xref: /freebsd/sys/kern/vfs_extattr.c (revision 64db83a8ab2d1f72a9b2174b39d2ef42b5b0580c)
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/bio.h>
48 #include <sys/buf.h>
49 #include <sys/sysent.h>
50 #include <sys/malloc.h>
51 #include <sys/mount.h>
52 #include <sys/sysproto.h>
53 #include <sys/namei.h>
54 #include <sys/filedesc.h>
55 #include <sys/kernel.h>
56 #include <sys/fcntl.h>
57 #include <sys/file.h>
58 #include <sys/linker.h>
59 #include <sys/stat.h>
60 #include <sys/unistd.h>
61 #include <sys/vnode.h>
62 #include <sys/proc.h>
63 #include <sys/dirent.h>
64 #include <sys/extattr.h>
65 
66 #include <machine/limits.h>
67 #include <miscfs/union/union.h>
68 #include <sys/sysctl.h>
69 #include <vm/vm.h>
70 #include <vm/vm_object.h>
71 #include <vm/vm_zone.h>
72 
73 static int change_dir __P((struct nameidata *ndp, struct proc *p));
74 static void checkdirs __P((struct vnode *olddp));
75 static int chroot_refuse_vdir_fds __P((struct filedesc *fdp));
76 static int getutimes __P((const struct timeval *, struct timespec *));
77 static int setfown __P((struct proc *, struct vnode *, uid_t, gid_t));
78 static int setfmode __P((struct proc *, struct vnode *, int));
79 static int setfflags __P((struct proc *, struct vnode *, int));
80 static int setutimes __P((struct proc *, struct vnode *,
81     const struct timespec *, int));
82 static int	usermount = 0;	/* if 1, non-root can mount fs. */
83 
84 int (*union_dircheckp) __P((struct proc *, struct vnode **, struct file *));
85 
86 SYSCTL_INT(_vfs, OID_AUTO, usermount, CTLFLAG_RW, &usermount, 0, "");
87 
88 /*
89  * Virtual File System System Calls
90  */
91 
92 /*
93  * Mount a file system.
94  */
95 #ifndef _SYS_SYSPROTO_H_
96 struct mount_args {
97 	char	*type;
98 	char	*path;
99 	int	flags;
100 	caddr_t	data;
101 };
102 #endif
103 /* ARGSUSED */
104 int
105 mount(p, uap)
106 	struct proc *p;
107 	register struct mount_args /* {
108 		syscallarg(char *) type;
109 		syscallarg(char *) path;
110 		syscallarg(int) flags;
111 		syscallarg(caddr_t) data;
112 	} */ *uap;
113 {
114 	struct vnode *vp;
115 	struct mount *mp;
116 	struct vfsconf *vfsp;
117 	int error, flag = 0, flag2 = 0;
118 	struct vattr va;
119 #ifdef COMPAT_43
120 	u_long fstypenum;
121 #endif
122 	struct nameidata nd;
123 	char fstypename[MFSNAMELEN];
124 
125 	if (usermount == 0 && (error = suser(p)))
126 		return (error);
127 	/*
128 	 * Do not allow NFS export by non-root users.
129 	 */
130 	if (SCARG(uap, flags) & MNT_EXPORTED) {
131 		error = suser(p);
132 		if (error)
133 			return (error);
134 	}
135 	/*
136 	 * Silently enforce MNT_NOSUID and MNT_NODEV for non-root users
137 	 */
138 	if (suser_xxx(p->p_ucred, 0, 0))
139 		SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
140 	/*
141 	 * Get vnode to be covered
142 	 */
143 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
144 	    SCARG(uap, path), p);
145 	if ((error = namei(&nd)) != 0)
146 		return (error);
147 	NDFREE(&nd, NDF_ONLY_PNBUF);
148 	vp = nd.ni_vp;
149 	if (SCARG(uap, flags) & MNT_UPDATE) {
150 		if ((vp->v_flag & VROOT) == 0) {
151 			vput(vp);
152 			return (EINVAL);
153 		}
154 		mp = vp->v_mount;
155 		flag = mp->mnt_flag;
156 		flag2 = mp->mnt_kern_flag;
157 		/*
158 		 * We only allow the filesystem to be reloaded if it
159 		 * is currently mounted read-only.
160 		 */
161 		if ((SCARG(uap, flags) & MNT_RELOAD) &&
162 		    ((mp->mnt_flag & MNT_RDONLY) == 0)) {
163 			vput(vp);
164 			return (EOPNOTSUPP);	/* Needs translation */
165 		}
166 		mp->mnt_flag |=
167 		    SCARG(uap, flags) & (MNT_RELOAD | MNT_FORCE | MNT_UPDATE);
168 		/*
169 		 * Only root, or the user that did the original mount is
170 		 * permitted to update it.
171 		 */
172 		if (mp->mnt_stat.f_owner != p->p_ucred->cr_uid &&
173 		    (error = suser(p))) {
174 			vput(vp);
175 			return (error);
176 		}
177 		if (vfs_busy(mp, LK_NOWAIT, 0, p)) {
178 			vput(vp);
179 			return (EBUSY);
180 		}
181 		VOP_UNLOCK(vp, 0, p);
182 		goto update;
183 	}
184 	/*
185 	 * If the user is not root, ensure that they own the directory
186 	 * onto which we are attempting to mount.
187 	 */
188 	if ((error = VOP_GETATTR(vp, &va, p->p_ucred, p)) ||
189 	    (va.va_uid != p->p_ucred->cr_uid &&
190 	     (error = suser(p)))) {
191 		vput(vp);
192 		return (error);
193 	}
194 	if ((error = vinvalbuf(vp, V_SAVE, p->p_ucred, p, 0, 0)) != 0)
195 		return (error);
196 	if (vp->v_type != VDIR) {
197 		vput(vp);
198 		return (ENOTDIR);
199 	}
200 #ifdef COMPAT_43
201 	/*
202 	 * Historically filesystem types were identified by number. If we
203 	 * get an integer for the filesystem type instead of a string, we
204 	 * check to see if it matches one of the historic filesystem types.
205 	 */
206 	fstypenum = (uintptr_t)SCARG(uap, type);
207 	if (fstypenum < maxvfsconf) {
208 		for (vfsp = vfsconf; vfsp; vfsp = vfsp->vfc_next)
209 			if (vfsp->vfc_typenum == fstypenum)
210 				break;
211 		if (vfsp == NULL) {
212 			vput(vp);
213 			return (ENODEV);
214 		}
215 		strncpy(fstypename, vfsp->vfc_name, MFSNAMELEN);
216 	} else
217 #endif /* COMPAT_43 */
218 	if ((error = copyinstr(SCARG(uap, type), fstypename, MFSNAMELEN, NULL)) != 0) {
219 		vput(vp);
220 		return (error);
221 	}
222 	for (vfsp = vfsconf; vfsp; vfsp = vfsp->vfc_next)
223 		if (!strcmp(vfsp->vfc_name, fstypename))
224 			break;
225 	if (vfsp == NULL) {
226 		linker_file_t lf;
227 
228 		/* Refuse to load modules if securelevel raised */
229 		if (securelevel > 0) {
230 			vput(vp);
231 			return EPERM;
232 		}
233 		/* Only load modules for root (very important!) */
234 		if ((error = suser(p)) != 0) {
235 			vput(vp);
236 			return error;
237 		}
238 		error = linker_load_file(fstypename, &lf);
239 		if (error || lf == NULL) {
240 			vput(vp);
241 			if (lf == NULL)
242 				error = ENODEV;
243 			return error;
244 		}
245 		lf->userrefs++;
246 		/* lookup again, see if the VFS was loaded */
247 		for (vfsp = vfsconf; vfsp; vfsp = vfsp->vfc_next)
248 			if (!strcmp(vfsp->vfc_name, fstypename))
249 				break;
250 		if (vfsp == NULL) {
251 			lf->userrefs--;
252 			linker_file_unload(lf);
253 			vput(vp);
254 			return (ENODEV);
255 		}
256 	}
257 	simple_lock(&vp->v_interlock);
258 	if ((vp->v_flag & VMOUNT) != 0 ||
259 	    vp->v_mountedhere != NULL) {
260 		simple_unlock(&vp->v_interlock);
261 		vput(vp);
262 		return (EBUSY);
263 	}
264 	vp->v_flag |= VMOUNT;
265 	simple_unlock(&vp->v_interlock);
266 
267 	/*
268 	 * Allocate and initialize the filesystem.
269 	 */
270 	mp = malloc(sizeof(struct mount), M_MOUNT, M_WAITOK);
271 	bzero((char *)mp, (u_long)sizeof(struct mount));
272 	lockinit(&mp->mnt_lock, PVFS, "vfslock", 0, LK_NOPAUSE);
273 	(void)vfs_busy(mp, LK_NOWAIT, 0, p);
274 	mp->mnt_op = vfsp->vfc_vfsops;
275 	mp->mnt_vfc = vfsp;
276 	vfsp->vfc_refcount++;
277 	mp->mnt_stat.f_type = vfsp->vfc_typenum;
278 	mp->mnt_flag |= vfsp->vfc_flags & MNT_VISFLAGMASK;
279 	strncpy(mp->mnt_stat.f_fstypename, vfsp->vfc_name, MFSNAMELEN);
280 	mp->mnt_vnodecovered = vp;
281 	mp->mnt_stat.f_owner = p->p_ucred->cr_uid;
282 	mp->mnt_iosize_max = DFLTPHYS;
283 	VOP_UNLOCK(vp, 0, p);
284 update:
285 	/*
286 	 * Set the mount level flags.
287 	 */
288 	if (SCARG(uap, flags) & MNT_RDONLY)
289 		mp->mnt_flag |= MNT_RDONLY;
290 	else if (mp->mnt_flag & MNT_RDONLY)
291 		mp->mnt_kern_flag |= MNTK_WANTRDWR;
292 	mp->mnt_flag &=~ (MNT_NOSUID | MNT_NOEXEC | MNT_NODEV |
293 	    MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | MNT_NOATIME |
294 	    MNT_NOSYMFOLLOW | MNT_IGNORE |
295 	    MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR);
296 	mp->mnt_flag |= SCARG(uap, flags) & (MNT_NOSUID | MNT_NOEXEC |
297 	    MNT_NODEV | MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | MNT_FORCE |
298 	    MNT_NOSYMFOLLOW | MNT_IGNORE |
299 	    MNT_NOATIME | MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR);
300 	/*
301 	 * Mount the filesystem.
302 	 * XXX The final recipients of VFS_MOUNT just overwrite the ndp they
303 	 * get.  No freeing of cn_pnbuf.
304 	 */
305 	error = VFS_MOUNT(mp, SCARG(uap, path), SCARG(uap, data), &nd, p);
306 	if (mp->mnt_flag & MNT_UPDATE) {
307 		vrele(vp);
308 		if (mp->mnt_kern_flag & MNTK_WANTRDWR)
309 			mp->mnt_flag &= ~MNT_RDONLY;
310 		mp->mnt_flag &=~ (MNT_UPDATE | MNT_RELOAD | MNT_FORCE);
311 		mp->mnt_kern_flag &=~ MNTK_WANTRDWR;
312 		if (error) {
313 			mp->mnt_flag = flag;
314 			mp->mnt_kern_flag = flag2;
315 		}
316 		if ((mp->mnt_flag & MNT_RDONLY) == 0) {
317 			if (mp->mnt_syncer == NULL)
318 				error = vfs_allocate_syncvnode(mp);
319 		} else {
320 			if (mp->mnt_syncer != NULL)
321 				vrele(mp->mnt_syncer);
322 			mp->mnt_syncer = NULL;
323 		}
324 		vfs_unbusy(mp, p);
325 		return (error);
326 	}
327 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
328 	/*
329 	 * Put the new filesystem on the mount list after root.
330 	 */
331 	cache_purge(vp);
332 	if (!error) {
333 		simple_lock(&vp->v_interlock);
334 		vp->v_flag &= ~VMOUNT;
335 		vp->v_mountedhere = mp;
336 		simple_unlock(&vp->v_interlock);
337 		simple_lock(&mountlist_slock);
338 		TAILQ_INSERT_TAIL(&mountlist, mp, mnt_list);
339 		simple_unlock(&mountlist_slock);
340 		checkdirs(vp);
341 		VOP_UNLOCK(vp, 0, p);
342 		if ((mp->mnt_flag & MNT_RDONLY) == 0)
343 			error = vfs_allocate_syncvnode(mp);
344 		vfs_unbusy(mp, p);
345 		if ((error = VFS_START(mp, 0, p)) != 0)
346 			vrele(vp);
347 	} else {
348 		simple_lock(&vp->v_interlock);
349 		vp->v_flag &= ~VMOUNT;
350 		simple_unlock(&vp->v_interlock);
351 		mp->mnt_vfc->vfc_refcount--;
352 		vfs_unbusy(mp, p);
353 		free((caddr_t)mp, M_MOUNT);
354 		vput(vp);
355 	}
356 	return (error);
357 }
358 
359 /*
360  * Scan all active processes to see if any of them have a current
361  * or root directory onto which the new filesystem has just been
362  * mounted. If so, replace them with the new mount point.
363  */
364 static void
365 checkdirs(olddp)
366 	struct vnode *olddp;
367 {
368 	struct filedesc *fdp;
369 	struct vnode *newdp;
370 	struct proc *p;
371 
372 	if (olddp->v_usecount == 1)
373 		return;
374 	if (VFS_ROOT(olddp->v_mountedhere, &newdp))
375 		panic("mount: lost mount");
376 	LIST_FOREACH(p, &allproc, p_list) {
377 		fdp = p->p_fd;
378 		if (fdp->fd_cdir == olddp) {
379 			vrele(fdp->fd_cdir);
380 			VREF(newdp);
381 			fdp->fd_cdir = newdp;
382 		}
383 		if (fdp->fd_rdir == olddp) {
384 			vrele(fdp->fd_rdir);
385 			VREF(newdp);
386 			fdp->fd_rdir = newdp;
387 		}
388 	}
389 	if (rootvnode == olddp) {
390 		vrele(rootvnode);
391 		VREF(newdp);
392 		rootvnode = newdp;
393 	}
394 	vput(newdp);
395 }
396 
397 /*
398  * Unmount a file system.
399  *
400  * Note: unmount takes a path to the vnode mounted on as argument,
401  * not special file (as before).
402  */
403 #ifndef _SYS_SYSPROTO_H_
404 struct unmount_args {
405 	char	*path;
406 	int	flags;
407 };
408 #endif
409 /* ARGSUSED */
410 int
411 unmount(p, uap)
412 	struct proc *p;
413 	register struct unmount_args /* {
414 		syscallarg(char *) path;
415 		syscallarg(int) flags;
416 	} */ *uap;
417 {
418 	register struct vnode *vp;
419 	struct mount *mp;
420 	int error;
421 	struct nameidata nd;
422 
423 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
424 	    SCARG(uap, path), p);
425 	if ((error = namei(&nd)) != 0)
426 		return (error);
427 	vp = nd.ni_vp;
428 	NDFREE(&nd, NDF_ONLY_PNBUF);
429 	mp = vp->v_mount;
430 
431 	/*
432 	 * Only root, or the user that did the original mount is
433 	 * permitted to unmount this filesystem.
434 	 */
435 	if ((mp->mnt_stat.f_owner != p->p_ucred->cr_uid) &&
436 	    (error = suser(p))) {
437 		vput(vp);
438 		return (error);
439 	}
440 
441 	/*
442 	 * Don't allow unmounting the root file system.
