 {
353 	struct vnode *coveredvp;
354 	int error;
355 
356 	coveredvp = mp->mnt_vnodecovered;
357 	if (vfs_busy(mp))
358 		return (EBUSY);
359 	mp->mnt_flag |= MNT_UNMOUNT;
360 	if (error = vfs_lock(mp))
361 		return (error);
362 
363 	mp->mnt_flag &=~ MNT_ASYNC;
364 	vnode_pager_umount(mp);	/* release cached vnodes */
365 	cache_purgevfs(mp);	/* remove cache entries for this file sys */
366 	if ((error = VFS_SYNC(mp, MNT_WAIT, p->p_ucred, p)) == 0 ||
367 	    (flags & MNT_FORCE))
368 		error = VFS_UNMOUNT(mp, flags, p);
369 	mp->mnt_flag &= ~MNT_UNMOUNT;
370 	vfs_unbusy(mp);
371 	if (error) {
372 		vfs_unlock(mp);
373 	} else {
374 		vrele(coveredvp);
375 		TAILQ_REMOVE(&mountlist, mp, mnt_list);
376 		mp->mnt_vnodecovered->v_mountedhere = (struct mount *)0;
377 		mp->mnt_op->vfs_refcount--;
378 		vfs_unlock(mp);
379 		if (mp->mnt_vnodelist.lh_first != NULL)
380 			panic("unmount: dangling vnode");
381 		free((caddr_t)mp, M_MOUNT);
382 	}
383 	return (error);
384 }
385 
386 /*
387  * Sync each mounted filesystem.
388  */
389 #ifdef DEBUG
390 int syncprt = 0;
391 struct ctldebug debug0 = { "syncprt", &syncprt };
392 #endif
393 
394 /* ARGSUSED */
395 sync(p, uap, retval)
396 	struct proc *p;
397 	void *uap;
398 	register_t *retval;
399 {
400 	register struct mount *mp, *nmp;
401 	int asyncflag;
402 
403 	for (mp = mountlist.tqh_first; mp != NULL; mp = nmp) {
404 		/*
405 		 * Get the next pointer in case we hang on vfs_busy
406 		 * while we are being unmounted.
407 		 */
408 		nmp = mp->mnt_list.tqe_next;
409 		/*
410 		 * The lock check below is to avoid races with mount
411 		 * and unmount.
412 		 */
413 		if ((mp->mnt_flag & (MNT_MLOCK|MNT_RDONLY|MNT_MPBUSY)) == 0 &&
414 		    !vfs_busy(mp)) {
415 			asyncflag = mp->mnt_flag & MNT_ASYNC;
416 			mp->mnt_flag &= ~MNT_ASYNC;
417 			VFS_SYNC(mp, MNT_NOWAIT, p->p_ucred, p);
418 			if (asyncflag)
419 				mp->mnt_flag |= MNT_ASYNC;
420 			/*
421 			 * Get the next pointer again, as the next filesystem
422 			 * might have been unmounted while we were sync'ing.
423 			 */
424 			nmp = mp->mnt_list.tqe_next;
425 			vfs_unbusy(mp);
426 		}
427 	}
428 #ifdef DEBUG
429 	if (syncprt)
430 		vfs_bufstats();
431 #endif /* DEBUG */
432 	return (0);
433 }
434 
435 /*
436  * Change filesystem quotas.
437  */
438 /* ARGSUSED */
439 quotactl(p, uap, retval)
440 	struct proc *p;
441 	register struct quotactl_args /* {
442 		syscallarg(char *) path;
443 		syscallarg(int) cmd;
444 		syscallarg(int) uid;
445 		syscallarg(caddr_t) arg;
446 	} */ *uap;
447 	register_t *retval;
448 {
449 	register struct mount *mp;
450 	int error;
451 	struct nameidata nd;
452 
453 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
454 	if (error = namei(&nd))
455 		return (error);
456 	mp = nd.ni_vp->v_mount;
457 	vrele(nd.ni_vp);
458 	return (VFS_QUOTACTL(mp, SCARG(uap, cmd), SCARG(uap, uid),
459 	    SCARG(uap, arg), p));
460 }
461 
462 /*
463  * Get filesystem statistics.
464  */
465 /* ARGSUSED */
466 statfs(p, uap, retval)
467 	struct proc *p;
468 	register struct statfs_args /* {
469 		syscallarg(char *) path;
470 		syscallarg(struct statfs *) buf;
471 	} */ *uap;
472 	register_t *retval;
473 {
474 	register struct mount *mp;
475 	register struct statfs *sp;
476 	int error;
477 	struct nameidata nd;
478 
479 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
480 	if (error = namei(&nd))
481 		return (error);
482 	mp = nd.ni_vp->v_mount;
483 	sp = &mp->mnt_stat;
484 	vrele(nd.ni_vp);
485 	if (error = VFS_STATFS(mp, sp, p))
486 		return (error);
487 	sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
488 	return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
489 }
490 
491 /*
492  * Get filesystem statistics.
493  */
494 /* ARGSUSED */
495 fstatfs(p, uap, retval)
496 	struct proc *p;
497 	register struct fstatfs_args /* {
498 		syscallarg(int) fd;
499 		syscallarg(struct statfs *) buf;
500 	} */ *uap;
501 	register_t *retval;
502 {
503 	struct file *fp;
504 	struct mount *mp;
505 	register struct statfs *sp;
506 	int error;
507 
508 	if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
509 		return (error);
510 	mp = ((struct vnode *)fp->f_data)->v_mount;
511 	sp = &mp->mnt_stat;
512 	if (error = VFS_STATFS(mp, sp, p))
513 		return (error);
514 	sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
515 	return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
516 }
517 
518 /*
519  * Get statistics on all filesystems.
520  */
521 getfsstat(p, uap, retval)
522 	struct proc *p;
523 	register struct getfsstat_args /* {
524 		syscallarg(struct statfs *) buf;
525 		syscallarg(long) bufsize;
526 		syscallarg(int) flags;
527 	} */ *uap;
528 	register_t *retval;
529 {
530 	register struct mount *mp, *nmp;
531 	register struct statfs *sp;
532 	caddr_t sfsp;
533 	long count, maxcount, error;
534 
535 	maxcount = SCARG(uap, bufsize) / sizeof(struct statfs);
536 	sfsp = (caddr_t)SCARG(uap, buf);
537 	for (count = 0, mp = mountlist.tqh_first; mp != NULL; mp = nmp) {
538 		nmp = mp->mnt_list.tqe_next;
539 		if (sfsp && count < maxcount &&
540 		    ((mp->mnt_flag & MNT_MLOCK) == 0)) {
541 			sp = &mp->mnt_stat;
542 			/*
543 			 * If MNT_NOWAIT is specified, do not refresh the
544 			 * fsstat cache. MNT_WAIT overrides MNT_NOWAIT.
545 			 */
546 			if (((SCARG(uap, flags) & MNT_NOWAIT) == 0 ||
547 			    (SCARG(uap, flags) & MNT_WAIT)) &&
548 			    (error = VFS_STATFS(mp, sp, p)))
549 				continue;
550 			sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
551 			if (error = copyout((caddr_t)sp, sfsp, sizeof(*sp)))
552 				return (error);
553 			sfsp += sizeof(*sp);
554 		}
555 		count++;
556 	}
557 	if (sfsp && count > maxcount)
558 		*retval = maxcount;
559 	else
560 		*retval = count;
561 	return (0);
562 }
563 
564 /*
565  * Change current working directory to a given file descriptor.
566  */
567 /* ARGSUSED */
568 fchdir(p, uap, retval)
569 	struct proc *p;
570 	struct fchdir_args /* {
571 		syscallarg(int) fd;
572 	} */ *uap;
573 	register_t *retval;
574 {
575 	register struct filedesc *fdp = p->p_fd;
576 	struct vnode *vp, *tdp;
577 	struct mount *mp;
578 	struct file *fp;
579 	int error;
580 
581 	if (error = getvnode(fdp, SCARG(uap, fd), &fp))
582 		return (error);
583 	vp = (struct vnode *)fp->f_data;
584 	VREF(vp);
585 	VOP_LOCK(vp);
586 	if (vp->v_type != VDIR)
587 		error = ENOTDIR;
588 	else
589 		error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
590 	while (!error && (mp = vp->v_mountedhere) != NULL) {
591 		if (mp->mnt_flag & MNT_MLOCK) {
592 			mp->mnt_flag |= MNT_MWAIT;
593 			sleep((caddr_t)mp, PVFS);
594 			continue;
595 		}
596 		if (error = VFS_ROOT(mp, &tdp))
597 			break;
598 		vput(vp);
599 		vp = tdp;
600 	}
601 	VOP_UNLOCK(vp);
602 	if (error) {
603 		vrele(vp);
604 		return (error);
605 	}
606 	vrele(fdp->fd_cdir);
607 	fdp->fd_cdir = vp;
608 	return (0);
609 }
610 
611 /*
612  * Change current working directory (``.'').
