xref: /openbsd/regress/usr.bin/diff/t9.1 (revision d415bd75)
1/*	$NetBSD: vfs_syscalls.c,v 1.57 1995/10/07 06:28:51 mycroft Exp $	*/
2
3/*
4 * Copyright (c) 1989, 1993
5 *	The Regents of the University of California.  All rights reserved.
6 * (c) UNIX System Laboratories, Inc.
7 * All or some portions of this file are derived from material licensed
8 * to the University of California by American Telephone and Telegraph
9 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
10 * the permission of UNIX System Laboratories, Inc.
11 *
12 * Redistribution and use in source and binary forms, with or without
13 * modification, are permitted provided that the following conditions
14 * are met:
15 * 1. Redistributions of source code must retain the above copyright
16 *    notice, this list of conditions and the following disclaimer.
17 * 2. Redistributions in binary form must reproduce the above copyright
18 *    notice, this list of conditions and the following disclaimer in the
19 *    documentation and/or other materials provided with the distribution.
20 * 3. All advertising materials mentioning features or use of this software
21 *    must display the following acknowledgement:
22 *	This product includes software developed by the University of
23 *	California, Berkeley and its contributors.
24 * 4. Neither the name of the University nor the names of its contributors
25 *    may be used to endorse or promote products derived from this software
26 *    without specific prior written permission.
27 *
28 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
29 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
30 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
31 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
32 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
33 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
34 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
35 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
36 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
37 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
38 * SUCH DAMAGE.
39 *
40 *	@(#)vfs_syscalls.c	8.28 (Berkeley) 12/10/94
41 */
42
43#include <sys/param.h>
44#include <sys/systm.h>
45#include <sys/namei.h>
46#include <sys/filedesc.h>
47#include <sys/kernel.h>
48#include <sys/file.h>
49#include <sys/stat.h>
50#include <sys/vnode.h>
51#include <sys/mount.h>
52#include <sys/proc.h>
53#include <sys/uio.h>
54#include <sys/malloc.h>
55#include <sys/dirent.h>
56
57#include <sys/syscallargs.h>
58
59#include <vm/vm.h>
60#include <sys/sysctl.h>
61
62static int change_dir __P((struct nameidata *ndp, struct proc *p));
63
64/*
65 * Virtual File System System Calls
66 */
67
68/*
69 * Mount a file system.
70 */
71/* ARGSUSED */
72sys_mount(p, v, retval)
73	struct proc *p;
74	void *v;
75	register_t *retval;
76{
77	register struct sys_mount_args /* {
78		syscallarg(char *) type;
79		syscallarg(char *) path;
80		syscallarg(int) flags;
81		syscallarg(caddr_t) data;
82	} */ *uap = v;
83	register struct vnode *vp;
84	register struct mount *mp;
85	int error, flag;
86	u_long fsindex;
87	char fstypename[MFSNAMELEN];
88	struct vattr va;
89	struct nameidata nd;
90
91	/*
92	 * Get vnode to be covered
93	 */
94	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
95	    SCARG(uap, path), p);
96	if (error = namei(&nd))
97		return (error);
98	vp = nd.ni_vp;
99	if (SCARG(uap, flags) & MNT_UPDATE) {
100		if ((vp->v_flag & VROOT) == 0) {
101			vput(vp);
102			return (EINVAL);
103		}
104		mp = vp->v_mount;
105		flag = mp->mnt_flag;
106		/*
107		 * We only allow the filesystem to be reloaded if it
108		 * is currently mounted read-only.
109		 */
110		if ((SCARG(uap, flags) & MNT_RELOAD) &&
111		    ((mp->mnt_flag & MNT_RDONLY) == 0)) {
112			vput(vp);
113			return (EOPNOTSUPP);	/* Needs translation */
114		}
115		mp->mnt_flag |=
116		    SCARG(uap, flags) & (MNT_RELOAD | MNT_FORCE | MNT_UPDATE);
117		/*
118		 * Only root, or the user that did the original mount is
119		 * permitted to update it.
120		 */
121		if (mp->mnt_stat.f_owner != p->p_ucred->cr_uid &&
122		    (error = suser(p->p_ucred, &p->p_acflag))) {
123			vput(vp);
124			return (error);
125		}
126		/*
127		 * Do not allow NFS export by non-root users. Silently
128		 * enforce MNT_NOSUID and MNT_NODEV for non-root users.
129		 */
130		if (p->p_ucred->cr_uid != 0) {
131			if (SCARG(uap, flags) & MNT_EXPORTED) {
132				vput(vp);
133				return (EPERM);
134			}
135			SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
136		}
137		VOP_UNLOCK(vp);
138		goto update;
139	}
140	/*
141	 * If the user is not root, ensure that they own the directory
142	 * onto which we are attempting to mount.
143	 */
144	if ((error = VOP_GETATTR(vp, &va, p->p_ucred, p)) ||
145	    (va.va_uid != p->p_ucred->cr_uid &&
146	     (error = suser(p->p_ucred, &p->p_acflag)))) {
147		vput(vp);
148		return (error);
149	}
150	/*
151	 * Do not allow NFS export by non-root users. Silently
152	 * enforce MNT_NOSUID and MNT_NODEV for non-root users.
153	 */
154	if (p->p_ucred->cr_uid != 0) {
155		if (SCARG(uap, flags) & MNT_EXPORTED) {
156			vput(vp);
157			return (EPERM);
158		}
159		SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
160	}
161	if (error = vinvalbuf(vp, V_SAVE, p->p_ucred, p, 0, 0))
162		return (error);
163	if (vp->v_type != VDIR) {
164		vput(vp);
165		return (ENOTDIR);
166	}
167	if (error = copyinstr(SCARG(uap, type), fstypename, MFSNAMELEN,
168	    (size_t *)0)) {
169#if defined(COMPAT_09) || defined(COMPAT_43)
170		/*
171		 * Historically filesystem types were identified by number.
172		 * If we get an integer for the filesystem type instead of a
173		 * string, we check to see if it matches one of the historic
174		 * filesystem types.
175		 */
176		fsindex = (u_long)SCARG(uap, type);
177		if (fsindex >= nvfssw || vfssw[fsindex] == NULL) {
178			vput(vp);
179			return (ENODEV);
180		}
181		strncpy(fstypename, vfssw[fsindex]->vfs_name, MFSNAMELEN);
182#else
183		vput(vp);
184		return (error);
185#endif
186	}
187	for (fsindex = 0; fsindex < nvfssw; fsindex++)
188		if (vfssw[fsindex] != NULL &&
189		    !strncmp(vfssw[fsindex]->vfs_name, fstypename, MFSNAMELEN))
190			break;
191	if (fsindex >= nvfssw) {
192		vput(vp);
193		return (ENODEV);
194	}
195	if (vp->v_mountedhere != NULL) {
196		vput(vp);
197		return (EBUSY);
198	}
199
200	/*
201	 * Allocate and initialize the file system.