443 	 */
444 	if (mp->mnt_flag & MNT_ROOTFS) {
445 		vput(vp);
446 		return (EINVAL);
447 	}
448 
449 	/*
450 	 * Must be the root of the filesystem
451 	 */
452 	if ((vp->v_flag & VROOT) == 0) {
453 		vput(vp);
454 		return (EINVAL);
455 	}
456 	vput(vp);
457 	return (dounmount(mp, SCARG(uap, flags), p));
458 }
459 
460 /*
461  * Do the actual file system unmount.
462  */
463 int
464 dounmount(mp, flags, p)
465 	register struct mount *mp;
466 	int flags;
467 	struct proc *p;
468 {
469 	struct vnode *coveredvp;
470 	int error;
471 	int async_flag;
472 
473 	simple_lock(&mountlist_slock);
474 	mp->mnt_kern_flag |= MNTK_UNMOUNT;
475 	lockmgr(&mp->mnt_lock, LK_DRAIN | LK_INTERLOCK, &mountlist_slock, p);
476 
477 	if (mp->mnt_flag & MNT_EXPUBLIC)
478 		vfs_setpublicfs(NULL, NULL, NULL);
479 
480 	vfs_msync(mp, MNT_WAIT);
481 	async_flag = mp->mnt_flag & MNT_ASYNC;
482 	mp->mnt_flag &=~ MNT_ASYNC;
483 	cache_purgevfs(mp);	/* remove cache entries for this file sys */
484 	if (mp->mnt_syncer != NULL)
485 		vrele(mp->mnt_syncer);
486 	if (((mp->mnt_flag & MNT_RDONLY) ||
487 	     (error = VFS_SYNC(mp, MNT_WAIT, p->p_ucred, p)) == 0) ||
488 	    (flags & MNT_FORCE))
489 		error = VFS_UNMOUNT(mp, flags, p);
490 	simple_lock(&mountlist_slock);
491 	if (error) {
492 		if ((mp->mnt_flag & MNT_RDONLY) == 0 && mp->mnt_syncer == NULL)
493 			(void) vfs_allocate_syncvnode(mp);
494 		mp->mnt_kern_flag &= ~MNTK_UNMOUNT;
495 		mp->mnt_flag |= async_flag;
496 		lockmgr(&mp->mnt_lock, LK_RELEASE | LK_INTERLOCK | LK_REENABLE,
497 		    &mountlist_slock, p);
498 		if (mp->mnt_kern_flag & MNTK_MWAIT)
499 			wakeup((caddr_t)mp);
500 		return (error);
501 	}
502 	TAILQ_REMOVE(&mountlist, mp, mnt_list);
503 	if ((coveredvp = mp->mnt_vnodecovered) != NULLVP) {
504 		coveredvp->v_mountedhere = (struct mount *)0;
505 		vrele(coveredvp);
506 	}
507 	mp->mnt_vfc->vfc_refcount--;
508 	if (!LIST_EMPTY(&mp->mnt_vnodelist))
509 		panic("unmount: dangling vnode");
510 	lockmgr(&mp->mnt_lock, LK_RELEASE | LK_INTERLOCK, &mountlist_slock, p);
511 	if (mp->mnt_kern_flag & MNTK_MWAIT)
512 		wakeup((caddr_t)mp);
513 	free((caddr_t)mp, M_MOUNT);
514 	return (0);
515 }
516 
517 /*
518  * Sync each mounted filesystem.
519  */
520 #ifndef _SYS_SYSPROTO_H_
521 struct sync_args {
522         int     dummy;
523 };
524 #endif
525 
526 #ifdef DEBUG
527 static int syncprt = 0;
528 SYSCTL_INT(_debug, OID_AUTO, syncprt, CTLFLAG_RW, &syncprt, 0, "");
529 #endif
530 
531 /* ARGSUSED */
532 int
533 sync(p, uap)
534 	struct proc *p;
535 	struct sync_args *uap;
536 {
537 	register struct mount *mp, *nmp;
538 	int asyncflag;
539 
540 	simple_lock(&mountlist_slock);
541 	for (mp = TAILQ_FIRST(&mountlist); mp != NULL; mp = nmp) {
542 		if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock, p)) {
543 			nmp = TAILQ_NEXT(mp, mnt_list);
544 			continue;
545 		}
546 		if ((mp->mnt_flag & MNT_RDONLY) == 0) {
547 			asyncflag = mp->mnt_flag & MNT_ASYNC;
548 			mp->mnt_flag &= ~MNT_ASYNC;
549 			vfs_msync(mp, MNT_NOWAIT);
550 			VFS_SYNC(mp, MNT_NOWAIT,
551 				((p != NULL) ? p->p_ucred : NOCRED), p);
552 			mp->mnt_flag |= asyncflag;
553 		}
554 		simple_lock(&mountlist_slock);
555 		nmp = TAILQ_NEXT(mp, mnt_list);
556 		vfs_unbusy(mp, p);
557 	}
558 	simple_unlock(&mountlist_slock);
559 #if 0
560 /*
561  * XXX don't call vfs_bufstats() yet because that routine
562  * was not imported in the Lite2 merge.
563  */
564 #ifdef DIAGNOSTIC
565 	if (syncprt)
566 		vfs_bufstats();
567 #endif /* DIAGNOSTIC */
568 #endif
569 	return (0);
570 }
571 
572 /* XXX PRISON: could be per prison flag */
573 static int prison_quotas;
574 #if 0
575 SYSCTL_INT(_kern_prison, OID_AUTO, quotas, CTLFLAG_RW, &prison_quotas, 0, "");
576 #endif
577 
578 /*
579  * Change filesystem quotas.
580  */
581 #ifndef _SYS_SYSPROTO_H_
582 struct quotactl_args {
583 	char *path;
584 	int cmd;
585 	int uid;
586 	caddr_t arg;
587 };
588 #endif
589 /* ARGSUSED */
590 int
591 quotactl(p, uap)
592 	struct proc *p;
593 	register struct quotactl_args /* {
594 		syscallarg(char *) path;
595 		syscallarg(int) cmd;
596 		syscallarg(int) uid;
597 		syscallarg(caddr_t) arg;
598 	} */ *uap;
599 {
600 	register struct mount *mp;
601 	int error;
602 	struct nameidata nd;
603 
604 	if (p->p_prison && !prison_quotas)
605 		return (EPERM);
606 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
607 	if ((error = namei(&nd)) != 0)
608 		return (error);
609 	mp = nd.ni_vp->v_mount;
610 	NDFREE(&nd, NDF_ONLY_PNBUF);
611 	vrele(nd.ni_vp);
612 	return (VFS_QUOTACTL(mp, SCARG(uap, cmd), SCARG(uap, uid),
613 	    SCARG(uap, arg), p));
614 }
615 
616 /*
617  * Get filesystem statistics.
618  */
619 #ifndef _SYS_SYSPROTO_H_
620 struct statfs_args {
621 	char *path;
622 	struct statfs *buf;
623 };
624 #endif
625 /* ARGSUSED */
626 int
627 statfs(p, uap)
628 	struct proc *p;
629 	register struct statfs_args /* {
630 		syscallarg(char *) path;
631 		syscallarg(struct statfs *) buf;
632 	} */ *uap;
633 {
634 	register struct mount *mp;
635 	register struct statfs *sp;
636 	int error;
637 	struct nameidata nd;
638 	struct statfs sb;
639 
640 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
641 	if ((error = namei(&nd)) != 0)
642 		return (error);
643 	mp = nd.ni_vp->v_mount;
644 	sp = &mp->mnt_stat;
645 	NDFREE(&nd, NDF_ONLY_PNBUF);
646 	vrele(nd.ni_vp);
647 	error = VFS_STATFS(mp, sp, p);
648 	if (error)
649 		return (error);
650 	sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
651 	if (suser_xxx(p->p_ucred, 0, 0)) {
652 		bcopy((caddr_t)sp, (caddr_t)&sb, sizeof(sb));
653 		sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0;
654 		sp = &sb;
655 	}
656 	return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
657 }
658 
659 /*
660  * Get filesystem statistics.
661  */
662 #ifndef _SYS_SYSPROTO_H_
663 struct fstatfs_args {
664 	int fd;
665 	struct statfs *buf;
666 };
667 #endif
668 /* ARGSUSED */
669 int
670 fstatfs(p, uap)
671 	struct proc *p;
672 	register struct fstatfs_args /* {
673 		syscallarg(int) fd;
674 		syscallarg(struct statfs *) buf;
675 	} */ *uap;
676 {
677 	struct file *fp;
678 	struct mount *mp;
679 	register struct statfs *sp;
680 	int error;
681 	struct statfs sb;
682 
683 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
684 		return (error);
685 	mp = ((struct vnode *)fp->f_data)->v_mount;
686 	sp = &mp->mnt_stat;
687 	error = VFS_STATFS(mp, sp, p);
688 	if (error)
689 		return (error);
690 	sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
691 	if (suser_xxx(p->p_ucred, 0, 0)) {
692 		bcopy((caddr_t)sp, (caddr_t)&sb, sizeof(sb));
693 		sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0;
694 		sp = &sb;
695 	}
696 	return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
697 }
698 
699 /*
700  * Get statistics on all filesystems.
701  */
702 #ifndef _SYS_SYSPROTO_H_
703 struct getfsstat_args {
704 	struct statfs *buf;
705 	long bufsize;
706 	int flags;
707 };
708 #endif
709 int
710 getfsstat(p, uap)
711 	struct proc *p;
712 	register struct getfsstat_args /* {
713 		syscallarg(struct statfs *) buf;
714 		syscallarg(long) bufsize;
715 		syscallarg(int) flags;
716 	} */ *uap;
717 {
718 	register struct mount *mp, *nmp;
719 	register struct statfs *sp;
720 	caddr_t sfsp;
721 	long count, maxcount, error;
722 
723 	maxcount = SCARG(uap, bufsize) / sizeof(struct statfs);
724 	sfsp = (caddr_t)SCARG(uap, buf);
725 	count = 0;
726 	simple_lock(&mountlist_slock);
727 	for (mp = TAILQ_FIRST(&mountlist); mp != NULL; mp = nmp) {
728 		if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock, p)) {
729 			nmp = TAILQ_NEXT(mp, mnt_list);
730 			continue;
731 		}
732 		if (sfsp && count < maxcount) {
733 			sp = &mp->mnt_stat;
734 			/*
735 			 * If MNT_NOWAIT or MNT_LAZY is specified, do not
736 			 * refresh the fsstat cache. MNT_NOWAIT or MNT_LAZY
737 			 * overrides MNT_WAIT.
738 			 */
739 			if (((SCARG(uap, flags) & (MNT_LAZY|MNT_NOWAIT)) == 0 ||
740 			    (SCARG(uap, flags) & MNT_WAIT)) &&
741 			    (error = VFS_STATFS(mp, sp, p))) {
742 				simple_lock(&mountlist_slock);
743 				nmp = TAILQ_NEXT(mp, mnt_list);
744 				vfs_unbusy(mp, p);
745 				continue;
746 			}
747 			sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
748 			error = copyout((caddr_t)sp, sfsp, sizeof(*sp));
749 			if (error) {
750 				vfs_unbusy(mp, p);
751 				return (error);
752 			}
753 			sfsp += sizeof(*sp);
754 		}
755 		count++;
756 		simple_lock(&mountlist_slock);
757 		nmp = TAILQ_NEXT(mp, mnt_list);
758 		vfs_unbusy(mp, p);
759 	}
760 	simple_unlock(&mountlist_slock);
761 	if (sfsp && count > maxcount)
762 		p->p_retval[0] = maxcount;
763 	else
764 		p->p_retval[0] = count;
765 	return (0);
766 }
767 
768 /*
769  * Change current working directory to a given file descriptor.
770  */
771 #ifndef _SYS_SYSPROTO_H_
772 struct fchdir_args {
773 	int	fd;
774 };
775 #endif
776 /* ARGSUSED */
777 int
778 fchdir(p, uap)
779 	struct proc *p;
780 	struct fchdir_args /* {
781 		syscallarg(int) fd;
782 	} */ *uap;
783 {
784 	register struct filedesc *fdp = p->p_fd;
785 	struct vnode *vp, *tdp;
786 	struct mount *mp;
787 	struct file *fp;
788 	int error;
789 
790 	if ((error = getvnode(fdp, SCARG(uap, fd), &fp)) != 0)
791 		return (error);
792 	vp = (struct vnode *)fp->f_data;
793 	VREF(vp);
794 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
795 	if (vp->v_type != VDIR)
796 		error = ENOTDIR;
797 	else
798 		error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
799 	while (!error && (mp = vp->v_mountedhere) != NULL) {
800 		if (vfs_busy(mp, 0, 0, p))
801 			continue;
802 		error = VFS_ROOT(mp, &tdp);
803 		vfs_unbusy(mp, p);
804 		if (error)
805 			break;
806 		vput(vp);
807 		vp = tdp;
808 	}
809 	if (error) {
810 		vput(vp);
811 		return (error);
812 	}
813 	VOP_UNLOCK(vp, 0, p);
814 	vrele(fdp->fd_cdir);
815 	fdp->fd_cdir = vp;
816 	return (0);
817 }
818 
819 /*
820  * Change current working directory (``.'').
821  */
822 #ifndef _SYS_SYSPROTO_H_
823 struct chdir_args {
824 	char	*path;
825 };
826 #endif
827 /* ARGSUSED */
828 int
829 chdir(p, uap)
830 	struct proc *p;
831 	struct chdir_args /* {
832 		syscallarg(char *) path;
833 	} */ *uap;
834 {
835 	register struct filedesc *fdp = p->p_fd;
836 	int error;
837 	struct nameidata nd;
838 
839 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
840 	    SCARG(uap, path), p);
841 	if ((error = change_dir(&nd, p)) != 0)
842 		return (error);
843 	NDFREE(&nd, NDF_ONLY_PNBUF);
844 	vrele(fdp->fd_cdir);
845 	fdp->fd_cdir = nd.ni_vp;
846 	return (0);
847 }
848 
849 /*
850  * Helper function for raised chroot(2) security function:  Refuse if
851  * any filedescriptors are open directories.
852  */
853 static int
854 chroot_refuse_vdir_fds(fdp)
855 	struct filedesc *fdp;
856 {
857 	struct vnode *vp;
858 	struct file *fp;
859 	int error;
860 	int fd;
861 
862 	for (fd = 0; fd < fdp->fd_nfiles ; fd++) {
863 		error = getvnode(fdp, fd, &fp);
864 		if (error)
865 			continue;
866 		vp = (struct vnode *)fp->f_data;
867 		if (vp->v_type != VDIR)
868 			continue;
869 		return(EPERM);
870 	}
871 	return (0);
872 }
873 
874 /*
875  * This sysctl determines if we will allow a process to chroot(2) if it
876  * has a directory open:
877  *	0: disallowed for all processes.
878  *	1: allowed for processes that were not already chroot(2)'ed.
879  *	2: allowed for all processes.
880  */
881 
882 static int chroot_allow_open_directories = 1;
883 
884 SYSCTL_INT(_kern, OID_AUTO, chroot_allow_open_directories, CTLFLAG_RW,
885      &chroot_allow_open_directories, 0, "");
886 
887 /*
888  * Change notion of root (``/'') directory.