613  */
614 /* ARGSUSED */
615 chdir(p, uap, retval)
616 	struct proc *p;
617 	struct chdir_args /* {
618 		syscallarg(char *) path;
619 	} */ *uap;
620 	register_t *retval;
621 {
622 	register struct filedesc *fdp = p->p_fd;
623 	int error;
624 	struct nameidata nd;
625 
626 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
627 	    SCARG(uap, path), p);
628 	if (error = change_dir(&nd, p))
629 		return (error);
630 	vrele(fdp->fd_cdir);
631 	fdp->fd_cdir = nd.ni_vp;
632 	return (0);
633 }
634 
635 /*
636  * Change notion of root (``/'') directory.
637  */
638 /* ARGSUSED */
639 chroot(p, uap, retval)
640 	struct proc *p;
641 	struct chroot_args /* {
642 		syscallarg(char *) path;
643 	} */ *uap;
644 	register_t *retval;
645 {
646 	register struct filedesc *fdp = p->p_fd;
647 	int error;
648 	struct nameidata nd;
649 
650 	if (error = suser(p->p_ucred, &p->p_acflag))
651 		return (error);
652 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
653 	    SCARG(uap, path), p);
654 	if (error = change_dir(&nd, p))
655 		return (error);
656 	if (fdp->fd_rdir != NULL)
657 		vrele(fdp->fd_rdir);
658 	fdp->fd_rdir = nd.ni_vp;
659 	return (0);
660 }
661 
662 /*
663  * Common routine for chroot and chdir.
664  */
665 static int
666 change_dir(ndp, p)
667 	register struct nameidata *ndp;
668 	struct proc *p;
669 {
670 	struct vnode *vp;
671 	int error;
672 
673 	if (error = namei(ndp))
674 		return (error);
675 	vp = ndp->ni_vp;
676 	if (vp->v_type != VDIR)
677 		error = ENOTDIR;
678 	else
679 		error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
680 	VOP_UNLOCK(vp);
681 	if (error)
682 		vrele(vp);
683 	return (error);
684 }
685 
686 /*
687  * Check permissions, allocate an open file structure,
688  * and call the device open routine if any.
689  */
690 open(p, uap, retval)
691 	struct proc *p;
692 	register struct open_args /* {
693 		syscallarg(char *) path;
694 		syscallarg(int) flags;
695 		syscallarg(int) mode;
696 	} */ *uap;
697 	register_t *retval;
698 {
699 	register struct filedesc *fdp = p->p_fd;
700 	register struct file *fp;
701 	register struct vnode *vp;
702 	int flags, cmode;
703 	struct file *nfp;
704 	int type, indx, error;
705 	struct flock lf;
706 	struct nameidata nd;
707 	extern struct fileops vnops;
708 
709 	if (error = falloc(p, &nfp, &indx))
710 		return (error);
711 	fp = nfp;
712 	flags = FFLAGS(SCARG(uap, flags));
713 	cmode = ((SCARG(uap, mode) &~ fdp->fd_cmask) & ALLPERMS) &~ S_ISTXT;
714 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
715 	p->p_dupfd = -indx - 1;			/* XXX check for fdopen */
716 	if (error = vn_open(&nd, flags, cmode)) {
717 		ffree(fp);
718 		if ((error == ENODEV || error == ENXIO) &&
719 		    p->p_dupfd >= 0 &&			/* XXX from fdopen */
720 		    (error =
721 			dupfdopen(fdp, indx, p->p_dupfd, flags, error)) == 0) {
722 			*retval = indx;
723 			return (0);
724 		}
725 		if (error == ERESTART)
726 			error = EINTR;
727 		fdp->fd_ofiles[indx] = NULL;
728 		return (error);
729 	}
730 	p->p_dupfd = 0;
731 	vp = nd.ni_vp;
732 	fp->f_flag = flags & FMASK;
733 	fp->f_type = DTYPE_VNODE;
734 	fp->f_ops = &vnops;
735 	fp->f_data = (caddr_t)vp;
736 	if (flags & (O_EXLOCK | O_SHLOCK)) {
737 		lf.l_whence = SEEK_SET;
738 		lf.l_start = 0;
739 		lf.l_len = 0;
740 		if (flags & O_EXLOCK)
741 			lf.l_type = F_WRLCK;
742 		else
743 			lf.l_type = F_RDLCK;
744 		type = F_FLOCK;
745 		if ((flags & FNONBLOCK) == 0)
746 			type |= F_WAIT;
747 		VOP_UNLOCK(vp);
748 		if (error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type)) {
749 			(void) vn_close(vp, fp->f_flag, fp->f_cred, p);
750 			ffree(fp);
751 			fdp->fd_ofiles[indx] = NULL;
752 			return (error);
753 		}
754 		VOP_LOCK(vp);
755 		fp->f_flag |= FHASLOCK;
756 	}
757 	VOP_UNLOCK(vp);
758 	*retval = indx;
759 	return (0);
760 }
761 
762 #ifdef COMPAT_43
763 /*
764  * Create a file.
765  */
766 compat_43_creat(p, uap, retval)
767 	struct proc *p;
768 	register struct compat_43_creat_args /* {
769 		syscallarg(char *) path;
770 		syscallarg(int) mode;
771 	} */ *uap;
772 	register_t *retval;
773 {
774 	struct open_args /* {
775 		syscallarg(char *) path;
776 		syscallarg(int) flags;
777 		syscallarg(int) mode;
778 	} */ nuap;
779 
780 	SCARG(&nuap, path) = SCARG(uap, path);
781 	SCARG(&nuap, mode) = SCARG(uap, mode);
782 	SCARG(&nuap, flags) = O_WRONLY | O_CREAT | O_TRUNC;
783 	return (open(p, &nuap, retval));
784 }
785 #endif /* COMPAT_43 */
786 
787 /*
788  * Create a special file.
789  */
790 /* ARGSUSED */
791 mknod(p, uap, retval)
792 	struct proc *p;
793 	register struct mknod_args /* {
794 		syscallarg(char *) path;
795 		syscallarg(int) mode;
796 		syscallarg(int) dev;
797 	} */ *uap;
798 	register_t *retval;
799 {
800 	register struct vnode *vp;
801 	struct vattr vattr;
802 	int error;
803 	int whiteout;
804 	struct nameidata nd;
805 
806 	if (error = suser(p->p_ucred, &p->p_acflag))
807 		return (error);
808 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
809 	if (error = namei(&nd))
810 		return (error);
811 	vp = nd.ni_vp;
812 	if (vp != NULL)
813 		error = EEXIST;
814 	else {
815 		VATTR_NULL(&vattr);
816 		vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
817 		vattr.va_rdev = SCARG(uap, dev);
818 		whiteout = 0;
819 
820 		switch (SCARG(uap, mode) & S_IFMT) {
821 		case S_IFMT:	/* used by badsect to flag bad sectors */
822 			vattr.va_type = VBAD;
823 			break;
824 		case S_IFCHR:
825 			vattr.va_type = VCHR;
826 			break;
827 		case S_IFBLK:
828 			vattr.va_type = VBLK;
829 			break;
830 		case S_IFWHT:
831 			whiteout = 1;
832 			break;
833 		default:
834 			error = EINVAL;
835 			break;
836 		}
837 	}
838 	if (!error) {
839 		VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
840 		if (whiteout) {
841 			error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, CREATE);
842 			if (error)
843 				VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
844 			vput(nd.ni_dvp);
845 		} else {
846 			error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp,
847 						&nd.ni_cnd, &vattr);
848 		}
849 	} else {
850 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
851 		if (nd.ni_dvp == vp)
852 			vrele(nd.ni_dvp);
853 		else
854 			vput(nd.ni_dvp);
855 		if (vp)
856 			vrele(vp);
857 	}
858 	return (error);
859 }
860 
861 /*
862  * Create a named pipe.
