xref: /freebsd/sys/kern/vfs_syscalls.c (revision 23f282aa31e9b6fceacd449020e936e98d6f2298)
1 /*
2  * Copyright (c) 1989, 1993
3  *	The Regents of the University of California.  All rights reserved.
4  * (c) UNIX System Laboratories, Inc.
5  * All or some portions of this file are derived from material licensed
6  * to the University of California by American Telephone and Telegraph
7  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
8  * the permission of UNIX System Laboratories, Inc.
9  *
10  * Redistribution and use in source and binary forms, with or without
11  * modification, are permitted provided that the following conditions
12  * are met:
13  * 1. Redistributions of source code must retain the above copyright
14  *    notice, this list of conditions and the following disclaimer.
15  * 2. Redistributions in binary form must reproduce the above copyright
16  *    notice, this list of conditions and the following disclaimer in the
17  *    documentation and/or other materials provided with the distribution.
18  * 3. All advertising materials mentioning features or use of this software
19  *    must display the following acknowledgement:
20  *	This product includes software developed by the University of
21  *	California, Berkeley and its contributors.
22  * 4. Neither the name of the University nor the names of its contributors
23  *    may be used to endorse or promote products derived from this software
24  *    without specific prior written permission.
25  *
26  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
27  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
28  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
29  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
30  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
31  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
32  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
33  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
34  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
35  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36  * SUCH DAMAGE.
37  *
38  *	@(#)vfs_syscalls.c	8.13 (Berkeley) 4/15/94
39  * $FreeBSD$
40  */
41 
42 /* For 4.3 integer FS ID compatibility */
43 #include "opt_compat.h"
44 
45 #include <sys/param.h>
46 #include <sys/systm.h>
47 #include <sys/buf.h>
48 #include <sys/sysent.h>
49 #include <sys/malloc.h>
50 #include <sys/mount.h>
51 #include <sys/sysproto.h>
52 #include <sys/namei.h>
53 #include <sys/filedesc.h>
54 #include <sys/kernel.h>
55 #include <sys/fcntl.h>
56 #include <sys/file.h>
57 #include <sys/linker.h>
58 #include <sys/stat.h>
59 #include <sys/unistd.h>
60 #include <sys/vnode.h>
61 #include <sys/proc.h>
62 #include <sys/dirent.h>
63 #include <sys/extattr.h>
64 
65 #include <machine/limits.h>
66 #include <miscfs/union/union.h>
67 #include <sys/sysctl.h>
68 #include <vm/vm.h>
69 #include <vm/vm_object.h>
70 #include <vm/vm_zone.h>
71 
72 static int change_dir __P((struct nameidata *ndp, struct proc *p));
73 static void checkdirs __P((struct vnode *olddp));
74 static int chroot_refuse_vdir_fds __P((struct filedesc *fdp));
75 static int getutimes __P((const struct timeval *, struct timespec *));
76 static int setfown __P((struct proc *, struct vnode *, uid_t, gid_t));
77 static int setfmode __P((struct proc *, struct vnode *, int));
78 static int setfflags __P((struct proc *, struct vnode *, int));
79 static int setutimes __P((struct proc *, struct vnode *,
80     const struct timespec *, int));
81 static int	usermount = 0;	/* if 1, non-root can mount fs. */
82 
83 int (*union_dircheckp) __P((struct proc *, struct vnode **, struct file *));
84 
85 SYSCTL_INT(_vfs, OID_AUTO, usermount, CTLFLAG_RW, &usermount, 0, "");
86 
87 /*
88  * Virtual File System System Calls
89  */
90 
91 /*
92  * Mount a file system.
93  */
94 #ifndef _SYS_SYSPROTO_H_
95 struct mount_args {
96 	char	*type;
97 	char	*path;
98 	int	flags;
99 	caddr_t	data;
100 };
101 #endif
102 /* ARGSUSED */
103 int
104 mount(p, uap)
105 	struct proc *p;
106 	register struct mount_args /* {
107 		syscallarg(char *) type;
108 		syscallarg(char *) path;
109 		syscallarg(int) flags;
110 		syscallarg(caddr_t) data;
111 	} */ *uap;
112 {
113 	struct vnode *vp;
114 	struct mount *mp;
115 	struct vfsconf *vfsp;
116 	int error, flag = 0, flag2 = 0;
117 	struct vattr va;
118 #ifdef COMPAT_43
119 	u_long fstypenum;
120 #endif
121 	struct nameidata nd;
122 	char fstypename[MFSNAMELEN];
123 
124 	if (usermount == 0 && (error = suser(p)))
125 		return (error);
126 	/*
127 	 * Do not allow NFS export by non-root users.
128 	 */
129 	if (SCARG(uap, flags) & MNT_EXPORTED) {
130 		error = suser(p);
131 		if (error)
132 			return (error);
133 	}
134 	/*
135 	 * Silently enforce MNT_NOSUID and MNT_NODEV for non-root users
136 	 */
137 	if (suser_xxx(p->p_ucred, 0, 0))
138 		SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
139 	/*
140 	 * Get vnode to be covered
141 	 */
142 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
143 	    SCARG(uap, path), p);
144 	if ((error = namei(&nd)) != 0)
145 		return (error);
146 	NDFREE(&nd, NDF_ONLY_PNBUF);
147 	vp = nd.ni_vp;
148 	if (SCARG(uap, flags) & MNT_UPDATE) {
149 		if ((vp->v_flag & VROOT) == 0) {
150 			vput(vp);
151 			return (EINVAL);
152 		}
153 		mp = vp->v_mount;
154 		flag = mp->mnt_flag;
155 		flag2 = mp->mnt_kern_flag;
156 		/*
157 		 * We only allow the filesystem to be reloaded if it
158 		 * is currently mounted read-only.
159 		 */
160 		if ((SCARG(uap, flags) & MNT_RELOAD) &&
161 		    ((mp->mnt_flag & MNT_RDONLY) == 0)) {
162 			vput(vp);
163 			return (EOPNOTSUPP);	/* Needs translation */
164 		}
165 		mp->mnt_flag |=
166 		    SCARG(uap, flags) & (MNT_RELOAD | MNT_FORCE | MNT_UPDATE);
167 		/*
168 		 * Only root, or the user that did the original mount is
169 		 * permitted to update it.
170 		 */
171 		if (mp->mnt_stat.f_owner != p->p_ucred->cr_uid &&
172 		    (error = suser(p))) {
173 			vput(vp);
174 			return (error);
175 		}
176 		if (vfs_busy(mp, LK_NOWAIT, 0, p)) {
177 			vput(vp);
178 			return (EBUSY);
179 		}
180 		VOP_UNLOCK(vp, 0, p);
181 		goto update;
182 	}
183 	/*
184 	 * If the user is not root, ensure that they own the directory
185 	 * onto which we are attempting to mount.
186 	 */
187 	if ((error = VOP_GETATTR(vp, &va, p->p_ucred, p)) ||
188 	    (va.va_uid != p->p_ucred->cr_uid &&
189 	     (error = suser(p)))) {
190 		vput(vp);
191 		return (error);
192 	}
193 	if ((error = vinvalbuf(vp, V_SAVE, p->p_ucred, p, 0, 0)) != 0)
194 		return (error);
195 	if (vp->v_type != VDIR) {
196 		vput(vp);
197 		return (ENOTDIR);
198 	}
199 #ifdef COMPAT_43
200 	/*
201 	 * Historically filesystem types were identified by number. If we
202 	 * get an integer for the filesystem type instead of a string, we
203 	 * check to see if it matches one of the historic filesystem types.
204 	 */
205 	fstypenum = (uintptr_t)SCARG(uap, type);
206 	if (fstypenum < maxvfsconf) {
207 		for (vfsp = vfsconf; vfsp; vfsp = vfsp->vfc_next)
208 			if (vfsp->vfc_typenum == fstypenum)
209 				break;
210 		if (vfsp == NULL) {
211 			vput(vp);
212 			return (ENODEV);
213 		}
214 		strncpy(fstypename, vfsp->vfc_name, MFSNAMELEN);
215 	} else
216 #endif /* COMPAT_43 */
217 	if ((error = copyinstr(SCARG(uap, type), fstypename, MFSNAMELEN, NULL)) != 0) {
218 		vput(vp);
219 		return (error);
220 	}
221 	for (vfsp = vfsconf; vfsp; vfsp = vfsp->vfc_next)
222 		if (!strcmp(vfsp->vfc_name, fstypename))
223 			break;
224 	if (vfsp == NULL) {
225 		linker_file_t lf;
226 
227 		/* Refuse to load modules if securelevel raised */
228 		if (securelevel > 0) {
229 			vput(vp);
230 			return EPERM;
231 		}
232 		/* Only load modules for root (very important!) */
233 		if ((error = suser(p)) != 0) {
234 			vput(vp);
235 			return error;
236 		}
237 		error = linker_load_file(fstypename, &lf);
238 		if (error || lf == NULL) {
239 			vput(vp);
240 			if (lf == NULL)
241 				error = ENODEV;
242 			return error;
243 		}
244 		lf->userrefs++;
245 		/* lookup again, see if the VFS was loaded */
246 		for (vfsp = vfsconf; vfsp; vfsp = vfsp->vfc_next)
247 			if (!strcmp(vfsp->vfc_name, fstypename))
248 				break;
249 		if (vfsp == NULL) {
250 			lf->userrefs--;
251 			linker_file_unload(lf);
252 			vput(vp);
253 			return (ENODEV);
254 		}
255 	}
256 	simple_lock(&vp->v_interlock);
257 	if ((vp->v_flag & VMOUNT) != 0 ||
258 	    vp->v_mountedhere != NULL) {
259 		simple_unlock(&vp->v_interlock);
260 		vput(vp);
261 		return (EBUSY);
262 	}
263 	vp->v_flag |= VMOUNT;
264 	simple_unlock(&vp->v_interlock);
265 
266 	/*
267 	 * Allocate and initialize the filesystem.
268 	 */
269 	mp = malloc(sizeof(struct mount), M_MOUNT, M_WAITOK);
270 	bzero((char *)mp, (u_long)sizeof(struct mount));
271 	lockinit(&mp->mnt_lock, PVFS, "vfslock", 0, LK_NOPAUSE);
272 	(void)vfs_busy(mp, LK_NOWAIT, 0, p);
273 	mp->mnt_op = vfsp->vfc_vfsops;
274 	mp->mnt_vfc = vfsp;
275 	vfsp->vfc_refcount++;
276 	mp->mnt_stat.f_type = vfsp->vfc_typenum;
277 	mp->mnt_flag |= vfsp->vfc_flags & MNT_VISFLAGMASK;
278 	strncpy(mp->mnt_stat.f_fstypename, vfsp->vfc_name, MFSNAMELEN);
279 	mp->mnt_vnodecovered = vp;
280 	mp->mnt_stat.f_owner = p->p_ucred->cr_uid;
281 	mp->mnt_iosize_max = DFLTPHYS;
282 	VOP_UNLOCK(vp, 0, p);
283 update:
284 	/*
285 	 * Set the mount level flags.
286 	 */
287 	if (SCARG(uap, flags) & MNT_RDONLY)
288 		mp->mnt_flag |= MNT_RDONLY;
289 	else if (mp->mnt_flag & MNT_RDONLY)
290 		mp->mnt_kern_flag |= MNTK_WANTRDWR;
291 	mp->mnt_flag &=~ (MNT_NOSUID | MNT_NOEXEC | MNT_NODEV |
292 	    MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | MNT_NOATIME |
293 	    MNT_NOSYMFOLLOW | MNT_IGNORE |
294 	    MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR);
295 	mp->mnt_flag |= SCARG(uap, flags) & (MNT_NOSUID | MNT_NOEXEC |
296 	    MNT_NODEV | MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | MNT_FORCE |
297 	    MNT_NOSYMFOLLOW | MNT_IGNORE |
298 	    MNT_NOATIME | MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR);
299 	/*
300 	 * Mount the filesystem.
301 	 * XXX The final recipients of VFS_MOUNT just overwrite the ndp they
302 	 * get.  No freeing of cn_pnbuf.
303 	 */
304 	error = VFS_MOUNT(mp, SCARG(uap, path), SCARG(uap, data), &nd, p);
305 	if (mp->mnt_flag & MNT_UPDATE) {
306 		vrele(vp);
307 		if (mp->mnt_kern_flag & MNTK_WANTRDWR)
308 			mp->mnt_flag &= ~MNT_RDONLY;
309 		mp->mnt_flag &=~ (MNT_UPDATE | MNT_RELOAD | MNT_FORCE);
310 		mp->mnt_kern_flag &=~ MNTK_WANTRDWR;
311 		if (error) {
312 			mp->mnt_flag = flag;
313 			mp->mnt_kern_flag = flag2;
314 		}
315 		if ((mp->mnt_flag & MNT_RDONLY) == 0) {
316 			if (mp->mnt_syncer == NULL)
317 				error = vfs_allocate_syncvnode(mp);
318 		} else {
319 			if (mp->mnt_syncer != NULL)
320 				vrele(mp->mnt_syncer);
321 			mp->mnt_syncer = NULL;
322 		}
323 		vfs_unbusy(mp, p);
324 		return (error);
325 	}
326 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
327 	/*
328 	 * Put the new filesystem on the mount list after root.
329 	 */
330 	cache_purge(vp);
331 	if (!error) {
332 		simple_lock(&vp->v_interlock);
333 		vp->v_flag &= ~VMOUNT;
334 		vp->v_mountedhere = mp;
335 		simple_unlock(&vp->v_interlock);
336 		simple_lock(&mountlist_slock);
337 		TAILQ_INSERT_TAIL(&mountlist, mp, mnt_list);
338 		simple_unlock(&mountlist_slock);
339 		checkdirs(vp);
340 		VOP_UNLOCK(vp, 0, p);
341 		if ((mp->mnt_flag & MNT_RDONLY) == 0)
342 			error = vfs_allocate_syncvnode(mp);
343 		vfs_unbusy(mp, p);
344 		if ((error = VFS_START(mp, 0, p)) != 0)
345 			vrele(vp);
346 	} else {
347 		simple_lock(&vp->v_interlock);
348 		vp->v_flag &= ~VMOUNT;
349 		simple_unlock(&vp->v_interlock);
350 		mp->mnt_vfc->vfc_refcount--;
351 		vfs_unbusy(mp, p);
352 		free((caddr_t)mp, M_MOUNT);
353 		vput(vp);
354 	}
355 	return (error);
356 }
357 
358 /*
359  * Scan all active processes to see if any of them have a current
360  * or root directory onto which the new filesystem has just been
361  * mounted. If so, replace them with the new mount point.
362  */
363 static void
364 checkdirs(olddp)
365 	struct vnode *olddp;
366 {
367 	struct filedesc *fdp;
368 	struct vnode *newdp;
369 	struct proc *p;
370 
371 	if (olddp->v_usecount == 1)
372 		return;
373 	if (VFS_ROOT(olddp->v_mountedhere, &newdp))
374 		panic("mount: lost mount");
375 	LIST_FOREACH(p, &allproc, p_list) {
376 		fdp = p->p_fd;
377 		if (fdp->fd_cdir == olddp) {
378 			vrele(fdp->fd_cdir);
379 			VREF(newdp);
380 			fdp->fd_cdir = newdp;
381 		}
382 		if (fdp->fd_rdir == olddp) {
383 			vrele(fdp->fd_rdir);
384 			VREF(newdp);
385 			fdp->fd_rdir = newdp;
386 		}
387 	}
388 	if (rootvnode == olddp) {
389 		vrele(rootvnode);
390 		VREF(newdp);
391 		rootvnode = newdp;
392 	}
393 	vput(newdp);
394 }
395 
396 /*
397  * Unmount a file system.