889  */
890 #ifndef _SYS_SYSPROTO_H_
891 struct chroot_args {
892 	char	*path;
893 };
894 #endif
895 /* ARGSUSED */
896 int
897 chroot(p, uap)
898 	struct proc *p;
899 	struct chroot_args /* {
900 		syscallarg(char *) path;
901 	} */ *uap;
902 {
903 	register struct filedesc *fdp = p->p_fd;
904 	int error;
905 	struct nameidata nd;
906 
907 	error = suser_xxx(0, p, PRISON_ROOT);
908 	if (error)
909 		return (error);
910 	if (chroot_allow_open_directories == 0 ||
911 	    (chroot_allow_open_directories == 1 && fdp->fd_rdir != rootvnode))
912 		error = chroot_refuse_vdir_fds(fdp);
913 	if (error)
914 		return (error);
915 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
916 	    SCARG(uap, path), p);
917 	if ((error = change_dir(&nd, p)) != 0)
918 		return (error);
919 	NDFREE(&nd, NDF_ONLY_PNBUF);
920 	vrele(fdp->fd_rdir);
921 	fdp->fd_rdir = nd.ni_vp;
922 	if (!fdp->fd_jdir) {
923 		fdp->fd_jdir = nd.ni_vp;
924                 VREF(fdp->fd_jdir);
925 	}
926 	return (0);
927 }
928 
929 /*
930  * Common routine for chroot and chdir.
931  */
932 static int
933 change_dir(ndp, p)
934 	register struct nameidata *ndp;
935 	struct proc *p;
936 {
937 	struct vnode *vp;
938 	int error;
939 
940 	error = namei(ndp);
941 	if (error)
942 		return (error);
943 	vp = ndp->ni_vp;
944 	if (vp->v_type != VDIR)
945 		error = ENOTDIR;
946 	else
947 		error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
948 	if (error)
949 		vput(vp);
950 	else
951 		VOP_UNLOCK(vp, 0, p);
952 	return (error);
953 }
954 
955 /*
956  * Check permissions, allocate an open file structure,
957  * and call the device open routine if any.
958  */
959 #ifndef _SYS_SYSPROTO_H_
960 struct open_args {
961 	char	*path;
962 	int	flags;
963 	int	mode;
964 };
965 #endif
966 int
967 open(p, uap)
968 	struct proc *p;
969 	register struct open_args /* {
970 		syscallarg(char *) path;
971 		syscallarg(int) flags;
972 		syscallarg(int) mode;
973 	} */ *uap;
974 {
975 	register struct filedesc *fdp = p->p_fd;
976 	register struct file *fp;
977 	register struct vnode *vp;
978 	int cmode, flags, oflags;
979 	struct file *nfp;
980 	int type, indx, error;
981 	struct flock lf;
982 	struct nameidata nd;
983 
984 	oflags = SCARG(uap, flags);
985 	if ((oflags & O_ACCMODE) == O_ACCMODE)
986 		return (EINVAL);
987 	flags = FFLAGS(oflags);
988 	error = falloc(p, &nfp, &indx);
989 	if (error)
990 		return (error);
991 	fp = nfp;
992 	cmode = ((SCARG(uap, mode) &~ fdp->fd_cmask) & ALLPERMS) &~ S_ISTXT;
993 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
994 	p->p_dupfd = -indx - 1;			/* XXX check for fdopen */
995 	error = vn_open(&nd, flags, cmode);
996 	if (error) {
997 		ffree(fp);
998 		if ((error == ENODEV || error == ENXIO) &&
999 		    p->p_dupfd >= 0 &&			/* XXX from fdopen */
1000 		    (error =
1001 			dupfdopen(fdp, indx, p->p_dupfd, flags, error)) == 0) {
1002 			p->p_retval[0] = indx;
1003 			return (0);
1004 		}
1005 		if (error == ERESTART)
1006 			error = EINTR;
1007 		fdp->fd_ofiles[indx] = NULL;
1008 		return (error);
1009 	}
1010 	p->p_dupfd = 0;
1011 	NDFREE(&nd, NDF_ONLY_PNBUF);
1012 	vp = nd.ni_vp;
1013 
1014 	fp->f_data = (caddr_t)vp;
1015 	fp->f_flag = flags & FMASK;
1016 	fp->f_ops = &vnops;
1017 	fp->f_type = (vp->v_type == VFIFO ? DTYPE_FIFO : DTYPE_VNODE);
1018 	if (flags & (O_EXLOCK | O_SHLOCK)) {
1019 		lf.l_whence = SEEK_SET;
1020 		lf.l_start = 0;
1021 		lf.l_len = 0;
1022 		if (flags & O_EXLOCK)
1023 			lf.l_type = F_WRLCK;
1024 		else
1025 			lf.l_type = F_RDLCK;
1026 		type = F_FLOCK;
1027 		if ((flags & FNONBLOCK) == 0)
1028 			type |= F_WAIT;
1029 		VOP_UNLOCK(vp, 0, p);
1030 		if ((error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type)) != 0) {
1031 			(void) vn_close(vp, fp->f_flag, fp->f_cred, p);
1032 			ffree(fp);
1033 			fdp->fd_ofiles[indx] = NULL;
1034 			return (error);
1035 		}
1036 		vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
1037 		fp->f_flag |= FHASLOCK;
1038 	}
1039 	/* assert that vn_open created a backing object if one is needed */
1040 	KASSERT(!vn_canvmio(vp) || vp->v_object != NULL,
1041 		("open: vmio vnode has no backing object after vn_open"));
1042 	VOP_UNLOCK(vp, 0, p);
1043 	p->p_retval[0] = indx;
1044 	return (0);
1045 }
1046 
1047 #ifdef COMPAT_43
1048 /*
1049  * Create a file.
1050  */
1051 #ifndef _SYS_SYSPROTO_H_
1052 struct ocreat_args {
1053 	char	*path;
1054 	int	mode;
1055 };
1056 #endif
1057 int
1058 ocreat(p, uap)
1059 	struct proc *p;
1060 	register struct ocreat_args /* {
1061 		syscallarg(char *) path;
1062 		syscallarg(int) mode;
1063 	} */ *uap;
1064 {
1065 	struct open_args /* {
1066 		syscallarg(char *) path;
1067 		syscallarg(int) flags;
1068 		syscallarg(int) mode;
1069 	} */ nuap;
1070 
1071 	SCARG(&nuap, path) = SCARG(uap, path);
1072 	SCARG(&nuap, mode) = SCARG(uap, mode);
1073 	SCARG(&nuap, flags) = O_WRONLY | O_CREAT | O_TRUNC;
1074 	return (open(p, &nuap));
1075 }
1076 #endif /* COMPAT_43 */
1077 
1078 /*
1079  * Create a special file.
1080  */
1081 #ifndef _SYS_SYSPROTO_H_
1082 struct mknod_args {
1083 	char	*path;
1084 	int	mode;
1085 	int	dev;
1086 };
1087 #endif
1088 /* ARGSUSED */
1089 int
1090 mknod(p, uap)
1091 	struct proc *p;
1092 	register struct mknod_args /* {
1093 		syscallarg(char *) path;
1094 		syscallarg(int) mode;
1095 		syscallarg(int) dev;
1096 	} */ *uap;
1097 {
1098 	register struct vnode *vp;
1099 	struct vattr vattr;
1100 	int error;
1101 	int whiteout = 0;
1102 	struct nameidata nd;
1103 
1104 	switch (SCARG(uap, mode) & S_IFMT) {
1105 	case S_IFCHR:
1106 	case S_IFBLK:
1107 		error = suser(p);
1108 		break;
1109 	default:
1110 		error = suser_xxx(0, p, PRISON_ROOT);
1111 		break;
1112 	}
1113 	if (error)
1114 		return (error);
1115 	bwillwrite();
1116 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1117 	if ((error = namei(&nd)) != 0)
1118 		return (error);
1119 	vp = nd.ni_vp;
1120 	if (vp != NULL)
1121 		error = EEXIST;
1122 	else {
1123 		VATTR_NULL(&vattr);
1124 		vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
1125 		vattr.va_rdev = SCARG(uap, dev);
1126 		whiteout = 0;
1127 
1128 		switch (SCARG(uap, mode) & S_IFMT) {
1129 		case S_IFMT:	/* used by badsect to flag bad sectors */
1130 			vattr.va_type = VBAD;
1131 			break;
1132 		case S_IFCHR:
1133 			vattr.va_type = VCHR;
1134 			break;
1135 		case S_IFBLK:
1136 			vattr.va_type = VBLK;
1137 			break;
1138 		case S_IFWHT:
1139 			whiteout = 1;
1140 			break;
1141 		default:
1142 			error = EINVAL;
1143 			break;
1144 		}
1145 	}
1146 	if (!error) {
1147 		VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1148 		if (whiteout)
1149 			error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, CREATE);
1150 		else {
1151 			error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp,
1152 						&nd.ni_cnd, &vattr);
1153 			if (error == 0)
1154 				vput(nd.ni_vp);
1155 		}
1156 		NDFREE(&nd, NDF_ONLY_PNBUF);
1157 		vput(nd.ni_dvp);
1158 	} else {
1159 		NDFREE(&nd, NDF_ONLY_PNBUF);
1160 		if (nd.ni_dvp == vp)
1161 			vrele(nd.ni_dvp);
1162 		else
1163 			vput(nd.ni_dvp);
1164 		if (vp)
1165 			vrele(vp);
1166 	}
1167 	ASSERT_VOP_UNLOCKED(nd.ni_dvp, "mknod");
1168 	ASSERT_VOP_UNLOCKED(nd.ni_vp, "mknod");
1169 	return (error);
1170 }
1171 
1172 /*
1173  * Create a named pipe.
1174  */
1175 #ifndef _SYS_SYSPROTO_H_
1176 struct mkfifo_args {
1177 	char	*path;
1178 	int	mode;
1179 };
1180 #endif
1181 /* ARGSUSED */
1182 int
1183 mkfifo(p, uap)
1184 	struct proc *p;
1185 	register struct mkfifo_args /* {
1186 		syscallarg(char *) path;
1187 		syscallarg(int) mode;
1188 	} */ *uap;
1189 {
1190 	struct vattr vattr;
1191 	int error;
1192 	struct nameidata nd;
1193 
1194 	bwillwrite();
1195 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1196 	if ((error = namei(&nd)) != 0)
1197 		return (error);
1198 	if (nd.ni_vp != NULL) {
1199 		NDFREE(&nd, NDF_ONLY_PNBUF);
1200 		if (nd.ni_dvp == nd.ni_vp)
1201 			vrele(nd.ni_dvp);
1202 		else
1203 			vput(nd.ni_dvp);
1204 		vrele(nd.ni_vp);
1205 		return (EEXIST);
1206 	}
1207 	VATTR_NULL(&vattr);
1208 	vattr.va_type = VFIFO;
1209 	vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
1210 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1211 	error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
1212 	if (error == 0)
1213 		vput(nd.ni_vp);
1214 	NDFREE(&nd, NDF_ONLY_PNBUF);
1215 	vput(nd.ni_dvp);
1216 	return (error);
1217 }
1218 
1219 /*
1220  * Make a hard file link.
1221  */
1222 #ifndef _SYS_SYSPROTO_H_
1223 struct link_args {
1224 	char	*path;
1225 	char	*link;
1226 };
1227 #endif
1228 /* ARGSUSED */
1229 int
1230 link(p, uap)
1231 	struct proc *p;
1232 	register struct link_args /* {
1233 		syscallarg(char *) path;
1234 		syscallarg(char *) link;
1235 	} */ *uap;
1236 {
1237 	register struct vnode *vp;
1238 	struct nameidata nd;
1239 	int error;
1240 
1241 	bwillwrite();
1242 	NDINIT(&nd, LOOKUP, FOLLOW|NOOBJ, UIO_USERSPACE, SCARG(uap, path), p);
1243 	if ((error = namei(&nd)) != 0)
1244 		return (error);
1245 	NDFREE(&nd, NDF_ONLY_PNBUF);
1246 	vp = nd.ni_vp;
1247 	if (vp->v_type == VDIR)
1248 		error = EPERM;		/* POSIX */
1249 	else {
1250 		NDINIT(&nd, CREATE, LOCKPARENT|NOOBJ, UIO_USERSPACE, SCARG(uap, link), p);
1251 		error = namei(&nd);
1252 		if (!error) {
1253 			if (nd.ni_vp != NULL) {
1254 				if (nd.ni_vp)
1255 					vrele(nd.ni_vp);
1256 				error = EEXIST;
1257 			} else {
1258 				VOP_LEASE(nd.ni_dvp, p, p->p_ucred,
1259 				    LEASE_WRITE);
1260 				VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1261 				error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd);
1262 			}
1263 			NDFREE(&nd, NDF_ONLY_PNBUF);
1264 			if (nd.ni_dvp == nd.ni_vp)
1265 				vrele(nd.ni_dvp);
1266 			else
1267 				vput(nd.ni_dvp);
1268 		}
1269 	}
1270 	vrele(vp);
1271 	ASSERT_VOP_UNLOCKED(nd.ni_dvp, "link");
1272 	ASSERT_VOP_UNLOCKED(nd.ni_vp, "link");
1273 	return (error);
1274 }
1275 
1276 /*
1277  * Make a symbolic link.
1278  */
1279 #ifndef _SYS_SYSPROTO_H_
1280 struct symlink_args {
1281 	char	*path;
1282 	char	*link;
1283 };
1284 #endif
1285 /* ARGSUSED */
1286 int
1287 symlink(p, uap)
1288 	struct proc *p;
1289 	register struct symlink_args /* {
1290 		syscallarg(char *) path;
1291 		syscallarg(char *) link;
1292 	} */ *uap;
1293 {
1294 	struct vattr vattr;
1295 	char *path;
1296 	int error;
1297 	struct nameidata nd;
1298 
1299 	path = zalloc(namei_zone);
1300 	if ((error = copyinstr(SCARG(uap, path), path, MAXPATHLEN, NULL)) != 0)
1301 		goto out;
1302 	bwillwrite();
1303 	NDINIT(&nd, CREATE, LOCKPARENT|NOOBJ, UIO_USERSPACE, SCARG(uap, link), p);
1304 	if ((error = namei(&nd)) != 0)
1305 		goto out;
1306 	if (nd.ni_vp) {
1307 		NDFREE(&nd, NDF_ONLY_PNBUF);
1308 		if (nd.ni_dvp == nd.ni_vp)
1309 			vrele(nd.ni_dvp);
1310 		else
1311 			vput(nd.ni_dvp);
1312 		vrele(nd.ni_vp);
1313 		error = EEXIST;
1314 		goto out;
1315 	}
1316 	VATTR_NULL(&vattr);
1317 	vattr.va_mode = ACCESSPERMS &~ p->p_fd->fd_cmask;
1318 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1319 	error = VOP_SYMLINK(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr, path);
1320 	NDFREE(&nd, NDF_ONLY_PNBUF);
1321 	if (error == 0)
1322 		vput(nd.ni_vp);
1323 	vput(nd.ni_dvp);
1324 	ASSERT_VOP_UNLOCKED(nd.ni_dvp, "symlink");
1325 	ASSERT_VOP_UNLOCKED(nd.ni_vp, "symlink");
1326 out:
1327 	zfree(namei_zone, path);
1328 	return (error);
1329 }
1330 
1331 /*
1332  * Delete a whiteout from the filesystem.