863  */
864 /* ARGSUSED */
865 mkfifo(p, uap, retval)
866 	struct proc *p;
867 	register struct mkfifo_args /* {
868 		syscallarg(char *) path;
869 		syscallarg(int) mode;
870 	} */ *uap;
871 	register_t *retval;
872 {
873 	struct vattr vattr;
874 	int error;
875 	struct nameidata nd;
876 
877 #ifndef FIFO
878 	return (EOPNOTSUPP);
879 #else
880 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
881 	if (error = namei(&nd))
882 		return (error);
883 	if (nd.ni_vp != NULL) {
884 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
885 		if (nd.ni_dvp == nd.ni_vp)
886 			vrele(nd.ni_dvp);
887 		else
888 			vput(nd.ni_dvp);
889 		vrele(nd.ni_vp);
890 		return (EEXIST);
891 	}
892 	VATTR_NULL(&vattr);
893 	vattr.va_type = VFIFO;
894 	vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
895 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
896 	return (VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr));
897 #endif /* FIFO */
898 }
899 
900 /*
901  * Make a hard file link.
902  */
903 /* ARGSUSED */
904 link(p, uap, retval)
905 	struct proc *p;
906 	register struct link_args /* {
907 		syscallarg(char *) path;
908 		syscallarg(char *) link;
909 	} */ *uap;
910 	register_t *retval;
911 {
912 	register struct vnode *vp;
913 	struct nameidata nd;
914 	int error;
915 
916 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
917 	if (error = namei(&nd))
918 		return (error);
919 	vp = nd.ni_vp;
920 	if (vp->v_type != VDIR ||
921 	    (error = suser(p->p_ucred, &p->p_acflag)) == 0) {
922 		nd.ni_cnd.cn_nameiop = CREATE;
923 		nd.ni_cnd.cn_flags = LOCKPARENT;
924 		nd.ni_dirp = SCARG(uap, link);
925 		if ((error = namei(&nd)) == 0) {
926 			if (nd.ni_vp != NULL)
927 				error = EEXIST;
928 			if (!error) {
929 				VOP_LEASE(nd.ni_dvp, p, p->p_ucred,
930 				    LEASE_WRITE);
931 				VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
932 				error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd);
933 			} else {
934 				VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
935 				if (nd.ni_dvp == nd.ni_vp)
936 					vrele(nd.ni_dvp);
937 				else
938 					vput(nd.ni_dvp);
939 				if (nd.ni_vp)
940 					vrele(nd.ni_vp);
941 			}
942 		}
943 	}
944 	vrele(vp);
945 	return (error);
946 }
947 
948 /*
949  * Make a symbolic link.
950  */
951 /* ARGSUSED */
952 symlink(p, uap, retval)
953 	struct proc *p;
954 	register struct symlink_args /* {
955 		syscallarg(char *) path;
956 		syscallarg(char *) link;
957 	} */ *uap;
958 	register_t *retval;
959 {
960 	struct vattr vattr;
961 	char *path;
962 	int error;
963 	struct nameidata nd;
964 
965 	MALLOC(path, char *, MAXPATHLEN, M_NAMEI, M_WAITOK);
966 	if (error = copyinstr(SCARG(uap, path), path, MAXPATHLEN, (u_int*)0))
967 		goto out;
968 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, link), p);
969 	if (error = namei(&nd))
970 		goto out;
971 	if (nd.ni_vp) {
972 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
973 		if (nd.ni_dvp == nd.ni_vp)
974 			vrele(nd.ni_dvp);
975 		else
976 			vput(nd.ni_dvp);
977 		vrele(nd.ni_vp);
978 		error = EEXIST;
979 		goto out;
980 	}
981 	VATTR_NULL(&vattr);
982 	vattr.va_mode = ACCESSPERMS &~ p->p_fd->fd_cmask;
983 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
984 	error = VOP_SYMLINK(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr, path);
985 out:
986 	FREE(path, M_NAMEI);
987 	return (error);
988 }
989 
990 /*
991  * Delete a whiteout from the filesystem.
992  */
993 /* ARGSUSED */
994 undelete(p, uap, retval)
995 	struct proc *p;
996 	register struct undelete_args /* {
997 		syscallarg(char *) path;
998 	} */ *uap;
999 	register_t *retval;
1000 {
1001 	int error;
1002 	struct nameidata nd;
1003 
1004 	NDINIT(&nd, DELETE, LOCKPARENT|DOWHITEOUT, UIO_USERSPACE,
1005 	    SCARG(uap, path), p);
1006 	error = namei(&nd);
1007 	if (error)
1008 		return (error);
1009 
1010 	if (nd.ni_vp != NULLVP || !(nd.ni_cnd.cn_flags & ISWHITEOUT)) {
1011 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1012 		if (nd.ni_dvp == nd.ni_vp)
1013 			vrele(nd.ni_dvp);
1014 		else
1015 			vput(nd.ni_dvp);
1016 		if (nd.ni_vp)
1017 			vrele(nd.ni_vp);
1018 		return (EEXIST);
1019 	}
1020 
1021 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1022 	if (error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, DELETE))
1023 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1024 	vput(nd.ni_dvp);
1025 	return (error);
1026 }
1027 
1028 /*
1029  * Delete a name from the filesystem.
1030  */
1031 /* ARGSUSED */
1032 unlink(p, uap, retval)
1033 	struct proc *p;
1034 	struct unlink_args /* {
1035 		syscallarg(char *) path;
1036 	} */ *uap;
1037 	register_t *retval;
1038 {
1039 	register struct vnode *vp;
1040 	int error;
1041 	struct nameidata nd;
1042 
1043 	NDINIT(&nd, DELETE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1044 	if (error = namei(&nd))
1045 		return (error);
1046 	vp = nd.ni_vp;
1047 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1048 	VOP_LOCK(vp);
1049 
1050 	if (vp->v_type != VDIR ||
1051 	    (error = suser(p->p_ucred, &p->p_acflag)) == 0) {
1052 		/*
1053 		 * The root of a mounted filesystem cannot be deleted.
1054 		 */
1055 		if (vp->v_flag & VROOT)
1056 			error = EBUSY;
1057 		else
1058 			(void)vnode_pager_uncache(vp);
1059 	}
1060 
1061 	if (!error) {
1062 		VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1063 		error = VOP_REMOVE(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
1064 	} else {
1065 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1066 		if (nd.ni_dvp == vp)
1067 			vrele(nd.ni_dvp);
1068 		else
1069 			vput(nd.ni_dvp);
1070 		if (vp != NULLVP)
1071 			vput(vp);
1072 	}
1073 	return (error);
1074 }
1075 
1076 /*
1077  * Reposition read/write file offset.
1078  */
1079 lseek(p, uap, retval)
1080 	struct proc *p;
1081 	register struct lseek_args /* {
1082 		syscallarg(int) fd;
1083 		syscallarg(int) pad;
1084 		syscallarg(off_t) offset;
1085 		syscallarg(int) whence;
1086 	} */ *uap;
1087 	register_t *retval;
1088 {
1089 	struct ucred *cred = p->p_ucred;
1090 	register struct filedesc *fdp = p->p_fd;
1091 	register struct file *fp;
1092 	struct vattr vattr;
1093 	int error;
1094 
1095 	if ((u_int)SCARG(uap, fd) >= fdp->fd_nfiles ||
1096 	    (fp = fdp->fd_ofiles[SCARG(uap, fd)]) == NULL)
1097 		return (EBADF);
1098 	if (fp->f_type != DTYPE_VNODE)
1099 		return (ESPIPE);
1100 	switch (SCARG(uap, whence)) {
1101 	case L_INCR:
1102 		fp->f_offset += SCARG(uap, offset);
1103 		break;
1104 	case L_XTND:
1105 		if (error =
1106 		    VOP_GETATTR((struct vnode *)fp->f_data, &vattr, cred, p))
1107 			return (error);
1108 		fp->f_offset = SCARG(uap, offset) + vattr.va_size;
1109 		break;
1110 	case L_SET:
1111 		fp->f_offset = SCARG(uap, offset);
1112 		break;
1113 	default:
1114 		return (EINVAL);
1115 	}
1116 	*(off_t *)retval = fp->f_offset;
1117 	return (0);
1118 }
1119 
1120 #if defined(COMPAT_43) || defined(COMPAT_SUNOS) || defined(COMPAT_SVR4)
1121 /*
1122  * Reposition read/write file offset.
1123  */
1124 compat_43_lseek(p, uap, retval)
1125 	struct proc *p;
1126 	register struct compat_43_lseek_args /* {
1127 		syscallarg(int) fd;
1128 		syscallarg(long) offset;
1129 		syscallarg(int) whence;
1130 	} */ *uap;
1131 	register_t *retval;
1132 {
1133 	struct lseek_args /* {
1134 		syscallarg(int) fd;
1135 		syscallarg(int) pad;
1136 		syscallarg(off_t) offset;
1137 		syscallarg(int) whence;
1138 	} */ nuap;
1139 	off_t qret;
1140 	int error;
1141 
1142 	SCARG(&nuap, fd) = SCARG(uap, fd);
1143 	SCARG(&nuap, offset) = SCARG(uap, offset);
1144 	SCARG(&nuap, whence) = SCARG(uap, whence);
1145 	error = lseek(p, &nuap, &qret);
1146 	*(long *)retval = qret;
1147 	return (error);
1148 }
1149 #endif /* COMPAT_43 || COMPAT_SUNOS */
1150 
1151 /*
1152  * Check access permissions.