202	 */
203	mp = (struct mount *)malloc((u_long)sizeof(struct mount),
204		M_MOUNT, M_WAITOK);
205	bzero((char *)mp, (u_long)sizeof(struct mount));
206	mp->mnt_op = vfssw[fsindex];
207	if (error = vfs_lock(mp)) {
208		free((caddr_t)mp, M_MOUNT);
209		vput(vp);
210		return (error);
211	}
212	/* Do this early in case we block later. */
213	vfssw[fsindex]->vfs_refcount++;
214	vp->v_mountedhere = mp;
215	mp->mnt_vnodecovered = vp;
216	mp->mnt_stat.f_owner = p->p_ucred->cr_uid;
217update:
218	/*
219	 * Set the mount level flags.
220	 */
221	if (SCARG(uap, flags) & MNT_RDONLY)
222		mp->mnt_flag |= MNT_RDONLY;
223	else if (mp->mnt_flag & MNT_RDONLY)
224		mp->mnt_flag |= MNT_WANTRDWR;
225	mp->mnt_flag &=~ (MNT_NOSUID | MNT_NOEXEC | MNT_NODEV |
226	    MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC);
227	mp->mnt_flag |= SCARG(uap, flags) & (MNT_NOSUID | MNT_NOEXEC |
228	    MNT_NODEV | MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC);
229	/*
230	 * Mount the filesystem.
231	 */
232	error = VFS_MOUNT(mp, SCARG(uap, path), SCARG(uap, data), &nd, p);
233	if (mp->mnt_flag & MNT_UPDATE) {
234		vrele(vp);
235		if (mp->mnt_flag & MNT_WANTRDWR)
236			mp->mnt_flag &= ~MNT_RDONLY;
237		mp->mnt_flag &=~
238		    (MNT_UPDATE | MNT_RELOAD | MNT_FORCE | MNT_WANTRDWR);
239		if (error)
240			mp->mnt_flag = flag;
241		return (error);
242	}
243	/*
244	 * Put the new filesystem on the mount list after root.
245	 */
246	cache_purge(vp);
247	if (!error) {
248		TAILQ_INSERT_TAIL(&mountlist, mp, mnt_list);
249		checkdirs(vp);
250		VOP_UNLOCK(vp);
251		vfs_unlock(mp);
252		(void) VFS_STATFS(mp, &mp->mnt_stat, p);
253		error = VFS_START(mp, 0, p);
254	} else {
255		mp->mnt_vnodecovered->v_mountedhere = (struct mount *)0;
256		vfssw[fsindex]->vfs_refcount--;
257		vfs_unlock(mp);
258		free((caddr_t)mp, M_MOUNT);
259		vput(vp);
260	}
261	return (error);
262}
263
264/*
265 * Scan all active processes to see if any of them have a current
266 * or root directory onto which the new filesystem has just been
267 * mounted. If so, replace them with the new mount point.
268 */
269checkdirs(olddp)
270	struct vnode *olddp;
271{
272	struct filedesc *fdp;
273	struct vnode *newdp;
274	struct proc *p;
275
276	if (olddp->v_usecount == 1)
277		return;
278	if (VFS_ROOT(olddp->v_mountedhere, &newdp))
279		panic("mount: lost mount");
280	for (p = allproc.lh_first; p != 0; p = p->p_list.le_next) {
281		fdp = p->p_fd;
282		if (fdp->fd_cdir == olddp) {
283			vrele(fdp->fd_cdir);
284			VREF(newdp);
285			fdp->fd_cdir = newdp;
286		}
287		if (fdp->fd_rdir == olddp) {
288			vrele(fdp->fd_rdir);
289			VREF(newdp);
290			fdp->fd_rdir = newdp;
291		}
292	}
293	if (rootvnode == olddp) {
294		vrele(rootvnode);
295		VREF(newdp);
296		rootvnode = newdp;
297	}
298	vput(newdp);
299}
300
301/*
302 * Unmount a file system.
303 *
304 * Note: unmount takes a path to the vnode mounted on as argument,
305 * not special file (as before).
306 */
307/* ARGSUSED */
308sys_unmount(p, v, retval)
309	struct proc *p;
310	void *v;
311	register_t *retval;
312{
313	register struct sys_unmount_args /* {
314		syscallarg(char *) path;
315		syscallarg(int) flags;
316	} */ *uap = v;
317	register struct vnode *vp;
318	struct mount *mp;
319	int error;
320	struct nameidata nd;
321
322	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
323	    SCARG(uap, path), p);
324	if (error = namei(&nd))
325		return (error);
326	vp = nd.ni_vp;
327	mp = vp->v_mount;
328
329	/*
330	 * Only root, or the user that did the original mount is
331	 * permitted to unmount this filesystem.
332	 */
333	if ((mp->mnt_stat.f_owner != p->p_ucred->cr_uid) &&
334	    (error = suser(p->p_ucred, &p->p_acflag))) {
335		vput(vp);
336		return (error);
337	}
338
339	/*
340	 * Don't allow unmounting the root file system.
341	 */
342	if (mp->mnt_flag & MNT_ROOTFS) {
343		vput(vp);
344		return (EINVAL);
345	}
346
347	/*
348	 * Must be the root of the filesystem
349	 */
350	if ((vp->v_flag & VROOT) == 0) {
351		vput(vp);
352		return (EINVAL);
353	}
354	vput(vp);
355	return (dounmount(mp, SCARG(uap, flags), p));
356}
357
358/*
359 * Do the actual file system unmount.
360 */
361dounmount(mp, flags, p)
362	register struct mount *mp;
363	int flags;
364	struct proc *p;
365{
366	struct vnode *coveredvp;
367	int error;
368
369	coveredvp = mp->mnt_vnodecovered;
370	if (vfs_busy(mp))
371		return (EBUSY);
372	mp->mnt_flag |= MNT_UNMOUNT;
373	if (error = vfs_lock(mp))
374		return (error);
375
376	mp->mnt_flag &=~ MNT_ASYNC;
377	vnode_pager_umount(mp);	/* release cached vnodes */
378	cache_purgevfs(mp);	/* remove cache entries for this file sys */
379	if ((error = VFS_SYNC(mp, MNT_WAIT, p->p_ucred, p)) == 0 ||
380	    (flags & MNT_FORCE))
381		error = VFS_UNMOUNT(mp, flags, p);
382	mp->mnt_flag &= ~MNT_UNMOUNT;
383	vfs_unbusy(mp);
384	if (error) {
385		vfs_unlock(mp);
386	} else {
387		TAILQ_REMOVE(&mountlist, mp, mnt_list);
388		if (coveredvp != NULLVP) {
389			vrele(coveredvp);
390			coveredvp->v_mountedhere = (struct mount *)0;
391		}
392		mp->mnt_op->vfs_refcount--;
393		vfs_unlock(mp);
394		if (mp->mnt_vnodelist.lh_first != NULL)
395			panic("unmount: dangling vnode");
396		free((caddr_t)mp, M_MOUNT);
397	}
398	return (error);
399}
400
401/*
402 * Sync each mounted filesystem.