1333  */
1334 /* ARGSUSED */
1335 int
1336 undelete(p, uap)
1337 	struct proc *p;
1338 	register struct undelete_args /* {
1339 		syscallarg(char *) path;
1340 	} */ *uap;
1341 {
1342 	int error;
1343 	struct nameidata nd;
1344 
1345 	bwillwrite();
1346 	NDINIT(&nd, DELETE, LOCKPARENT|DOWHITEOUT, UIO_USERSPACE,
1347 	    SCARG(uap, path), p);
1348 	error = namei(&nd);
1349 	if (error)
1350 		return (error);
1351 
1352 	if (nd.ni_vp != NULLVP || !(nd.ni_cnd.cn_flags & ISWHITEOUT)) {
1353 		NDFREE(&nd, NDF_ONLY_PNBUF);
1354 		if (nd.ni_dvp == nd.ni_vp)
1355 			vrele(nd.ni_dvp);
1356 		else
1357 			vput(nd.ni_dvp);
1358 		if (nd.ni_vp)
1359 			vrele(nd.ni_vp);
1360 		return (EEXIST);
1361 	}
1362 
1363 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1364 	error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, DELETE);
1365 	NDFREE(&nd, NDF_ONLY_PNBUF);
1366 	vput(nd.ni_dvp);
1367 	ASSERT_VOP_UNLOCKED(nd.ni_dvp, "undelete");
1368 	ASSERT_VOP_UNLOCKED(nd.ni_vp, "undelete");
1369 	return (error);
1370 }
1371 
1372 /*
1373  * Delete a name from the filesystem.
1374  */
1375 #ifndef _SYS_SYSPROTO_H_
1376 struct unlink_args {
1377 	char	*path;
1378 };
1379 #endif
1380 /* ARGSUSED */
1381 int
1382 unlink(p, uap)
1383 	struct proc *p;
1384 	struct unlink_args /* {
1385 		syscallarg(char *) path;
1386 	} */ *uap;
1387 {
1388 	register struct vnode *vp;
1389 	int error;
1390 	struct nameidata nd;
1391 
1392 	bwillwrite();
1393 	NDINIT(&nd, DELETE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1394 	if ((error = namei(&nd)) != 0)
1395 		return (error);
1396 	vp = nd.ni_vp;
1397 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1398 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
1399 
1400 	if (vp->v_type == VDIR)
1401 		error = EPERM;		/* POSIX */
1402 	else {
1403 		/*
1404 		 * The root of a mounted filesystem cannot be deleted.
1405 		 *
1406 		 * XXX: can this only be a VDIR case?
1407 		 */
1408 		if (vp->v_flag & VROOT)
1409 			error = EBUSY;
1410 	}
1411 
1412 	if (!error) {
1413 		VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1414 		error = VOP_REMOVE(nd.ni_dvp, vp, &nd.ni_cnd);
1415 	}
1416 	NDFREE(&nd, NDF_ONLY_PNBUF);
1417 	if (nd.ni_dvp == vp)
1418 		vrele(nd.ni_dvp);
1419 	else
1420 		vput(nd.ni_dvp);
1421 	if (vp != NULLVP)
1422 		vput(vp);
1423 	ASSERT_VOP_UNLOCKED(nd.ni_dvp, "unlink");
1424 	ASSERT_VOP_UNLOCKED(nd.ni_vp, "unlink");
1425 	return (error);
1426 }
1427 
1428 /*
1429  * Reposition read/write file offset.
1430  */
1431 #ifndef _SYS_SYSPROTO_H_
1432 struct lseek_args {
1433 	int	fd;
1434 	int	pad;
1435 	off_t	offset;
1436 	int	whence;
1437 };
1438 #endif
1439 int
1440 lseek(p, uap)
1441 	struct proc *p;
1442 	register struct lseek_args /* {
1443 		syscallarg(int) fd;
1444 		syscallarg(int) pad;
1445 		syscallarg(off_t) offset;
1446 		syscallarg(int) whence;
1447 	} */ *uap;
1448 {
1449 	struct ucred *cred = p->p_ucred;
1450 	register struct filedesc *fdp = p->p_fd;
1451 	register struct file *fp;
1452 	struct vattr vattr;
1453 	int error;
1454 
1455 	if ((u_int)SCARG(uap, fd) >= fdp->fd_nfiles ||
1456 	    (fp = fdp->fd_ofiles[SCARG(uap, fd)]) == NULL)
1457 		return (EBADF);
1458 	if (fp->f_type != DTYPE_VNODE)
1459 		return (ESPIPE);
1460 	switch (SCARG(uap, whence)) {
1461 	case L_INCR:
1462 		fp->f_offset += SCARG(uap, offset);
1463 		break;
1464 	case L_XTND:
1465 		error=VOP_GETATTR((struct vnode *)fp->f_data, &vattr, cred, p);
1466 		if (error)
1467 			return (error);
1468 		fp->f_offset = SCARG(uap, offset) + vattr.va_size;
1469 		break;
1470 	case L_SET:
1471 		fp->f_offset = SCARG(uap, offset);
1472 		break;
1473 	default:
1474 		return (EINVAL);
1475 	}
1476 	*(off_t *)(p->p_retval) = fp->f_offset;
1477 	return (0);
1478 }
1479 
1480 #if defined(COMPAT_43) || defined(COMPAT_SUNOS)
1481 /*
1482  * Reposition read/write file offset.
1483  */
1484 #ifndef _SYS_SYSPROTO_H_
1485 struct olseek_args {
1486 	int	fd;
1487 	long	offset;
1488 	int	whence;
1489 };
1490 #endif
1491 int
1492 olseek(p, uap)
1493 	struct proc *p;
1494 	register struct olseek_args /* {
1495 		syscallarg(int) fd;
1496 		syscallarg(long) offset;
1497 		syscallarg(int) whence;
1498 	} */ *uap;
1499 {
1500 	struct lseek_args /* {
1501 		syscallarg(int) fd;
1502 		syscallarg(int) pad;
1503 		syscallarg(off_t) offset;
1504 		syscallarg(int) whence;
1505 	} */ nuap;
1506 	int error;
1507 
1508 	SCARG(&nuap, fd) = SCARG(uap, fd);
1509 	SCARG(&nuap, offset) = SCARG(uap, offset);
1510 	SCARG(&nuap, whence) = SCARG(uap, whence);
1511 	error = lseek(p, &nuap);
1512 	return (error);
1513 }
1514 #endif /* COMPAT_43 */
1515 
1516 /*
1517  * Check access permissions.
1518  */
1519 #ifndef _SYS_SYSPROTO_H_
1520 struct access_args {
1521 	char	*path;
1522 	int	flags;
1523 };
1524 #endif
1525 int
1526 access(p, uap)
1527 	struct proc *p;
1528 	register struct access_args /* {
1529 		syscallarg(char *) path;
1530 		syscallarg(int) flags;
1531 	} */ *uap;
1532 {
1533 	register struct ucred *cred = p->p_ucred;
1534 	register struct vnode *vp;
1535 	int error, flags, t_gid, t_uid;
1536 	struct nameidata nd;
1537 
1538 	t_uid = cred->cr_uid;
1539 	t_gid = cred->cr_groups[0];
1540 	cred->cr_uid = p->p_cred->p_ruid;
1541 	cred->cr_groups[0] = p->p_cred->p_rgid;
1542 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1543 	    SCARG(uap, path), p);
1544 	if ((error = namei(&nd)) != 0)
1545 		goto out1;
1546 	vp = nd.ni_vp;
1547 
1548 	/* Flags == 0 means only check for existence. */
1549 	if (SCARG(uap, flags)) {
1550 		flags = 0;
1551 		if (SCARG(uap, flags) & R_OK)
1552 			flags |= VREAD;
1553 		if (SCARG(uap, flags) & W_OK)
1554 			flags |= VWRITE;
1555 		if (SCARG(uap, flags) & X_OK)
1556 			flags |= VEXEC;
1557 		if ((flags & VWRITE) == 0 || (error = vn_writechk(vp)) == 0)
1558 			error = VOP_ACCESS(vp, flags, cred, p);
1559 	}
1560 	NDFREE(&nd, NDF_ONLY_PNBUF);
1561 	vput(vp);
1562 out1:
1563 	cred->cr_uid = t_uid;
1564 	cred->cr_groups[0] = t_gid;
1565 	return (error);
1566 }
1567 
1568 #if defined(COMPAT_43) || defined(COMPAT_SUNOS)
1569 /*
1570  * Get file status; this version follows links.
1571  */
1572 #ifndef _SYS_SYSPROTO_H_
1573 struct ostat_args {
1574 	char	*path;
1575 	struct ostat *ub;
1576 };
1577 #endif
1578 /* ARGSUSED */
1579 int
1580 ostat(p, uap)
1581 	struct proc *p;
1582 	register struct ostat_args /* {
1583 		syscallarg(char *) path;
1584 		syscallarg(struct ostat *) ub;
1585 	} */ *uap;
1586 {
1587 	struct stat sb;
1588 	struct ostat osb;
1589 	int error;
1590 	struct nameidata nd;
1591 
1592 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1593 	    SCARG(uap, path), p);
1594 	if ((error = namei(&nd)) != 0)
1595 		return (error);
1596 	NDFREE(&nd, NDF_ONLY_PNBUF);
1597 	error = vn_stat(nd.ni_vp, &sb, p);
1598 	vput(nd.ni_vp);
1599 	if (error)
1600 		return (error);
1601 	cvtstat(&sb, &osb);
1602 	error = copyout((caddr_t)&osb, (caddr_t)SCARG(uap, ub), sizeof (osb));
1603 	return (error);
1604 }
1605 
1606 /*
1607  * Get file status; this version does not follow links.
1608  */
1609 #ifndef _SYS_SYSPROTO_H_
1610 struct olstat_args {
1611 	char	*path;
1612 	struct ostat *ub;
1613 };
1614 #endif
1615 /* ARGSUSED */
1616 int
1617 olstat(p, uap)
1618 	struct proc *p;
1619 	register struct olstat_args /* {
1620 		syscallarg(char *) path;
1621 		syscallarg(struct ostat *) ub;
1622 	} */ *uap;
1623 {
1624 	struct vnode *vp;
1625 	struct stat sb;
1626 	struct ostat osb;
1627 	int error;
1628 	struct nameidata nd;
1629 
1630 	NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1631 	    SCARG(uap, path), p);
1632 	if ((error = namei(&nd)) != 0)
1633 		return (error);
1634 	vp = nd.ni_vp;
1635 	error = vn_stat(vp, &sb, p);
1636 	NDFREE(&nd, NDF_ONLY_PNBUF);
1637 	vput(vp);
1638 	if (error)
1639 		return (error);
1640 	cvtstat(&sb, &osb);
1641 	error = copyout((caddr_t)&osb, (caddr_t)SCARG(uap, ub), sizeof (osb));
1642 	return (error);
1643 }
1644 
1645 /*
1646  * Convert from an old to a new stat structure.
1647  */
1648 void
1649 cvtstat(st, ost)
1650 	struct stat *st;
1651 	struct ostat *ost;
1652 {
1653 
1654 	ost->st_dev = st->st_dev;
1655 	ost->st_ino = st->st_ino;
1656 	ost->st_mode = st->st_mode;
1657 	ost->st_nlink = st->st_nlink;
1658 	ost->st_uid = st->st_uid;
1659 	ost->st_gid = st->st_gid;
1660 	ost->st_rdev = st->st_rdev;
1661 	if (st->st_size < (quad_t)1 << 32)
1662 		ost->st_size = st->st_size;
1663 	else
1664 		ost->st_size = -2;
1665 	ost->st_atime = st->st_atime;
1666 	ost->st_mtime = st->st_mtime;
1667 	ost->st_ctime = st->st_ctime;
1668 	ost->st_blksize = st->st_blksize;
1669 	ost->st_blocks = st->st_blocks;
1670 	ost->st_flags = st->st_flags;
1671 	ost->st_gen = st->st_gen;
1672 }
1673 #endif /* COMPAT_43 || COMPAT_SUNOS */
1674 
1675 /*
1676  * Get file status; this version follows links.
1677  */
1678 #ifndef _SYS_SYSPROTO_H_
1679 struct stat_args {
1680 	char	*path;
1681 	struct stat *ub;
1682 };
1683 #endif
1684 /* ARGSUSED */
1685 int
1686 stat(p, uap)
1687 	struct proc *p;
1688 	register struct stat_args /* {
1689 		syscallarg(char *) path;
1690 		syscallarg(struct stat *) ub;
1691 	} */ *uap;
1692 {
1693 	struct stat sb;
1694 	int error;
1695 	struct nameidata nd;
1696 
1697 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1698 	    SCARG(uap, path), p);
1699 	if ((error = namei(&nd)) != 0)
1700 		return (error);
1701 	error = vn_stat(nd.ni_vp, &sb, p);
1702 	NDFREE(&nd, NDF_ONLY_PNBUF);
1703 	vput(nd.ni_vp);
1704 	if (error)
1705 		return (error);
1706 	error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
1707 	return (error);
1708 }
1709 
1710 /*
1711  * Get file status; this version does not follow links.
1712  */
1713 #ifndef _SYS_SYSPROTO_H_
1714 struct lstat_args {
1715 	char	*path;
1716 	struct stat *ub;
1717 };
1718 #endif
1719 /* ARGSUSED */
1720 int
1721 lstat(p, uap)
1722 	struct proc *p;
1723 	register struct lstat_args /* {
1724 		syscallarg(char *) path;
1725 		syscallarg(struct stat *) ub;
1726 	} */ *uap;
1727 {
1728 	int error;
1729 	struct vnode *vp;
1730 	struct stat sb;
1731 	struct nameidata nd;
1732 
1733 	NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1734 	    SCARG(uap, path), p);
1735 	if ((error = namei(&nd)) != 0)
1736 		return (error);
1737 	vp = nd.ni_vp;
1738 	error = vn_stat(vp, &sb, p);
1739 	NDFREE(&nd, NDF_ONLY_PNBUF);
1740 	vput(vp);
1741 	if (error)
1742 		return (error);
1743 	error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
1744 	return (error);
1745 }
1746 
1747 void
1748 cvtnstat(sb, nsb)
1749 	struct stat *sb;
1750 	struct nstat *nsb;
1751 {
1752 	nsb->st_dev = sb->st_dev;
1753 	nsb->st_ino = sb->st_ino;
1754 	nsb->st_mode = sb->st_mode;
1755 	nsb->st_nlink = sb->st_nlink;
1756 	nsb->st_uid = sb->st_uid;
1757 	nsb->st_gid = sb->st_gid;
1758 	nsb->st_rdev = sb->st_rdev;
1759 	nsb->st_atimespec = sb->st_atimespec;
1760 	nsb->st_mtimespec = sb->st_mtimespec;
1761 	nsb->st_ctimespec = sb->st_ctimespec;
1762 	nsb->st_size = sb->st_size;
1763 	nsb->st_blocks = sb->st_blocks;
1764 	nsb->st_blksize = sb->st_blksize;
1765 	nsb->st_flags = sb->st_flags;
1766 	nsb->st_gen = sb->st_gen;
1767 	nsb->st_qspare[0] = sb->st_qspare[0];
1768 	nsb->st_qspare[1] = sb->st_qspare[1];
1769 }
1770 
1771 #ifndef _SYS_SYSPROTO_H_
1772 struct nstat_args {
1773 	char	*path;
1774 	struct nstat *ub;
1775 };
1776 #endif
1777 /* ARGSUSED */
1778 int
1779 nstat(p, uap)
1780 	struct proc *p;
1781 	register struct nstat_args /* {
1782 		syscallarg(char *) path;
1783 		syscallarg(struct nstat *) ub;
1784 	} */ *uap;
1785 {
1786 	struct stat sb;
1787 	struct nstat nsb;
1788 	int error;
1789 	struct nameidata nd;
1790 
1791 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1792 	    SCARG(uap, path), p);
1793 	if ((error = namei(&nd)) != 0)
1794 		return (error);
1795 	NDFREE(&nd, NDF_ONLY_PNBUF);
1796 	error = vn_stat(nd.ni_vp, &sb, p);
1797 	vput(nd.ni_vp);
1798 	if (error)
1799 		return (error);
1800 	cvtnstat(&sb, &nsb);
1801 	error = copyout((caddr_t)&nsb, (caddr_t)SCARG(uap, ub), sizeof (nsb));
1802 	return (error);
1803 }
1804 
1805 /*
1806  * Get file status; this version does not follow links.