1153  */
1154 access(p, uap, retval)
1155 	struct proc *p;
1156 	register struct access_args /* {
1157 		syscallarg(char *) path;
1158 		syscallarg(int) flags;
1159 	} */ *uap;
1160 	register_t *retval;
1161 {
1162 	register struct ucred *cred = p->p_ucred;
1163 	register struct vnode *vp;
1164 	int error, flags, t_gid, t_uid;
1165 	struct nameidata nd;
1166 
1167 	t_uid = cred->cr_uid;
1168 	t_gid = cred->cr_groups[0];
1169 	cred->cr_uid = p->p_cred->p_ruid;
1170 	cred->cr_groups[0] = p->p_cred->p_rgid;
1171 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1172 	    SCARG(uap, path), p);
1173 	if (error = namei(&nd))
1174 		goto out1;
1175 	vp = nd.ni_vp;
1176 
1177 	/* Flags == 0 means only check for existence. */
1178 	if (SCARG(uap, flags)) {
1179 		flags = 0;
1180 		if (SCARG(uap, flags) & R_OK)
1181 			flags |= VREAD;
1182 		if (SCARG(uap, flags) & W_OK)
1183 			flags |= VWRITE;
1184 		if (SCARG(uap, flags) & X_OK)
1185 			flags |= VEXEC;
1186 		if ((flags & VWRITE) == 0 || (error = vn_writechk(vp)) == 0)
1187 			error = VOP_ACCESS(vp, flags, cred, p);
1188 	}
1189 	vput(vp);
1190 out1:
1191 	cred->cr_uid = t_uid;
1192 	cred->cr_groups[0] = t_gid;
1193 	return (error);
1194 }
1195 
1196 #if defined(COMPAT_43) || defined(COMPAT_SUNOS) || defined(COMPAT_IBCS2)
1197 /*
1198  * Get file status; this version follows links.
1199  */
1200 /* ARGSUSED */
1201 compat_43_stat(p, uap, retval)
1202 	struct proc *p;
1203 	register struct compat_43_stat_args /* {
1204 		syscallarg(char *) path;
1205 		syscallarg(struct ostat *) ub;
1206 	} */ *uap;
1207 	register_t *retval;
1208 {
1209 	struct stat sb;
1210 	struct ostat osb;
1211 	int error;
1212 	struct nameidata nd;
1213 
1214 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1215 	    SCARG(uap, path), p);
1216 	if (error = namei(&nd))
1217 		return (error);
1218 	error = vn_stat(nd.ni_vp, &sb, p);
1219 	vput(nd.ni_vp);
1220 	if (error)
1221 		return (error);
1222 	cvtstat(&sb, &osb);
1223 	error = copyout((caddr_t)&osb, (caddr_t)SCARG(uap, ub), sizeof (osb));
1224 	return (error);
1225 }
1226 
1227 /*
1228  * Get file status; this version does not follow links.
1229  */
1230 /* ARGSUSED */
1231 compat_43_lstat(p, uap, retval)
1232 	struct proc *p;
1233 	register struct compat_43_lstat_args /* {
1234 		syscallarg(char *) path;
1235 		syscallarg(struct ostat *) ub;
1236 	} */ *uap;
1237 	register_t *retval;
1238 {
1239 	struct vnode *vp, *dvp;
1240 	struct stat sb, sb1;
1241 	struct ostat osb;
1242 	int error;
1243 	struct nameidata nd;
1244 
1245 	NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | LOCKPARENT, UIO_USERSPACE,
1246 	    SCARG(uap, path), p);
1247 	if (error = namei(&nd))
1248 		return (error);
1249 	/*
1250 	 * For symbolic links, always return the attributes of its
1251 	 * containing directory, except for mode, size, and links.
1252 	 */
1253 	vp = nd.ni_vp;
1254 	dvp = nd.ni_dvp;
1255 	if (vp->v_type != VLNK) {
1256 		if (dvp == vp)
1257 			vrele(dvp);
1258 		else
1259 			vput(dvp);
1260 		error = vn_stat(vp, &sb, p);
1261 		vput(vp);
1262 		if (error)
1263 			return (error);
1264 	} else {
1265 		error = vn_stat(dvp, &sb, p);
1266 		vput(dvp);
1267 		if (error) {
1268 			vput(vp);
1269 			return (error);
1270 		}
1271 		error = vn_stat(vp, &sb1, p);
1272 		vput(vp);
1273 		if (error)
1274 			return (error);
1275 		sb.st_mode &= ~S_IFDIR;
1276 		sb.st_mode |= S_IFLNK;
1277 		sb.st_nlink = sb1.st_nlink;
1278 		sb.st_size = sb1.st_size;
1279 		sb.st_blocks = sb1.st_blocks;
1280 	}
1281 	cvtstat(&sb, &osb);
1282 	error = copyout((caddr_t)&osb, (caddr_t)SCARG(uap, ub), sizeof (osb));
1283 	return (error);
1284 }
1285 
1286 /*
1287  * Convert from an old to a new stat structure.
1288  */
1289 cvtstat(st, ost)
1290 	struct stat *st;
1291 	struct ostat *ost;
1292 {
1293 
1294 	ost->st_dev = st->st_dev;
1295 	ost->st_ino = st->st_ino;
1296 	ost->st_mode = st->st_mode;
1297 	ost->st_nlink = st->st_nlink;
1298 	ost->st_uid = st->st_uid;
1299 	ost->st_gid = st->st_gid;
1300 	ost->st_rdev = st->st_rdev;
1301 	if (st->st_size < (quad_t)1 << 32)
1302 		ost->st_size = st->st_size;
1303 	else
1304 		ost->st_size = -2;
1305 	ost->st_atime = st->st_atime;
1306 	ost->st_mtime = st->st_mtime;
1307 	ost->st_ctime = st->st_ctime;
1308 	ost->st_blksize = st->st_blksize;
1309 	ost->st_blocks = st->st_blocks;
1310 	ost->st_flags = st->st_flags;
1311 	ost->st_gen = st->st_gen;
1312 }
1313 #endif /* COMPAT_43 || COMPAT_SUNOS || COMPAT_IBCS2 */
1314 
1315 /*
1316  * Get file status; this version follows links.
1317  */
1318 /* ARGSUSED */
1319 stat(p, uap, retval)
1320 	struct proc *p;
1321 	register struct stat_args /* {
1322 		syscallarg(char *) path;
1323 		syscallarg(struct stat *) ub;
1324 	} */ *uap;
1325 	register_t *retval;
1326 {
1327 	struct stat sb;
1328 	int error;
1329 	struct nameidata nd;
1330 
1331 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1332 	    SCARG(uap, path), p);
1333 	if (error = namei(&nd))
1334 		return (error);
1335 	error = vn_stat(nd.ni_vp, &sb, p);
1336 	vput(nd.ni_vp);
1337 	if (error)
1338 		return (error);
1339 	error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
1340 	return (error);
1341 }
1342 
1343 /*
1344  * Get file status; this version does not follow links.
1345  */
1346 /* ARGSUSED */
1347 lstat(p, uap, retval)
1348 	struct proc *p;
1349 	register struct lstat_args /* {
1350 		syscallarg(char *) path;
1351 		syscallarg(struct stat *) ub;
1352 	} */ *uap;
1353 	register_t *retval;
1354 {
1355 	int error;
1356 	struct vnode *vp, *dvp;
1357 	struct stat sb, sb1;
1358 	struct nameidata nd;
1359 
1360 	NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | LOCKPARENT, UIO_USERSPACE,
1361 	    SCARG(uap, path), p);
1362 	if (error = namei(&nd))
1363 		return (error);
1364 	/*
1365 	 * For symbolic links, always return the attributes of its
1366 	 * containing directory, except for mode, size, and links.