403 */
404#ifdef DEBUG
405int syncprt = 0;
406struct ctldebug debug0 = { "syncprt", &syncprt };
407#endif
408
409/* ARGSUSED */
410sys_sync(p, v, retval)
411	struct proc *p;
412	void *v;
413	register_t *retval;
414{
415	register struct mount *mp, *nmp;
416	int asyncflag;
417
418	for (mp = mountlist.cqh_first; mp != (void *)&mountlist; mp = nmp) {
419		/*
420		 * Get the next pointer in case we hang on vfs_busy
421		 * while we are being unmounted.
422		 */
423		nmp = mp->mnt_list.cqe_next;
424		/*
425		 * The lock check below is to avoid races with mount
426		 * and unmount.
427		 */
428		if ((mp->mnt_flag & (MNT_MLOCK|MNT_RDONLY|MNT_MPBUSY)) == 0 &&
429		    !vfs_busy(mp)) {
430			asyncflag = mp->mnt_flag & MNT_ASYNC;
431			mp->mnt_flag &= ~MNT_ASYNC;
432			VFS_SYNC(mp, MNT_NOWAIT, p->p_ucred, p);
433			if (asyncflag)
434				mp->mnt_flag |= MNT_ASYNC;
435			/*
436			 * Get the next pointer again, as the next filesystem
437			 * might have been unmounted while we were sync'ing.
438			 */
439			nmp = mp->mnt_list.cqe_next;
440			vfs_unbusy(mp);
441		}
442	}
443#ifdef DEBUG
444	if (syncprt)
445		vfs_bufstats();
446#endif /* DEBUG */
447	return (0);
448}
449
450/*
451 * Change filesystem quotas.
452 */
453/* ARGSUSED */
454sys_quotactl(p, v, retval)
455	struct proc *p;
456	void *v;
457	register_t *retval;
458{
459	register struct sys_quotactl_args /* {
460		syscallarg(char *) path;
461		syscallarg(int) cmd;
462		syscallarg(int) uid;
463		syscallarg(caddr_t) arg;
464	} */ *uap = v;
465	register struct mount *mp;
466	int error;
467	struct nameidata nd;
468
469	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
470	if (error = namei(&nd))
471		return (error);
472	mp = nd.ni_vp->v_mount;
473	vrele(nd.ni_vp);
474	return (VFS_QUOTACTL(mp, SCARG(uap, cmd), SCARG(uap, uid),
475	    SCARG(uap, arg), p));
476}
477
478/*
479 * Get filesystem statistics.
480 */
481/* ARGSUSED */
482sys_statfs(p, v, retval)
483	struct proc *p;
484	void *v;
485	register_t *retval;
486{
487	register struct sys_statfs_args /* {
488		syscallarg(char *) path;
489		syscallarg(struct statfs *) buf;
490	} */ *uap = v;
491	register struct mount *mp;
492	register struct statfs *sp;
493	int error;
494	struct nameidata nd;
495
496	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
497	if (error = namei(&nd))
498		return (error);
499	mp = nd.ni_vp->v_mount;
500	sp = &mp->mnt_stat;
501	vrele(nd.ni_vp);
502	if (error = VFS_STATFS(mp, sp, p))
503		return (error);
504	sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
505	return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
506}
507
508/*
509 * Get filesystem statistics.
510 */
511/* ARGSUSED */
512sys_fstatfs(p, v, retval)
513	struct proc *p;
514	void *v;
515	register_t *retval;
516{
517	register struct sys_fstatfs_args /* {
518		syscallarg(int) fd;
519		syscallarg(struct statfs *) buf;
520	} */ *uap = v;
521	struct file *fp;
522	struct mount *mp;
523	register struct statfs *sp;
524	int error;
525
526	if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
527		return (error);
528	mp = ((struct vnode *)fp->f_data)->v_mount;
529	sp = &mp->mnt_stat;
530	if (error = VFS_STATFS(mp, sp, p))
531		return (error);
532	sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
533	return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
534}
535
536/*
537 * Get statistics on all filesystems.
538 */
539sys_getfsstat(p, v, retval)
540	struct proc *p;
541	void *v;
542	register_t *retval;
543{
544	register struct sys_getfsstat_args /* {
545		syscallarg(struct statfs *) buf;
546		syscallarg(long) bufsize;
547		syscallarg(int) flags;
548	} */ *uap = v;
549	register struct mount *mp, *nmp;
550	register struct statfs *sp;
551	caddr_t sfsp;
552	long count, maxcount, error;
553
554	maxcount = SCARG(uap, bufsize) / sizeof(struct statfs);
555	sfsp = (caddr_t)SCARG(uap, buf);
556	for (count = 0,
557	     mp = mountlist.cqh_first; mp != (void *)&mountlist; mp = nmp) {
558		nmp = mp->mnt_list.cqe_next;
559		if (sfsp && count < maxcount &&
560		    ((mp->mnt_flag & MNT_MLOCK) == 0)) {
561			sp = &mp->mnt_stat;
562			/*
563			 * If MNT_NOWAIT is specified, do not refresh the
564			 * fsstat cache. MNT_WAIT overrides MNT_NOWAIT.
565			 */
566			if (((SCARG(uap, flags) & MNT_NOWAIT) == 0 ||
567			    (SCARG(uap, flags) & MNT_WAIT)) &&
568			    (error = VFS_STATFS(mp, sp, p)))
569				continue;
570			sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
571			if (error = copyout((caddr_t)sp, sfsp, sizeof(*sp)))
572				return (error);
573			sfsp += sizeof(*sp);
574		}
575		count++;
576	}
577	if (sfsp && count > maxcount)
578		*retval = maxcount;
579	else
580		*retval = count;
581	return (0);
582}
583
584/*
585 * Change current working directory to a given file descriptor.
586 */
587/* ARGSUSED */
588sys_fchdir(p, v, retval)
589	struct proc *p;
590	void *v;
591	register_t *retval;
592{
593	struct sys_fchdir_args /* {
594		syscallarg(int) fd;
595	} */ *uap = v;
596	register struct filedesc *fdp = p->p_fd;
597	struct vnode *vp, *tdp;
598	struct mount *mp;
599	struct file *fp;
600	int error;
601
602	if (error = getvnode(fdp, SCARG(uap, fd), &fp))
603		return (error);
604	vp = (struct vnode *)fp->f_data;
605	VREF(vp);
606	VOP_LOCK(vp);
607	if (vp->v_type != VDIR)
608		error = ENOTDIR;
609	else
610		error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
611	while (!error && (mp = vp->v_mountedhere) != NULL) {
612		if (mp->mnt_flag & MNT_MLOCK) {
613			mp->mnt_flag |= MNT_MWAIT;
614			sleep((caddr_t)mp, PVFS);
615			continue;
616		}
617		if (error = VFS_ROOT(mp, &tdp))
618			break;
619		vput(vp);
620		vp = tdp;
621	}
622	VOP_UNLOCK(vp);
623	if (error) {
624		vrele(vp);
625		return (error);
626	}
627	vrele(fdp->fd_cdir);
628	fdp->fd_cdir = vp;
629	return (0);
630}
631
632/*
633 * Change current working directory (``.'').