1807  */
1808 #ifndef _SYS_SYSPROTO_H_
1809 struct lstat_args {
1810 	char	*path;
1811 	struct stat *ub;
1812 };
1813 #endif
1814 /* ARGSUSED */
1815 int
1816 nlstat(p, uap)
1817 	struct proc *p;
1818 	register struct nlstat_args /* {
1819 		syscallarg(char *) path;
1820 		syscallarg(struct nstat *) ub;
1821 	} */ *uap;
1822 {
1823 	int error;
1824 	struct vnode *vp;
1825 	struct stat sb;
1826 	struct nstat nsb;
1827 	struct nameidata nd;
1828 
1829 	NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1830 	    SCARG(uap, path), p);
1831 	if ((error = namei(&nd)) != 0)
1832 		return (error);
1833 	vp = nd.ni_vp;
1834 	NDFREE(&nd, NDF_ONLY_PNBUF);
1835 	error = vn_stat(vp, &sb, p);
1836 	vput(vp);
1837 	if (error)
1838 		return (error);
1839 	cvtnstat(&sb, &nsb);
1840 	error = copyout((caddr_t)&nsb, (caddr_t)SCARG(uap, ub), sizeof (nsb));
1841 	return (error);
1842 }
1843 
1844 /*
1845  * Get configurable pathname variables.
1846  */
1847 #ifndef _SYS_SYSPROTO_H_
1848 struct pathconf_args {
1849 	char	*path;
1850 	int	name;
1851 };
1852 #endif
1853 /* ARGSUSED */
1854 int
1855 pathconf(p, uap)
1856 	struct proc *p;
1857 	register struct pathconf_args /* {
1858 		syscallarg(char *) path;
1859 		syscallarg(int) name;
1860 	} */ *uap;
1861 {
1862 	int error;
1863 	struct nameidata nd;
1864 
1865 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1866 	    SCARG(uap, path), p);
1867 	if ((error = namei(&nd)) != 0)
1868 		return (error);
1869 	NDFREE(&nd, NDF_ONLY_PNBUF);
1870 	error = VOP_PATHCONF(nd.ni_vp, SCARG(uap, name), p->p_retval);
1871 	vput(nd.ni_vp);
1872 	return (error);
1873 }
1874 
1875 /*
1876  * Return target name of a symbolic link.
1877  */
1878 #ifndef _SYS_SYSPROTO_H_
1879 struct readlink_args {
1880 	char	*path;
1881 	char	*buf;
1882 	int	count;
1883 };
1884 #endif
1885 /* ARGSUSED */
1886 int
1887 readlink(p, uap)
1888 	struct proc *p;
1889 	register struct readlink_args /* {
1890 		syscallarg(char *) path;
1891 		syscallarg(char *) buf;
1892 		syscallarg(int) count;
1893 	} */ *uap;
1894 {
1895 	register struct vnode *vp;
1896 	struct iovec aiov;
1897 	struct uio auio;
1898 	int error;
1899 	struct nameidata nd;
1900 
1901 	NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1902 	    SCARG(uap, path), p);
1903 	if ((error = namei(&nd)) != 0)
1904 		return (error);
1905 	NDFREE(&nd, NDF_ONLY_PNBUF);
1906 	vp = nd.ni_vp;
1907 	if (vp->v_type != VLNK)
1908 		error = EINVAL;
1909 	else {
1910 		aiov.iov_base = SCARG(uap, buf);
1911 		aiov.iov_len = SCARG(uap, count);
1912 		auio.uio_iov = &aiov;
1913 		auio.uio_iovcnt = 1;
1914 		auio.uio_offset = 0;
1915 		auio.uio_rw = UIO_READ;
1916 		auio.uio_segflg = UIO_USERSPACE;
1917 		auio.uio_procp = p;
1918 		auio.uio_resid = SCARG(uap, count);
1919 		error = VOP_READLINK(vp, &auio, p->p_ucred);
1920 	}
1921 	vput(vp);
1922 	p->p_retval[0] = SCARG(uap, count) - auio.uio_resid;
1923 	return (error);
1924 }
1925 
1926 static int
1927 setfflags(p, vp, flags)
1928 	struct proc *p;
1929 	struct vnode *vp;
1930 	int flags;
1931 {
1932 	int error;
1933 	struct vattr vattr;
1934 
1935 	/*
1936 	 * Prevent non-root users from setting flags on devices.  When
1937 	 * a device is reused, users can retain ownership of the device
1938 	 * if they are allowed to set flags and programs assume that
1939 	 * chown can't fail when done as root.
1940 	 */
1941 	if ((vp->v_type == VCHR || vp->v_type == VBLK) &&
1942 	    ((error = suser_xxx(p->p_ucred, p, PRISON_ROOT)) != 0))
1943 		return (error);
1944 
1945 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1946 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
1947 	VATTR_NULL(&vattr);
1948 	vattr.va_flags = flags;
1949 	error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1950 	VOP_UNLOCK(vp, 0, p);
1951 	return (error);
1952 }
1953 
1954 /*
1955  * Change flags of a file given a path name.
1956  */
1957 #ifndef _SYS_SYSPROTO_H_
1958 struct chflags_args {
1959 	char	*path;
1960 	int	flags;
1961 };
1962 #endif
1963 /* ARGSUSED */
1964 int
1965 chflags(p, uap)
1966 	struct proc *p;
1967 	register struct chflags_args /* {
1968 		syscallarg(char *) path;
1969 		syscallarg(int) flags;
1970 	} */ *uap;
1971 {
1972 	int error;
1973 	struct nameidata nd;
1974 
1975 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1976 	if ((error = namei(&nd)) != 0)
1977 		return (error);
1978 	NDFREE(&nd, NDF_ONLY_PNBUF);
1979 	error = setfflags(p, nd.ni_vp, SCARG(uap, flags));
1980 	vrele(nd.ni_vp);
1981 	return error;
1982 }
1983 
1984 /*
1985  * Change flags of a file given a file descriptor.
1986  */
1987 #ifndef _SYS_SYSPROTO_H_
1988 struct fchflags_args {
1989 	int	fd;
1990 	int	flags;
1991 };
1992 #endif
1993 /* ARGSUSED */
1994 int
1995 fchflags(p, uap)
1996 	struct proc *p;
1997 	register struct fchflags_args /* {
1998 		syscallarg(int) fd;
1999 		syscallarg(int) flags;
2000 	} */ *uap;
2001 {
2002 	struct file *fp;
2003 	int error;
2004 
2005 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2006 		return (error);
2007 	return setfflags(p, (struct vnode *) fp->f_data, SCARG(uap, flags));
2008 }
2009 
2010 static int
2011 setfmode(p, vp, mode)
2012 	struct proc *p;
2013 	struct vnode *vp;
2014 	int mode;
2015 {
2016 	int error;
2017 	struct vattr vattr;
2018 
2019 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2020 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2021 	VATTR_NULL(&vattr);
2022 	vattr.va_mode = mode & ALLPERMS;
2023 	error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2024 	VOP_UNLOCK(vp, 0, p);
2025 	return error;
2026 }
2027 
2028 /*
2029  * Change mode of a file given path name.
2030  */
2031 #ifndef _SYS_SYSPROTO_H_
2032 struct chmod_args {
2033 	char	*path;
2034 	int	mode;
2035 };
2036 #endif
2037 /* ARGSUSED */
2038 int
2039 chmod(p, uap)
2040 	struct proc *p;
2041 	register struct chmod_args /* {
2042 		syscallarg(char *) path;
2043 		syscallarg(int) mode;
2044 	} */ *uap;
2045 {
2046 	int error;
2047 	struct nameidata nd;
2048 
2049 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2050 	if ((error = namei(&nd)) != 0)
2051 		return (error);
2052 	NDFREE(&nd, NDF_ONLY_PNBUF);
2053 	error = setfmode(p, nd.ni_vp, SCARG(uap, mode));
2054 	vrele(nd.ni_vp);
2055 	return error;
2056 }
2057 
2058 /*
2059  * Change mode of a file given path name (don't follow links.)
2060  */
2061 #ifndef _SYS_SYSPROTO_H_
2062 struct lchmod_args {
2063 	char	*path;
2064 	int	mode;
2065 };
2066 #endif
2067 /* ARGSUSED */
2068 int
2069 lchmod(p, uap)
2070 	struct proc *p;
2071 	register struct lchmod_args /* {
2072 		syscallarg(char *) path;
2073 		syscallarg(int) mode;
2074 	} */ *uap;
2075 {
2076 	int error;
2077 	struct nameidata nd;
2078 
2079 	NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2080 	if ((error = namei(&nd)) != 0)
2081 		return (error);
2082 	NDFREE(&nd, NDF_ONLY_PNBUF);
2083 	error = setfmode(p, nd.ni_vp, SCARG(uap, mode));
2084 	vrele(nd.ni_vp);
2085 	return error;
2086 }
2087 
2088 /*
2089  * Change mode of a file given a file descriptor.
2090  */
2091 #ifndef _SYS_SYSPROTO_H_
2092 struct fchmod_args {
2093 	int	fd;
2094 	int	mode;
2095 };
2096 #endif
2097 /* ARGSUSED */
2098 int
2099 fchmod(p, uap)
2100 	struct proc *p;
2101 	register struct fchmod_args /* {
2102 		syscallarg(int) fd;
2103 		syscallarg(int) mode;
2104 	} */ *uap;
2105 {
2106 	struct file *fp;
2107 	int error;
2108 
2109 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2110 		return (error);
2111 	return setfmode(p, (struct vnode *)fp->f_data, SCARG(uap, mode));
2112 }
2113 
2114 static int
2115 setfown(p, vp, uid, gid)
2116 	struct proc *p;
2117 	struct vnode *vp;
2118 	uid_t uid;
2119 	gid_t gid;
2120 {
2121 	int error;
2122 	struct vattr vattr;
2123 
2124 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2125 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2126 	VATTR_NULL(&vattr);
2127 	vattr.va_uid = uid;
2128 	vattr.va_gid = gid;
2129 	error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2130 	VOP_UNLOCK(vp, 0, p);
2131 	return error;
2132 }
2133 
2134 /*
2135  * Set ownership given a path name.
2136  */
2137 #ifndef _SYS_SYSPROTO_H_
2138 struct chown_args {
2139 	char	*path;
2140 	int	uid;
2141 	int	gid;
2142 };
2143 #endif
2144 /* ARGSUSED */
2145 int
2146 chown(p, uap)
2147 	struct proc *p;
2148 	register struct chown_args /* {
2149 		syscallarg(char *) path;
2150 		syscallarg(int) uid;
2151 		syscallarg(int) gid;
2152 	} */ *uap;
2153 {
2154 	int error;
2155 	struct nameidata nd;
2156 
2157 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2158 	if ((error = namei(&nd)) != 0)
2159 		return (error);
2160 	NDFREE(&nd, NDF_ONLY_PNBUF);
2161 	error = setfown(p, nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid));
2162 	vrele(nd.ni_vp);
2163 	return (error);
2164 }
2165 
2166 /*
2167  * Set ownership given a path name, do not cross symlinks.
2168  */
2169 #ifndef _SYS_SYSPROTO_H_
2170 struct lchown_args {
2171 	char	*path;
2172 	int	uid;
2173 	int	gid;
2174 };
2175 #endif
2176 /* ARGSUSED */
2177 int
2178 lchown(p, uap)
2179 	struct proc *p;
2180 	register struct lchown_args /* {
2181 		syscallarg(char *) path;
2182 		syscallarg(int) uid;
2183 		syscallarg(int) gid;
2184 	} */ *uap;
2185 {
2186 	int error;
2187 	struct nameidata nd;
2188 
2189 	NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2190 	if ((error = namei(&nd)) != 0)
2191 		return (error);
2192 	NDFREE(&nd, NDF_ONLY_PNBUF);
2193 	error = setfown(p, nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid));
2194 	vrele(nd.ni_vp);
2195 	return (error);
2196 }
2197 
2198 /*
2199  * Set ownership given a file descriptor.
2200  */
2201 #ifndef _SYS_SYSPROTO_H_
2202 struct fchown_args {
2203 	int	fd;
2204 	int	uid;
2205 	int	gid;
2206 };
2207 #endif
2208 /* ARGSUSED */
2209 int
2210 fchown(p, uap)
2211 	struct proc *p;
2212 	register struct fchown_args /* {
2213 		syscallarg(int) fd;
2214 		syscallarg(int) uid;
2215 		syscallarg(int) gid;
2216 	} */ *uap;
2217 {
2218 	struct file *fp;
2219 	int error;
2220 
2221 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2222 		return (error);
2223 	return setfown(p, (struct vnode *)fp->f_data,
2224 		SCARG(uap, uid), SCARG(uap, gid));
2225 }
2226 
2227 static int
2228 getutimes(usrtvp, tsp)
2229 	const struct timeval *usrtvp;
2230 	struct timespec *tsp;
2231 {
2232 	struct timeval tv[2];
2233 	int error;
2234 
2235 	if (usrtvp == NULL) {
2236 		microtime(&tv[0]);
2237 		TIMEVAL_TO_TIMESPEC(&tv[0], &tsp[0]);
2238 		tsp[1] = tsp[0];
2239 	} else {
2240 		if ((error = copyin(usrtvp, tv, sizeof (tv))) != 0)
2241 			return (error);
2242 		TIMEVAL_TO_TIMESPEC(&tv[0], &tsp[0]);
2243 		TIMEVAL_TO_TIMESPEC(&tv[1], &tsp[1]);
2244 	}
2245 	return 0;
2246 }
2247 
2248 static int
2249 setutimes(p, vp, ts, nullflag)
2250 	struct proc *p;
2251 	struct vnode *vp;
2252 	const struct timespec *ts;
2253 	int nullflag;
2254 {
2255 	int error;
2256 	struct vattr vattr;
2257 
2258 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2259 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2260 	VATTR_NULL(&vattr);
2261 	vattr.va_atime = ts[0];
2262 	vattr.va_mtime = ts[1];
2263 	if (nullflag)
2264 		vattr.va_vaflags |= VA_UTIMES_NULL;
2265 	error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2266 	VOP_UNLOCK(vp, 0, p);
2267 	return error;
2268 }
2269 
2270 /*
2271  * Set the access and modification times of a file.
2272  */
2273 #ifndef _SYS_SYSPROTO_H_
2274 struct utimes_args {
2275 	char	*path;
2276 	struct	timeval *tptr;
2277 };
2278 #endif
2279 /* ARGSUSED */
2280 int
2281 utimes(p, uap)
2282 	struct proc *p;
2283 	register struct utimes_args /* {
2284 		syscallarg(char *) path;
2285 		syscallarg(struct timeval *) tptr;
2286 	} */ *uap;
2287 {
2288 	struct timespec ts[2];
2289 	struct timeval *usrtvp;
2290 	int error;
2291 	struct nameidata nd;
2292 
2293 	usrtvp = SCARG(uap, tptr);
2294 	if ((error = getutimes(usrtvp, ts)) != 0)
2295 		return (error);
2296 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2297 	if ((error = namei(&nd)) != 0)
2298 		return (error);
2299 	NDFREE(&nd, NDF_ONLY_PNBUF);
2300 	error = setutimes(p, nd.ni_vp, ts, usrtvp == NULL);
2301 	vrele(nd.ni_vp);
2302 	return (error);
2303 }
2304 
2305 /*
2306  * Set the access and modification times of a file.