1367 	 */
1368 	vp = nd.ni_vp;
1369 	dvp = nd.ni_dvp;
1370 	if (vp->v_type != VLNK) {
1371 		if (dvp == vp)
1372 			vrele(dvp);
1373 		else
1374 			vput(dvp);
1375 		error = vn_stat(vp, &sb, p);
1376 		vput(vp);
1377 		if (error)
1378 			return (error);
1379 	} else {
1380 		error = vn_stat(dvp, &sb, p);
1381 		vput(dvp);
1382 		if (error) {
1383 			vput(vp);
1384 			return (error);
1385 		}
1386 		error = vn_stat(vp, &sb1, p);
1387 		vput(vp);
1388 		if (error)
1389 			return (error);
1390 		sb.st_mode &= ~S_IFDIR;
1391 		sb.st_mode |= S_IFLNK;
1392 		sb.st_nlink = sb1.st_nlink;
1393 		sb.st_size = sb1.st_size;
1394 		sb.st_blocks = sb1.st_blocks;
1395 	}
1396 	error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
1397 	return (error);
1398 }
1399 
1400 /*
1401  * Get configurable pathname variables.
1402  */
1403 /* ARGSUSED */
1404 pathconf(p, uap, retval)
1405 	struct proc *p;
1406 	register struct pathconf_args /* {
1407 		syscallarg(char *) path;
1408 		syscallarg(int) name;
1409 	} */ *uap;
1410 	register_t *retval;
1411 {
1412 	int error;
1413 	struct nameidata nd;
1414 
1415 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1416 	    SCARG(uap, path), p);
1417 	if (error = namei(&nd))
1418 		return (error);
1419 	error = VOP_PATHCONF(nd.ni_vp, SCARG(uap, name), retval);
1420 	vput(nd.ni_vp);
1421 	return (error);
1422 }
1423 
1424 /*
1425  * Return target name of a symbolic link.
1426  */
1427 /* ARGSUSED */
1428 readlink(p, uap, retval)
1429 	struct proc *p;
1430 	register struct readlink_args /* {
1431 		syscallarg(char *) path;
1432 		syscallarg(char *) buf;
1433 		syscallarg(int) count;
1434 	} */ *uap;
1435 	register_t *retval;
1436 {
1437 	register struct vnode *vp;
1438 	struct iovec aiov;
1439 	struct uio auio;
1440 	int error;
1441 	struct nameidata nd;
1442 
1443 	NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF, UIO_USERSPACE,
1444 	    SCARG(uap, path), p);
1445 	if (error = namei(&nd))
1446 		return (error);
1447 	vp = nd.ni_vp;
1448 	if (vp->v_type != VLNK)
1449 		error = EINVAL;
1450 	else {
1451 		aiov.iov_base = SCARG(uap, buf);
1452 		aiov.iov_len = SCARG(uap, count);
1453 		auio.uio_iov = &aiov;
1454 		auio.uio_iovcnt = 1;
1455 		auio.uio_offset = 0;
1456 		auio.uio_rw = UIO_READ;
1457 		auio.uio_segflg = UIO_USERSPACE;
1458 		auio.uio_procp = p;
1459 		auio.uio_resid = SCARG(uap, count);
1460 		error = VOP_READLINK(vp, &auio, p->p_ucred);
1461 	}
1462 	vput(vp);
1463 	*retval = SCARG(uap, count) - auio.uio_resid;
1464 	return (error);
1465 }
1466 
1467 /*
1468  * Change flags of a file given a path name.
1469  */
1470 /* ARGSUSED */
1471 chflags(p, uap, retval)
1472 	struct proc *p;
1473 	register struct chflags_args /* {
1474 		syscallarg(char *) path;
1475 		syscallarg(int) flags;
1476 	} */ *uap;
1477 	register_t *retval;
1478 {
1479 	register struct vnode *vp;
1480 	struct vattr vattr;
1481 	int error;
1482 	struct nameidata nd;
1483 
1484 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1485 	if (error = namei(&nd))
1486 		return (error);
1487 	vp = nd.ni_vp;
1488 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1489 	VOP_LOCK(vp);
1490 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1491 		error = EROFS;
1492 	else {
1493 		VATTR_NULL(&vattr);
1494 		vattr.va_flags = SCARG(uap, flags);
1495 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1496 	}
1497 	vput(vp);
1498 	return (error);
1499 }
1500 
1501 /*
1502  * Change flags of a file given a file descriptor.
1503  */
1504 /* ARGSUSED */
1505 fchflags(p, uap, retval)
1506 	struct proc *p;
1507 	register struct fchflags_args /* {
1508 		syscallarg(int) fd;
1509 		syscallarg(int) flags;
1510 	} */ *uap;
1511 	register_t *retval;
1512 {
1513 	struct vattr vattr;
1514 	struct vnode *vp;
1515 	struct file *fp;
1516 	int error;
1517 
1518 	if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
1519 		return (error);
1520 	vp = (struct vnode *)fp->f_data;
1521 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1522 	VOP_LOCK(vp);
1523 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1524 		error = EROFS;
1525 	else {
1526 		VATTR_NULL(&vattr);
1527 		vattr.va_flags = SCARG(uap, flags);
1528 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1529 	}
1530 	VOP_UNLOCK(vp);
1531 	return (error);
1532 }
1533 
1534 /*
1535  * Change mode of a file given path name.
1536  */
1537 /* ARGSUSED */
1538 chmod(p, uap, retval)
1539 	struct proc *p;
1540 	register struct chmod_args /* {
1541 		syscallarg(char *) path;
1542 		syscallarg(int) mode;
1543 	} */ *uap;
1544 	register_t *retval;
1545 {
1546 	register struct vnode *vp;
1547 	struct vattr vattr;
1548 	int error;
1549 	struct nameidata nd;
1550 
1551 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1552 	if (error = namei(&nd))
1553 		return (error);
1554 	vp = nd.ni_vp;
1555 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1556 	VOP_LOCK(vp);
1557 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1558 		error = EROFS;
1559 	else {
1560 		VATTR_NULL(&vattr);
1561 		vattr.va_mode = SCARG(uap, mode) & ALLPERMS;
1562 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1563 	}
1564 	vput(vp);
1565 	return (error);
1566 }
1567 
1568 /*
1569  * Change mode of a file given a file descriptor.
1570  */
1571 /* ARGSUSED */
1572 fchmod(p, uap, retval)
1573 	struct proc *p;
1574 	register struct fchmod_args /* {
1575 		syscallarg(int) fd;
1576 		syscallarg(int) mode;
1577 	} */ *uap;
1578 	register_t *retval;
1579 {
1580 	struct vattr vattr;
1581 	struct vnode *vp;
1582 	struct file *fp;
1583 	int error;
1584 
1585 	if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
1586 		return (error);
1587 	vp = (struct vnode *)fp->f_data;
1588 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1589 	VOP_LOCK(vp);
1590 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1591 		error = EROFS;
1592 	else {
1593 		VATTR_NULL(&vattr);
1594 		vattr.va_mode = SCARG(uap, mode) & ALLPERMS;
1595 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1596 	}
1597 	VOP_UNLOCK(vp);
1598 	return (error);
1599 }
1600 
1601 /*
1602  * Set ownership given a path name.
1603  */
1604 /* ARGSUSED */
1605 chown(p, uap, retval)
1606 	struct proc *p;
1607 	register struct chown_args /* {
1608 		syscallarg(char *) path;
1609 		syscallarg(int) uid;
1610 		syscallarg(int) gid;
1611 	} */ *uap;
1612 	register_t *retval;
1613 {
1614 	register struct vnode *vp;
1615 	struct vattr vattr;
1616 	int error;
1617 	struct nameidata nd;
1618 
1619 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1620 	if (error = namei(&nd))
1621 		return (error);
1622 	vp = nd.ni_vp;
1623 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1624 	VOP_LOCK(vp);
1625 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1626 		error = EROFS;
1627 	else {
1628 		VATTR_NULL(&vattr);
1629 		vattr.va_uid = SCARG(uap, uid);
1630 		vattr.va_gid = SCARG(uap, gid);
1631 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1632 	}
1633 	vput(vp);
1634 	return (error);
1635 }
1636 
1637 /*
1638  * Set ownership given a file descriptor.
1639  */
1640 /* ARGSUSED */
1641 fchown(p, uap, retval)
1642 	struct proc *p;
1643 	register struct fchown_args /* {
1644 		syscallarg(int) fd;
1645 		syscallarg(int) uid;
1646 		syscallarg(int) gid;
1647 	} */ *uap;
1648 	register_t *retval;
1649 {
1650 	struct vattr vattr;
1651 	struct vnode *vp;
1652 	struct file *fp;
1653 	int error;
1654 
1655 	if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
1656 		return (error);
1657 	vp = (struct vnode *)fp->f_data;
1658 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1659 	VOP_LOCK(vp);
1660 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1661 		error = EROFS;
1662 	else {
1663 		VATTR_NULL(&vattr);
1664 		vattr.va_uid = SCARG(uap, uid);
1665 		vattr.va_gid = SCARG(uap, gid);
1666 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1667 	}
1668 	VOP_UNLOCK(vp);
1669 	return (error);
1670 }
1671 
1672 /*
1673  * Set the access and modification times of a file.