398  *
399  * Note: unmount takes a path to the vnode mounted on as argument,
400  * not special file (as before).
401  */
402 #ifndef _SYS_SYSPROTO_H_
403 struct unmount_args {
404 	char	*path;
405 	int	flags;
406 };
407 #endif
408 /* ARGSUSED */
409 int
410 unmount(p, uap)
411 	struct proc *p;
412 	register struct unmount_args /* {
413 		syscallarg(char *) path;
414 		syscallarg(int) flags;
415 	} */ *uap;
416 {
417 	register struct vnode *vp;
418 	struct mount *mp;
419 	int error;
420 	struct nameidata nd;
421 
422 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
423 	    SCARG(uap, path), p);
424 	if ((error = namei(&nd)) != 0)
425 		return (error);
426 	vp = nd.ni_vp;
427 	NDFREE(&nd, NDF_ONLY_PNBUF);
428 	mp = vp->v_mount;
429 
430 	/*
431 	 * Only root, or the user that did the original mount is
432 	 * permitted to unmount this filesystem.
433 	 */
434 	if ((mp->mnt_stat.f_owner != p->p_ucred->cr_uid) &&
435 	    (error = suser(p))) {
436 		vput(vp);
437 		return (error);
438 	}
439 
440 	/*
441 	 * Don't allow unmounting the root file system.
442 	 */
443 	if (mp->mnt_flag & MNT_ROOTFS) {
444 		vput(vp);
445 		return (EINVAL);
446 	}
447 
448 	/*
449 	 * Must be the root of the filesystem
450 	 */
451 	if ((vp->v_flag & VROOT) == 0) {
452 		vput(vp);
453 		return (EINVAL);
454 	}
455 	vput(vp);
456 	return (dounmount(mp, SCARG(uap, flags), p));
457 }
458 
459 /*
460  * Do the actual file system unmount.
461  */
462 int
463 dounmount(mp, flags, p)
464 	register struct mount *mp;
465 	int flags;
466 	struct proc *p;
467 {
468 	struct vnode *coveredvp;
469 	int error;
470 	int async_flag;
471 
472 	simple_lock(&mountlist_slock);
473 	mp->mnt_kern_flag |= MNTK_UNMOUNT;
474 	lockmgr(&mp->mnt_lock, LK_DRAIN | LK_INTERLOCK, &mountlist_slock, p);
475 
476 	if (mp->mnt_flag & MNT_EXPUBLIC)
477 		vfs_setpublicfs(NULL, NULL, NULL);
478 
479 	vfs_msync(mp, MNT_WAIT);
480 	async_flag = mp->mnt_flag & MNT_ASYNC;
481 	mp->mnt_flag &=~ MNT_ASYNC;
482 	cache_purgevfs(mp);	/* remove cache entries for this file sys */
483 	if (mp->mnt_syncer != NULL)
484 		vrele(mp->mnt_syncer);
485 	if (((mp->mnt_flag & MNT_RDONLY) ||
486 	     (error = VFS_SYNC(mp, MNT_WAIT, p->p_ucred, p)) == 0) ||
487 	    (flags & MNT_FORCE))
488 		error = VFS_UNMOUNT(mp, flags, p);
489 	simple_lock(&mountlist_slock);
490 	if (error) {
491 		if ((mp->mnt_flag & MNT_RDONLY) == 0 && mp->mnt_syncer == NULL)
492 			(void) vfs_allocate_syncvnode(mp);
493 		mp->mnt_kern_flag &= ~MNTK_UNMOUNT;
494 		mp->mnt_flag |= async_flag;
495 		lockmgr(&mp->mnt_lock, LK_RELEASE | LK_INTERLOCK | LK_REENABLE,
496 		    &mountlist_slock, p);
497 		if (mp->mnt_kern_flag & MNTK_MWAIT)
498 			wakeup((caddr_t)mp);
499 		return (error);
500 	}
501 	TAILQ_REMOVE(&mountlist, mp, mnt_list);
502 	if ((coveredvp = mp->mnt_vnodecovered) != NULLVP) {
503 		coveredvp->v_mountedhere = (struct mount *)0;
504 		vrele(coveredvp);
505 	}
506 	mp->mnt_vfc->vfc_refcount--;
507 	if (!LIST_EMPTY(&mp->mnt_vnodelist))
508 		panic("unmount: dangling vnode");
509 	lockmgr(&mp->mnt_lock, LK_RELEASE | LK_INTERLOCK, &mountlist_slock, p);
510 	if (mp->mnt_kern_flag & MNTK_MWAIT)
511 		wakeup((caddr_t)mp);
512 	free((caddr_t)mp, M_MOUNT);
513 	return (0);
514 }
515 
516 /*
517  * Sync each mounted filesystem.
518  */
519 #ifndef _SYS_SYSPROTO_H_
520 struct sync_args {
521         int     dummy;
522 };
523 #endif
524 
525 #ifdef DEBUG
526 static int syncprt = 0;
527 SYSCTL_INT(_debug, OID_AUTO, syncprt, CTLFLAG_RW, &syncprt, 0, "");
528 #endif
529 
530 /* ARGSUSED */
531 int
532 sync(p, uap)
533 	struct proc *p;
534 	struct sync_args *uap;
535 {
536 	register struct mount *mp, *nmp;
537 	int asyncflag;
538 
539 	simple_lock(&mountlist_slock);
540 	for (mp = TAILQ_FIRST(&mountlist); mp != NULL; mp = nmp) {
541 		if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock, p)) {
542 			nmp = TAILQ_NEXT(mp, mnt_list);
543 			continue;
544 		}
545 		if ((mp->mnt_flag & MNT_RDONLY) == 0) {
546 			asyncflag = mp->mnt_flag & MNT_ASYNC;
547 			mp->mnt_flag &= ~MNT_ASYNC;
548 			vfs_msync(mp, MNT_NOWAIT);
549 			VFS_SYNC(mp, MNT_NOWAIT,
550 				((p != NULL) ? p->p_ucred : NOCRED), p);
551 			mp->mnt_flag |= asyncflag;
552 		}
553 		simple_lock(&mountlist_slock);
554 		nmp = TAILQ_NEXT(mp, mnt_list);
555 		vfs_unbusy(mp, p);
556 	}
557 	simple_unlock(&mountlist_slock);
558 #if 0
559 /*
560  * XXX don't call vfs_bufstats() yet because that routine
561  * was not imported in the Lite2 merge.
562  */
563 #ifdef DIAGNOSTIC
564 	if (syncprt)
565 		vfs_bufstats();
566 #endif /* DIAGNOSTIC */
567 #endif
568 	return (0);
569 }
570 
571 /* XXX PRISON: could be per prison flag */
572 static int prison_quotas;
573 #if 0
574 SYSCTL_INT(_kern_prison, OID_AUTO, quotas, CTLFLAG_RW, &prison_quotas, 0, "");
575 #endif
576 
577 /*
578  * Change filesystem quotas.
579  */
580 #ifndef _SYS_SYSPROTO_H_
581 struct quotactl_args {
582 	char *path;
583 	int cmd;
584 	int uid;
585 	caddr_t arg;
586 };
587 #endif
588 /* ARGSUSED */
589 int
590 quotactl(p, uap)
591 	struct proc *p;
592 	register struct quotactl_args /* {
593 		syscallarg(char *) path;
594 		syscallarg(int) cmd;
595 		syscallarg(int) uid;
596 		syscallarg(caddr_t) arg;
597 	} */ *uap;
598 {
599 	register struct mount *mp;
600 	int error;
601 	struct nameidata nd;
602 
603 	if (p->p_prison && !prison_quotas)
604 		return (EPERM);
605 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
606 	if ((error = namei(&nd)) != 0)
607 		return (error);
608 	mp = nd.ni_vp->v_mount;
609 	NDFREE(&nd, NDF_ONLY_PNBUF);
610 	vrele(nd.ni_vp);
611 	return (VFS_QUOTACTL(mp, SCARG(uap, cmd), SCARG(uap, uid),
612 	    SCARG(uap, arg), p));
613 }
614 
615 /*
616  * Get filesystem statistics.
617  */
618 #ifndef _SYS_SYSPROTO_H_
619 struct statfs_args {
620 	char *path;
621 	struct statfs *buf;
622 };
623 #endif
624 /* ARGSUSED */
625 int
626 statfs(p, uap)
627 	struct proc *p;
628 	register struct statfs_args /* {
629 		syscallarg(char *) path;
630 		syscallarg(struct statfs *) buf;
631 	} */ *uap;
632 {
633 	register struct mount *mp;
634 	register struct statfs *sp;
635 	int error;
636 	struct nameidata nd;
637 	struct statfs sb;
638 
639 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
640 	if ((error = namei(&nd)) != 0)
641 		return (error);
642 	mp = nd.ni_vp->v_mount;
643 	sp = &mp->mnt_stat;
644 	NDFREE(&nd, NDF_ONLY_PNBUF);
645 	vrele(nd.ni_vp);
646 	error = VFS_STATFS(mp, sp, p);
647 	if (error)
648 		return (error);
649 	sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
650 	if (suser_xxx(p->p_ucred, 0, 0)) {
651 		bcopy((caddr_t)sp, (caddr_t)&sb, sizeof(sb));
652 		sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0;
653 		sp = &sb;
654 	}
655 	return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
656 }
657 
658 /*
659  * Get filesystem statistics.
660  */
661 #ifndef _SYS_SYSPROTO_H_
662 struct fstatfs_args {
663 	int fd;
664 	struct statfs *buf;
665 };
666 #endif
667 /* ARGSUSED */
668 int
669 fstatfs(p, uap)
670 	struct proc *p;
671 	register struct fstatfs_args /* {
672 		syscallarg(int) fd;
673 		syscallarg(struct statfs *) buf;
674 	} */ *uap;
675 {
676 	struct file *fp;
677 	struct mount *mp;
678 	register struct statfs *sp;
679 	int error;
680 	struct statfs sb;
681 
682 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
683 		return (error);
684 	mp = ((struct vnode *)fp->f_data)->v_mount;
685 	sp = &mp->mnt_stat;
686 	error = VFS_STATFS(mp, sp, p);
687 	if (error)
688 		return (error);
689 	sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
690 	if (suser_xxx(p->p_ucred, 0, 0)) {
691 		bcopy((caddr_t)sp, (caddr_t)&sb, sizeof(sb));
692 		sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0;
693 		sp = &sb;
694 	}
695 	return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
696 }
697 
698 /*
699  * Get statistics on all filesystems.
700  */
701 #ifndef _SYS_SYSPROTO_H_
702 struct getfsstat_args {
703 	struct statfs *buf;
704 	long bufsize;
705 	int flags;
706 };
707 #endif
708 int
709 getfsstat(p, uap)
710 	struct proc *p;
711 	register struct getfsstat_args /* {
712 		syscallarg(struct statfs *) buf;
713 		syscallarg(long) bufsize;
714 		syscallarg(int) flags;
715 	} */ *uap;
716 {
717 	register struct mount *mp, *nmp;
718 	register struct statfs *sp;
719 	caddr_t sfsp;
720 	long count, maxcount, error;
721 
722 	maxcount = SCARG(uap, bufsize) / sizeof(struct statfs);
723 	sfsp = (caddr_t)SCARG(uap, buf);
724 	count = 0;
725 	simple_lock(&mountlist_slock);
726 	for (mp = TAILQ_FIRST(&mountlist); mp != NULL; mp = nmp) {
727 		if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock, p)) {
728 			nmp = TAILQ_NEXT(mp, mnt_list);
729 			continue;
730 		}
731 		if (sfsp && count < maxcount) {
732 			sp = &mp->mnt_stat;
733 			/*
734 			 * If MNT_NOWAIT or MNT_LAZY is specified, do not
735 			 * refresh the fsstat cache. MNT_NOWAIT or MNT_LAZY
736 			 * overrides MNT_WAIT.
737 			 */
738 			if (((SCARG(uap, flags) & (MNT_LAZY|MNT_NOWAIT)) == 0 ||
739 			    (SCARG(uap, flags) & MNT_WAIT)) &&
740 			    (error = VFS_STATFS(mp, sp, p))) {
741 				simple_lock(&mountlist_slock);
742 				nmp = TAILQ_NEXT(mp, mnt_list);
743 				vfs_unbusy(mp, p);
744 				continue;
745 			}
746 			sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
747 			error = copyout((caddr_t)sp, sfsp, sizeof(*sp));
748 			if (error) {
749 				vfs_unbusy(mp, p);
750 				return (error);
751 			}
752 			sfsp += sizeof(*sp);
753 		}
754 		count++;
755 		simple_lock(&mountlist_slock);
756 		nmp = TAILQ_NEXT(mp, mnt_list);
757 		vfs_unbusy(mp, p);
758 	}
759 	simple_unlock(&mountlist_slock);
760 	if (sfsp && count > maxcount)
761 		p->p_retval[0] = maxcount;
762 	else
763 		p->p_retval[0] = count;
764 	return (0);
765 }
766 
767 /*
768  * Change current working directory to a given file descriptor.
769  */
770 #ifndef _SYS_SYSPROTO_H_
771 struct fchdir_args {
772 	int	fd;
773 };
774 #endif
775 /* ARGSUSED */
776 int
777 fchdir(p, uap)
778 	struct proc *p;
779 	struct fchdir_args /* {
780 		syscallarg(int) fd;
781 	} */ *uap;
782 {
783 	register struct filedesc *fdp = p->p_fd;
784 	struct vnode *vp, *tdp;
785 	struct mount *mp;
786 	struct file *fp;
787 	int error;
788 
789 	if ((error = getvnode(fdp, SCARG(uap, fd), &fp)) != 0)
790 		return (error);
791 	vp = (struct vnode *)fp->f_data;
792 	VREF(vp);
793 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
794 	if (vp->v_type != VDIR)
795 		error = ENOTDIR;
796 	else
797 		error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
798 	while (!error && (mp = vp->v_mountedhere) != NULL) {
799 		if (vfs_busy(mp, 0, 0, p))
800 			continue;
801 		error = VFS_ROOT(mp, &tdp);
802 		vfs_unbusy(mp, p);
803 		if (error)
804 			break;
805 		vput(vp);
806 		vp = tdp;
807 	}
808 	if (error) {
809 		vput(vp);
810 		return (error);
811 	}
812 	VOP_UNLOCK(vp, 0, p);
813 	vrele(fdp->fd_cdir);
814 	fdp->fd_cdir = vp;
815 	return (0);
816 }
817 
818 /*
819  * Change current working directory (``.'').
820  */
821 #ifndef _SYS_SYSPROTO_H_
822 struct chdir_args {
823 	char	*path;
824 };
825 #endif
826 /* ARGSUSED */
827 int
828 chdir(p, uap)
829 	struct proc *p;
830 	struct chdir_args /* {
831 		syscallarg(char *) path;
832 	} */ *uap;
833 {
834 	register struct filedesc *fdp = p->p_fd;
835 	int error;
836 	struct nameidata nd;
837 
838 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
839 	    SCARG(uap, path), p);
840 	if ((error = change_dir(&nd, p)) != 0)
841 		return (error);
842 	NDFREE(&nd, NDF_ONLY_PNBUF);
843 	vrele(fdp->fd_cdir);
844 	fdp->fd_cdir = nd.ni_vp;
845 	return (0);
846 }
847 
848 /*
849  * Helper function for raised chroot(2) security function:  Refuse if
850  * any filedescriptors are open directories.