634 */
635/* ARGSUSED */
636sys_chdir(p, v, retval)
637	struct proc *p;
638	void *v;
639	register_t *retval;
640{
641	struct sys_chdir_args /* {
642		syscallarg(char *) path;
643	} */ *uap = v;
644	register struct filedesc *fdp = p->p_fd;
645	int error;
646	struct nameidata nd;
647
648	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
649	    SCARG(uap, path), p);
650	if (error = change_dir(&nd, p))
651		return (error);
652	vrele(fdp->fd_cdir);
653	fdp->fd_cdir = nd.ni_vp;
654	return (0);
655}
656
657/*
658 * Change notion of root (``/'') directory.
659 */
660/* ARGSUSED */
661sys_chroot(p, v, retval)
662	struct proc *p;
663	void *v;
664	register_t *retval;
665{
666	struct sys_chroot_args /* {
667		syscallarg(char *) path;
668	} */ *uap = v;
669	register struct filedesc *fdp = p->p_fd;
670	int error;
671	struct nameidata nd;
672
673	if (error = suser(p->p_ucred, &p->p_acflag))
674		return (error);
675	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
676	    SCARG(uap, path), p);
677	if (error = change_dir(&nd, p))
678		return (error);
679	if (fdp->fd_rdir != NULL)
680		vrele(fdp->fd_rdir);
681	fdp->fd_rdir = nd.ni_vp;
682	return (0);
683}
684
685/*
686 * Common routine for chroot and chdir.
687 */
688static int
689change_dir(ndp, p)
690	register struct nameidata *ndp;
691	struct proc *p;
692{
693	struct vnode *vp;
694	int error;
695
696	if (error = namei(ndp))
697		return (error);
698	vp = ndp->ni_vp;
699	if (vp->v_type != VDIR)
700		error = ENOTDIR;
701	else
702		error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
703	VOP_UNLOCK(vp);
704	if (error)
705		vrele(vp);
706	return (error);
707}
708
709/*
710 * Check permissions, allocate an open file structure,
711 * and call the device open routine if any.
712 */
713sys_open(p, v, retval)
714	struct proc *p;
715	void *v;
716	register_t *retval;
717{
718	register struct sys_open_args /* {
719		syscallarg(char *) path;
720		syscallarg(int) flags;
721		syscallarg(int) mode;
722	} */ *uap = v;
723	register struct filedesc *fdp = p->p_fd;
724	register struct file *fp;
725	register struct vnode *vp;
726	int flags, cmode;
727	struct file *nfp;
728	int type, indx, error;
729	struct flock lf;
730	struct nameidata nd;
731	extern struct fileops vnops;
732
733	if (error = falloc(p, &nfp, &indx))
734		return (error);
735	fp = nfp;
736	flags = FFLAGS(SCARG(uap, flags));
737	cmode = ((SCARG(uap, mode) &~ fdp->fd_cmask) & ALLPERMS) &~ S_ISTXT;
738	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
739	p->p_dupfd = -indx - 1;			/* XXX check for fdopen */
740	if (error = vn_open(&nd, flags, cmode)) {
741		ffree(fp);
742		if ((error == ENODEV || error == ENXIO) &&
743		    p->p_dupfd >= 0 &&			/* XXX from fdopen */
744		    (error =
745			dupfdopen(fdp, indx, p->p_dupfd, flags, error)) == 0) {
746			*retval = indx;
747			return (0);
748		}
749		if (error == ERESTART)
750			error = EINTR;
751		fdp->fd_ofiles[indx] = NULL;
752		return (error);
753	}
754	p->p_dupfd = 0;
755	vp = nd.ni_vp;
756	fp->f_flag = flags & FMASK;
757	fp->f_type = DTYPE_VNODE;
758	fp->f_ops = &vnops;
759	fp->f_data = (caddr_t)vp;
760	if (flags & (O_EXLOCK | O_SHLOCK)) {
761		lf.l_whence = SEEK_SET;
762		lf.l_start = 0;
763		lf.l_len = 0;
764		if (flags & O_EXLOCK)
765			lf.l_type = F_WRLCK;
766		else
767			lf.l_type = F_RDLCK;
768		type = F_FLOCK;
769		if ((flags & FNONBLOCK) == 0)
770			type |= F_WAIT;
771		VOP_UNLOCK(vp);
772		if (error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type)) {
773			(void) vn_close(vp, fp->f_flag, fp->f_cred, p);
774			ffree(fp);
775			fdp->fd_ofiles[indx] = NULL;
776			return (error);
777		}
778		VOP_LOCK(vp);
779		fp->f_flag |= FHASLOCK;
780	}
781	VOP_UNLOCK(vp);
782	*retval = indx;
783	return (0);
784}
785
786/*
787 * Create a special file.
788 */
789/* ARGSUSED */
790sys_mknod(p, v, retval)
791	struct proc *p;
792	void *v;
793	register_t *retval;
794{
795	register struct sys_mknod_args /* {
796		syscallarg(char *) path;
797		syscallarg(int) mode;
798		syscallarg(int) dev;
799	} */ *uap = v;
800	register struct vnode *vp;
801	struct vattr vattr;
802	int error;
803	int whiteout;
804	struct nameidata nd;
805
806	if (error = suser(p->p_ucred, &p->p_acflag))
807		return (error);
808	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
809	if (error = namei(&nd))
810		return (error);
811	vp = nd.ni_vp;
812	if (vp != NULL)
813		error = EEXIST;
814	else {
815		VATTR_NULL(&vattr);
816		vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
817		vattr.va_rdev = SCARG(uap, dev);
818		whiteout = 0;
819
820		switch (SCARG(uap, mode) & S_IFMT) {
821		case S_IFMT:	/* used by badsect to flag bad sectors */
822			vattr.va_type = VBAD;
823			break;
824		case S_IFCHR:
825			vattr.va_type = VCHR;
826			break;
827		case S_IFBLK:
828			vattr.va_type = VBLK;
829			break;
830		case S_IFWHT:
831			whiteout = 1;
832			break;
833		default:
834			error = EINVAL;
835			break;
836		}
837	}
838	if (!error) {
839		VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
840		if (whiteout) {
841			error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, CREATE);
842			if (error)
843				VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
844			vput(nd.ni_dvp);
845		} else {
846			error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp,
847						&nd.ni_cnd, &vattr);
848		}
849	} else {
850		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
851		if (nd.ni_dvp == vp)
852			vrele(nd.ni_dvp);
853		else
854			vput(nd.ni_dvp);
855		if (vp)
856			vrele(vp);
857	}
858	return (error);
859}
860
861/*
862 * Create a named pipe.