1674  */
1675 /* ARGSUSED */
1676 utimes(p, uap, retval)
1677 	struct proc *p;
1678 	register struct utimes_args /* {
1679 		syscallarg(char *) path;
1680 		syscallarg(struct timeval *) tptr;
1681 	} */ *uap;
1682 	register_t *retval;
1683 {
1684 	register struct vnode *vp;
1685 	struct timeval tv[2];
1686 	struct vattr vattr;
1687 	int error;
1688 	struct nameidata nd;
1689 
1690 	VATTR_NULL(&vattr);
1691 	if (SCARG(uap, tptr) == NULL) {
1692 		microtime(&tv[0]);
1693 		tv[1] = tv[0];
1694 		vattr.va_vaflags |= VA_UTIMES_NULL;
1695 	} else if (error = copyin((caddr_t)SCARG(uap, tptr), (caddr_t)tv,
1696 	    sizeof (tv)))
1697   		return (error);
1698 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1699 	if (error = namei(&nd))
1700 		return (error);
1701 	vp = nd.ni_vp;
1702 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1703 	VOP_LOCK(vp);
1704 	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1705 		error = EROFS;
1706 	else {
1707 		vattr.va_atime.ts_sec = tv[0].tv_sec;
1708 		vattr.va_atime.ts_nsec = tv[0].tv_usec * 1000;
1709 		vattr.va_mtime.ts_sec = tv[1].tv_sec;
1710 		vattr.va_mtime.ts_nsec = tv[1].tv_usec * 1000;
1711 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1712 	}
1713 	vput(vp);
1714 	return (error);
1715 }
1716 
1717 /*
1718  * Truncate a file given its path name.
1719  */
1720 /* ARGSUSED */
1721 truncate(p, uap, retval)
1722 	struct proc *p;
1723 	register struct truncate_args /* {
1724 		syscallarg(char *) path;
1725 		syscallarg(int) pad;
1726 		syscallarg(off_t) length;
1727 	} */ *uap;
1728 	register_t *retval;
1729 {
1730 	register struct vnode *vp;
1731 	struct vattr vattr;
1732 	int error;
1733 	struct nameidata nd;
1734 
1735 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1736 	if (error = namei(&nd))
1737 		return (error);
1738 	vp = nd.ni_vp;
1739 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1740 	VOP_LOCK(vp);
1741 	if (vp->v_type == VDIR)
1742 		error = EISDIR;
1743 	else if ((error = vn_writechk(vp)) == 0 &&
1744 	    (error = VOP_ACCESS(vp, VWRITE, p->p_ucred, p)) == 0) {
1745 		VATTR_NULL(&vattr);
1746 		vattr.va_size = SCARG(uap, length);
1747 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1748 	}
1749 	vput(vp);
1750 	return (error);
1751 }
1752 
1753 /*
1754  * Truncate a file given a file descriptor.
1755  */
1756 /* ARGSUSED */
1757 ftruncate(p, uap, retval)
1758 	struct proc *p;
1759 	register struct ftruncate_args /* {
1760 		syscallarg(int) fd;
1761 		syscallarg(int) pad;
1762 		syscallarg(off_t) length;
1763 	} */ *uap;
1764 	register_t *retval;
1765 {
1766 	struct vattr vattr;
1767 	struct vnode *vp;
1768 	struct file *fp;
1769 	int error;
1770 
1771 	if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
1772 		return (error);
1773 	if ((fp->f_flag & FWRITE) == 0)
1774 		return (EINVAL);
1775 	vp = (struct vnode *)fp->f_data;
1776 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1777 	VOP_LOCK(vp);
1778 	if (vp->v_type == VDIR)
1779 		error = EISDIR;
1780 	else if ((error = vn_writechk(vp)) == 0) {
1781 		VATTR_NULL(&vattr);
1782 		vattr.va_size = SCARG(uap, length);
1783 		error = VOP_SETATTR(vp, &vattr, fp->f_cred, p);
1784 	}
1785 	VOP_UNLOCK(vp);
1786 	return (error);
1787 }
1788 
1789 #if defined(COMPAT_43) || defined(COMPAT_SUNOS)
1790 /*
1791  * Truncate a file given its path name.
1792  */
1793 /* ARGSUSED */
1794 compat_43_truncate(p, uap, retval)
1795 	struct proc *p;
1796 	register struct compat_43_truncate_args /* {
1797 		syscallarg(char *) path;
1798 		syscallarg(long) length;
1799 	} */ *uap;
1800 	register_t *retval;
1801 {
1802 	struct truncate_args /* {
1803 		syscallarg(char *) path;
1804 		syscallarg(int) pad;
1805 		syscallarg(off_t) length;
1806 	} */ nuap;
1807 
1808 	SCARG(&nuap, path) = SCARG(uap, path);
1809 	SCARG(&nuap, length) = SCARG(uap, length);
1810 	return (truncate(p, &nuap, retval));
1811 }
1812 
1813 /*
1814  * Truncate a file given a file descriptor.
1815  */
1816 /* ARGSUSED */
1817 compat_43_ftruncate(p, uap, retval)
1818 	struct proc *p;
1819 	register struct compat_43_ftruncate_args /* {
1820 		syscallarg(int) fd;
1821 		syscallarg(long) length;
1822 	} */ *uap;
1823 	register_t *retval;
1824 {
1825 	struct ftruncate_args /* {
1826 		syscallarg(int) fd;
1827 		syscallarg(int) pad;
1828 		syscallarg(off_t) length;
1829 	} */ nuap;
1830 
1831 	SCARG(&nuap, fd) = SCARG(uap, fd);
1832 	SCARG(&nuap, length) = SCARG(uap, length);
1833 	return (ftruncate(p, &nuap, retval));
1834 }
1835 #endif /* COMPAT_43 || COMPAT_SUNOS */
1836 
1837 /*
1838  * Sync an open file.
1839  */
1840 /* ARGSUSED */
1841 fsync(p, uap, retval)
1842 	struct proc *p;
1843 	struct fsync_args /* {
1844 		syscallarg(int) fd;
1845 	} */ *uap;
1846 	register_t *retval;
1847 {
1848 	register struct vnode *vp;
1849 	struct file *fp;
1850 	int error;
1851 
1852 	if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
1853 		return (error);
1854 	vp = (struct vnode *)fp->f_data;
1855 	VOP_LOCK(vp);
1856 	error = VOP_FSYNC(vp, fp->f_cred, MNT_WAIT, p);
1857 	VOP_UNLOCK(vp);
1858 	return (error);
1859 }
1860 
1861 /*
1862  * Rename files.  Source and destination must either both be directories,
1863  * or both not be directories.  If target is a directory, it must be empty.
1864  */
1865 /* ARGSUSED */
1866 rename(p, uap, retval)
1867 	struct proc *p;
1868 	register struct rename_args /* {
1869 		syscallarg(char *) from;
1870 		syscallarg(char *) to;
1871 	} */ *uap;
1872 	register_t *retval;
1873 {
1874 	register struct vnode *tvp, *fvp, *tdvp;
1875 	struct nameidata fromnd, tond;
1876 	int error;
1877 
1878 	NDINIT(&fromnd, DELETE, WANTPARENT | SAVESTART, UIO_USERSPACE,
1879 	    SCARG(uap, from), p);
1880 	if (error = namei(&fromnd))
1881 		return (error);
1882 	fvp = fromnd.ni_vp;
1883 	NDINIT(&tond, RENAME, LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART,
1884 	    UIO_USERSPACE, SCARG(uap, to), p);
1885 	if (error = namei(&tond)) {
1886 		VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
1887 		vrele(fromnd.ni_dvp);
1888 		vrele(fvp);
1889 		goto out1;
1890 	}
1891 	tdvp = tond.ni_dvp;
1892 	tvp = tond.ni_vp;
1893 	if (tvp != NULL) {
1894 		if (fvp->v_type == VDIR && tvp->v_type != VDIR) {
1895 			error = ENOTDIR;
1896 			goto out;
1897 		} else if (fvp->v_type != VDIR && tvp->v_type == VDIR) {
1898 			error = EISDIR;
1899 			goto out;
1900 		}
1901 	}
1902 	if (fvp == tdvp)
1903 		error = EINVAL;
1904 	/*
1905 	 * If source is the same as the destination (that is the
1906 	 * same inode number with the same name in the same directory),
1907 	 * then there is nothing to do.