851  */
852 static int
853 chroot_refuse_vdir_fds(fdp)
854 	struct filedesc *fdp;
855 {
856 	struct vnode *vp;
857 	struct file *fp;
858 	int error;
859 	int fd;
860 
861 	for (fd = 0; fd < fdp->fd_nfiles ; fd++) {
862 		error = getvnode(fdp, fd, &fp);
863 		if (error)
864 			continue;
865 		vp = (struct vnode *)fp->f_data;
866 		if (vp->v_type != VDIR)
867 			continue;
868 		return(EPERM);
869 	}
870 	return (0);
871 }
872 
873 /*
874  * This sysctl determines if we will allow a process to chroot(2) if it
875  * has a directory open:
876  *	0: disallowed for all processes.
877  *	1: allowed for processes that were not already chroot(2)'ed.
878  *	2: allowed for all processes.
879  */
880 
881 static int chroot_allow_open_directories = 1;
882 
883 SYSCTL_INT(_kern, OID_AUTO, chroot_allow_open_directories, CTLFLAG_RW,
884      &chroot_allow_open_directories, 0, "");
885 
886 /*
887  * Change notion of root (``/'') directory.
888  */
889 #ifndef _SYS_SYSPROTO_H_
890 struct chroot_args {
891 	char	*path;
892 };
893 #endif
894 /* ARGSUSED */
895 int
896 chroot(p, uap)
897 	struct proc *p;
898 	struct chroot_args /* {
899 		syscallarg(char *) path;
900 	} */ *uap;
901 {
902 	register struct filedesc *fdp = p->p_fd;
903 	int error;
904 	struct nameidata nd;
905 
906 	error = suser_xxx(0, p, PRISON_ROOT);
907 	if (error)
908 		return (error);
909 	if (chroot_allow_open_directories == 0 ||
910 	    (chroot_allow_open_directories == 1 && fdp->fd_rdir != rootvnode))
911 		error = chroot_refuse_vdir_fds(fdp);
912 	if (error)
913 		return (error);
914 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
915 	    SCARG(uap, path), p);
916 	if ((error = change_dir(&nd, p)) != 0)
917 		return (error);
918 	NDFREE(&nd, NDF_ONLY_PNBUF);
919 	vrele(fdp->fd_rdir);
920 	fdp->fd_rdir = nd.ni_vp;
921 	if (!fdp->fd_jdir) {
922 		fdp->fd_jdir = nd.ni_vp;
923                 VREF(fdp->fd_jdir);
924 	}
925 	return (0);
926 }
927 
928 /*
929  * Common routine for chroot and chdir.
930  */
931 static int
932 change_dir(ndp, p)
933 	register struct nameidata *ndp;
934 	struct proc *p;
935 {
936 	struct vnode *vp;
937 	int error;
938 
939 	error = namei(ndp);
940 	if (error)
941 		return (error);
942 	vp = ndp->ni_vp;
943 	if (vp->v_type != VDIR)
944 		error = ENOTDIR;
945 	else
946 		error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
947 	if (error)
948 		vput(vp);
949 	else
950 		VOP_UNLOCK(vp, 0, p);
951 	return (error);
952 }
953 
954 /*
955  * Check permissions, allocate an open file structure,
956  * and call the device open routine if any.
957  */
958 #ifndef _SYS_SYSPROTO_H_
959 struct open_args {
960 	char	*path;
961 	int	flags;
962 	int	mode;
963 };
964 #endif
965 int
966 open(p, uap)
967 	struct proc *p;
968 	register struct open_args /* {
969 		syscallarg(char *) path;
970 		syscallarg(int) flags;
971 		syscallarg(int) mode;
972 	} */ *uap;
973 {
974 	register struct filedesc *fdp = p->p_fd;
975 	register struct file *fp;
976 	register struct vnode *vp;
977 	int cmode, flags, oflags;
978 	struct file *nfp;
979 	int type, indx, error;
980 	struct flock lf;
981 	struct nameidata nd;
982 
983 	oflags = SCARG(uap, flags);
984 	if ((oflags & O_ACCMODE) == O_ACCMODE)
985 		return (EINVAL);
986 	flags = FFLAGS(oflags);
987 	error = falloc(p, &nfp, &indx);
988 	if (error)
989 		return (error);
990 	fp = nfp;
991 	cmode = ((SCARG(uap, mode) &~ fdp->fd_cmask) & ALLPERMS) &~ S_ISTXT;
992 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
993 	p->p_dupfd = -indx - 1;			/* XXX check for fdopen */
994 	error = vn_open(&nd, flags, cmode);
995 	if (error) {
996 		ffree(fp);
997 		if ((error == ENODEV || error == ENXIO) &&
998 		    p->p_dupfd >= 0 &&			/* XXX from fdopen */
999 		    (error =
1000 			dupfdopen(fdp, indx, p->p_dupfd, flags, error)) == 0) {
1001 			p->p_retval[0] = indx;
1002 			return (0);
1003 		}
1004 		if (error == ERESTART)
1005 			error = EINTR;
1006 		fdp->fd_ofiles[indx] = NULL;
1007 		return (error);
1008 	}
1009 	p->p_dupfd = 0;
1010 	NDFREE(&nd, NDF_ONLY_PNBUF);
1011 	vp = nd.ni_vp;
1012 
1013 	fp->f_data = (caddr_t)vp;
1014 	fp->f_flag = flags & FMASK;
1015 	fp->f_ops = &vnops;
1016 	fp->f_type = (vp->v_type == VFIFO ? DTYPE_FIFO : DTYPE_VNODE);
1017 	if (flags & (O_EXLOCK | O_SHLOCK)) {
1018 		lf.l_whence = SEEK_SET;
1019 		lf.l_start = 0;
1020 		lf.l_len = 0;
1021 		if (flags & O_EXLOCK)
1022 			lf.l_type = F_WRLCK;
1023 		else
1024 			lf.l_type = F_RDLCK;
1025 		type = F_FLOCK;
1026 		if ((flags & FNONBLOCK) == 0)
1027 			type |= F_WAIT;
1028 		VOP_UNLOCK(vp, 0, p);
1029 		if ((error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type)) != 0) {
1030 			(void) vn_close(vp, fp->f_flag, fp->f_cred, p);
1031 			ffree(fp);
1032 			fdp->fd_ofiles[indx] = NULL;
1033 			return (error);
1034 		}
1035 		vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
1036 		fp->f_flag |= FHASLOCK;
1037 	}
1038 	/* assert that vn_open created a backing object if one is needed */
1039 	KASSERT(!vn_canvmio(vp) || vp->v_object != NULL,
1040 		("open: vmio vnode has no backing object after vn_open"));
1041 	VOP_UNLOCK(vp, 0, p);
1042 	p->p_retval[0] = indx;
1043 	return (0);
1044 }
1045 
1046 #ifdef COMPAT_43
1047 /*
1048  * Create a file.
1049  */
1050 #ifndef _SYS_SYSPROTO_H_
1051 struct ocreat_args {
1052 	char	*path;
1053 	int	mode;
1054 };
1055 #endif
1056 int
1057 ocreat(p, uap)
1058 	struct proc *p;
1059 	register struct ocreat_args /* {
1060 		syscallarg(char *) path;
1061 		syscallarg(int) mode;
1062 	} */ *uap;
1063 {
1064 	struct open_args /* {
1065 		syscallarg(char *) path;
1066 		syscallarg(int) flags;
1067 		syscallarg(int) mode;
1068 	} */ nuap;
1069 
1070 	SCARG(&nuap, path) = SCARG(uap, path);
1071 	SCARG(&nuap, mode) = SCARG(uap, mode);
1072 	SCARG(&nuap, flags) = O_WRONLY | O_CREAT | O_TRUNC;
1073 	return (open(p, &nuap));
1074 }
1075 #endif /* COMPAT_43 */
1076 
1077 /*
1078  * Create a special file.
1079  */
1080 #ifndef _SYS_SYSPROTO_H_
1081 struct mknod_args {
1082 	char	*path;
1083 	int	mode;
1084 	int	dev;
1085 };
1086 #endif
1087 /* ARGSUSED */
1088 int
1089 mknod(p, uap)
1090 	struct proc *p;
1091 	register struct mknod_args /* {
1092 		syscallarg(char *) path;
1093 		syscallarg(int) mode;
1094 		syscallarg(int) dev;
1095 	} */ *uap;
1096 {
1097 	register struct vnode *vp;
1098 	struct vattr vattr;
1099 	int error;
1100 	int whiteout = 0;
1101 	struct nameidata nd;
1102 
1103 	switch (SCARG(uap, mode) & S_IFMT) {
1104 	case S_IFCHR:
1105 	case S_IFBLK:
1106 		error = suser(p);
1107 		break;
1108 	default:
1109 		error = suser_xxx(0, p, PRISON_ROOT);
1110 		break;
1111 	}
1112 	if (error)
1113 		return (error);
1114 	bwillwrite();
1115 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1116 	if ((error = namei(&nd)) != 0)
1117 		return (error);
1118 	vp = nd.ni_vp;
1119 	if (vp != NULL)
1120 		error = EEXIST;
1121 	else {
1122 		VATTR_NULL(&vattr);
1123 		vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
1124 		vattr.va_rdev = SCARG(uap, dev);
1125 		whiteout = 0;
1126 
1127 		switch (SCARG(uap, mode) & S_IFMT) {
1128 		case S_IFMT:	/* used by badsect to flag bad sectors */
1129 			vattr.va_type = VBAD;
1130 			break;
1131 		case S_IFCHR:
1132 			vattr.va_type = VCHR;
1133 			break;
1134 		case S_IFBLK:
1135 			vattr.va_type = VBLK;
1136 			break;
1137 		case S_IFWHT:
1138 			whiteout = 1;
1139 			break;
1140 		default:
1141 			error = EINVAL;
1142 			break;
1143 		}
1144 	}
1145 	if (!error) {
1146 		VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1147 		if (whiteout)
1148 			error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, CREATE);
1149 		else {
1150 			error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp,
1151 						&nd.ni_cnd, &vattr);
1152 			if (error == 0)
1153 				vput(nd.ni_vp);
1154 		}
1155 		NDFREE(&nd, NDF_ONLY_PNBUF);
1156 		vput(nd.ni_dvp);
1157 	} else {
1158 		NDFREE(&nd, NDF_ONLY_PNBUF);
1159 		if (nd.ni_dvp == vp)
1160 			vrele(nd.ni_dvp);
1161 		else
1162 			vput(nd.ni_dvp);
1163 		if (vp)
1164 			vrele(vp);
1165 	}
1166 	ASSERT_VOP_UNLOCKED(nd.ni_dvp, "mknod");
1167 	ASSERT_VOP_UNLOCKED(nd.ni_vp, "mknod");
1168 	return (error);
1169 }
1170 
1171 /*
1172  * Create a named pipe.
1173  */
1174 #ifndef _SYS_SYSPROTO_H_
1175 struct mkfifo_args {
1176 	char	*path;
1177 	int	mode;
1178 };
1179 #endif
1180 /* ARGSUSED */
1181 int
1182 mkfifo(p, uap)
1183 	struct proc *p;
1184 	register struct mkfifo_args /* {
1185 		syscallarg(char *) path;
1186 		syscallarg(int) mode;
1187 	} */ *uap;
1188 {
1189 	struct vattr vattr;
1190 	int error;
1191 	struct nameidata nd;
1192 
1193 	bwillwrite();
1194 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1195 	if ((error = namei(&nd)) != 0)
1196 		return (error);
1197 	if (nd.ni_vp != NULL) {
1198 		NDFREE(&nd, NDF_ONLY_PNBUF);
1199 		if (nd.ni_dvp == nd.ni_vp)
1200 			vrele(nd.ni_dvp);
1201 		else
1202 			vput(nd.ni_dvp);
1203 		vrele(nd.ni_vp);
1204 		return (EEXIST);
1205 	}
1206 	VATTR_NULL(&vattr);
1207 	vattr.va_type = VFIFO;
1208 	vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
1209 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1210 	error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
1211 	if (error == 0)
1212 		vput(nd.ni_vp);
1213 	NDFREE(&nd, NDF_ONLY_PNBUF);
1214 	vput(nd.ni_dvp);
1215 	return (error);
1216 }
1217 
1218 /*
1219  * Make a hard file link.
1220  */
1221 #ifndef _SYS_SYSPROTO_H_
1222 struct link_args {
1223 	char	*path;
1224 	char	*link;
1225 };
1226 #endif
1227 /* ARGSUSED */
1228 int
1229 link(p, uap)
1230 	struct proc *p;
1231 	register struct link_args /* {
1232 		syscallarg(char *) path;
1233 		syscallarg(char *) link;
1234 	} */ *uap;
1235 {
1236 	register struct vnode *vp;
1237 	struct nameidata nd;
1238 	int error;
1239 
1240 	bwillwrite();
1241 	NDINIT(&nd, LOOKUP, FOLLOW|NOOBJ, UIO_USERSPACE, SCARG(uap, path), p);
1242 	if ((error = namei(&nd)) != 0)
1243 		return (error);
1244 	NDFREE(&nd, NDF_ONLY_PNBUF);
1245 	vp = nd.ni_vp;
1246 	if (vp->v_type == VDIR)
1247 		error = EPERM;		/* POSIX */
1248 	else {
1249 		NDINIT(&nd, CREATE, LOCKPARENT|NOOBJ, UIO_USERSPACE, SCARG(uap, link), p);
1250 		error = namei(&nd);
1251 		if (!error) {
1252 			if (nd.ni_vp != NULL) {
1253 				if (nd.ni_vp)
1254 					vrele(nd.ni_vp);
1255 				error = EEXIST;
1256 			} else {
1257 				VOP_LEASE(nd.ni_dvp, p, p->p_ucred,
1258 				    LEASE_WRITE);
1259 				VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1260 				error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd);
1261 			}
1262 			NDFREE(&nd, NDF_ONLY_PNBUF);
1263 			if (nd.ni_dvp == nd.ni_vp)
1264 				vrele(nd.ni_dvp);
1265 			else
1266 				vput(nd.ni_dvp);
1267 		}
1268 	}
1269 	vrele(vp);
1270 	ASSERT_VOP_UNLOCKED(nd.ni_dvp, "link");
1271 	ASSERT_VOP_UNLOCKED(nd.ni_vp, "link");
1272 	return (error);
1273 }
1274 
1275 /*
1276  * Make a symbolic link.
1277  */
1278 #ifndef _SYS_SYSPROTO_H_
1279 struct symlink_args {
1280 	char	*path;
1281 	char	*link;
1282 };
1283 #endif
1284 /* ARGSUSED */
1285 int
1286 symlink(p, uap)
1287 	struct proc *p;
1288 	register struct symlink_args /* {
1289 		syscallarg(char *) path;
1290 		syscallarg(char *) link;
1291 	} */ *uap;
1292 {
1293 	struct vattr vattr;
1294 	char *path;
1295 	int error;
1296 	struct nameidata nd;
1297 
1298 	path = zalloc(namei_zone);
1299 	if ((error = copyinstr(SCARG(uap, path), path, MAXPATHLEN, NULL)) != 0)
1300 		goto out;
1301 	bwillwrite();
1302 	NDINIT(&nd, CREATE, LOCKPARENT|NOOBJ, UIO_USERSPACE, SCARG(uap, link), p);
1303 	if ((error = namei(&nd)) != 0)
1304 		goto out;
1305 	if (nd.ni_vp) {
1306 		NDFREE(&nd, NDF_ONLY_PNBUF);
1307 		if (nd.ni_dvp == nd.ni_vp)
1308 			vrele(nd.ni_dvp);
1309 		else
1310 			vput(nd.ni_dvp);
1311 		vrele(nd.ni_vp);
1312 		error = EEXIST;
1313 		goto out;
1314 	}
1315 	VATTR_NULL(&vattr);
1316 	vattr.va_mode = ACCESSPERMS &~ p->p_fd->fd_cmask;
1317 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1318 	error = VOP_SYMLINK(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr, path);
1319 	NDFREE(&nd, NDF_ONLY_PNBUF);
1320 	if (error == 0)
1321 		vput(nd.ni_vp);
1322 	vput(nd.ni_dvp);
1323 	ASSERT_VOP_UNLOCKED(nd.ni_dvp, "symlink");
1324 	ASSERT_VOP_UNLOCKED(nd.ni_vp, "symlink");
1325 out:
1326 	zfree(namei_zone, path);
1327 	return (error);
1328 }
1329 
1330 /*
1331  * Delete a whiteout from the filesystem.