863 */
864/* ARGSUSED */
865sys_mkfifo(p, v, retval)
866	struct proc *p;
867	void *v;
868	register_t *retval;
869{
870	register struct sys_mkfifo_args /* {
871		syscallarg(char *) path;
872		syscallarg(int) mode;
873	} */ *uap = v;
874	struct vattr vattr;
875	int error;
876	struct nameidata nd;
877
878#ifndef FIFO
879	return (EOPNOTSUPP);
880#else
881	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
882	if (error = namei(&nd))
883		return (error);
884	if (nd.ni_vp != NULL) {
885		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
886		if (nd.ni_dvp == nd.ni_vp)
887			vrele(nd.ni_dvp);
888		else
889			vput(nd.ni_dvp);
890		vrele(nd.ni_vp);
891		return (EEXIST);
892	}
893	VATTR_NULL(&vattr);
894	vattr.va_type = VFIFO;
895	vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
896	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
897	return (VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr));
898#endif /* FIFO */
899}
900
901/*
902 * Make a hard file link.
903 */
904/* ARGSUSED */
905sys_link(p, v, retval)
906	struct proc *p;
907	void *v;
908	register_t *retval;
909{
910	register struct sys_link_args /* {
911		syscallarg(char *) path;
912		syscallarg(char *) link;
913	} */ *uap = v;
914	register struct vnode *vp;
915	struct nameidata nd;
916	int error;
917
918	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
919	if (error = namei(&nd))
920		return (error);
921	vp = nd.ni_vp;
922	if (vp->v_type != VDIR ||
923	    (error = suser(p->p_ucred, &p->p_acflag)) == 0) {
924		nd.ni_cnd.cn_nameiop = CREATE;
925		nd.ni_cnd.cn_flags = LOCKPARENT;
926		nd.ni_dirp = SCARG(uap, link);
927		if ((error = namei(&nd)) == 0) {
928			if (nd.ni_vp != NULL)
929				error = EEXIST;
930			if (!error) {
931				VOP_LEASE(nd.ni_dvp, p, p->p_ucred,
932				    LEASE_WRITE);
933				VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
934				error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd);
935			} else {
936				VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
937				if (nd.ni_dvp == nd.ni_vp)
938					vrele(nd.ni_dvp);
939				else
940					vput(nd.ni_dvp);
941				if (nd.ni_vp)
942					vrele(nd.ni_vp);
943			}
944		}
945	}
946	vrele(vp);
947	return (error);
948}
949
950/*
951 * Make a symbolic link.
952 */
953/* ARGSUSED */
954sys_symlink(p, v, retval)
955	struct proc *p;
956	void *v;
957	register_t *retval;
958{
959	register struct sys_symlink_args /* {
960		syscallarg(char *) path;
961		syscallarg(char *) link;
962	} */ *uap = v;
963	struct vattr vattr;
964	char *path;
965	int error;
966	struct nameidata nd;
967
968	MALLOC(path, char *, MAXPATHLEN, M_NAMEI, M_WAITOK);
969	if (error = copyinstr(SCARG(uap, path), path, MAXPATHLEN, (size_t *)0))
970		goto out;
971	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, link), p);
972	if (error = namei(&nd))
973		goto out;
974	if (nd.ni_vp) {
975		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
976		if (nd.ni_dvp == nd.ni_vp)
977			vrele(nd.ni_dvp);
978		else
979			vput(nd.ni_dvp);
980		vrele(nd.ni_vp);
981		error = EEXIST;
982		goto out;
983	}
984	VATTR_NULL(&vattr);
985	vattr.va_mode = ACCESSPERMS &~ p->p_fd->fd_cmask;
986	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
987	error = VOP_SYMLINK(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr, path);
988out:
989	FREE(path, M_NAMEI);
990	return (error);
991}
992
993/*
994 * Delete a whiteout from the filesystem.
995 */
996/* ARGSUSED */
997sys_undelete(p, v, retval)
998	struct proc *p;
999	void *v;
1000	register_t *retval;
1001{
1002	register struct sys_undelete_args /* {
1003		syscallarg(char *) path;
1004	} */ *uap = v;
1005	int error;
1006	struct nameidata nd;
1007
1008	NDINIT(&nd, DELETE, LOCKPARENT|DOWHITEOUT, UIO_USERSPACE,
1009	    SCARG(uap, path), p);
1010	error = namei(&nd);
1011	if (error)
1012		return (error);
1013
1014	if (nd.ni_vp != NULLVP || !(nd.ni_cnd.cn_flags & ISWHITEOUT)) {
1015		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1016		if (nd.ni_dvp == nd.ni_vp)
1017			vrele(nd.ni_dvp);
1018		else
1019			vput(nd.ni_dvp);
1020		if (nd.ni_vp)
1021			vrele(nd.ni_vp);
1022		return (EEXIST);
1023	}
1024
1025	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1026	if (error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, DELETE))
1027		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1028	vput(nd.ni_dvp);
1029	return (error);
1030}
1031
1032/*
1033 * Delete a name from the filesystem.
1034 */
1035/* ARGSUSED */
1036sys_unlink(p, v, retval)
1037	struct proc *p;
1038	void *v;
1039	register_t *retval;
1040{
1041	struct sys_unlink_args /* {
1042		syscallarg(char *) path;
1043	} */ *uap = v;
1044	register struct vnode *vp;
1045	int error;
1046	struct nameidata nd;
1047
1048	NDINIT(&nd, DELETE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1049	if (error = namei(&nd))
1050		return (error);
1051	vp = nd.ni_vp;
1052	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1053	VOP_LOCK(vp);
1054
1055	if (vp->v_type != VDIR ||
1056	    (error = suser(p->p_ucred, &p->p_acflag)) == 0) {
1057		/*
1058		 * The root of a mounted filesystem cannot be deleted.
1059		 */
1060		if (vp->v_flag & VROOT)
1061			error = EBUSY;
1062		else
1063			(void)vnode_pager_uncache(vp);
1064	}
1065
1066	if (!error) {
1067		VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1068		error = VOP_REMOVE(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
1069	} else {
1070		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1071		if (nd.ni_dvp == vp)
1072			vrele(nd.ni_dvp);
1073		else
1074			vput(nd.ni_dvp);
1075		if (vp != NULLVP)
1076			vput(vp);
1077	}
1078	return (error);
1079}
1080
1081/*
1082 * Reposition read/write file offset.