1908 	 */
1909 	if (fvp == tvp && fromnd.ni_dvp == tdvp &&
1910 	    fromnd.ni_cnd.cn_namelen == tond.ni_cnd.cn_namelen &&
1911 	    !bcmp(fromnd.ni_cnd.cn_nameptr, tond.ni_cnd.cn_nameptr,
1912 	      fromnd.ni_cnd.cn_namelen))
1913 		error = -1;
1914 out:
1915 	if (!error) {
1916 		VOP_LEASE(tdvp, p, p->p_ucred, LEASE_WRITE);
1917 		if (fromnd.ni_dvp != tdvp)
1918 			VOP_LEASE(fromnd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1919 		if (tvp)
1920 			VOP_LEASE(tvp, p, p->p_ucred, LEASE_WRITE);
1921 		error = VOP_RENAME(fromnd.ni_dvp, fromnd.ni_vp, &fromnd.ni_cnd,
1922 				   tond.ni_dvp, tond.ni_vp, &tond.ni_cnd);
1923 	} else {
1924 		VOP_ABORTOP(tond.ni_dvp, &tond.ni_cnd);
1925 		if (tdvp == tvp)
1926 			vrele(tdvp);
1927 		else
1928 			vput(tdvp);
1929 		if (tvp)
1930 			vput(tvp);
1931 		VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
1932 		vrele(fromnd.ni_dvp);
1933 		vrele(fvp);
1934 	}
1935 	vrele(tond.ni_startdir);
1936 	FREE(tond.ni_cnd.cn_pnbuf, M_NAMEI);
1937 out1:
1938 	if (fromnd.ni_startdir)
1939 		vrele(fromnd.ni_startdir);
1940 	FREE(fromnd.ni_cnd.cn_pnbuf, M_NAMEI);
1941 	if (error == -1)
1942 		return (0);
1943 	return (error);
1944 }
1945 
1946 /*
1947  * Make a directory file.
1948  */
1949 /* ARGSUSED */
1950 mkdir(p, uap, retval)
1951 	struct proc *p;
1952 	register struct mkdir_args /* {
1953 		syscallarg(char *) path;
1954 		syscallarg(int) mode;
1955 	} */ *uap;
1956 	register_t *retval;
1957 {
1958 	register struct vnode *vp;
1959 	struct vattr vattr;
1960 	int error;
1961 	struct nameidata nd;
1962 
1963 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1964 	if (error = namei(&nd))
1965 		return (error);
1966 	vp = nd.ni_vp;
1967 	if (vp != NULL) {
1968 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1969 		if (nd.ni_dvp == vp)
1970 			vrele(nd.ni_dvp);
1971 		else
1972 			vput(nd.ni_dvp);
1973 		vrele(vp);
1974 		return (EEXIST);
1975 	}
1976 	VATTR_NULL(&vattr);
1977 	vattr.va_type = VDIR;
1978 	vattr.va_mode = (SCARG(uap, mode) & ACCESSPERMS) &~ p->p_fd->fd_cmask;
1979 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1980 	error = VOP_MKDIR(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
1981 	if (!error)
1982 		vput(nd.ni_vp);
1983 	return (error);
1984 }
1985 
1986 /*
1987  * Remove a directory file.
1988  */
1989 /* ARGSUSED */
1990 rmdir(p, uap, retval)
1991 	struct proc *p;
1992 	struct rmdir_args /* {
1993 		syscallarg(char *) path;
1994 	} */ *uap;
1995 	register_t *retval;
1996 {
1997 	register struct vnode *vp;
1998 	int error;
1999 	struct nameidata nd;
2000 
2001 	NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
2002 	    SCARG(uap, path), p);
2003 	if (error = namei(&nd))
2004 		return (error);
2005 	vp = nd.ni_vp;
2006 	if (vp->v_type != VDIR) {
2007 		error = ENOTDIR;
2008 		goto out;
2009 	}
2010 	/*
2011 	 * No rmdir "." please.
2012 	 */
2013 	if (nd.ni_dvp == vp) {
2014 		error = EINVAL;
2015 		goto out;
2016 	}
2017 	/*
2018 	 * The root of a mounted filesystem cannot be deleted.
2019 	 */
2020 	if (vp->v_flag & VROOT)
2021 		error = EBUSY;
2022 out:
2023 	if (!error) {
2024 		VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
2025 		VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2026 		error = VOP_RMDIR(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
2027 	} else {
2028 		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
2029 		if (nd.ni_dvp == vp)
2030 			vrele(nd.ni_dvp);
2031 		else
2032 			vput(nd.ni_dvp);
2033 		vput(vp);
2034 	}
2035 	return (error);
2036 }
2037 
2038 #if defined(COMPAT_43) || defined(COMPAT_HPUX)
2039 /*
2040  * Read a block of directory entries in a file system independent format.
2041  */
2042 compat_43_getdirentries(p, uap, retval)
2043 	struct proc *p;
2044 	register struct compat_43_getdirentries_args /* {
2045 		syscallarg(int) fd;
2046 		syscallarg(char *) buf;
2047 		syscallarg(u_int) count;
2048 		syscallarg(long *) basep;
2049 	} */ *uap;
2050 	register_t *retval;
2051 {
2052 	register struct vnode *vp;
2053 	struct file *fp;
2054 	struct uio auio, kuio;
2055 	struct iovec aiov, kiov;
2056 	struct dirent *dp, *edp;
2057 	caddr_t dirbuf;
2058 	int error, eofflag, readcnt;
2059 	long loff;
2060 
2061 	if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
2062 		return (error);
2063 	if ((fp->f_flag & FREAD) == 0)
2064 		return (EBADF);
2065 	vp = (struct vnode *)fp->f_data;
2066 unionread:
2067 	if (vp->v_type != VDIR)
2068 		return (EINVAL);
2069 	aiov.iov_base = SCARG(uap, buf);
2070 	aiov.iov_len = SCARG(uap, count);
2071 	auio.uio_iov = &aiov;
2072 	auio.uio_iovcnt = 1;
2073 	auio.uio_rw = UIO_READ;
2074 	auio.uio_segflg = UIO_USERSPACE;
2075 	auio.uio_procp = p;
2076 	auio.uio_resid = SCARG(uap, count);
2077 	VOP_LOCK(vp);
2078 	loff = auio.uio_offset = fp->f_offset;
2079 #	if (BYTE_ORDER != LITTLE_ENDIAN)
2080 		if (vp->v_mount->mnt_maxsymlinklen <= 0) {
2081 			error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag,
2082 			    (u_long *)0, 0);
2083 			fp->f_offset = auio.uio_offset;
2084 		} else
2085 #	endif
2086 	{
2087 		kuio = auio;
2088 		kuio.uio_iov = &kiov;
2089 		kuio.uio_segflg = UIO_SYSSPACE;
2090 		kiov.iov_len = SCARG(uap, count);
2091 		MALLOC(dirbuf, caddr_t, SCARG(uap, count), M_TEMP, M_WAITOK);
2092 		kiov.iov_base = dirbuf;
2093 		error = VOP_READDIR(vp, &kuio, fp->f_cred, &eofflag,
2094 			    (u_long *)0, 0);
2095 		fp->f_offset = kuio.uio_offset;
2096 		if (error == 0) {
2097 			readcnt = SCARG(uap, count) - kuio.uio_resid;
2098 			edp = (struct dirent *)&dirbuf[readcnt];
2099 			for (dp = (struct dirent *)dirbuf; dp < edp; ) {
2100 #				if (BYTE_ORDER == LITTLE_ENDIAN)
2101 					/*
2102 					 * The expected low byte of
2103 					 * dp->d_namlen is our dp->d_type.
2104 					 * The high MBZ byte of dp->d_namlen
2105 					 * is our dp->d_namlen.
2106 					 */
2107 					dp->d_type = dp->d_namlen;
2108 					dp->d_namlen = 0;
2109 #				else
2110 					/*
2111 					 * The dp->d_type is the high byte
2112 					 * of the expected dp->d_namlen,
2113 					 * so must be zero'ed.