1332  */
1333 /* ARGSUSED */
1334 int
1335 undelete(p, uap)
1336 	struct proc *p;
1337 	register struct undelete_args /* {
1338 		syscallarg(char *) path;
1339 	} */ *uap;
1340 {
1341 	int error;
1342 	struct nameidata nd;
1343 
1344 	bwillwrite();
1345 	NDINIT(&nd, DELETE, LOCKPARENT|DOWHITEOUT, UIO_USERSPACE,
1346 	    SCARG(uap, path), p);
1347 	error = namei(&nd);
1348 	if (error)
1349 		return (error);
1350 
1351 	if (nd.ni_vp != NULLVP || !(nd.ni_cnd.cn_flags & ISWHITEOUT)) {
1352 		NDFREE(&nd, NDF_ONLY_PNBUF);
1353 		if (nd.ni_dvp == nd.ni_vp)
1354 			vrele(nd.ni_dvp);
1355 		else
1356 			vput(nd.ni_dvp);
1357 		if (nd.ni_vp)
1358 			vrele(nd.ni_vp);
1359 		return (EEXIST);
1360 	}
1361 
1362 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1363 	error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, DELETE);
1364 	NDFREE(&nd, NDF_ONLY_PNBUF);
1365 	vput(nd.ni_dvp);
1366 	ASSERT_VOP_UNLOCKED(nd.ni_dvp, "undelete");
1367 	ASSERT_VOP_UNLOCKED(nd.ni_vp, "undelete");
1368 	return (error);
1369 }
1370 
1371 /*
1372  * Delete a name from the filesystem.
1373  */
1374 #ifndef _SYS_SYSPROTO_H_
1375 struct unlink_args {
1376 	char	*path;
1377 };
1378 #endif
1379 /* ARGSUSED */
1380 int
1381 unlink(p, uap)
1382 	struct proc *p;
1383 	struct unlink_args /* {
1384 		syscallarg(char *) path;
1385 	} */ *uap;
1386 {
1387 	register struct vnode *vp;
1388 	int error;
1389 	struct nameidata nd;
1390 
1391 	bwillwrite();
1392 	NDINIT(&nd, DELETE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1393 	if ((error = namei(&nd)) != 0)
1394 		return (error);
1395 	vp = nd.ni_vp;
1396 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1397 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
1398 
1399 	if (vp->v_type == VDIR)
1400 		error = EPERM;		/* POSIX */
1401 	else {
1402 		/*
1403 		 * The root of a mounted filesystem cannot be deleted.
1404 		 *
1405 		 * XXX: can this only be a VDIR case?
1406 		 */
1407 		if (vp->v_flag & VROOT)
1408 			error = EBUSY;
1409 	}
1410 
1411 	if (!error) {
1412 		VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1413 		error = VOP_REMOVE(nd.ni_dvp, vp, &nd.ni_cnd);
1414 	}
1415 	NDFREE(&nd, NDF_ONLY_PNBUF);
1416 	if (nd.ni_dvp == vp)
1417 		vrele(nd.ni_dvp);
1418 	else
1419 		vput(nd.ni_dvp);
1420 	if (vp != NULLVP)
1421 		vput(vp);
1422 	ASSERT_VOP_UNLOCKED(nd.ni_dvp, "unlink");
1423 	ASSERT_VOP_UNLOCKED(nd.ni_vp, "unlink");
1424 	return (error);
1425 }
1426 
1427 /*
1428  * Reposition read/write file offset.
1429  */
1430 #ifndef _SYS_SYSPROTO_H_
1431 struct lseek_args {
1432 	int	fd;
1433 	int	pad;
1434 	off_t	offset;
1435 	int	whence;
1436 };
1437 #endif
1438 int
1439 lseek(p, uap)
1440 	struct proc *p;
1441 	register struct lseek_args /* {
1442 		syscallarg(int) fd;
1443 		syscallarg(int) pad;
1444 		syscallarg(off_t) offset;
1445 		syscallarg(int) whence;
1446 	} */ *uap;
1447 {
1448 	struct ucred *cred = p->p_ucred;
1449 	register struct filedesc *fdp = p->p_fd;
1450 	register struct file *fp;
1451 	struct vattr vattr;
1452 	int error;
1453 
1454 	if ((u_int)SCARG(uap, fd) >= fdp->fd_nfiles ||
1455 	    (fp = fdp->fd_ofiles[SCARG(uap, fd)]) == NULL)
1456 		return (EBADF);
1457 	if (fp->f_type != DTYPE_VNODE)
1458 		return (ESPIPE);
1459 	switch (SCARG(uap, whence)) {
1460 	case L_INCR:
1461 		fp->f_offset += SCARG(uap, offset);
1462 		break;
1463 	case L_XTND:
1464 		error=VOP_GETATTR((struct vnode *)fp->f_data, &vattr, cred, p);
1465 		if (error)
1466 			return (error);
1467 		fp->f_offset = SCARG(uap, offset) + vattr.va_size;
1468 		break;
1469 	case L_SET:
1470 		fp->f_offset = SCARG(uap, offset);
1471 		break;
1472 	default:
1473 		return (EINVAL);
1474 	}
1475 	*(off_t *)(p->p_retval) = fp->f_offset;
1476 	return (0);
1477 }
1478 
1479 #if defined(COMPAT_43) || defined(COMPAT_SUNOS)
1480 /*
1481  * Reposition read/write file offset.
1482  */
1483 #ifndef _SYS_SYSPROTO_H_
1484 struct olseek_args {
1485 	int	fd;
1486 	long	offset;
1487 	int	whence;
1488 };
1489 #endif
1490 int
1491 olseek(p, uap)
1492 	struct proc *p;
1493 	register struct olseek_args /* {
1494 		syscallarg(int) fd;
1495 		syscallarg(long) offset;
1496 		syscallarg(int) whence;
1497 	} */ *uap;
1498 {
1499 	struct lseek_args /* {
1500 		syscallarg(int) fd;
1501 		syscallarg(int) pad;
1502 		syscallarg(off_t) offset;
1503 		syscallarg(int) whence;
1504 	} */ nuap;
1505 	int error;
1506 
1507 	SCARG(&nuap, fd) = SCARG(uap, fd);
1508 	SCARG(&nuap, offset) = SCARG(uap, offset);
1509 	SCARG(&nuap, whence) = SCARG(uap, whence);
1510 	error = lseek(p, &nuap);
1511 	return (error);
1512 }
1513 #endif /* COMPAT_43 */
1514 
1515 /*
1516  * Check access permissions.
1517  */
1518 #ifndef _SYS_SYSPROTO_H_
1519 struct access_args {
1520 	char	*path;
1521 	int	flags;
1522 };
1523 #endif
1524 int
1525 access(p, uap)
1526 	struct proc *p;
1527 	register struct access_args /* {
1528 		syscallarg(char *) path;
1529 		syscallarg(int) flags;
1530 	} */ *uap;
1531 {
1532 	register struct ucred *cred = p->p_ucred;
1533 	register struct vnode *vp;
1534 	int error, flags, t_gid, t_uid;
1535 	struct nameidata nd;
1536 
1537 	t_uid = cred->cr_uid;
1538 	t_gid = cred->cr_groups[0];
1539 	cred->cr_uid = p->p_cred->p_ruid;
1540 	cred->cr_groups[0] = p->p_cred->p_rgid;
1541 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1542 	    SCARG(uap, path), p);
1543 	if ((error = namei(&nd)) != 0)
1544 		goto out1;
1545 	vp = nd.ni_vp;
1546 
1547 	/* Flags == 0 means only check for existence. */
1548 	if (SCARG(uap, flags)) {
1549 		flags = 0;
1550 		if (SCARG(uap, flags) & R_OK)
1551 			flags |= VREAD;
1552 		if (SCARG(uap, flags) & W_OK)
1553 			flags |= VWRITE;
1554 		if (SCARG(uap, flags) & X_OK)
1555 			flags |= VEXEC;
1556 		if ((flags & VWRITE) == 0 || (error = vn_writechk(vp)) == 0)
1557 			error = VOP_ACCESS(vp, flags, cred, p);
1558 	}
1559 	NDFREE(&nd, NDF_ONLY_PNBUF);
1560 	vput(vp);
1561 out1:
1562 	cred->cr_uid = t_uid;
1563 	cred->cr_groups[0] = t_gid;
1564 	return (error);
1565 }
1566 
1567 #if defined(COMPAT_43) || defined(COMPAT_SUNOS)
1568 /*
1569  * Get file status; this version follows links.
1570  */
1571 #ifndef _SYS_SYSPROTO_H_
1572 struct ostat_args {
1573 	char	*path;
1574 	struct ostat *ub;
1575 };
1576 #endif
1577 /* ARGSUSED */
1578 int
1579 ostat(p, uap)
1580 	struct proc *p;
1581 	register struct ostat_args /* {
1582 		syscallarg(char *) path;
1583 		syscallarg(struct ostat *) ub;
1584 	} */ *uap;
1585 {
1586 	struct stat sb;
1587 	struct ostat osb;
1588 	int error;
1589 	struct nameidata nd;
1590 
1591 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1592 	    SCARG(uap, path), p);
1593 	if ((error = namei(&nd)) != 0)
1594 		return (error);
1595 	NDFREE(&nd, NDF_ONLY_PNBUF);
1596 	error = vn_stat(nd.ni_vp, &sb, p);
1597 	vput(nd.ni_vp);
1598 	if (error)
1599 		return (error);
1600 	cvtstat(&sb, &osb);
1601 	error = copyout((caddr_t)&osb, (caddr_t)SCARG(uap, ub), sizeof (osb));
1602 	return (error);
1603 }
1604 
1605 /*
1606  * Get file status; this version does not follow links.
1607  */
1608 #ifndef _SYS_SYSPROTO_H_
1609 struct olstat_args {
1610 	char	*path;
1611 	struct ostat *ub;
1612 };
1613 #endif
1614 /* ARGSUSED */
1615 int
1616 olstat(p, uap)
1617 	struct proc *p;
1618 	register struct olstat_args /* {
1619 		syscallarg(char *) path;
1620 		syscallarg(struct ostat *) ub;
1621 	} */ *uap;
1622 {
1623 	struct vnode *vp;
1624 	struct stat sb;
1625 	struct ostat osb;
1626 	int error;
1627 	struct nameidata nd;
1628 
1629 	NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1630 	    SCARG(uap, path), p);
1631 	if ((error = namei(&nd)) != 0)
1632 		return (error);
1633 	vp = nd.ni_vp;
1634 	error = vn_stat(vp, &sb, p);
1635 	NDFREE(&nd, NDF_ONLY_PNBUF);
1636 	vput(vp);
1637 	if (error)
1638 		return (error);
1639 	cvtstat(&sb, &osb);
1640 	error = copyout((caddr_t)&osb, (caddr_t)SCARG(uap, ub), sizeof (osb));
1641 	return (error);
1642 }
1643 
1644 /*
1645  * Convert from an old to a new stat structure.
1646  */
1647 void
1648 cvtstat(st, ost)
1649 	struct stat *st;
1650 	struct ostat *ost;
1651 {
1652 
1653 	ost->st_dev = st->st_dev;
1654 	ost->st_ino = st->st_ino;
1655 	ost->st_mode = st->st_mode;
1656 	ost->st_nlink = st->st_nlink;
1657 	ost->st_uid = st->st_uid;
1658 	ost->st_gid = st->st_gid;
1659 	ost->st_rdev = st->st_rdev;
1660 	if (st->st_size < (quad_t)1 << 32)
1661 		ost->st_size = st->st_size;
1662 	else
1663 		ost->st_size = -2;
1664 	ost->st_atime = st->st_atime;
1665 	ost->st_mtime = st->st_mtime;
1666 	ost->st_ctime = st->st_ctime;
1667 	ost->st_blksize = st->st_blksize;
1668 	ost->st_blocks = st->st_blocks;
1669 	ost->st_flags = st->st_flags;
1670 	ost->st_gen = st->st_gen;
1671 }
1672 #endif /* COMPAT_43 || COMPAT_SUNOS */
1673 
1674 /*
1675  * Get file status; this version follows links.
1676  */
1677 #ifndef _SYS_SYSPROTO_H_
1678 struct stat_args {
1679 	char	*path;
1680 	struct stat *ub;
1681 };
1682 #endif
1683 /* ARGSUSED */
1684 int
1685 stat(p, uap)
1686 	struct proc *p;
1687 	register struct stat_args /* {
1688 		syscallarg(char *) path;
1689 		syscallarg(struct stat *) ub;
1690 	} */ *uap;
1691 {
1692 	struct stat sb;
1693 	int error;
1694 	struct nameidata nd;
1695 
1696 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1697 	    SCARG(uap, path), p);
1698 	if ((error = namei(&nd)) != 0)
1699 		return (error);
1700 	error = vn_stat(nd.ni_vp, &sb, p);
1701 	NDFREE(&nd, NDF_ONLY_PNBUF);
1702 	vput(nd.ni_vp);
1703 	if (error)
1704 		return (error);
1705 	error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
1706 	return (error);
1707 }
1708 
1709 /*
1710  * Get file status; this version does not follow links.
1711  */
1712 #ifndef _SYS_SYSPROTO_H_
1713 struct lstat_args {
1714 	char	*path;
1715 	struct stat *ub;
1716 };
1717 #endif
1718 /* ARGSUSED */
1719 int
1720 lstat(p, uap)
1721 	struct proc *p;
1722 	register struct lstat_args /* {
1723 		syscallarg(char *) path;
1724 		syscallarg(struct stat *) ub;
1725 	} */ *uap;
1726 {
1727 	int error;
1728 	struct vnode *vp;
1729 	struct stat sb;
1730 	struct nameidata nd;
1731 
1732 	NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1733 	    SCARG(uap, path), p);
1734 	if ((error = namei(&nd)) != 0)
1735 		return (error);
1736 	vp = nd.ni_vp;
1737 	error = vn_stat(vp, &sb, p);
1738 	NDFREE(&nd, NDF_ONLY_PNBUF);
1739 	vput(vp);
1740 	if (error)
1741 		return (error);
1742 	error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
1743 	return (error);
1744 }
1745 
1746 void
1747 cvtnstat(sb, nsb)
1748 	struct stat *sb;
1749 	struct nstat *nsb;
1750 {
1751 	nsb->st_dev = sb->st_dev;
1752 	nsb->st_ino = sb->st_ino;
1753 	nsb->st_mode = sb->st_mode;
1754 	nsb->st_nlink = sb->st_nlink;
1755 	nsb->st_uid = sb->st_uid;
1756 	nsb->st_gid = sb->st_gid;
1757 	nsb->st_rdev = sb->st_rdev;
1758 	nsb->st_atimespec = sb->st_atimespec;
1759 	nsb->st_mtimespec = sb->st_mtimespec;
1760 	nsb->st_ctimespec = sb->st_ctimespec;
1761 	nsb->st_size = sb->st_size;
1762 	nsb->st_blocks = sb->st_blocks;
1763 	nsb->st_blksize = sb->st_blksize;
1764 	nsb->st_flags = sb->st_flags;
1765 	nsb->st_gen = sb->st_gen;
1766 	nsb->st_qspare[0] = sb->st_qspare[0];
1767 	nsb->st_qspare[1] = sb->st_qspare[1];
1768 }
1769 
1770 #ifndef _SYS_SYSPROTO_H_
1771 struct nstat_args {
1772 	char	*path;
1773 	struct nstat *ub;
1774 };
1775 #endif
1776 /* ARGSUSED */
1777 int
1778 nstat(p, uap)
1779 	struct proc *p;
1780 	register struct nstat_args /* {
1781 		syscallarg(char *) path;
1782 		syscallarg(struct nstat *) ub;
1783 	} */ *uap;
1784 {
1785 	struct stat sb;
1786 	struct nstat nsb;
1787 	int error;
1788 	struct nameidata nd;
1789 
1790 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1791 	    SCARG(uap, path), p);
1792 	if ((error = namei(&nd)) != 0)
1793 		return (error);
1794 	NDFREE(&nd, NDF_ONLY_PNBUF);
1795 	error = vn_stat(nd.ni_vp, &sb, p);
1796 	vput(nd.ni_vp);
1797 	if (error)
1798 		return (error);
1799 	cvtnstat(&sb, &nsb);
1800 	error = copyout((caddr_t)&nsb, (caddr_t)SCARG(uap, ub), sizeof (nsb));
1801 	return (error);
1802 }
1803 
1804 /*
1805  * Get file status; this version does not follow links.