2307  */
2308 #ifndef _SYS_SYSPROTO_H_
2309 struct lutimes_args {
2310 	char	*path;
2311 	struct	timeval *tptr;
2312 };
2313 #endif
2314 /* ARGSUSED */
2315 int
2316 lutimes(p, uap)
2317 	struct proc *p;
2318 	register struct lutimes_args /* {
2319 		syscallarg(char *) path;
2320 		syscallarg(struct timeval *) tptr;
2321 	} */ *uap;
2322 {
2323 	struct timespec ts[2];
2324 	struct timeval *usrtvp;
2325 	int error;
2326 	struct nameidata nd;
2327 
2328 	usrtvp = SCARG(uap, tptr);
2329 	if ((error = getutimes(usrtvp, ts)) != 0)
2330 		return (error);
2331 	NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2332 	if ((error = namei(&nd)) != 0)
2333 		return (error);
2334 	NDFREE(&nd, NDF_ONLY_PNBUF);
2335 	error = setutimes(p, nd.ni_vp, ts, usrtvp == NULL);
2336 	vrele(nd.ni_vp);
2337 	return (error);
2338 }
2339 
2340 /*
2341  * Set the access and modification times of a file.
2342  */
2343 #ifndef _SYS_SYSPROTO_H_
2344 struct futimes_args {
2345 	int	fd;
2346 	struct	timeval *tptr;
2347 };
2348 #endif
2349 /* ARGSUSED */
2350 int
2351 futimes(p, uap)
2352 	struct proc *p;
2353 	register struct futimes_args /* {
2354 		syscallarg(int ) fd;
2355 		syscallarg(struct timeval *) tptr;
2356 	} */ *uap;
2357 {
2358 	struct timespec ts[2];
2359 	struct file *fp;
2360 	struct timeval *usrtvp;
2361 	int error;
2362 
2363 	usrtvp = SCARG(uap, tptr);
2364 	if ((error = getutimes(usrtvp, ts)) != 0)
2365 		return (error);
2366 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2367 		return (error);
2368 	return setutimes(p, (struct vnode *)fp->f_data, ts, usrtvp == NULL);
2369 }
2370 
2371 /*
2372  * Truncate a file given its path name.
2373  */
2374 #ifndef _SYS_SYSPROTO_H_
2375 struct truncate_args {
2376 	char	*path;
2377 	int	pad;
2378 	off_t	length;
2379 };
2380 #endif
2381 /* ARGSUSED */
2382 int
2383 truncate(p, uap)
2384 	struct proc *p;
2385 	register struct truncate_args /* {
2386 		syscallarg(char *) path;
2387 		syscallarg(int) pad;
2388 		syscallarg(off_t) length;
2389 	} */ *uap;
2390 {
2391 	register struct vnode *vp;
2392 	struct vattr vattr;
2393 	int error;
2394 	struct nameidata nd;
2395 
2396 	if (uap->length < 0)
2397 		return(EINVAL);
2398 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2399 	if ((error = namei(&nd)) != 0)
2400 		return (error);
2401 	vp = nd.ni_vp;
2402 	NDFREE(&nd, NDF_ONLY_PNBUF);
2403 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2404 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2405 	if (vp->v_type == VDIR)
2406 		error = EISDIR;
2407 	else if ((error = vn_writechk(vp)) == 0 &&
2408 	    (error = VOP_ACCESS(vp, VWRITE, p->p_ucred, p)) == 0) {
2409 		VATTR_NULL(&vattr);
2410 		vattr.va_size = SCARG(uap, length);
2411 		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2412 	}
2413 	vput(vp);
2414 	return (error);
2415 }
2416 
2417 /*
2418  * Truncate a file given a file descriptor.
2419  */
2420 #ifndef _SYS_SYSPROTO_H_
2421 struct ftruncate_args {
2422 	int	fd;
2423 	int	pad;
2424 	off_t	length;
2425 };
2426 #endif
2427 /* ARGSUSED */
2428 int
2429 ftruncate(p, uap)
2430 	struct proc *p;
2431 	register struct ftruncate_args /* {
2432 		syscallarg(int) fd;
2433 		syscallarg(int) pad;
2434 		syscallarg(off_t) length;
2435 	} */ *uap;
2436 {
2437 	struct vattr vattr;
2438 	struct vnode *vp;
2439 	struct file *fp;
2440 	int error;
2441 
2442 	if (uap->length < 0)
2443 		return(EINVAL);
2444 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2445 		return (error);
2446 	if ((fp->f_flag & FWRITE) == 0)
2447 		return (EINVAL);
2448 	vp = (struct vnode *)fp->f_data;
2449 	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2450 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2451 	if (vp->v_type == VDIR)
2452 		error = EISDIR;
2453 	else if ((error = vn_writechk(vp)) == 0) {
2454 		VATTR_NULL(&vattr);
2455 		vattr.va_size = SCARG(uap, length);
2456 		error = VOP_SETATTR(vp, &vattr, fp->f_cred, p);
2457 	}
2458 	VOP_UNLOCK(vp, 0, p);
2459 	return (error);
2460 }
2461 
2462 #if defined(COMPAT_43) || defined(COMPAT_SUNOS)
2463 /*
2464  * Truncate a file given its path name.
2465  */
2466 #ifndef _SYS_SYSPROTO_H_
2467 struct otruncate_args {
2468 	char	*path;
2469 	long	length;
2470 };
2471 #endif
2472 /* ARGSUSED */
2473 int
2474 otruncate(p, uap)
2475 	struct proc *p;
2476 	register struct otruncate_args /* {
2477 		syscallarg(char *) path;
2478 		syscallarg(long) length;
2479 	} */ *uap;
2480 {
2481 	struct truncate_args /* {
2482 		syscallarg(char *) path;
2483 		syscallarg(int) pad;
2484 		syscallarg(off_t) length;
2485 	} */ nuap;
2486 
2487 	SCARG(&nuap, path) = SCARG(uap, path);
2488 	SCARG(&nuap, length) = SCARG(uap, length);
2489 	return (truncate(p, &nuap));
2490 }
2491 
2492 /*
2493  * Truncate a file given a file descriptor.
2494  */
2495 #ifndef _SYS_SYSPROTO_H_
2496 struct oftruncate_args {
2497 	int	fd;
2498 	long	length;
2499 };
2500 #endif
2501 /* ARGSUSED */
2502 int
2503 oftruncate(p, uap)
2504 	struct proc *p;
2505 	register struct oftruncate_args /* {
2506 		syscallarg(int) fd;
2507 		syscallarg(long) length;
2508 	} */ *uap;
2509 {
2510 	struct ftruncate_args /* {
2511 		syscallarg(int) fd;
2512 		syscallarg(int) pad;
2513 		syscallarg(off_t) length;
2514 	} */ nuap;
2515 
2516 	SCARG(&nuap, fd) = SCARG(uap, fd);
2517 	SCARG(&nuap, length) = SCARG(uap, length);
2518 	return (ftruncate(p, &nuap));
2519 }
2520 #endif /* COMPAT_43 || COMPAT_SUNOS */
2521 
2522 /*
2523  * Sync an open file.
2524  */
2525 #ifndef _SYS_SYSPROTO_H_
2526 struct fsync_args {
2527 	int	fd;
2528 };
2529 #endif
2530 /* ARGSUSED */
2531 int
2532 fsync(p, uap)
2533 	struct proc *p;
2534 	struct fsync_args /* {
2535 		syscallarg(int) fd;
2536 	} */ *uap;
2537 {
2538 	register struct vnode *vp;
2539 	struct file *fp;
2540 	int error;
2541 
2542 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2543 		return (error);
2544 	vp = (struct vnode *)fp->f_data;
2545 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2546 	if (vp->v_object)
2547 		vm_object_page_clean(vp->v_object, 0, 0, 0);
2548 	if ((error = VOP_FSYNC(vp, fp->f_cred, MNT_WAIT, p)) == 0 &&
2549 	    vp->v_mount && (vp->v_mount->mnt_flag & MNT_SOFTDEP) &&
2550 	    bioops.io_fsync)
2551 		error = (*bioops.io_fsync)(vp);
2552 	VOP_UNLOCK(vp, 0, p);
2553 	return (error);
2554 }
2555 
2556 /*
2557  * Rename files.  Source and destination must either both be directories,
2558  * or both not be directories.  If target is a directory, it must be empty.
2559  */
2560 #ifndef _SYS_SYSPROTO_H_
2561 struct rename_args {
2562 	char	*from;
2563 	char	*to;
2564 };
2565 #endif
2566 /* ARGSUSED */
2567 int
2568 rename(p, uap)
2569 	struct proc *p;
2570 	register struct rename_args /* {
2571 		syscallarg(char *) from;
2572 		syscallarg(char *) to;
2573 	} */ *uap;
2574 {
2575 	register struct vnode *tvp, *fvp, *tdvp;
2576 	struct nameidata fromnd, tond;
2577 	int error;
2578 
2579 	bwillwrite();
2580 	NDINIT(&fromnd, DELETE, WANTPARENT | SAVESTART, UIO_USERSPACE,
2581 	    SCARG(uap, from), p);
2582 	if ((error = namei(&fromnd)) != 0)
2583 		return (error);
2584 	fvp = fromnd.ni_vp;
2585 	NDINIT(&tond, RENAME, LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART | NOOBJ,
2586 	    UIO_USERSPACE, SCARG(uap, to), p);
2587 	if (fromnd.ni_vp->v_type == VDIR)
2588 		tond.ni_cnd.cn_flags |= WILLBEDIR;
2589 	if ((error = namei(&tond)) != 0) {
2590 		/* Translate error code for rename("dir1", "dir2/."). */
2591 		if (error == EISDIR && fvp->v_type == VDIR)
2592 			error = EINVAL;
2593 		NDFREE(&fromnd, NDF_ONLY_PNBUF);
2594 		vrele(fromnd.ni_dvp);
2595 		vrele(fvp);
2596 		goto out1;
2597 	}
2598 	tdvp = tond.ni_dvp;
2599 	tvp = tond.ni_vp;
2600 	if (tvp != NULL) {
2601 		if (fvp->v_type == VDIR && tvp->v_type != VDIR) {
2602 			error = ENOTDIR;
2603 			goto out;
2604 		} else if (fvp->v_type != VDIR && tvp->v_type == VDIR) {
2605 			error = EISDIR;
2606 			goto out;
2607 		}
2608 	}
2609 	if (fvp == tdvp)
2610 		error = EINVAL;
2611 	/*
2612 	 * If source is the same as the destination (that is the
2613 	 * same inode number with the same name in the same directory),
2614 	 * then there is nothing to do.
2615 	 */
2616 	if (fvp == tvp && fromnd.ni_dvp == tdvp &&
2617 	    fromnd.ni_cnd.cn_namelen == tond.ni_cnd.cn_namelen &&
2618 	    !bcmp(fromnd.ni_cnd.cn_nameptr, tond.ni_cnd.cn_nameptr,
2619 	      fromnd.ni_cnd.cn_namelen))
2620 		error = -1;
2621 out:
2622 	if (!error) {
2623 		VOP_LEASE(tdvp, p, p->p_ucred, LEASE_WRITE);
2624 		if (fromnd.ni_dvp != tdvp) {
2625 			VOP_LEASE(fromnd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
2626 		}
2627 		if (tvp) {
2628 			VOP_LEASE(tvp, p, p->p_ucred, LEASE_WRITE);
2629 		}
2630 		error = VOP_RENAME(fromnd.ni_dvp, fromnd.ni_vp, &fromnd.ni_cnd,
2631 				   tond.ni_dvp, tond.ni_vp, &tond.ni_cnd);
2632 		NDFREE(&fromnd, NDF_ONLY_PNBUF);
2633 		NDFREE(&tond, NDF_ONLY_PNBUF);
2634 	} else {
2635 		NDFREE(&fromnd, NDF_ONLY_PNBUF);
2636 		NDFREE(&tond, NDF_ONLY_PNBUF);
2637 		if (tdvp == tvp)
2638 			vrele(tdvp);
2639 		else
2640 			vput(tdvp);
2641 		if (tvp)
2642 			vput(tvp);
2643 		vrele(fromnd.ni_dvp);
2644 		vrele(fvp);
2645 	}
2646 	vrele(tond.ni_startdir);
2647 	ASSERT_VOP_UNLOCKED(fromnd.ni_dvp, "rename");
2648 	ASSERT_VOP_UNLOCKED(fromnd.ni_vp, "rename");
2649 	ASSERT_VOP_UNLOCKED(tond.ni_dvp, "rename");
2650 	ASSERT_VOP_UNLOCKED(tond.ni_vp, "rename");
2651 out1:
2652 	if (fromnd.ni_startdir)
2653 		vrele(fromnd.ni_startdir);
2654 	if (error == -1)
2655 		return (0);
2656 	return (error);
2657 }
2658 
2659 /*
2660  * Make a directory file.
2661  */
2662 #ifndef _SYS_SYSPROTO_H_
2663 struct mkdir_args {
2664 	char	*path;
2665 	int	mode;
2666 };
2667 #endif
2668 /* ARGSUSED */
2669 int
2670 mkdir(p, uap)
2671 	struct proc *p;
2672 	register struct mkdir_args /* {
2673 		syscallarg(char *) path;
2674 		syscallarg(int) mode;
2675 	} */ *uap;
2676 {
2677 	register struct vnode *vp;
2678 	struct vattr vattr;
2679 	int error;
2680 	struct nameidata nd;
2681 
2682 	bwillwrite();
2683 	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
2684 	nd.ni_cnd.cn_flags |= WILLBEDIR;
2685 	if ((error = namei(&nd)) != 0)
2686 		return (error);
2687 	vp = nd.ni_vp;
2688 	if (vp != NULL) {
2689 		NDFREE(&nd, NDF_ONLY_PNBUF);
2690 		if (nd.ni_dvp == vp)
2691 			vrele(nd.ni_dvp);
2692 		else
2693 			vput(nd.ni_dvp);
2694 		vrele(vp);
2695 		return (EEXIST);
2696 	}
2697 	VATTR_NULL(&vattr);
2698 	vattr.va_type = VDIR;
2699 	vattr.va_mode = (SCARG(uap, mode) & ACCESSPERMS) &~ p->p_fd->fd_cmask;
2700 	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
2701 	error = VOP_MKDIR(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
2702 	NDFREE(&nd, NDF_ONLY_PNBUF);
2703 	vput(nd.ni_dvp);
2704 	if (!error)
2705 		vput(nd.ni_vp);
2706 	ASSERT_VOP_UNLOCKED(nd.ni_dvp, "mkdir");
2707 	ASSERT_VOP_UNLOCKED(nd.ni_vp, "mkdir");
2708 	return (error);
2709 }
2710 
2711 /*
2712  * Remove a directory file.