2114 					 */
2115 					dp->d_type = 0;
2116 #				endif
2117 				if (dp->d_reclen > 0) {
2118 					dp = (struct dirent *)
2119 					    ((char *)dp + dp->d_reclen);
2120 				} else {
2121 					error = EIO;
2122 					break;
2123 				}
2124 			}
2125 			if (dp >= edp)
2126 				error = uiomove(dirbuf, readcnt, &auio);
2127 		}
2128 		FREE(dirbuf, M_TEMP);
2129 	}
2130 	VOP_UNLOCK(vp);
2131 	if (error)
2132 		return (error);
2133 
2134 #ifdef UNION
2135 {
2136 	extern int (**union_vnodeop_p)();
2137 	extern struct vnode *union_dircache __P((struct vnode *));
2138 
2139 	if ((SCARG(uap, count) == auio.uio_resid) &&
2140 	    (vp->v_op == union_vnodeop_p)) {
2141 		struct vnode *lvp;
2142 
2143 		lvp = union_dircache(vp);
2144 		if (lvp != NULLVP) {
2145 			struct vattr va;
2146 
2147 			/*
2148 			 * If the directory is opaque,
2149 			 * then don't show lower entries
2150 			 */
2151 			error = VOP_GETATTR(vp, &va, fp->f_cred, p);
2152 			if (va.va_flags & OPAQUE) {
2153 				vput(lvp);
2154 				lvp = NULL;
2155 			}
2156 		}
2157 
2158 		if (lvp != NULLVP) {
2159 			error = VOP_OPEN(lvp, FREAD, fp->f_cred, p);
2160 			VOP_UNLOCK(lvp);
2161 
2162 			if (error) {
2163 				vrele(lvp);
2164 				return (error);
2165 			}
2166 			fp->f_data = (caddr_t) lvp;
2167 			fp->f_offset = 0;
2168 			error = vn_close(vp, FREAD, fp->f_cred, p);
2169 			if (error)
2170 				return (error);
2171 			vp = lvp;
2172 			goto unionread;
2173 		}
2174 	}
2175 }
2176 #endif /* UNION */
2177 
2178 	if ((SCARG(uap, count) == auio.uio_resid) &&
2179 	    (vp->v_flag & VROOT) &&
2180 	    (vp->v_mount->mnt_flag & MNT_UNION)) {
2181 		struct vnode *tvp = vp;
2182 		vp = vp->v_mount->mnt_vnodecovered;
2183 		VREF(vp);
2184 		fp->f_data = (caddr_t) vp;
2185 		fp->f_offset = 0;
2186 		vrele(tvp);
2187 		goto unionread;
2188 	}
2189 	error = copyout((caddr_t)&loff, (caddr_t)SCARG(uap, basep),
2190 	    sizeof(long));
2191 	*retval = SCARG(uap, count) - auio.uio_resid;
2192 	return (error);
2193 }
2194 #endif /* COMPAT_43 */
2195 
2196 /*
2197  * Read a block of directory entries in a file system independent format.
2198  */
2199 getdirentries(p, uap, retval)
2200 	struct proc *p;
2201 	register struct getdirentries_args /* {
2202 		syscallarg(int) fd;
2203 		syscallarg(char *) buf;
2204 		syscallarg(u_int) count;
2205 		syscallarg(long *) basep;
2206 	} */ *uap;
2207 	register_t *retval;
2208 {
2209 	register struct vnode *vp;
2210 	struct file *fp;
2211 	struct uio auio;
2212 	struct iovec aiov;
2213 	long loff;
2214 	int error, eofflag;
2215 
2216 	if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
2217 		return (error);
2218 	if ((fp->f_flag & FREAD) == 0)
2219 		return (EBADF);
2220 	vp = (struct vnode *)fp->f_data;
2221 unionread:
2222 	if (vp->v_type != VDIR)
2223 		return (EINVAL);
2224 	aiov.iov_base = SCARG(uap, buf);
2225 	aiov.iov_len = SCARG(uap, count);
2226 	auio.uio_iov = &aiov;
2227 	auio.uio_iovcnt = 1;
2228 	auio.uio_rw = UIO_READ;
2229 	auio.uio_segflg = UIO_USERSPACE;
2230 	auio.uio_procp = p;
2231 	auio.uio_resid = SCARG(uap, count);
2232 	VOP_LOCK(vp);
2233 	loff = auio.uio_offset = fp->f_offset;
2234 	error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, (u_long *)0, 0);
2235 	fp->f_offset = auio.uio_offset;
2236 	VOP_UNLOCK(vp);
2237 	if (error)
2238 		return (error);
2239 
2240 #ifdef UNION
2241 {
2242 	extern int (**union_vnodeop_p)();
2243 	extern struct vnode *union_dircache __P((struct vnode *));
2244 
2245 	if ((SCARG(uap, count) == auio.uio_resid) &&
2246 	    (vp->v_op == union_vnodeop_p)) {
2247 		struct vnode *lvp;
2248 
2249 		lvp = union_dircache(vp);
2250 		if (lvp != NULLVP) {
2251 			struct vattr va;
2252 
2253 			/*
2254 			 * If the directory is opaque,
2255 			 * then don't show lower entries
2256 			 */
2257 			error = VOP_GETATTR(vp, &va, fp->f_cred, p);
2258 			if (va.va_flags & OPAQUE) {
2259 				vput(lvp);
2260 				lvp = NULL;
2261 			}
2262 		}
2263 
2264 		if (lvp != NULLVP) {
2265 			error = VOP_OPEN(lvp, FREAD, fp->f_cred, p);
2266 			VOP_UNLOCK(lvp);
2267 
2268 			if (error) {
2269 				vrele(lvp);
2270 				return (error);
2271 			}
2272 			fp->f_data = (caddr_t) lvp;
2273 			fp->f_offset = 0;
2274 			error = vn_close(vp, FREAD, fp->f_cred, p);
2275 			if (error)
2276 				return (error);
2277 			vp = lvp;
2278 			goto unionread;
2279 		}
2280 	}
2281 }
2282 #endif /* UNION */
2283 
2284 	if ((SCARG(uap, count) == auio.uio_resid) &&
2285 	    (vp->v_flag & VROOT) &&
2286 	    (vp->v_mount->mnt_flag & MNT_UNION)) {
2287 		struct vnode *tvp = vp;
2288 		vp = vp->v_mount->mnt_vnodecovered;
2289 		VREF(vp);
2290 		fp->f_data = (caddr_t) vp;
2291 		fp->f_offset = 0;
2292 		vrele(tvp);
2293 		goto unionread;
2294 	}
2295 	error = copyout((caddr_t)&loff, (caddr_t)SCARG(uap, basep),
2296 	    sizeof(long));
2297 	*retval = SCARG(uap, count) - auio.uio_resid;
2298 	return (error);
2299 }
2300 
2301 /*
2302  * Set the mode mask for creation of filesystem nodes.
2303  */
2304 mode_t				/* XXX */
2305 umask(p, uap, retval)
2306 	struct proc *p;
2307 	struct umask_args /* {
2308 		syscallarg(int) newmask;
2309 	} */ *uap;
2310 	register_t *retval;
2311 {
2312 	register struct filedesc *fdp;
2313 
2314 	fdp = p->p_fd;
2315 	*retval = fdp->fd_cmask;
2316 	fdp->fd_cmask = SCARG(uap, newmask) & ALLPERMS;
2317 	return (0);
2318 }
2319 
2320 /*
2321  * Void all references to file by ripping underlying filesystem
2322  * away from vnode.
2323  */
2324 /* ARGSUSED */
2325 revoke(p, uap, retval)
2326 	struct proc *p;
2327 	register struct revoke_args /* {
2328 		syscallarg(char *) path;
2329 	} */ *uap;
2330 	register_t *retval;
2331 {
2332 	register struct vnode *vp;
2333 	struct vattr vattr;
2334 	int error;
2335 	struct nameidata nd;
2336 
2337 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2338 	if (error = namei(&nd))
2339 		return (error);
2340 	vp = nd.ni_vp;
2341 	if (vp->v_type != VCHR && vp->v_type != VBLK) {
2342 		error = EINVAL;
2343 		goto out;
2344 	}
2345 	if (error = VOP_GETATTR(vp, &vattr, p->p_ucred, p))
2346 		goto out;
2347 	if (p->p_ucred->cr_uid != vattr.va_uid &&
2348 	    (error = suser(p->p_ucred, &p->p_acflag)))
2349 		goto out;
2350 	if (vp->v_usecount > 1 || (vp->v_flag & VALIASED))
2351 		vgoneall(vp);
2352 out:
2353 	vrele(vp);
2354 	return (error);
2355 }
2356 
2357 /*
2358  * Convert a user file descriptor to a kernel file entry.
2359  */
2360 getvnode(fdp, fd, fpp)
2361 	struct filedesc *fdp;
2362 	struct file **fpp;
2363 	int fd;
2364 {
2365 	struct file *fp;
2366 
2367 	if ((u_int)fd >= fdp->fd_nfiles ||
2368 	    (fp = fdp->fd_ofiles[fd]) == NULL)
2369 		return (EBADF);
2370 	if (fp->f_type != DTYPE_VNODE)
2371 		return (EINVAL);
2372 	*fpp = fp;
2373 	return (0);
2374 }
2375