1806  */
1807 #ifndef _SYS_SYSPROTO_H_
1808 struct lstat_args {
1809 	char	*path;
1810 	struct stat *ub;
1811 };
1812 #endif
1813 /* ARGSUSED */
1814 int
1815 nlstat(p, uap)
1816 	struct proc *p;
1817 	register struct nlstat_args /* {
1818 		syscallarg(char *) path;
1819 		syscallarg(struct nstat *) ub;
1820 	} */ *uap;
1821 {
1822 	int error;
1823 	struct vnode *vp;
1824 	struct stat sb;
1825 	struct nstat nsb;
1826 	struct nameidata nd;
1827 
1828 	NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1829 	    SCARG(uap, path), p);
1830 	if ((error = namei(&nd)) != 0)
1831 		return (error);
1832 	vp = nd.ni_vp;
1833 	NDFREE(&nd, NDF_ONLY_PNBUF);
1834 	error = vn_stat(vp, &sb, p);
1835 	vput(vp);
1836 	if (error)
1837 		return (error);
1838 	cvtnstat(&sb, &nsb);
1839 	error = copyout((caddr_t)&nsb, (caddr_t)SCARG(uap, ub), sizeof (nsb));
1840 	return (error);
1841 }
1842 
1843 /*
1844  * Get configurable pathname variables.
1845  */
1846 #ifndef _SYS_SYSPROTO_H_
1847 struct pathconf_args {
1848 	char	*path;
1849 	int	name;
1850 };
1851 #endif
1852 /* ARGSUSED */
1853 int
1854 pathconf(p, uap)
1855 	struct proc *p;
1856 	register struct pathconf_args /* {
1857 		syscallarg(char *) path;
1858 		syscallarg(int) name;
1859 	} */ *uap;
1860 {
1861 	int error;
1862 	struct nameidata nd;
1863 
1864 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1865 	    SCARG(uap, path), p);
1866 	if ((error = namei(&nd)) != 0)
1867 		return (error);
1868 	NDFREE(&nd, NDF_ONLY_PNBUF);
1869 	error = VOP_PATHCONF(nd.ni_vp, SCARG(uap, name), p->p_retval);
1870 	vput(nd.ni_vp);
1871 	return (error);
1872 }
1873 
1874 /*
1875  * Return target name of a symbolic link.
1876  */
1877 #ifndef _SYS_SYSPROTO_H_
1878 struct readlink_args {
1879 	char	*path;
1880 	char	*buf;
1881 	int	count;
1882 };
1883 #endif
1884 /* ARGSUSED */
1885 int
1886 readlink(p, uap)
1887 	struct proc *p;
1888 	register struct readlink_args /* {
1889 		syscallarg(char *) path;
1890 		syscallarg(char *) buf;
1891 		syscallarg(int) count;
1892 	} */ *uap;
1893 {
1894 	register struct vnode *vp;
1895 	struct iovec aiov;
1896 	struct uio auio;
1897 	int error;
1898 	struct nameidata nd;
1899 
1900 	NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1901 	    SCARG(uap, path), p);
1902 	if ((error = namei(&nd)) != 0)
1903 		return (error);
1904 	NDFREE(&nd, NDF_ONLY_PNBUF);
1905 	vp = nd.ni_vp;
1906 	if (vp->v_type != VLNK)
1907 		error = EINVAL;
1908 	else {
1909 		aiov.iov_base = SCARG(uap, buf);
1910 		aiov.iov_len = SCARG(uap, count);
1911 		auio.uio_iov = &aiov;
1912 		auio.uio_iovcnt = 1;
1913 		auio.uio_offset = 0;
1914 		auio.uio_rw = UIO_READ;
1915 		auio.uio_segflg = UIO_USERSPACE;
1916 		auio.uio_procp = p;
1917 		auio.uio_resid = SCARG(uap, count);
1918 		error = VOP_READLINK(vp, &auio, p->p_ucred);
1919 	}
1920 	vput(vp);
1921 	p->p_retval[0] = SCARG(uap, count) - auio.uio_resid;
1922 	return (error);
1923 }
1924 
1925 static int
1926 setfflags(p, vp, flags)
1927 	struct proc *p;
1928 	struct vnode *vp;
1929 	int flags;
1930 {
1931 	int error;
1932 	struct vattr vattr;
1933 
1934 	/*
1935 	 * Prevent non-root users from setting flags on devices.  When
1936 	 * a device is reused, users can retain ownership of the device
1937 	 * if they are allowed to set flags and programs assume that
1938 	 * chown can't fail when done as root.
1939 	 */
1940 	if ((vp->v_type == VCHR || vp->v_type == VBLK) &&
1941 	    ((error = suser_xxx(p->p_ucred, p, PRISON_ROOT)) != 0))
1942 		return (error);
1943 
1944 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1945 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
1946 	VATTR_NULL(&vattr);
1947 	vattr.va_flags = flags;
1948 	error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1949 	VOP_UNLOCK(vp, 0, p);
1950 	return (error);
1951 }
1952 
1953 /*
1954  * Change flags of a file given a path name.
1955  */
1956 #ifndef _SYS_SYSPROTO_H_
1957 struct chflags_args {
1958 	char	*path;
1959 	int	flags;
1960 };
1961 #endif
1962 /* ARGSUSED */
1963 int
1964 chflags(p, uap)
1965 	struct proc *p;
1966 	register struct chflags_args /* {
1967 		syscallarg(char *) path;
1968 		syscallarg(int) flags;
1969 	} */ *uap;
1970 {
1971 	int error;
1972 	struct nameidata nd;
1973 
1974 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1975 	if ((error = namei(&nd)) != 0)
1976 		return (error);
1977 	NDFREE(&nd, NDF_ONLY_PNBUF);
1978 	error = setfflags(p, nd.ni_vp, SCARG(uap, flags));
1979 	vrele(nd.ni_vp);
1980 	return error;
1981 }
1982 
1983 /*
1984  * Change flags of a file given a file descriptor.
1985  */
1986 #ifndef _SYS_SYSPROTO_H_
1987 struct fchflags_args {
1988 	int	fd;
1989 	int	flags;
1990 };
1991 #endif
1992 /* ARGSUSED */
1993 int
1994 fchflags(p, uap)
1995 	struct proc *p;
1996 	register struct fchflags_args /* {
1997 		syscallarg(int) fd;
1998 		syscallarg(int) flags;
1999 	} */ *uap;
2000 {
2001 	struct file *fp;
2002 	int error;
2003 
2004 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2005 		return (error);
2006 	return setfflags(p, (struct vnode *) fp->f_data, SCARG(uap, flags));
2007 }
2008 
2009 static int
2010 setfmode(p, vp, mode)
2011 	struct proc *p;
2012 	struct vnode *vp;
2013 	int mode;
2014 {
2015 	int error;
2016 	struct vattr vattr;
2017 
2018 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2019 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2020 	VATTR_NULL(&vattr);
2021 	vattr.va_mode = mode & ALLPERMS;
2022 	error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2023 	VOP_UNLOCK(vp, 0, p);
2024 	return error;
2025 }
2026 
2027 /*
2028  * Change mode of a file given path name.
2029  */
2030 #ifndef _SYS_SYSPROTO_H_
2031 struct chmod_args {
2032 	char	*path;
2033 	int	mode;
2034 };
2035 #endif
2036 /* ARGSUSED */
2037 int
2038 chmod(p, uap)
2039 	struct proc *p;
2040 	register struct chmod_args /* {
2041 		syscallarg(char *) path;
2042 		syscallarg(int) mode;
2043 	} */ *uap;
2044 {
2045 	int error;
2046 	struct nameidata nd;
2047 
2048 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2049 	if ((error = namei(&nd)) != 0)
2050 		return (error);
2051 	NDFREE(&nd, NDF_ONLY_PNBUF);
2052 	error = setfmode(p, nd.ni_vp, SCARG(uap, mode));
2053 	vrele(nd.ni_vp);
2054 	return error;
2055 }
2056 
2057 /*
2058  * Change mode of a file given path name (don't follow links.)
2059  */
2060 #ifndef _SYS_SYSPROTO_H_
2061 struct lchmod_args {
2062 	char	*path;
2063 	int	mode;
2064 };
2065 #endif
2066 /* ARGSUSED */
2067 int
2068 lchmod(p, uap)
2069 	struct proc *p;
2070 	register struct lchmod_args /* {
2071 		syscallarg(char *) path;
2072 		syscallarg(int) mode;
2073 	} */ *uap;
2074 {
2075 	int error;
2076 	struct nameidata nd;
2077 
2078 	NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2079 	if ((error = namei(&nd)) != 0)
2080 		return (error);
2081 	NDFREE(&nd, NDF_ONLY_PNBUF);
2082 	error = setfmode(p, nd.ni_vp, SCARG(uap, mode));
2083 	vrele(nd.ni_vp);
2084 	return error;
2085 }
2086 
2087 /*
2088  * Change mode of a file given a file descriptor.
2089  */
2090 #ifndef _SYS_SYSPROTO_H_
2091 struct fchmod_args {
2092 	int	fd;
2093 	int	mode;
2094 };
2095 #endif
2096 /* ARGSUSED */
2097 int
2098 fchmod(p, uap)
2099 	struct proc *p;
2100 	register struct fchmod_args /* {
2101 		syscallarg(int) fd;
2102 		syscallarg(int) mode;
2103 	} */ *uap;
2104 {
2105 	struct file *fp;
2106 	int error;
2107 
2108 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2109 		return (error);
2110 	return setfmode(p, (struct vnode *)fp->f_data, SCARG(uap, mode));
2111 }
2112 
2113 static int
2114 setfown(p, vp, uid, gid)
2115 	struct proc *p;
2116 	struct vnode *vp;
2117 	uid_t uid;
2118 	gid_t gid;
2119 {
2120 	int error;
2121 	struct vattr vattr;
2122 
2123 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2124 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2125 	VATTR_NULL(&vattr);
2126 	vattr.va_uid = uid;
2127 	vattr.va_gid = gid;
2128 	error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2129 	VOP_UNLOCK(vp, 0, p);
2130 	return error;
2131 }
2132 
2133 /*
2134  * Set ownership given a path name.
2135  */
2136 #ifndef _SYS_SYSPROTO_H_
2137 struct chown_args {
2138 	char	*path;
2139 	int	uid;
2140 	int	gid;
2141 };
2142 #endif
2143 /* ARGSUSED */
2144 int
2145 chown(p, uap)
2146 	struct proc *p;
2147 	register struct chown_args /* {
2148 		syscallarg(char *) path;
2149 		syscallarg(int) uid;
2150 		syscallarg(int) gid;
2151 	} */ *uap;
2152 {
2153 	int error;
2154 	struct nameidata nd;
2155 
2156 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2157 	if ((error = namei(&nd)) != 0)
2158 		return (error);
2159 	NDFREE(&nd, NDF_ONLY_PNBUF);
2160 	error = setfown(p, nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid));
2161 	vrele(nd.ni_vp);
2162 	return (error);
2163 }
2164 
2165 /*
2166  * Set ownership given a path name, do not cross symlinks.
2167  */
2168 #ifndef _SYS_SYSPROTO_H_
2169 struct lchown_args {
2170 	char	*path;
2171 	int	uid;
2172 	int	gid;
2173 };
2174 #endif
2175 /* ARGSUSED */
2176 int
2177 lchown(p, uap)
2178 	struct proc *p;
2179 	register struct lchown_args /* {
2180 		syscallarg(char *) path;
2181 		syscallarg(int) uid;
2182 		syscallarg(int) gid;
2183 	} */ *uap;
2184 {
2185 	int error;
2186 	struct nameidata nd;
2187 
2188 	NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2189 	if ((error = namei(&nd)) != 0)
2190 		return (error);
2191 	NDFREE(&nd, NDF_ONLY_PNBUF);
2192 	error = setfown(p, nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid));
2193 	vrele(nd.ni_vp);
2194 	return (error);
2195 }
2196 
2197 /*
2198  * Set ownership given a file descriptor.
2199  */
2200 #ifndef _SYS_SYSPROTO_H_
2201 struct fchown_args {
2202 	int	fd;
2203 	int	uid;
2204 	int	gid;
2205 };
2206 #endif
2207 /* ARGSUSED */
2208 int
2209 fchown(p, uap)
2210 	struct proc *p;
2211 	register struct fchown_args /* {
2212 		syscallarg(int) fd;
2213 		syscallarg(int) uid;
2214 		syscallarg(int) gid;
2215 	} */ *uap;
2216 {
2217 	struct file *fp;
2218 	int error;
2219 
2220 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2221 		return (error);
2222 	return setfown(p, (struct vnode *)fp->f_data,
2223 		SCARG(uap, uid), SCARG(uap, gid));
2224 }
2225 
2226 static int
2227 getutimes(usrtvp, tsp)
2228 	const struct timeval *usrtvp;
2229 	struct timespec *tsp;
2230 {
2231 	struct timeval tv[2];
2232 	int error;
2233 
2234 	if (usrtvp == NULL) {
2235 		microtime(&tv[0]);
2236 		TIMEVAL_TO_TIMESPEC(&tv[0], &tsp[0]);
2237 		tsp[1] = tsp[0];
2238 	} else {
2239 		if ((error = copyin(usrtvp, tv, sizeof (tv))) != 0)
2240 			return (error);
2241 		TIMEVAL_TO_TIMESPEC(&tv[0], &tsp[0]);
2242 		TIMEVAL_TO_TIMESPEC(&tv[1], &tsp[1]);
2243 	}
2244 	return 0;
2245 }
2246 
2247 static int
2248 setutimes(p, vp, ts, nullflag)
2249 	struct proc *p;
2250 	struct vnode *vp;
2251 	const struct timespec *ts;
2252 	int nullflag;
2253 {
2254 	int error;
2255 	struct vattr vattr;
2256 
2257 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2258 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2259 	VATTR_NULL(&vattr);
2260 	vattr.va_atime = ts[0];
2261 	vattr.va_mtime = ts[1];
2262 	if (nullflag)
2263 		vattr.va_vaflags |= VA_UTIMES_NULL;
2264 	error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2265 	VOP_UNLOCK(vp, 0, p);
2266 	return error;
2267 }
2268 
2269 /*
2270  * Set the access and modification times of a file.