2713  */
2714 #ifndef _SYS_SYSPROTO_H_
2715 struct rmdir_args {
2716 	char	*path;
2717 };
2718 #endif
2719 /* ARGSUSED */
2720 int
2721 rmdir(p, uap)
2722 	struct proc *p;
2723 	struct rmdir_args /* {
2724 		syscallarg(char *) path;
2725 	} */ *uap;
2726 {
2727 	register struct vnode *vp;
2728 	int error;
2729 	struct nameidata nd;
2730 
2731 	bwillwrite();
2732 	NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
2733 	    SCARG(uap, path), p);
2734 	if ((error = namei(&nd)) != 0)
2735 		return (error);
2736 	vp = nd.ni_vp;
2737 	if (vp->v_type != VDIR) {
2738 		error = ENOTDIR;
2739 		goto out;
2740 	}
2741 	/*
2742 	 * No rmdir "." please.
2743 	 */
2744 	if (nd.ni_dvp == vp) {
2745 		error = EINVAL;
2746 		goto out;
2747 	}
2748 	/*
2749 	 * The root of a mounted filesystem cannot be deleted.
2750 	 */
2751 	if (vp->v_flag & VROOT)
2752 		error = EBUSY;
2753 	else {
2754 		VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
2755 		VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2756 		error = VOP_RMDIR(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
2757 	}
2758 out:
2759 	NDFREE(&nd, NDF_ONLY_PNBUF);
2760 	if (nd.ni_dvp == vp)
2761 		vrele(nd.ni_dvp);
2762 	else
2763 		vput(nd.ni_dvp);
2764 	if (vp != NULLVP)
2765 		vput(vp);
2766 	ASSERT_VOP_UNLOCKED(nd.ni_dvp, "rmdir");
2767 	ASSERT_VOP_UNLOCKED(nd.ni_vp, "rmdir");
2768 	return (error);
2769 }
2770 
2771 #ifdef COMPAT_43
2772 /*
2773  * Read a block of directory entries in a file system independent format.
2774  */
2775 #ifndef _SYS_SYSPROTO_H_
2776 struct ogetdirentries_args {
2777 	int	fd;
2778 	char	*buf;
2779 	u_int	count;
2780 	long	*basep;
2781 };
2782 #endif
2783 int
2784 ogetdirentries(p, uap)
2785 	struct proc *p;
2786 	register struct ogetdirentries_args /* {
2787 		syscallarg(int) fd;
2788 		syscallarg(char *) buf;
2789 		syscallarg(u_int) count;
2790 		syscallarg(long *) basep;
2791 	} */ *uap;
2792 {
2793 	struct vnode *vp;
2794 	struct file *fp;
2795 	struct uio auio, kuio;
2796 	struct iovec aiov, kiov;
2797 	struct dirent *dp, *edp;
2798 	caddr_t dirbuf;
2799 	int error, eofflag, readcnt;
2800 	long loff;
2801 
2802 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2803 		return (error);
2804 	if ((fp->f_flag & FREAD) == 0)
2805 		return (EBADF);
2806 	vp = (struct vnode *)fp->f_data;
2807 unionread:
2808 	if (vp->v_type != VDIR)
2809 		return (EINVAL);
2810 	aiov.iov_base = SCARG(uap, buf);
2811 	aiov.iov_len = SCARG(uap, count);
2812 	auio.uio_iov = &aiov;
2813 	auio.uio_iovcnt = 1;
2814 	auio.uio_rw = UIO_READ;
2815 	auio.uio_segflg = UIO_USERSPACE;
2816 	auio.uio_procp = p;
2817 	auio.uio_resid = SCARG(uap, count);
2818 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2819 	loff = auio.uio_offset = fp->f_offset;
2820 #	if (BYTE_ORDER != LITTLE_ENDIAN)
2821 		if (vp->v_mount->mnt_maxsymlinklen <= 0) {
2822 			error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag,
2823 			    NULL, NULL);
2824 			fp->f_offset = auio.uio_offset;
2825 		} else
2826 #	endif
2827 	{
2828 		kuio = auio;
2829 		kuio.uio_iov = &kiov;
2830 		kuio.uio_segflg = UIO_SYSSPACE;
2831 		kiov.iov_len = SCARG(uap, count);
2832 		MALLOC(dirbuf, caddr_t, SCARG(uap, count), M_TEMP, M_WAITOK);
2833 		kiov.iov_base = dirbuf;
2834 		error = VOP_READDIR(vp, &kuio, fp->f_cred, &eofflag,
2835 			    NULL, NULL);
2836 		fp->f_offset = kuio.uio_offset;
2837 		if (error == 0) {
2838 			readcnt = SCARG(uap, count) - kuio.uio_resid;
2839 			edp = (struct dirent *)&dirbuf[readcnt];
2840 			for (dp = (struct dirent *)dirbuf; dp < edp; ) {
2841 #				if (BYTE_ORDER == LITTLE_ENDIAN)
2842 					/*
2843 					 * The expected low byte of
2844 					 * dp->d_namlen is our dp->d_type.
2845 					 * The high MBZ byte of dp->d_namlen
2846 					 * is our dp->d_namlen.
2847 					 */
2848 					dp->d_type = dp->d_namlen;
2849 					dp->d_namlen = 0;
2850 #				else
2851 					/*
2852 					 * The dp->d_type is the high byte
2853 					 * of the expected dp->d_namlen,
2854 					 * so must be zero'ed.
2855 					 */
2856 					dp->d_type = 0;
2857 #				endif
2858 				if (dp->d_reclen > 0) {
2859 					dp = (struct dirent *)
2860 					    ((char *)dp + dp->d_reclen);
2861 				} else {
2862 					error = EIO;
2863 					break;
2864 				}
2865 			}
2866 			if (dp >= edp)
2867 				error = uiomove(dirbuf, readcnt, &auio);
2868 		}
2869 		FREE(dirbuf, M_TEMP);
2870 	}
2871 	VOP_UNLOCK(vp, 0, p);
2872 	if (error)
2873 		return (error);
2874 	if (SCARG(uap, count) == auio.uio_resid) {
2875 		if (union_dircheckp) {
2876 			error = union_dircheckp(p, &vp, fp);
2877 			if (error == -1)
2878 				goto unionread;
2879 			if (error)
2880 				return (error);
2881 		}
2882 		if ((vp->v_flag & VROOT) &&
2883 		    (vp->v_mount->mnt_flag & MNT_UNION)) {
2884 			struct vnode *tvp = vp;
2885 			vp = vp->v_mount->mnt_vnodecovered;
2886 			VREF(vp);
2887 			fp->f_data = (caddr_t) vp;
2888 			fp->f_offset = 0;
2889 			vrele(tvp);
2890 			goto unionread;
2891 		}
2892 	}
2893 	error = copyout((caddr_t)&loff, (caddr_t)SCARG(uap, basep),
2894 	    sizeof(long));
2895 	p->p_retval[0] = SCARG(uap, count) - auio.uio_resid;
2896 	return (error);
2897 }
2898 #endif /* COMPAT_43 */
2899 
2900 /*
2901  * Read a block of directory entries in a file system independent format.
2902  */
2903 #ifndef _SYS_SYSPROTO_H_
2904 struct getdirentries_args {
2905 	int	fd;
2906 	char	*buf;
2907 	u_int	count;
2908 	long	*basep;
2909 };
2910 #endif
2911 int
2912 getdirentries(p, uap)
2913 	struct proc *p;
2914 	register struct getdirentries_args /* {
2915 		syscallarg(int) fd;
2916 		syscallarg(char *) buf;
2917 		syscallarg(u_int) count;
2918 		syscallarg(long *) basep;
2919 	} */ *uap;
2920 {
2921 	struct vnode *vp;
2922 	struct file *fp;
2923 	struct uio auio;
2924 	struct iovec aiov;
2925 	long loff;
2926 	int error, eofflag;
2927 
2928 	if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2929 		return (error);
2930 	if ((fp->f_flag & FREAD) == 0)
2931 		return (EBADF);
2932 	vp = (struct vnode *)fp->f_data;
2933 unionread:
2934 	if (vp->v_type != VDIR)
2935 		return (EINVAL);
2936 	aiov.iov_base = SCARG(uap, buf);
2937 	aiov.iov_len = SCARG(uap, count);
2938 	auio.uio_iov = &aiov;
2939 	auio.uio_iovcnt = 1;
2940 	auio.uio_rw = UIO_READ;
2941 	auio.uio_segflg = UIO_USERSPACE;
2942 	auio.uio_procp = p;
2943 	auio.uio_resid = SCARG(uap, count);
2944 	/* vn_lock(vp, LK_SHARED | LK_RETRY, p); */
2945 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2946 	loff = auio.uio_offset = fp->f_offset;
2947 	error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, NULL, NULL);
2948 	fp->f_offset = auio.uio_offset;
2949 	VOP_UNLOCK(vp, 0, p);
2950 	if (error)
2951 		return (error);
2952 	if (SCARG(uap, count) == auio.uio_resid) {
2953 		if (union_dircheckp) {
2954 			error = union_dircheckp(p, &vp, fp);
2955 			if (error == -1)
2956 				goto unionread;
2957 			if (error)
2958 				return (error);
2959 		}
2960 		if ((vp->v_flag & VROOT) &&
2961 		    (vp->v_mount->mnt_flag & MNT_UNION)) {
2962 			struct vnode *tvp = vp;
2963 			vp = vp->v_mount->mnt_vnodecovered;
2964 			VREF(vp);
2965 			fp->f_data = (caddr_t) vp;
2966 			fp->f_offset = 0;
2967 			vrele(tvp);
2968 			goto unionread;
2969 		}
2970 	}
2971 	if (SCARG(uap, basep) != NULL) {
2972 		error = copyout((caddr_t)&loff, (caddr_t)SCARG(uap, basep),
2973 		    sizeof(long));
2974 	}
2975 	p->p_retval[0] = SCARG(uap, count) - auio.uio_resid;
2976 	return (error);
2977 }
2978 #ifndef _SYS_SYSPROTO_H_
2979 struct getdents_args {
2980 	int fd;
2981 	char *buf;
2982 	size_t count;
2983 };
2984 #endif
2985 int
2986 getdents(p, uap)
2987 	struct proc *p;
2988 	register struct getdents_args /* {
2989 		syscallarg(int) fd;
2990 		syscallarg(char *) buf;
2991 		syscallarg(u_int) count;
2992 	} */ *uap;
2993 {
2994 	struct getdirentries_args ap;
2995 	ap.fd = uap->fd;
2996 	ap.buf = uap->buf;
2997 	ap.count = uap->count;
2998 	ap.basep = NULL;
2999 	return getdirentries(p, &ap);
3000 }
3001 
3002 /*
3003  * Set the mode mask for creation of filesystem nodes.
3004  *
3005  * MP SAFE
3006  */
3007 #ifndef _SYS_SYSPROTO_H_
3008 struct umask_args {
3009 	int	newmask;
3010 };
3011 #endif
3012 int
3013 umask(p, uap)
3014 	struct proc *p;
3015 	struct umask_args /* {
3016 		syscallarg(int) newmask;
3017 	} */ *uap;
3018 {
3019 	register struct filedesc *fdp;
3020 
3021 	fdp = p->p_fd;
3022 	p->p_retval[0] = fdp->fd_cmask;
3023 	fdp->fd_cmask = SCARG(uap, newmask) & ALLPERMS;
3024 	return (0);
3025 }
3026 
3027 /*
3028  * Void all references to file by ripping underlying filesystem
3029  * away from vnode.
3030  */
3031 #ifndef _SYS_SYSPROTO_H_
3032 struct revoke_args {
3033 	char	*path;
3034 };
3035 #endif
3036 /* ARGSUSED */
3037 int
3038 revoke(p, uap)
3039 	struct proc *p;
3040 	register struct revoke_args /* {
3041 		syscallarg(char *) path;
3042 	} */ *uap;
3043 {
3044 	register struct vnode *vp;
3045 	struct vattr vattr;
3046 	int error;
3047 	struct nameidata nd;
3048 
3049 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
3050 	if ((error = namei(&nd)) != 0)
3051 		return (error);
3052 	vp = nd.ni_vp;
3053 	NDFREE(&nd, NDF_ONLY_PNBUF);
3054 	if (vp->v_type != VCHR && vp->v_type != VBLK) {
3055 		error = EINVAL;
3056 		goto out;
3057 	}
3058 	if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)) != 0)
3059 		goto out;
3060 	if (p->p_ucred->cr_uid != vattr.va_uid &&
3061 	    (error = suser_xxx(0, p, PRISON_ROOT)))
3062 		goto out;
3063 	if (vcount(vp) > 1)
3064 		VOP_REVOKE(vp, REVOKEALL);
3065 out:
3066 	vrele(vp);
3067 	return (error);
3068 }
3069 
3070 /*
3071  * Convert a user file descriptor to a kernel file entry.
3072  */
3073 int
3074 getvnode(fdp, fd, fpp)
3075 	struct filedesc *fdp;
3076 	int fd;
3077 	struct file **fpp;
3078 {
3079 	struct file *fp;
3080 
3081 	if ((u_int)fd >= fdp->fd_nfiles ||
3082 	    (fp = fdp->fd_ofiles[fd]) == NULL)
3083 		return (EBADF);
3084 	if (fp->f_type != DTYPE_VNODE && fp->f_type != DTYPE_FIFO)
3085 		return (EINVAL);
3086 	*fpp = fp;
3087 	return (0);
3088 }
3089 /*
3090  * Get (NFS) file handle
3091  */
3092 #ifndef _SYS_SYSPROTO_H_
3093 struct getfh_args {
3094 	char	*fname;
3095 	fhandle_t *fhp;
3096 };
3097 #endif
3098 int
3099 getfh(p, uap)
3100 	struct proc *p;
3101 	register struct getfh_args *uap;
3102 {
3103 	struct nameidata nd;
3104 	fhandle_t fh;
3105 	register struct vnode *vp;
3106 	int error;
3107 
3108 	/*
3109 	 * Must be super user
3110 	 */
3111 	error = suser(p);
3112 	if (error)
3113 		return (error);
3114 	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE, uap->fname, p);
3115 	error = namei(&nd);
3116 	if (error)
3117 		return (error);
3118 	NDFREE(&nd, NDF_ONLY_PNBUF);
3119 	vp = nd.ni_vp;
3120 	bzero(&fh, sizeof(fh));
3121 	fh.fh_fsid = vp->v_mount->mnt_stat.f_fsid;
3122 	error = VFS_VPTOFH(vp, &fh.fh_fid);
3123 	vput(vp);
3124 	if (error)
3125 		return (error);
3126 	error = copyout(&fh, uap->fhp, sizeof (fh));
3127 	return (error);
3128 }
3129 
3130 /*
3131  * syscall for the rpc.lockd to use to translate a NFS file handle into
3132  * an open descriptor.
3133  *
3134  * warning: do not remove the suser() call or this becomes one giant
3135  * security hole.