1083 */
1084sys_lseek(p, v, retval)
1085	struct proc *p;
1086	void *v;
1087	register_t *retval;
1088{
1089	register struct sys_lseek_args /* {
1090		syscallarg(int) fd;
1091		syscallarg(int) pad;
1092		syscallarg(off_t) offset;
1093		syscallarg(int) whence;
1094	} */ *uap = v;
1095	struct ucred *cred = p->p_ucred;
1096	register struct filedesc *fdp = p->p_fd;
1097	register struct file *fp;
1098	struct vattr vattr;
1099	int error;
1100
1101	if ((u_int)SCARG(uap, fd) >= fdp->fd_nfiles ||
1102	    (fp = fdp->fd_ofiles[SCARG(uap, fd)]) == NULL)
1103		return (EBADF);
1104	if (fp->f_type != DTYPE_VNODE)
1105		return (ESPIPE);
1106	switch (SCARG(uap, whence)) {
1107	case L_INCR:
1108		fp->f_offset += SCARG(uap, offset);
1109		break;
1110	case L_XTND:
1111		if (error =
1112		    VOP_GETATTR((struct vnode *)fp->f_data, &vattr, cred, p))
1113			return (error);
1114		fp->f_offset = SCARG(uap, offset) + vattr.va_size;
1115		break;
1116	case L_SET:
1117		fp->f_offset = SCARG(uap, offset);
1118		break;
1119	default:
1120		return (EINVAL);
1121	}
1122	*(off_t *)retval = fp->f_offset;
1123	return (0);
1124}
1125
1126/*
1127 * Check access permissions.
1128 */
1129sys_access(p, v, retval)
1130	struct proc *p;
1131	void *v;
1132	register_t *retval;
1133{
1134	register struct sys_access_args /* {
1135		syscallarg(char *) path;
1136		syscallarg(int) flags;
1137	} */ *uap = v;
1138	register struct ucred *cred = p->p_ucred;
1139	register struct vnode *vp;
1140	int error, flags, t_gid, t_uid;
1141	struct nameidata nd;
1142
1143	t_uid = cred->cr_uid;
1144	t_gid = cred->cr_gid;
1145	cred->cr_uid = p->p_cred->p_ruid;
1146	cred->cr_gid = p->p_cred->p_rgid;
1147	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1148	    SCARG(uap, path), p);
1149	if (error = namei(&nd))
1150		goto out1;
1151	vp = nd.ni_vp;
1152
1153	/* Flags == 0 means only check for existence. */
1154	if (SCARG(uap, flags)) {
1155		flags = 0;
1156		if (SCARG(uap, flags) & R_OK)
1157			flags |= VREAD;
1158		if (SCARG(uap, flags) & W_OK)
1159			flags |= VWRITE;
1160		if (SCARG(uap, flags) & X_OK)
1161			flags |= VEXEC;
1162		if ((flags & VWRITE) == 0 || (error = vn_writechk(vp)) == 0)
1163			error = VOP_ACCESS(vp, flags, cred, p);
1164	}
1165	vput(vp);
1166out1:
1167	cred->cr_uid = t_uid;
1168	cred->cr_gid = t_gid;
1169	return (error);
1170}
1171
1172/*
1173 * Get file status; this version follows links.
1174 */
1175/* ARGSUSED */
1176sys_stat(p, v, retval)
1177	struct proc *p;
1178	void *v;
1179	register_t *retval;
1180{
1181	register struct sys_stat_args /* {
1182		syscallarg(char *) path;
1183		syscallarg(struct stat *) ub;
1184	} */ *uap = v;
1185	struct stat sb;
1186	int error;
1187	struct nameidata nd;
1188
1189	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1190	    SCARG(uap, path), p);
1191	if (error = namei(&nd))
1192		return (error);
1193	error = vn_stat(nd.ni_vp, &sb, p);
1194	vput(nd.ni_vp);
1195	if (error)
1196		return (error);
1197	error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
1198	return (error);
1199}
1200
1201/*
1202 * Get file status; this version does not follow links.
1203 */
1204/* ARGSUSED */
1205sys_lstat(p, v, retval)
1206	struct proc *p;
1207	void *v;
1208	register_t *retval;
1209{
1210	register struct sys_lstat_args /* {
1211		syscallarg(char *) path;
1212		syscallarg(struct stat *) ub;
1213	} */ *uap = v;
1214	int error;
1215	struct vnode *vp, *dvp;
1216	struct stat sb, sb1;
1217	struct nameidata nd;
1218
1219	NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | LOCKPARENT, UIO_USERSPACE,
1220	    SCARG(uap, path), p);
1221	if (error = namei(&nd))
1222		return (error);
1223	/*
1224	 * For symbolic links, always return the attributes of its
1225	 * containing directory, except for mode, size, and links.
1226	 */
1227	vp = nd.ni_vp;
1228	dvp = nd.ni_dvp;
1229	if (vp->v_type != VLNK) {
1230		if (dvp == vp)
1231			vrele(dvp);
1232		else
1233			vput(dvp);
1234		error = vn_stat(vp, &sb, p);
1235		vput(vp);
1236		if (error)
1237			return (error);
1238	} else {
1239		error = vn_stat(dvp, &sb, p);
1240		vput(dvp);
1241		if (error) {
1242			vput(vp);
1243			return (error);
1244		}
1245		error = vn_stat(vp, &sb1, p);
1246		vput(vp);
1247		if (error)
1248			return (error);
1249		sb.st_mode &= ~S_IFDIR;
1250		sb.st_mode |= S_IFLNK;
1251		sb.st_nlink = sb1.st_nlink;
1252		sb.st_size = sb1.st_size;
1253		sb.st_blocks = sb1.st_blocks;
1254	}
1255	error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
1256	return (error);
1257}
1258
1259/*
1260 * Get configurable pathname variables.
1261 */
1262/* ARGSUSED */
1263sys_pathconf(p, v, retval)
1264	struct proc *p;
1265	void *v;
1266	register_t *retval;
1267{
1268	register struct sys_pathconf_args /* {
1269		syscallarg(char *) path;
1270		syscallarg(int) name;
1271	} */ *uap = v;
1272	int error;
1273	struct nameidata nd;
1274
1275	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1276	    SCARG(uap, path), p);
1277	if (error = namei(&nd))
1278		return (error);
1279	error = VOP_PATHCONF(nd.ni_vp, SCARG(uap, name), retval);
1280	vput(nd.ni_vp);
1281	return (error);
1282}
1283
1284/*
1285 * Return target name of a symbolic link.
1286 */
1287/* ARGSUSED */
1288sys_readlink(p, v, retval)
1289	struct proc *p;
1290	void *v;
1291	register_t *retval;
1292{
1293	register struct sys_readlink_args /* {
1294		syscallarg(char *) path;
1295		syscallarg(char *) buf;
1296		syscallarg(int) count;
1297	} */ *uap = v;
1298	register struct vnode *vp;
1299	struct iovec aiov;
1300	struct uio auio;
1301	int error;
1302	struct nameidata nd;
1303
1304	NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF, UIO_USERSPACE,
1305	    SCARG(uap, path), p);
1306	if (error = namei(&nd))
1307		return (error);
1308	vp = nd.ni_vp;
1309	if (vp->v_type != VLNK)
1310		error = EINVAL;
1311	else {
1312		aiov.iov_base = SCARG(uap, buf);
1313		aiov.iov_len = SCARG(uap, count);
1314		auio.uio_iov = &aiov;
1315		auio.uio_iovcnt = 1;
1316		auio.uio_offset = 0;
1317		auio.uio_rw = UIO_READ;
1318		auio.uio_segflg = UIO_USERSPACE;
1319		auio.uio_procp = p;
1320		auio.uio_resid = SCARG(uap, count);
1321		error = VOP_READLINK(vp, &auio, p->p_ucred);
1322	}
1323	vput(vp);
1324	*retval = SCARG(uap, count) - auio.uio_resid;
1325	return (error);
1326}
1327
1328/*
1329 * Change flags of a file given a path name.