2271  */
2272 #ifndef _SYS_SYSPROTO_H_
2273 struct utimes_args {
2274 	char	*path;
2275 	struct	timeval *tptr;
2276 };
2277 #endif
2278 /* ARGSUSED */
2279 int
2280 utimes(p, uap)
2281 	struct proc *p;
2282 	register struct utimes_args /* {
2283 		syscallarg(char *) path;
2284 		syscallarg(struct timeval *) tptr;
2285 	} */ *uap;
2286 {
2287 	struct timespec ts[2];
2288 	struct timeval *usrtvp;
2289 	int error;
2290 	struct nameidata nd;
2291 
2292 	usrtvp = SCARG(uap, tptr);
2293 	if ((error = getutimes(usrtvp, ts)) != 0)
2294 		return (error);
2295 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2296 	if ((error = namei(&nd)) != 0)
2297 		return (error);
2298 	NDFREE(&nd, NDF_ONLY_PNBUF);
2299 	error = setutimes(p, nd.ni_vp, ts, usrtvp == NULL);
2300 	vrele(nd.ni_vp);
2301 	return (error);
2302 }
2303 
2304 /*
2305  * Set the access and modification times of a file.
2306  */
2307 #ifndef _SYS_SYSPROTO_H_
2308 struct lutimes_args {
2309 	char	*path;
2310 	struct	timeval *tptr;
2311 };
2312 #endif
2313 /* ARGSUSED */
2314 int
2315 lutimes(p, uap)
2316 	struct proc *p;
2317 	register struct lutimes_args /* {
2318 		syscallarg(char *) path;
2319 		syscallarg(struct timeval *) tptr;
2320 	} */ *uap;
2321 {
2322 	struct timespec ts[2];
2323 	struct timeval *usrtvp;
2324 	int error;
2325 	struct nameidata nd;
2326 
2327 	usrtvp = SCARG(uap, tptr);
2328 	if ((error = getutimes(usrtvp, ts)) != 0)
2329 		return (error);
2330 	NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2331 	if ((error = namei(&nd)) != 0)
2332 		return (error);
2333 	NDFREE(&nd, NDF_ONLY_PNBUF);
2334 	error = setutimes(p, nd.ni_vp, ts, usrtvp == NULL);
2335 	vrele(nd.ni_vp);
2336 	return (error);
2337 }
2338 
2339 /*
2340  * Set the access and modification times of a file.
2341  */
2342 #ifndef _SYS_SYSPROTO_H_
2343 struct futimes_args {
2344 	int	fd;
2345 	struct	timeval *tptr;
2346 };
2347 #endif
2348 /* ARGSUSED */
2349 int
2350 futimes(p, uap)
2351 	struct proc *p;
2352 	register struct futimes_args /* {
2353 		syscallarg(int ) fd;
2354 		syscallarg(struct timeval *) tptr;
2355 	} */ *uap;
2356 {
2357 	struct timespec ts[2];
2358 	struct file *fp;
2359 	struct timeval *usrtvp;
2360 	int error;
2361 
2362 	usrtvp = SCARG(uap, tptr);
2363 	if ((error = getutimes(usrtvp, ts)) != 0)
2364 		return (error);
2365 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2366 		return (error);
2367 	return setutimes(p, (struct vnode *)fp->f_data, ts, usrtvp == NULL);
2368 }
2369 
2370 /*
2371  * Truncate a file given its path name.
2372  */
2373 #ifndef _SYS_SYSPROTO_H_
2374 struct truncate_args {
2375 	char	*path;
2376 	int	pad;
2377 	off_t	length;
2378 };
2379 #endif
2380 /* ARGSUSED */
2381 int
2382 truncate(p, uap)
2383 	struct proc *p;
2384 	register struct truncate_args /* {
2385 		syscallarg(char *) path;
2386 		syscallarg(int) pad;
2387 		syscallarg(off_t) length;
2388 	} */ *uap;
2389 {
2390 	register struct vnode *vp;
2391 	struct vattr vattr;
2392 	int error;
2393 	struct nameidata nd;
2394 
2395 	if (uap->length < 0)
2396 		return(EINVAL);
2397 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2398 	if ((error = namei(&nd)) != 0)
2399 		return (error);
2400 	vp = nd.ni_vp;
2401 	NDFREE(&nd, NDF_ONLY_PNBUF);
2402 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2403 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2404 	if (vp->v_type == VDIR)
2405 		error = EISDIR;
2406 	else if ((error = vn_writechk(vp)) == 0 &&
2407 	    (error = VOP_ACCESS(vp, VWRITE, p->p_ucred, p)) == 0) {
2408 		VATTR_NULL(&vattr);
2409 		vattr.va_size = SCARG(uap, length);
2410 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2411 	}
2412 	vput(vp);
2413 	return (error);
2414 }
2415 
2416 /*
2417  * Truncate a file given a file descriptor.
2418  */
2419 #ifndef _SYS_SYSPROTO_H_
2420 struct ftruncate_args {
2421 	int	fd;
2422 	int	pad;
2423 	off_t	length;
2424 };
2425 #endif
2426 /* ARGSUSED */
2427 int
2428 ftruncate(p, uap)
2429 	struct proc *p;
2430 	register struct ftruncate_args /* {
2431 		syscallarg(int) fd;
2432 		syscallarg(int) pad;
2433 		syscallarg(off_t) length;
2434 	} */ *uap;
2435 {
2436 	struct vattr vattr;
2437 	struct vnode *vp;
2438 	struct file *fp;
2439 	int error;
2440 
2441 	if (uap->length < 0)
2442 		return(EINVAL);
2443 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2444 		return (error);
2445 	if ((fp->f_flag & FWRITE) == 0)
2446 		return (EINVAL);
2447 	vp = (struct vnode *)fp->f_data;
2448 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2449 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2450 	if (vp->v_type == VDIR)
2451 		error = EISDIR;
2452 	else if ((error = vn_writechk(vp)) == 0) {
2453 		VATTR_NULL(&vattr);
2454 		vattr.va_size = SCARG(uap, length);
2455 		error = VOP_SETATTR(vp, &vattr, fp->f_cred, p);
2456 	}
2457 	VOP_UNLOCK(vp, 0, p);
2458 	return (error);
2459 }
2460 
2461 #if defined(COMPAT_43) || defined(COMPAT_SUNOS)
2462 /*
2463  * Truncate a file given its path name.
2464  */
2465 #ifndef _SYS_SYSPROTO_H_
2466 struct otruncate_args {
2467 	char	*path;
2468 	long	length;
2469 };
2470 #endif
2471 /* ARGSUSED */
2472 int
2473 otruncate(p, uap)
2474 	struct proc *p;
2475 	register struct otruncate_args /* {
2476 		syscallarg(char *) path;
2477 		syscallarg(long) length;
2478 	} */ *uap;
2479 {
2480 	struct truncate_args /* {
2481 		syscallarg(char *) path;
2482 		syscallarg(int) pad;
2483 		syscallarg(off_t) length;
2484 	} */ nuap;
2485 
2486 	SCARG(&nuap, path) = SCARG(uap, path);
2487 	SCARG(&nuap, length) = SCARG(uap, length);
2488 	return (truncate(p, &nuap));
2489 }
2490 
2491 /*
2492  * Truncate a file given a file descriptor.
2493  */
2494 #ifndef _SYS_SYSPROTO_H_
2495 struct oftruncate_args {
2496 	int	fd;
2497 	long	length;
2498 };
2499 #endif
2500 /* ARGSUSED */
2501 int
2502 oftruncate(p, uap)
2503 	struct proc *p;
2504 	register struct oftruncate_args /* {
2505 		syscallarg(int) fd;
2506 		syscallarg(long) length;
2507 	} */ *uap;
2508 {
2509 	struct ftruncate_args /* {
2510 		syscallarg(int) fd;
2511 		syscallarg(int) pad;
2512 		syscallarg(off_t) length;
2513 	} */ nuap;
2514 
2515 	SCARG(&nuap, fd) = SCARG(uap, fd);
2516 	SCARG(&nuap, length) = SCARG(uap, length);
2517 	return (ftruncate(p, &nuap));
2518 }
2519 #endif /* COMPAT_43 || COMPAT_SUNOS */
2520 
2521 /*
2522  * Sync an open file.
2523  */
2524 #ifndef _SYS_SYSPROTO_H_
2525 struct fsync_args {
2526 	int	fd;
2527 };
2528 #endif
2529 /* ARGSUSED */
2530 int
2531 fsync(p, uap)
2532 	struct proc *p;
2533 	struct fsync_args /* {
2534 		syscallarg(int) fd;
2535 	} */ *uap;
2536 {
2537 	register struct vnode *vp;
2538 	struct file *fp;
2539 	int error;
2540 
2541 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2542 		return (error);
2543 	vp = (struct vnode *)fp->f_data;
2544 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2545 	if (vp->v_object)
2546 		vm_object_page_clean(vp->v_object, 0, 0, 0);
2547 	if ((error = VOP_FSYNC(vp, fp->f_cred, MNT_WAIT, p)) == 0 &&
2548 	    vp->v_mount && (vp->v_mount->mnt_flag & MNT_SOFTDEP) &&
2549 	    bioops.io_fsync)
2550 		error = (*bioops.io_fsync)(vp);
2551 	VOP_UNLOCK(vp, 0, p);
2552 	return (error);
2553 }
2554 
2555 /*
2556  * Rename files.  Source and destination must either both be directories,
2557  * or both not be directories.  If target is a directory, it must be empty.
2558  */
2559 #ifndef _SYS_SYSPROTO_H_
2560 struct rename_args {
2561 	char	*from;
2562 	char	*to;
2563 };
2564 #endif
2565 /* ARGSUSED */
2566 int
2567 rename(p, uap)
2568 	struct proc *p;
2569 	register struct rename_args /* {
2570 		syscallarg(char *) from;
2571 		syscallarg(char *) to;
2572 	} */ *uap;
2573 {
2574 	register struct vnode *tvp, *fvp, *tdvp;
2575 	struct nameidata fromnd, tond;
2576 	int error;
2577 
2578 	bwillwrite();
2579 	NDINIT(&fromnd, DELETE, WANTPARENT | SAVESTART, UIO_USERSPACE,
2580 	    SCARG(uap, from), p);
2581 	if ((error = namei(&fromnd)) != 0)
2582 		return (error);
2583 	fvp = fromnd.ni_vp;
2584 	NDINIT(&tond, RENAME, LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART | NOOBJ,
2585 	    UIO_USERSPACE, SCARG(uap, to), p);
2586 	if (fromnd.ni_vp->v_type == VDIR)
2587 		tond.ni_cnd.cn_flags |= WILLBEDIR;
2588 	if ((error = namei(&tond)) != 0) {
2589 		/* Translate error code for rename("dir1", "dir2/."). */
2590 		if (error == EISDIR && fvp->v_type == VDIR)
2591 			error = EINVAL;
2592 		NDFREE(&fromnd, NDF_ONLY_PNBUF);
2593 		vrele(fromnd.ni_dvp);
2594 		vrele(fvp);
2595 		goto out1;
2596 	}
2597 	tdvp = tond.ni_dvp;
2598 	tvp = tond.ni_vp;
2599 	if (tvp != NULL) {
2600 		if (fvp->v_type == VDIR && tvp->v_type != VDIR) {
2601 			error = ENOTDIR;
2602 			goto out;
2603 		} else if (fvp->v_type != VDIR && tvp->v_type == VDIR) {
2604 			error = EISDIR;
2605 			goto out;
2606 		}
2607 	}
2608 	if (fvp == tdvp)
2609 		error = EINVAL;
2610 	/*
2611 	 * If source is the same as the destination (that is the
2612 	 * same inode number with the same name in the same directory),
2613 	 * then there is nothing to do.
2614 	 */
2615 	if (fvp == tvp && fromnd.ni_dvp == tdvp &&
2616 	    fromnd.ni_cnd.cn_namelen == tond.ni_cnd.cn_namelen &&
2617 	    !bcmp(fromnd.ni_cnd.cn_nameptr, tond.ni_cnd.cn_nameptr,
2618 	      fromnd.ni_cnd.cn_namelen))
2619 		error = -1;
2620 out:
2621 	if (!error) {
2622 		VOP_LEASE(tdvp, p, p->p_ucred, LEASE_WRITE);
2623 		if (fromnd.ni_dvp != tdvp) {
2624 			VOP_LEASE(fromnd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
2625 		}
2626 		if (tvp) {
2627 			VOP_LEASE(tvp, p, p->p_ucred, LEASE_WRITE);
2628 		}
2629 		error = VOP_RENAME(fromnd.ni_dvp, fromnd.ni_vp, &fromnd.ni_cnd,
2630 				   tond.ni_dvp, tond.ni_vp, &tond.ni_cnd);
2631 		NDFREE(&fromnd, NDF_ONLY_PNBUF);
2632 		NDFREE(&tond, NDF_ONLY_PNBUF);
2633 	} else {
2634 		NDFREE(&fromnd, NDF_ONLY_PNBUF);
2635 		NDFREE(&tond, NDF_ONLY_PNBUF);
2636 		if (tdvp == tvp)
2637 			vrele(tdvp);
2638 		else
2639 			vput(tdvp);
2640 		if (tvp)
2641 			vput(tvp);
2642 		vrele(fromnd.ni_dvp);
2643 		vrele(fvp);
2644 	}
2645 	vrele(tond.ni_startdir);
2646 	ASSERT_VOP_UNLOCKED(fromnd.ni_dvp, "rename");
2647 	ASSERT_VOP_UNLOCKED(fromnd.ni_vp, "rename");
2648 	ASSERT_VOP_UNLOCKED(tond.ni_dvp, "rename");
2649 	ASSERT_VOP_UNLOCKED(tond.ni_vp, "rename");
2650 out1:
2651 	if (fromnd.ni_startdir)
2652 		vrele(fromnd.ni_startdir);
2653 	if (error == -1)
2654 		return (0);
2655 	return (error);
2656 }
2657 
2658 /*
2659  * Make a directory file.
2660  */
2661 #ifndef _SYS_SYSPROTO_H_
2662 struct mkdir_args {
2663 	char	*path;
2664 	int	mode;
2665 };
2666 #endif
2667 /* ARGSUSED */
2668 int
2669 mkdir(p, uap)
2670 	struct proc *p;
2671 	register struct mkdir_args /* {
2672 		syscallarg(char *) path;
2673 		syscallarg(int) mode;
2674 	} */ *uap;
2675 {
2676 	register struct vnode *vp;
2677 	struct vattr vattr;
2678 	int error;
2679 	struct nameidata nd;
2680 
2681 	bwillwrite();
2682 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
2683 	nd.ni_cnd.cn_flags |= WILLBEDIR;
2684 	if ((error = namei(&nd)) != 0)
2685 		return (error);
2686 	vp = nd.ni_vp;
2687 	if (vp != NULL) {
2688 		NDFREE(&nd, NDF_ONLY_PNBUF);
2689 		if (nd.ni_dvp == vp)
2690 			vrele(nd.ni_dvp);
2691 		else
2692 			vput(nd.ni_dvp);
2693 		vrele(vp);
2694 		return (EEXIST);
2695 	}
2696 	VATTR_NULL(&vattr);
2697 	vattr.va_type = VDIR;
2698 	vattr.va_mode = (SCARG(uap, mode) & ACCESSPERMS) &~ p->p_fd->fd_cmask;
2699 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
2700 	error = VOP_MKDIR(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
2701 	NDFREE(&nd, NDF_ONLY_PNBUF);
2702 	vput(nd.ni_dvp);
2703 	if (!error)
2704 		vput(nd.ni_vp);
2705 	ASSERT_VOP_UNLOCKED(nd.ni_dvp, "mkdir");
2706 	ASSERT_VOP_UNLOCKED(nd.ni_vp, "mkdir");
2707 	return (error);
2708 }
2709 
2710 /*
2711  * Remove a directory file.