3136  */
3137 #ifndef _SYS_SYSPROTO_H_
3138 struct fhopen_args {
3139 	const struct fhandle *u_fhp;
3140 	int flags;
3141 };
3142 #endif
3143 int
3144 fhopen(p, uap)
3145 	struct proc *p;
3146 	struct fhopen_args /* {
3147 		syscallarg(const struct fhandle *) u_fhp;
3148 		syscallarg(int) flags;
3149 	} */ *uap;
3150 {
3151 	struct mount *mp;
3152 	struct vnode *vp;
3153 	struct fhandle fhp;
3154 	struct vattr vat;
3155 	struct vattr *vap = &vat;
3156 	struct flock lf;
3157 	struct file *fp;
3158 	register struct filedesc *fdp = p->p_fd;
3159 	int fmode, mode, error, type;
3160 	struct file *nfp;
3161 	int indx;
3162 
3163 	/*
3164 	 * Must be super user
3165 	 */
3166 	error = suser(p);
3167 	if (error)
3168 		return (error);
3169 
3170 	fmode = FFLAGS(SCARG(uap, flags));
3171 	/* why not allow a non-read/write open for our lockd? */
3172 	if (((fmode & (FREAD | FWRITE)) == 0) || (fmode & O_CREAT))
3173 		return (EINVAL);
3174 	error = copyin(SCARG(uap,u_fhp), &fhp, sizeof(fhp));
3175 	if (error)
3176 		return(error);
3177 	/* find the mount point */
3178 	mp = vfs_getvfs(&fhp.fh_fsid);
3179 	if (mp == NULL)
3180 		return (ESTALE);
3181 	/* now give me my vnode, it gets returned to me locked */
3182 	error = VFS_FHTOVP(mp, &fhp.fh_fid, &vp);
3183 	if (error)
3184 		return (error);
3185  	/*
3186 	 * from now on we have to make sure not
3187 	 * to forget about the vnode
3188 	 * any error that causes an abort must vput(vp)
3189 	 * just set error = err and 'goto bad;'.
3190 	 */
3191 
3192 	/*
3193 	 * from vn_open
3194 	 */
3195 	if (vp->v_type == VLNK) {
3196 		error = EMLINK;
3197 		goto bad;
3198 	}
3199 	if (vp->v_type == VSOCK) {
3200 		error = EOPNOTSUPP;
3201 		goto bad;
3202 	}
3203 	mode = 0;
3204 	if (fmode & (FWRITE | O_TRUNC)) {
3205 		if (vp->v_type == VDIR) {
3206 			error = EISDIR;
3207 			goto bad;
3208 		}
3209 		error = vn_writechk(vp);
3210 		if (error)
3211 			goto bad;
3212 		mode |= VWRITE;
3213 	}
3214 	if (fmode & FREAD)
3215 		mode |= VREAD;
3216 	if (mode) {
3217 		error = VOP_ACCESS(vp, mode, p->p_ucred, p);
3218 		if (error)
3219 			goto bad;
3220 	}
3221 	if (fmode & O_TRUNC) {
3222 		VOP_UNLOCK(vp, 0, p);				/* XXX */
3223 		VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
3224 		vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);	/* XXX */
3225 		VATTR_NULL(vap);
3226 		vap->va_size = 0;
3227 		error = VOP_SETATTR(vp, vap, p->p_ucred, p);
3228 		if (error)
3229 			goto bad;
3230 	}
3231 	error = VOP_OPEN(vp, fmode, p->p_ucred, p);
3232 	if (error)
3233 		goto bad;
3234 	/*
3235 	 * Make sure that a VM object is created for VMIO support.
3236 	 */
3237 	if (vn_canvmio(vp) == TRUE) {
3238 		if ((error = vfs_object_create(vp, p, p->p_ucred)) != 0)
3239 			goto bad;
3240 	}
3241 	if (fmode & FWRITE)
3242 		vp->v_writecount++;
3243 
3244 	/*
3245 	 * end of vn_open code
3246 	 */
3247 
3248 	if ((error = falloc(p, &nfp, &indx)) != 0)
3249 		goto bad;
3250 	fp = nfp;
3251 	nfp->f_data = (caddr_t)vp;
3252 	nfp->f_flag = fmode & FMASK;
3253 	nfp->f_ops = &vnops;
3254 	nfp->f_type = DTYPE_VNODE;
3255 	if (fmode & (O_EXLOCK | O_SHLOCK)) {
3256 		lf.l_whence = SEEK_SET;
3257 		lf.l_start = 0;
3258 		lf.l_len = 0;
3259 		if (fmode & O_EXLOCK)
3260 			lf.l_type = F_WRLCK;
3261 		else
3262 			lf.l_type = F_RDLCK;
3263 		type = F_FLOCK;
3264 		if ((fmode & FNONBLOCK) == 0)
3265 			type |= F_WAIT;
3266 		VOP_UNLOCK(vp, 0, p);
3267 		if ((error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type)) != 0) {
3268 			(void) vn_close(vp, fp->f_flag, fp->f_cred, p);
3269 			ffree(fp);
3270 			fdp->fd_ofiles[indx] = NULL;
3271 			return (error);
3272 		}
3273 		vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
3274 		fp->f_flag |= FHASLOCK;
3275 	}
3276 	if ((vp->v_type == VREG) && (vp->v_object == NULL))
3277 		vfs_object_create(vp, p, p->p_ucred);
3278 
3279 	VOP_UNLOCK(vp, 0, p);
3280 	p->p_retval[0] = indx;
3281 	return (0);
3282 
3283 bad:
3284 	vput(vp);
3285 	return (error);
3286 }
3287 
3288 #ifndef _SYS_SYSPROTO_H_
3289 struct fhstat_args {
3290 	struct fhandle *u_fhp;
3291 	struct stat *sb;
3292 };
3293 #endif
3294 int
3295 fhstat(p, uap)
3296 	struct proc *p;
3297 	register struct fhstat_args /* {
3298 		syscallarg(struct fhandle *) u_fhp;
3299 		syscallarg(struct stat *) sb;
3300 	} */ *uap;
3301 {
3302 	struct stat sb;
3303 	fhandle_t fh;
3304 	struct mount *mp;
3305 	struct vnode *vp;
3306 	int error;
3307 
3308 	/*
3309 	 * Must be super user
3310 	 */
3311 	error = suser(p);
3312 	if (error)
3313 		return (error);
3314 
3315 	error = copyin(SCARG(uap, u_fhp), &fh, sizeof(fhandle_t));
3316 	if (error)
3317 		return (error);
3318 
3319 	if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL)
3320 		return (ESTALE);
3321 	if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)))
3322 		return (error);
3323 	error = vn_stat(vp, &sb, p);
3324 	vput(vp);
3325 	if (error)
3326 		return (error);
3327 	error = copyout(&sb, SCARG(uap, sb), sizeof(sb));
3328 	return (error);
3329 }
3330 
3331 #ifndef _SYS_SYSPROTO_H_
3332 struct fhstatfs_args {
3333 	struct fhandle *u_fhp;
3334 	struct statfs *buf;
3335 };
3336 #endif
3337 int
3338 fhstatfs(p, uap)
3339 	struct proc *p;
3340 	struct fhstatfs_args /* {
3341 		syscallarg(struct fhandle) *u_fhp;
3342 		syscallarg(struct statfs) *buf;
3343 	} */ *uap;
3344 {
3345 	struct statfs *sp;
3346 	struct mount *mp;
3347 	struct vnode *vp;
3348 	struct statfs sb;
3349 	fhandle_t fh;
3350 	int error;
3351 
3352 	/*
3353 	 * Must be super user
3354 	 */
3355 	if ((error = suser(p)))
3356 		return (error);
3357 
3358 	if ((error = copyin(SCARG(uap, u_fhp), &fh, sizeof(fhandle_t))) != 0)
3359 		return (error);
3360 
3361 	if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL)
3362 		return (ESTALE);
3363 	if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)))
3364 		return (error);
3365 	mp = vp->v_mount;
3366 	sp = &mp->mnt_stat;
3367 	vput(vp);
3368 	if ((error = VFS_STATFS(mp, sp, p)) != 0)
3369 		return (error);
3370 	sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
3371 	if (suser_xxx(p->p_ucred, 0, 0)) {
3372 		bcopy((caddr_t)sp, (caddr_t)&sb, sizeof(sb));
3373 		sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0;
3374 		sp = &sb;
3375 	}
3376 	return (copyout(sp, SCARG(uap, buf), sizeof(*sp)));
3377 }
3378 
3379 /*
3380  * Syscall to push extended attribute configuration information into the
3381  * VFS.  Accepts a path, which it converts to a mountpoint, as well as
3382  * a command (int cmd), and attribute name and misc data.  For now, the
3383  * attribute name is left in userspace for consumption by the VFS_op.
3384  * It will probably be changed to be copied into sysspace by the
3385  * syscall in the future, once issues with various consumers of the
3386  * attribute code have raised their hands.
3387  *
3388  * Currently this is used only by UFS Extended Attributes.
3389  */
3390 int
3391 extattrctl(p, uap)
3392 	struct proc *p;
3393 	struct extattrctl_args *uap;
3394 {
3395 	struct nameidata nd;
3396 	struct mount *mp;
3397 	int error;
3398 
3399 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
3400 	if ((error = namei(&nd)) != 0)
3401 		return (error);
3402 	mp = nd.ni_vp->v_mount;
3403 	NDFREE(&nd, 0);
3404 	return (VFS_EXTATTRCTL(mp, SCARG(uap, cmd), SCARG(uap, attrname),
3405 	    SCARG(uap, arg), p));
3406 }
3407 
3408 /*
3409  * Syscall to set a named extended attribute on a file or directory.
3410  * Accepts attribute name, and a uio structure pointing to the data to set.
3411  * The uio is consumed in the style of writev().  The real work happens
3412  * in VOP_SETEXTATTR().
3413  */
3414 int
3415 extattr_set_file(p, uap)
3416 	struct proc *p;
3417 	struct extattr_set_file_args *uap;
3418 {
3419 	struct nameidata nd;
3420 	struct uio auio;
3421 	struct iovec *iov, *needfree = NULL, aiov[UIO_SMALLIOV];
3422 	char attrname[EXTATTR_MAXNAMELEN];
3423 	u_int iovlen, cnt;
3424 	int error, i;
3425 
3426 	error = copyin(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN);
3427 	if (error)
3428 		return (error);
3429 	NDINIT(&nd, LOOKUP, LOCKLEAF | FOLLOW, UIO_USERSPACE, SCARG(uap, path),
3430 	    p);
3431 	if ((error = namei(&nd)) != 0)
3432 		return(error);
3433 	iovlen = uap->iovcnt * sizeof(struct iovec);
3434 	if (uap->iovcnt > UIO_SMALLIOV) {
3435 		if (uap->iovcnt > UIO_MAXIOV) {
3436 			error = EINVAL;
3437 			goto done;
3438 		}
3439 		MALLOC(iov, struct iovec *, iovlen, M_IOV, M_WAITOK);
3440 		needfree = iov;
3441 	} else
3442 		iov = aiov;
3443 	auio.uio_iov = iov;
3444 	auio.uio_iovcnt = uap->iovcnt;
3445 	auio.uio_rw = UIO_WRITE;
3446 	auio.uio_segflg = UIO_USERSPACE;
3447 	auio.uio_procp = p;
3448 	auio.uio_offset = 0;
3449 	if ((error = copyin((caddr_t)uap->iovp, (caddr_t)iov, iovlen)))
3450 		goto done;
3451 	auio.uio_resid = 0;
3452 	for (i = 0; i < uap->iovcnt; i++) {
3453 		if (iov->iov_len > INT_MAX - auio.uio_resid) {
3454 			error = EINVAL;
3455 			goto done;
3456 		}
3457 		auio.uio_resid += iov->iov_len;
3458 		iov++;
3459 	}
3460 	cnt = auio.uio_resid;
3461 	error = VOP_SETEXTATTR(nd.ni_vp, attrname, &auio, p->p_cred->pc_ucred,
3462 	    p);
3463 	if (auio.uio_resid != cnt && (error == ERESTART ||
3464 	    error == EINTR || error == EWOULDBLOCK))
3465 		error = 0;
3466 	cnt -= auio.uio_resid;
3467 	p->p_retval[0] = cnt;
3468 done:
3469 	if (needfree)
3470 		FREE(needfree, M_IOV);
3471 	NDFREE(&nd, 0);
3472 	return (error);
3473 }
3474 
3475 /*
3476  * Syscall to get a named extended attribute on a file or directory.
3477  * Accepts attribute name, and a uio structure pointing to a buffer for the
3478  * data.  The uio is consumed in the style of readv().  The real work
3479  * happens in VOP_GETEXTATTR();
3480  */
3481 int
3482 extattr_get_file(p, uap)
3483 	struct proc *p;
3484 	struct extattr_get_file_args *uap;
3485 {
3486 	struct nameidata nd;
3487 	struct uio auio;
3488 	struct iovec *iov, *needfree, aiov[UIO_SMALLIOV];
3489 	char attrname[EXTATTR_MAXNAMELEN];
3490 	u_int iovlen, cnt;
3491 	int error, i;
3492 
3493 	error = copyin(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN);
3494 	if (error)
3495 		return (error);
3496 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
3497 	if ((error = namei(&nd)) != 0)
3498 		return (error);
3499 	iovlen = uap->iovcnt * sizeof (struct iovec);
3500 	if (uap->iovcnt > UIO_SMALLIOV) {
3501 		if (uap->iovcnt > UIO_MAXIOV) {
3502 			NDFREE(&nd, 0);
3503 			return (EINVAL);
3504 		}
3505 		MALLOC(iov, struct iovec *, iovlen, M_IOV, M_WAITOK);
3506 		needfree = iov;
3507 	} else {
3508 		iov = aiov;
3509 		needfree = NULL;
3510 	}
3511 	auio.uio_iov = iov;
3512 	auio.uio_iovcnt = uap->iovcnt;
3513 	auio.uio_rw = UIO_READ;
3514 	auio.uio_segflg = UIO_USERSPACE;
3515 	auio.uio_procp = p;
3516 	auio.uio_offset = 0;
3517 	if ((error = copyin((caddr_t)uap->iovp, (caddr_t)iov, iovlen)))
3518 		goto done;
3519 	auio.uio_resid = 0;
3520 	for (i = 0; i < uap->iovcnt; i++) {
3521 		if (iov->iov_len > INT_MAX - auio.uio_resid) {
3522 			error = EINVAL;
3523 			goto done;
3524 		}
3525 		auio.uio_resid += iov->iov_len;
3526 		iov++;
3527 	}
3528 	cnt = auio.uio_resid;
3529 	error = VOP_GETEXTATTR(nd.ni_vp, attrname, &auio, p->p_cred->pc_ucred,
3530 	    p);
3531 	if (auio.uio_resid != cnt && (error == ERESTART ||
3532 	    error == EINTR || error == EWOULDBLOCK))
3533 		error = 0;
3534 	cnt -= auio.uio_resid;
3535 	p->p_retval[0] = cnt;
3536 done:
3537 	if (needfree)
3538 		FREE(needfree, M_IOV);
3539 	NDFREE(&nd, 0);
3540 	return(error);
3541 }
3542 
3543 /*
3544  * Syscall to delete a named extended attribute from a file or directory.
3545  * Accepts attribute name.  The real work happens in VOP_SETEXTATTR().
3546  */
3547 int
3548 extattr_delete_file(p, uap)
3549 	struct proc *p;
3550 	struct extattr_delete_file_args *uap;
3551 {
3552 	struct nameidata nd;
3553 	char attrname[EXTATTR_MAXNAMELEN];
3554 	int	error;
3555 
3556 	error = copyin(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN);
3557 	if (error)
3558 		return(error);
3559 	NDINIT(&nd, LOOKUP, LOCKLEAF | FOLLOW, UIO_USERSPACE, SCARG(uap, path),
3560 	    p);
3561 	if ((error = namei(&nd)) != 0)
3562 		return(error);
3563 	error = VOP_SETEXTATTR(nd.ni_vp, attrname, NULL, p->p_cred->pc_ucred,
3564 	    p);
3565 	NDFREE(&nd, 0);
3566 	return(error);
3567 }
3568