1330 */
1331/* ARGSUSED */
1332sys_chflags(p, v, retval)
1333	struct proc *p;
1334	void *v;
1335	register_t *retval;
1336{
1337	register struct sys_chflags_args /* {
1338		syscallarg(char *) path;
1339		syscallarg(int) flags;
1340	} */ *uap = v;
1341	register struct vnode *vp;
1342	struct vattr vattr;
1343	int error;
1344	struct nameidata nd;
1345
1346	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1347	if (error = namei(&nd))
1348		return (error);
1349	vp = nd.ni_vp;
1350	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1351	VOP_LOCK(vp);
1352	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1353		error = EROFS;
1354	else {
1355		VATTR_NULL(&vattr);
1356		vattr.va_flags = SCARG(uap, flags);
1357		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1358	}
1359	vput(vp);
1360	return (error);
1361}
1362
1363/*
1364 * Change flags of a file given a file descriptor.
1365 */
1366/* ARGSUSED */
1367sys_fchflags(p, v, retval)
1368	struct proc *p;
1369	void *v;
1370	register_t *retval;
1371{
1372	register struct sys_fchflags_args /* {
1373		syscallarg(int) fd;
1374		syscallarg(int) flags;
1375	} */ *uap = v;
1376	struct vattr vattr;
1377	struct vnode *vp;
1378	struct file *fp;
1379	int error;
1380
1381	if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
1382		return (error);
1383	vp = (struct vnode *)fp->f_data;
1384	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1385	VOP_LOCK(vp);
1386	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1387		error = EROFS;
1388	else {
1389		VATTR_NULL(&vattr);
1390		vattr.va_flags = SCARG(uap, flags);
1391		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1392	}
1393	VOP_UNLOCK(vp);
1394	return (error);
1395}
1396
1397/*
1398 * Change mode of a file given path name.
1399 */
1400/* ARGSUSED */
1401sys_chmod(p, v, retval)
1402	struct proc *p;
1403	void *v;
1404	register_t *retval;
1405{
1406	register struct sys_chmod_args /* {
1407		syscallarg(char *) path;
1408		syscallarg(int) mode;
1409	} */ *uap = v;
1410	register struct vnode *vp;
1411	struct vattr vattr;
1412	int error;
1413	struct nameidata nd;
1414
1415	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1416	if (error = namei(&nd))
1417		return (error);
1418	vp = nd.ni_vp;
1419	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1420	VOP_LOCK(vp);
1421	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1422		error = EROFS;
1423	else {
1424		VATTR_NULL(&vattr);
1425		vattr.va_mode = SCARG(uap, mode) & ALLPERMS;
1426		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1427	}
1428	vput(vp);
1429	return (error);
1430}
1431
1432/*
1433 * Change mode of a file given a file descriptor.
1434 */
1435/* ARGSUSED */
1436sys_fchmod(p, v, retval)
1437	struct proc *p;
1438	void *v;
1439	register_t *retval;
1440{
1441	register struct sys_fchmod_args /* {
1442		syscallarg(int) fd;
1443		syscallarg(int) mode;
1444	} */ *uap = v;
1445	struct vattr vattr;
1446	struct vnode *vp;
1447	struct file *fp;
1448	int error;
1449
1450	if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
1451		return (error);
1452	vp = (struct vnode *)fp->f_data;
1453	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1454	VOP_LOCK(vp);
1455	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1456		error = EROFS;
1457	else {
1458		VATTR_NULL(&vattr);
1459		vattr.va_mode = SCARG(uap, mode) & ALLPERMS;
1460		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1461	}
1462	VOP_UNLOCK(vp);
1463	return (error);
1464}
1465
1466/*
1467 * Set ownership given a path name.
1468 */
1469/* ARGSUSED */
1470sys_chown(p, v, retval)
1471	struct proc *p;
1472	void *v;
1473	register_t *retval;
1474{
1475	register struct sys_chown_args /* {
1476		syscallarg(char *) path;
1477		syscallarg(int) uid;
1478		syscallarg(int) gid;
1479	} */ *uap = v;
1480	register struct vnode *vp;
1481	struct vattr vattr;
1482	int error;
1483	struct nameidata nd;
1484
1485	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1486	if (error = namei(&nd))
1487		return (error);
1488	vp = nd.ni_vp;
1489	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1490	VOP_LOCK(vp);
1491	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1492		error = EROFS;
1493	else {
1494		VATTR_NULL(&vattr);
1495		vattr.va_uid = SCARG(uap, uid);
1496		vattr.va_gid = SCARG(uap, gid);
1497		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1498	}
1499	vput(vp);
1500	return (error);
1501}
1502
1503/*
1504 * Set ownership given a file descriptor.
1505 */
1506/* ARGSUSED */
1507sys_fchown(p, v, retval)
1508	struct proc *p;
1509	void *v;
1510	register_t *retval;
1511{
1512	register struct sys_fchown_args /* {
1513		syscallarg(int) fd;
1514		syscallarg(int) uid;
1515		syscallarg(int) gid;
1516	} */ *uap = v;
1517	struct vattr vattr;
1518	struct vnode *vp;
1519	struct file *fp;
1520	int error;
1521
1522	if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
1523		return (error);
1524	vp = (struct vnode *)fp->f_data;
1525	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1526	VOP_LOCK(vp);
1527	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1528		error = EROFS;
1529	else {
1530		VATTR_NULL(&vattr);
1531		vattr.va_uid = SCARG(uap, uid);
1532		vattr.va_gid = SCARG(uap, gid);
1533		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1534	}
1535	VOP_UNLOCK(vp);
1536	return (error);
1537}
1538
1539/*
1540 * Set the access and modification times of a file.