2712  */
2713 #ifndef _SYS_SYSPROTO_H_
2714 struct rmdir_args {
2715 	char	*path;
2716 };
2717 #endif
2718 /* ARGSUSED */
2719 int
2720 rmdir(p, uap)
2721 	struct proc *p;
2722 	struct rmdir_args /* {
2723 		syscallarg(char *) path;
2724 	} */ *uap;
2725 {
2726 	register struct vnode *vp;
2727 	int error;
2728 	struct nameidata nd;
2729 
2730 	bwillwrite();
2731 	NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
2732 	    SCARG(uap, path), p);
2733 	if ((error = namei(&nd)) != 0)
2734 		return (error);
2735 	vp = nd.ni_vp;
2736 	if (vp->v_type != VDIR) {
2737 		error = ENOTDIR;
2738 		goto out;
2739 	}
2740 	/*
2741 	 * No rmdir "." please.
2742 	 */
2743 	if (nd.ni_dvp == vp) {
2744 		error = EINVAL;
2745 		goto out;
2746 	}
2747 	/*
2748 	 * The root of a mounted filesystem cannot be deleted.
2749 	 */
2750 	if (vp->v_flag & VROOT)
2751 		error = EBUSY;
2752 	else {
2753 		VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
2754 		VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2755 		error = VOP_RMDIR(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
2756 	}
2757 out:
2758 	NDFREE(&nd, NDF_ONLY_PNBUF);
2759 	if (nd.ni_dvp == vp)
2760 		vrele(nd.ni_dvp);
2761 	else
2762 		vput(nd.ni_dvp);
2763 	if (vp != NULLVP)
2764 		vput(vp);
2765 	ASSERT_VOP_UNLOCKED(nd.ni_dvp, "rmdir");
2766 	ASSERT_VOP_UNLOCKED(nd.ni_vp, "rmdir");
2767 	return (error);
2768 }
2769 
2770 #ifdef COMPAT_43
2771 /*
2772  * Read a block of directory entries in a file system independent format.
2773  */
2774 #ifndef _SYS_SYSPROTO_H_
2775 struct ogetdirentries_args {
2776 	int	fd;
2777 	char	*buf;
2778 	u_int	count;
2779 	long	*basep;
2780 };
2781 #endif
2782 int
2783 ogetdirentries(p, uap)
2784 	struct proc *p;
2785 	register struct ogetdirentries_args /* {
2786 		syscallarg(int) fd;
2787 		syscallarg(char *) buf;
2788 		syscallarg(u_int) count;
2789 		syscallarg(long *) basep;
2790 	} */ *uap;
2791 {
2792 	struct vnode *vp;
2793 	struct file *fp;
2794 	struct uio auio, kuio;
2795 	struct iovec aiov, kiov;
2796 	struct dirent *dp, *edp;
2797 	caddr_t dirbuf;
2798 	int error, eofflag, readcnt;
2799 	long loff;
2800 
2801 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2802 		return (error);
2803 	if ((fp->f_flag & FREAD) == 0)
2804 		return (EBADF);
2805 	vp = (struct vnode *)fp->f_data;
2806 unionread:
2807 	if (vp->v_type != VDIR)
2808 		return (EINVAL);
2809 	aiov.iov_base = SCARG(uap, buf);
2810 	aiov.iov_len = SCARG(uap, count);
2811 	auio.uio_iov = &aiov;
2812 	auio.uio_iovcnt = 1;
2813 	auio.uio_rw = UIO_READ;
2814 	auio.uio_segflg = UIO_USERSPACE;
2815 	auio.uio_procp = p;
2816 	auio.uio_resid = SCARG(uap, count);
2817 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2818 	loff = auio.uio_offset = fp->f_offset;
2819 #	if (BYTE_ORDER != LITTLE_ENDIAN)
2820 		if (vp->v_mount->mnt_maxsymlinklen <= 0) {
2821 			error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag,
2822 			    NULL, NULL);
2823 			fp->f_offset = auio.uio_offset;
2824 		} else
2825 #	endif
2826 	{
2827 		kuio = auio;
2828 		kuio.uio_iov = &kiov;
2829 		kuio.uio_segflg = UIO_SYSSPACE;
2830 		kiov.iov_len = SCARG(uap, count);
2831 		MALLOC(dirbuf, caddr_t, SCARG(uap, count), M_TEMP, M_WAITOK);
2832 		kiov.iov_base = dirbuf;
2833 		error = VOP_READDIR(vp, &kuio, fp->f_cred, &eofflag,
2834 			    NULL, NULL);
2835 		fp->f_offset = kuio.uio_offset;
2836 		if (error == 0) {
2837 			readcnt = SCARG(uap, count) - kuio.uio_resid;
2838 			edp = (struct dirent *)&dirbuf[readcnt];
2839 			for (dp = (struct dirent *)dirbuf; dp < edp; ) {
2840 #				if (BYTE_ORDER == LITTLE_ENDIAN)
2841 					/*
2842 					 * The expected low byte of
2843 					 * dp->d_namlen is our dp->d_type.
2844 					 * The high MBZ byte of dp->d_namlen
2845 					 * is our dp->d_namlen.
2846 					 */
2847 					dp->d_type = dp->d_namlen;
2848 					dp->d_namlen = 0;
2849 #				else
2850 					/*
2851 					 * The dp->d_type is the high byte
2852 					 * of the expected dp->d_namlen,
2853 					 * so must be zero'ed.
2854 					 */
2855 					dp->d_type = 0;
2856 #				endif
2857 				if (dp->d_reclen > 0) {
2858 					dp = (struct dirent *)
2859 					    ((char *)dp + dp->d_reclen);
2860 				} else {
2861 					error = EIO;
2862 					break;
2863 				}
2864 			}
2865 			if (dp >= edp)
2866 				error = uiomove(dirbuf, readcnt, &auio);
2867 		}
2868 		FREE(dirbuf, M_TEMP);
2869 	}
2870 	VOP_UNLOCK(vp, 0, p);
2871 	if (error)
2872 		return (error);
2873 	if (SCARG(uap, count) == auio.uio_resid) {
2874 		if (union_dircheckp) {
2875 			error = union_dircheckp(p, &vp, fp);
2876 			if (error == -1)
2877 				goto unionread;
2878 			if (error)
2879 				return (error);
2880 		}
2881 		if ((vp->v_flag & VROOT) &&
2882 		    (vp->v_mount->mnt_flag & MNT_UNION)) {
2883 			struct vnode *tvp = vp;
2884 			vp = vp->v_mount->mnt_vnodecovered;
2885 			VREF(vp);
2886 			fp->f_data = (caddr_t) vp;
2887 			fp->f_offset = 0;
2888 			vrele(tvp);
2889 			goto unionread;
2890 		}
2891 	}
2892 	error = copyout((caddr_t)&loff, (caddr_t)SCARG(uap, basep),
2893 	    sizeof(long));
2894 	p->p_retval[0] = SCARG(uap, count) - auio.uio_resid;
2895 	return (error);
2896 }
2897 #endif /* COMPAT_43 */
2898 
2899 /*
2900  * Read a block of directory entries in a file system independent format.
2901  */
2902 #ifndef _SYS_SYSPROTO_H_
2903 struct getdirentries_args {
2904 	int	fd;
2905 	char	*buf;
2906 	u_int	count;
2907 	long	*basep;
2908 };
2909 #endif
2910 int
2911 getdirentries(p, uap)
2912 	struct proc *p;
2913 	register struct getdirentries_args /* {
2914 		syscallarg(int) fd;
2915 		syscallarg(char *) buf;
2916 		syscallarg(u_int) count;
2917 		syscallarg(long *) basep;
2918 	} */ *uap;
2919 {
2920 	struct vnode *vp;
2921 	struct file *fp;
2922 	struct uio auio;
2923 	struct iovec aiov;
2924 	long loff;
2925 	int error, eofflag;
2926 
2927 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2928 		return (error);
2929 	if ((fp->f_flag & FREAD) == 0)
2930 		return (EBADF);
2931 	vp = (struct vnode *)fp->f_data;
2932 unionread:
2933 	if (vp->v_type != VDIR)
2934 		return (EINVAL);
2935 	aiov.iov_base = SCARG(uap, buf);
2936 	aiov.iov_len = SCARG(uap, count);
2937 	auio.uio_iov = &aiov;
2938 	auio.uio_iovcnt = 1;
2939 	auio.uio_rw = UIO_READ;
2940 	auio.uio_segflg = UIO_USERSPACE;
2941 	auio.uio_procp = p;
2942 	auio.uio_resid = SCARG(uap, count);
2943 	/* vn_lock(vp, LK_SHARED | LK_RETRY, p); */
2944 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2945 	loff = auio.uio_offset = fp->f_offset;
2946 	error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, NULL, NULL);
2947 	fp->f_offset = auio.uio_offset;
2948 	VOP_UNLOCK(vp, 0, p);
2949 	if (error)
2950 		return (error);
2951 	if (SCARG(uap, count) == auio.uio_resid) {
2952 		if (union_dircheckp) {
2953 			error = union_dircheckp(p, &vp, fp);
2954 			if (error == -1)
2955 				goto unionread;
2956 			if (error)
2957 				return (error);
2958 		}
2959 		if ((vp->v_flag & VROOT) &&
2960 		    (vp->v_mount->mnt_flag & MNT_UNION)) {
2961 			struct vnode *tvp = vp;
2962 			vp = vp->v_mount->mnt_vnodecovered;
2963 			VREF(vp);
2964 			fp->f_data = (caddr_t) vp;
2965 			fp->f_offset = 0;
2966 			vrele(tvp);
2967 			goto unionread;
2968 		}
2969 	}
2970 	if (SCARG(uap, basep) != NULL) {
2971 		error = copyout((caddr_t)&loff, (caddr_t)SCARG(uap, basep),
2972 		    sizeof(long));
2973 	}
2974 	p->p_retval[0] = SCARG(uap, count) - auio.uio_resid;
2975 	return (error);
2976 }
2977 #ifndef _SYS_SYSPROTO_H_
2978 struct getdents_args {
2979 	int fd;
2980 	char *buf;
2981 	size_t count;
2982 };
2983 #endif
2984 int
2985 getdents(p, uap)
2986 	struct proc *p;
2987 	register struct getdents_args /* {
2988 		syscallarg(int) fd;
2989 		syscallarg(char *) buf;
2990 		syscallarg(u_int) count;
2991 	} */ *uap;
2992 {
2993 	struct getdirentries_args ap;
2994 	ap.fd = uap->fd;
2995 	ap.buf = uap->buf;
2996 	ap.count = uap->count;
2997 	ap.basep = NULL;
2998 	return getdirentries(p, &ap);
2999 }
3000 
3001 /*
3002  * Set the mode mask for creation of filesystem nodes.
3003  *
3004  * MP SAFE
3005  */
3006 #ifndef _SYS_SYSPROTO_H_
3007 struct umask_args {
3008 	int	newmask;
3009 };
3010 #endif
3011 int
3012 umask(p, uap)
3013 	struct proc *p;
3014 	struct umask_args /* {
3015 		syscallarg(int) newmask;
3016 	} */ *uap;
3017 {
3018 	register struct filedesc *fdp;
3019 
3020 	fdp = p->p_fd;
3021 	p->p_retval[0] = fdp->fd_cmask;
3022 	fdp->fd_cmask = SCARG(uap, newmask) & ALLPERMS;
3023 	return (0);
3024 }
3025 
3026 /*
3027  * Void all references to file by ripping underlying filesystem
3028  * away from vnode.
3029  */
3030 #ifndef _SYS_SYSPROTO_H_
3031 struct revoke_args {
3032 	char	*path;
3033 };
3034 #endif
3035 /* ARGSUSED */
3036 int
3037 revoke(p, uap)
3038 	struct proc *p;
3039 	register struct revoke_args /* {
3040 		syscallarg(char *) path;
3041 	} */ *uap;
3042 {
3043 	register struct vnode *vp;
3044 	struct vattr vattr;
3045 	int error;
3046 	struct nameidata nd;
3047 
3048 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
3049 	if ((error = namei(&nd)) != 0)
3050 		return (error);
3051 	vp = nd.ni_vp;
3052 	NDFREE(&nd, NDF_ONLY_PNBUF);
3053 	if (vp->v_type != VCHR && vp->v_type != VBLK) {
3054 		error = EINVAL;
3055 		goto out;
3056 	}
3057 	if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)) != 0)
3058 		goto out;
3059 	if (p->p_ucred->cr_uid != vattr.va_uid &&
3060 	    (error = suser_xxx(0, p, PRISON_ROOT)))
3061 		goto out;
3062 	if (vcount(vp) > 1)
3063 		VOP_REVOKE(vp, REVOKEALL);
3064 out:
3065 	vrele(vp);
3066 	return (error);
3067 }
3068 
3069 /*
3070  * Convert a user file descriptor to a kernel file entry.
3071  */
3072 int
3073 getvnode(fdp, fd, fpp)
3074 	struct filedesc *fdp;
3075 	int fd;
3076 	struct file **fpp;
3077 {
3078 	struct file *fp;
3079 
3080 	if ((u_int)fd >= fdp->fd_nfiles ||
3081 	    (fp = fdp->fd_ofiles[fd]) == NULL)
3082 		return (EBADF);
3083 	if (fp->f_type != DTYPE_VNODE && fp->f_type != DTYPE_FIFO)
3084 		return (EINVAL);
3085 	*fpp = fp;
3086 	return (0);
3087 }
3088 /*
3089  * Get (NFS) file handle
3090  */
3091 #ifndef _SYS_SYSPROTO_H_
3092 struct getfh_args {
3093 	char	*fname;
3094 	fhandle_t *fhp;
3095 };
3096 #endif
3097 int
3098 getfh(p, uap)
3099 	struct proc *p;
3100 	register struct getfh_args *uap;
3101 {
3102 	struct nameidata nd;
3103 	fhandle_t fh;
3104 	register struct vnode *vp;
3105 	int error;
3106 
3107 	/*
3108 	 * Must be super user
3109 	 */
3110 	error = suser(p);
3111 	if (error)
3112 		return (error);
3113 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE, uap->fname, p);
3114 	error = namei(&nd);
3115 	if (error)
3116 		return (error);
3117 	NDFREE(&nd, NDF_ONLY_PNBUF);
3118 	vp = nd.ni_vp;
3119 	bzero(&fh, sizeof(fh));
3120 	fh.fh_fsid = vp->v_mount->mnt_stat.f_fsid;
3121 	error = VFS_VPTOFH(vp, &fh.fh_fid);
3122 	vput(vp);
3123 	if (error)
3124 		return (error);
3125 	error = copyout(&fh, uap->fhp, sizeof (fh));
3126 	return (error);
3127 }
3128 
3129 /*
3130  * syscall for the rpc.lockd to use to translate a NFS file handle into
3131  * an open descriptor.
3132  *
3133  * warning: do not remove the suser() call or this becomes one giant
3134  * security hole.
3135  */
3136 #ifndef _SYS_SYSPROTO_H_
3137 struct fhopen_args {
3138 	const struct fhandle *u_fhp;
3139 	int flags;
3140 };
3141 #endif
3142 int
3143 fhopen(p, uap)
3144 	struct proc *p;
3145 	struct fhopen_args /* {
3146 		syscallarg(const struct fhandle *) u_fhp;
3147 		syscallarg(int) flags;
3148 	} */ *uap;
3149 {
3150 	struct mount *mp;
3151 	struct vnode *vp;
3152 	struct fhandle fhp;
3153 	struct vattr vat;
3154 	struct vattr *vap = &vat;
3155 	struct flock lf;
3156 	struct file *fp;
3157 	register struct filedesc *fdp = p->p_fd;
3158 	int fmode, mode, error, type;
3159 	struct file *nfp;
3160 	int indx;
3161 
3162 	/*
3163 	 * Must be super user
3164 	 */
3165 	error = suser(p);
3166 	if (error)
3167 		return (error);
3168 
3169 	fmode = FFLAGS(SCARG(uap, flags));
3170 	/* why not allow a non-read/write open for our lockd? */
3171 	if (((fmode & (FREAD | FWRITE)) == 0) || (fmode & O_CREAT))
3172 		return (EINVAL);
3173 	error = copyin(SCARG(uap,u_fhp), &fhp, sizeof(fhp));
3174 	if (error)
3175 		return(error);
3176 	/* find the mount point */
3177 	mp = vfs_getvfs(&fhp.fh_fsid);
3178 	if (mp == NULL)
3179 		return (ESTALE);
3180 	/* now give me my vnode, it gets returned to me locked */
3181 	error = VFS_FHTOVP(mp, &fhp.fh_fid, &vp);
3182 	if (error)
3183 		return (error);
3184  	/*
3185 	 * from now on we have to make sure not
3186 	 * to forget about the vnode
3187 	 * any error that causes an abort must vput(vp)
3188 	 * just set error = err and 'goto bad;'.
3189 	 */
3190 
3191 	/*
3192 	 * from vn_open
3193 	 */
3194 	if (vp->v_type == VLNK) {
3195 		error = EMLINK;
3196 		goto bad;
3197 	}
3198 	if (vp->v_type == VSOCK) {
3199 		error = EOPNOTSUPP;
3200 		goto bad;
3201 	}
3202 	mode = 0;
3203 	if (fmode & (FWRITE | O_TRUNC)) {
3204 		if (vp->v_type == VDIR) {
3205 			error = EISDIR;
3206 			goto bad;
3207 		}
3208 		error = vn_writechk(vp);
3209 		if (error)
3210 			goto bad;
3211 		mode |= VWRITE;
3212 	}
3213 	if (fmode & FREAD)
3214 		mode |= VREAD;
3215 	if (mode) {
3216 		error = VOP_ACCESS(vp, mode, p->p_ucred, p);
3217 		if (error)
3218 			goto bad;
3219 	}
3220 	if (fmode & O_TRUNC) {
3221 		VOP_UNLOCK(vp, 0, p);				/* XXX */
3222 		VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
3223 		vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);	/* XXX */
3224 		VATTR_NULL(vap);
3225 		vap->va_size = 0;
3226 		error = VOP_SETATTR(vp, vap, p->p_ucred, p);
3227 		if (error)
3228 			goto bad;
3229 	}
3230 	error = VOP_OPEN(vp, fmode, p->p_ucred, p);
3231 	if (error)
3232 		goto bad;
3233 	/*
3234 	 * Make sure that a VM object is created for VMIO support.
3235 	 */
3236 	if (vn_canvmio(vp) == TRUE) {
3237 		if ((error = vfs_object_create(vp, p, p->p_ucred)) != 0)
3238 			goto bad;
3239 	}
3240 	if (fmode & FWRITE)
3241 		vp->v_writecount++;
3242 
3243 	/*
3244 	 * end of vn_open code
3245 	 */
3246 
3247 	if ((error = falloc(p, &nfp, &indx)) != 0)
3248 		goto bad;
3249 	fp = nfp;
3250 	nfp->f_data = (caddr_t)vp;
3251 	nfp->f_flag = fmode & FMASK;
3252 	nfp->f_ops = &vnops;
3253 	nfp->f_type = DTYPE_VNODE;
3254 	if (fmode & (O_EXLOCK | O_SHLOCK)) {
3255 		lf.l_whence = SEEK_SET;
3256 		lf.l_start = 0;
3257 		lf.l_len = 0;
3258 		if (fmode & O_EXLOCK)
3259 			lf.l_type = F_WRLCK;
3260 		else
3261 			lf.l_type = F_RDLCK;
3262 		type = F_FLOCK;
3263 		if ((fmode & FNONBLOCK) == 0)
3264 			type |= F_WAIT;
3265 		VOP_UNLOCK(vp, 0, p);
3266 		if ((error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type)) != 0) {
3267 			(void) vn_close(vp, fp->f_flag, fp->f_cred, p);
3268 			ffree(fp);
3269 			fdp->fd_ofiles[indx] = NULL;
3270 			return (error);
3271 		}
3272 		vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
3273 		fp->f_flag |= FHASLOCK;
3274 	}
3275 	if ((vp->v_type == VREG) && (vp->v_object == NULL))
3276 		vfs_object_create(vp, p, p->p_ucred);
3277 
3278 	VOP_UNLOCK(vp, 0, p);
3279 	p->p_retval[0] = indx;
3280 	return (0);
3281 
3282 bad:
3283 	vput(vp);
3284 	return (error);
3285 }
3286 
3287 #ifndef _SYS_SYSPROTO_H_
3288 struct fhstat_args {
3289 	struct fhandle *u_fhp;
3290 	struct stat *sb;
3291 };
3292 #endif
3293 int
3294 fhstat(p, uap)
3295 	struct proc *p;
3296 	register struct fhstat_args /* {
3297 		syscallarg(struct fhandle *) u_fhp;
3298 		syscallarg(struct stat *) sb;
3299 	} */ *uap;
3300 {
3301 	struct stat sb;
3302 	fhandle_t fh;
3303 	struct mount *mp;
3304 	struct vnode *vp;
3305 	int error;
3306 
3307 	/*
3308 	 * Must be super user
3309 	 */
3310 	error = suser(p);
3311 	if (error)
3312 		return (error);
3313 
3314 	error = copyin(SCARG(uap, u_fhp), &fh, sizeof(fhandle_t));
3315 	if (error)
3316 		return (error);
3317 
3318 	if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL)
3319 		return (ESTALE);
3320 	if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)))
3321 		return (error);
3322 	error = vn_stat(vp, &sb, p);
3323 	vput(vp);
3324 	if (error)
3325 		return (error);
3326 	error = copyout(&sb, SCARG(uap, sb), sizeof(sb));
3327 	return (error);
3328 }
3329 
3330 #ifndef _SYS_SYSPROTO_H_
3331 struct fhstatfs_args {
3332 	struct fhandle *u_fhp;
3333 	struct statfs *buf;
3334 };
3335 #endif
3336 int
3337 fhstatfs(p, uap)
3338 	struct proc *p;
3339 	struct fhstatfs_args /* {
3340 		syscallarg(struct fhandle) *u_fhp;
3341 		syscallarg(struct statfs) *buf;
3342 	} */ *uap;
3343 {
3344 	struct statfs *sp;
3345 	struct mount *mp;
3346 	struct vnode *vp;
3347 	struct statfs sb;
3348 	fhandle_t fh;
3349 	int error;
3350 
3351 	/*
3352 	 * Must be super user
3353 	 */
3354 	if ((error = suser(p)))
3355 		return (error);
3356 
3357 	if ((error = copyin(SCARG(uap, u_fhp), &fh, sizeof(fhandle_t))) != 0)
3358 		return (error);
3359 
3360 	if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL)
3361 		return (ESTALE);
3362 	if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)))
3363 		return (error);
3364 	mp = vp->v_mount;
3365 	sp = &mp->mnt_stat;
3366 	vput(vp);
3367 	if ((error = VFS_STATFS(mp, sp, p)) != 0)
3368 		return (error);
3369 	sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
3370 	if (suser_xxx(p->p_ucred, 0, 0)) {
3371 		bcopy((caddr_t)sp, (caddr_t)&sb, sizeof(sb));
3372 		sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0;
3373 		sp = &sb;
3374 	}
3375 	return (copyout(sp, SCARG(uap, buf), sizeof(*sp)));
3376 }
3377 
3378 /*
3379  * Syscall to push extended attribute configuration information into the
3380  * VFS.  Accepts a path, which it converts to a mountpoint, as well as
3381  * a command (int cmd), and attribute name and misc data.  For now, the
3382  * attribute name is left in userspace for consumption by the VFS_op.
3383  * It will probably be changed to be copied into sysspace by the
3384  * syscall in the future, once issues with various consumers of the
3385  * attribute code have raised their hands.
3386  *
3387  * Currently this is used only by UFS Extended Attributes.
3388  */
3389 int
3390 extattrctl(p, uap)
3391 	struct proc *p;
3392 	struct extattrctl_args *uap;
3393 {
3394 	struct nameidata nd;
3395 	struct mount *mp;
3396 	int error;
3397 
3398 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
3399 	if ((error = namei(&nd)) != 0)
3400 		return (error);
3401 	mp = nd.ni_vp->v_mount;
3402 	NDFREE(&nd, 0);
3403 	return (VFS_EXTATTRCTL(mp, SCARG(uap, cmd), SCARG(uap, attrname),
3404 	    SCARG(uap, arg), p));
3405 }
3406 
3407 /*
3408  * Syscall to set a named extended attribute on a file or directory.
3409  * Accepts attribute name, and a uio structure pointing to the data to set.
3410  * The uio is consumed in the style of writev().  The real work happens
3411  * in VOP_SETEXTATTR().
3412  */
3413 int
3414 extattr_set_file(p, uap)
3415 	struct proc *p;
3416 	struct extattr_set_file_args *uap;
3417 {
3418 	struct nameidata nd;
3419 	struct uio auio;
3420 	struct iovec *iov, *needfree = NULL, aiov[UIO_SMALLIOV];
3421 	char attrname[EXTATTR_MAXNAMELEN];
3422 	u_int iovlen, cnt;
3423 	int error, i;
3424 
3425 	error = copyin(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN);
3426 	if (error)
3427 		return (error);
3428 	NDINIT(&nd, LOOKUP, LOCKLEAF | FOLLOW, UIO_USERSPACE, SCARG(uap, path),
3429 	    p);
3430 	if ((error = namei(&nd)) != 0)
3431 		return(error);
3432 	iovlen = uap->iovcnt * sizeof(struct iovec);
3433 	if (uap->iovcnt > UIO_SMALLIOV) {
3434 		if (uap->iovcnt > UIO_MAXIOV) {
3435 			error = EINVAL;
3436 			goto done;
3437 		}
3438 		MALLOC(iov, struct iovec *, iovlen, M_IOV, M_WAITOK);
3439 		needfree = iov;
3440 	} else
3441 		iov = aiov;
3442 	auio.uio_iov = iov;
3443 	auio.uio_iovcnt = uap->iovcnt;
3444 	auio.uio_rw = UIO_WRITE;
3445 	auio.uio_segflg = UIO_USERSPACE;
3446 	auio.uio_procp = p;
3447 	auio.uio_offset = 0;
3448 	if ((error = copyin((caddr_t)uap->iovp, (caddr_t)iov, iovlen)))
3449 		goto done;
3450 	auio.uio_resid = 0;
3451 	for (i = 0; i < uap->iovcnt; i++) {
3452 		if (iov->iov_len > INT_MAX - auio.uio_resid) {
3453 			error = EINVAL;
3454 			goto done;
3455 		}
3456 		auio.uio_resid += iov->iov_len;
3457 		iov++;
3458 	}
3459 	cnt = auio.uio_resid;
3460 	error = VOP_SETEXTATTR(nd.ni_vp, attrname, &auio, p->p_cred->pc_ucred,
3461 	    p);
3462 	if (auio.uio_resid != cnt && (error == ERESTART ||
3463 	    error == EINTR || error == EWOULDBLOCK))
3464 		error = 0;
3465 	cnt -= auio.uio_resid;
3466 	p->p_retval[0] = cnt;
3467 done:
3468 	if (needfree)
3469 		FREE(needfree, M_IOV);
3470 	NDFREE(&nd, 0);
3471 	return (error);
3472 }
3473 
3474 /*
3475  * Syscall to get a named extended attribute on a file or directory.
3476  * Accepts attribute name, and a uio structure pointing to a buffer for the
3477  * data.  The uio is consumed in the style of readv().  The real work
3478  * happens in VOP_GETEXTATTR();
3479  */
3480 int
3481 extattr_get_file(p, uap)
3482 	struct proc *p;
3483 	struct extattr_get_file_args *uap;
3484 {
3485 	struct nameidata nd;
3486 	struct uio auio;
3487 	struct iovec *iov, *needfree, aiov[UIO_SMALLIOV];
3488 	char attrname[EXTATTR_MAXNAMELEN];
3489 	u_int iovlen, cnt;
3490 	int error, i;
3491 
3492 	error = copyin(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN);
3493 	if (error)
3494 		return (error);
3495 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
3496 	if ((error = namei(&nd)) != 0)
3497 		return (error);
3498 	iovlen = uap->iovcnt * sizeof (struct iovec);
3499 	if (uap->iovcnt > UIO_SMALLIOV) {
3500 		if (uap->iovcnt > UIO_MAXIOV) {
3501 			NDFREE(&nd, 0);
3502 			return (EINVAL);
3503 		}
3504 		MALLOC(iov, struct iovec *, iovlen, M_IOV, M_WAITOK);
3505 		needfree = iov;
3506 	} else {
3507 		iov = aiov;
3508 		needfree = NULL;
3509 	}
3510 	auio.uio_iov = iov;
3511 	auio.uio_iovcnt = uap->iovcnt;
3512 	auio.uio_rw = UIO_READ;
3513 	auio.uio_segflg = UIO_USERSPACE;
3514 	auio.uio_procp = p;
3515 	auio.uio_offset = 0;
3516 	if ((error = copyin((caddr_t)uap->iovp, (caddr_t)iov, iovlen)))
3517 		goto done;
3518 	auio.uio_resid = 0;
3519 	for (i = 0; i < uap->iovcnt; i++) {
3520 		if (iov->iov_len > INT_MAX - auio.uio_resid) {
3521 			error = EINVAL;
3522 			goto done;
3523 		}
3524 		auio.uio_resid += iov->iov_len;
3525 		iov++;
3526 	}
3527 	cnt = auio.uio_resid;
3528 	error = VOP_GETEXTATTR(nd.ni_vp, attrname, &auio, p->p_cred->pc_ucred,
3529 	    p);
3530 	if (auio.uio_resid != cnt && (error == ERESTART ||
3531 	    error == EINTR || error == EWOULDBLOCK))
3532 		error = 0;
3533 	cnt -= auio.uio_resid;
3534 	p->p_retval[0] = cnt;
3535 done:
3536 	if (needfree)
3537 		FREE(needfree, M_IOV);
3538 	NDFREE(&nd, 0);
3539 	return(error);
3540 }
3541 
3542 /*
3543  * Syscall to delete a named extended attribute from a file or directory.
3544  * Accepts attribute name.  The real work happens in VOP_SETEXTATTR().
3545  */
3546 int
3547 extattr_delete_file(p, uap)
3548 	struct proc *p;
3549 	struct extattr_delete_file_args *uap;
3550 {
3551 	struct nameidata nd;
3552 	char attrname[EXTATTR_MAXNAMELEN];
3553 	int	error;
3554 
3555 	error = copyin(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN);
3556 	if (error)
3557 		return(error);
3558 	NDINIT(&nd, LOOKUP, LOCKLEAF | FOLLOW, UIO_USERSPACE, SCARG(uap, path),
3559 	    p);
3560 	if ((error = namei(&nd)) != 0)
3561 		return(error);
3562 	error = VOP_SETEXTATTR(nd.ni_vp, attrname, NULL, p->p_cred->pc_ucred,
3563 	    p);
3564 	NDFREE(&nd, 0);
3565 	return(error);
3566 }
3567