1541 */
1542/* ARGSUSED */
1543sys_utimes(p, v, retval)
1544	struct proc *p;
1545	void *v;
1546	register_t *retval;
1547{
1548	register struct sys_utimes_args /* {
1549		syscallarg(char *) path;
1550		syscallarg(struct timeval *) tptr;
1551	} */ *uap = v;
1552	register struct vnode *vp;
1553	struct timeval tv[2];
1554	struct vattr vattr;
1555	int error;
1556	struct nameidata nd;
1557
1558	VATTR_NULL(&vattr);
1559	if (SCARG(uap, tptr) == NULL) {
1560		microtime(&tv[0]);
1561		tv[1] = tv[0];
1562		vattr.va_vaflags |= VA_UTIMES_NULL;
1563	} else if (error = copyin((caddr_t)SCARG(uap, tptr), (caddr_t)tv,
1564	    sizeof (tv)))
1565  		return (error);
1566	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1567	if (error = namei(&nd))
1568		return (error);
1569	vp = nd.ni_vp;
1570	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1571	VOP_LOCK(vp);
1572	if (vp->v_mount->mnt_flag & MNT_RDONLY)
1573		error = EROFS;
1574	else {
1575		vattr.va_atime.ts_sec = tv[0].tv_sec;
1576		vattr.va_atime.ts_nsec = tv[0].tv_usec * 1000;
1577		vattr.va_mtime.ts_sec = tv[1].tv_sec;
1578		vattr.va_mtime.ts_nsec = tv[1].tv_usec * 1000;
1579		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1580	}
1581	vput(vp);
1582	return (error);
1583}
1584
1585/*
1586 * Truncate a file given its path name.
1587 */
1588/* ARGSUSED */
1589sys_truncate(p, v, retval)
1590	struct proc *p;
1591	void *v;
1592	register_t *retval;
1593{
1594	register struct sys_truncate_args /* {
1595		syscallarg(char *) path;
1596		syscallarg(int) pad;
1597		syscallarg(off_t) length;
1598	} */ *uap = v;
1599	register struct vnode *vp;
1600	struct vattr vattr;
1601	int error;
1602	struct nameidata nd;
1603
1604	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1605	if (error = namei(&nd))
1606		return (error);
1607	vp = nd.ni_vp;
1608	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1609	VOP_LOCK(vp);
1610	if (vp->v_type == VDIR)
1611		error = EISDIR;
1612	else if ((error = vn_writechk(vp)) == 0 &&
1613	    (error = VOP_ACCESS(vp, VWRITE, p->p_ucred, p)) == 0) {
1614		VATTR_NULL(&vattr);
1615		vattr.va_size = SCARG(uap, length);
1616		error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
1617	}
1618	vput(vp);
1619	return (error);
1620}
1621
1622/*
1623 * Truncate a file given a file descriptor.
1624 */
1625/* ARGSUSED */
1626sys_ftruncate(p, v, retval)
1627	struct proc *p;
1628	void *v;
1629	register_t *retval;
1630{
1631	register struct sys_ftruncate_args /* {
1632		syscallarg(int) fd;
1633		syscallarg(int) pad;
1634		syscallarg(off_t) length;
1635	} */ *uap = v;
1636	struct vattr vattr;
1637	struct vnode *vp;
1638	struct file *fp;
1639	int error;
1640
1641	if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
1642		return (error);
1643	if ((fp->f_flag & FWRITE) == 0)
1644		return (EINVAL);
1645	vp = (struct vnode *)fp->f_data;
1646	VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1647	VOP_LOCK(vp);
1648	if (vp->v_type == VDIR)
1649		error = EISDIR;
1650	else if ((error = vn_writechk(vp)) == 0) {
1651		VATTR_NULL(&vattr);
1652		vattr.va_size = SCARG(uap, length);
1653		error = VOP_SETATTR(vp, &vattr, fp->f_cred, p);
1654	}
1655	VOP_UNLOCK(vp);
1656	return (error);
1657}
1658
1659/*
1660 * Sync an open file.
1661 */
1662/* ARGSUSED */
1663sys_fsync(p, v, retval)
1664	struct proc *p;
1665	void *v;
1666	register_t *retval;
1667{
1668	struct sys_fsync_args /* {
1669		syscallarg(int) fd;
1670	} */ *uap = v;
1671	register struct vnode *vp;
1672	struct file *fp;
1673	int error;
1674
1675	if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
1676		return (error);
1677	vp = (struct vnode *)fp->f_data;
1678	VOP_LOCK(vp);
1679	error = VOP_FSYNC(vp, fp->f_cred, MNT_WAIT, p);
1680	VOP_UNLOCK(vp);
1681	return (error);
1682}
1683
1684/*
1685 * Rename files.  Source and destination must either both be directories,
1686 * or both not be directories.  If target is a directory, it must be empty.
1687 */
1688/* ARGSUSED */
1689sys_rename(p, v, retval)
1690	struct proc *p;
1691	void *v;
1692	register_t *retval;
1693{
1694	register struct sys_rename_args /* {
1695		syscallarg(char *) from;
1696		syscallarg(char *) to;
1697	} */ *uap = v;
1698	register struct vnode *tvp, *fvp, *tdvp;
1699	struct nameidata fromnd, tond;
1700	int error;
1701
1702	NDINIT(&fromnd, DELETE, WANTPARENT | SAVESTART, UIO_USERSPACE,
1703	    SCARG(uap, from), p);
1704	if (error = namei(&fromnd))
1705		return (error);
1706	fvp = fromnd.ni_vp;
1707	NDINIT(&tond, RENAME, LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART,
1708	    UIO_USERSPACE, SCARG(uap, to), p);
1709	if (error = namei(&tond)) {
1710		VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
1711		vrele(fromnd.ni_dvp);
1712		vrele(fvp);
1713		goto out1;
1714	}
1715	tdvp = tond.ni_dvp;
1716	tvp = tond.ni_vp;
1717	if (tvp != NULL) {
1718		if (fvp->v_type == VDIR && tvp->v_type != VDIR) {
1719			error = ENOTDIR;
1720			goto out;
1721		} else if (fvp->v_type != VDIR && tvp->v_type == VDIR) {
1722			error = EISDIR;
1723			goto out;
1724		}
1725	}
1726	if (fvp == tdvp)
1727		error = EINVAL;
1728	/*
1729	 * If source is the same as the destination (that is the
1730	 * same inode number with the same name in the same directory),
1731	 * then there is nothing to do.
1732	 */
1733	if (fvp == tvp && fromnd.ni_dvp == tdvp &&
1734	    fromnd.ni_cnd.cn_namelen == tond.ni_cnd.cn_namelen &&
1735	    !bcmp(fromnd.ni_cnd.cn_nameptr, tond.ni_cnd.cn_nameptr,
1736	      fromnd.ni_cnd.cn_namelen))
1737		error = -1;
1738out:
1739	if (!error) {
1740		VOP_LEASE(tdvp, p, p->p_ucred, LEASE_WRITE);
1741		if (fromnd.ni_dvp != tdvp)
1742			VOP_LEASE(fromnd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1743		if (tvp)
1744			VOP_LEASE(tvp, p, p->p_ucred, LEASE_WRITE);
1745		error = VOP_RENAME(fromnd.ni_dvp, fromnd.ni_vp, &fromnd.ni_cnd,
1746				   tond.ni_dvp, tond.ni_vp, &tond.ni_cnd);
1747	} else {
1748		VOP_ABORTOP(tond.ni_dvp, &tond.ni_cnd);
1749		if (tdvp == tvp)
1750			vrele(tdvp);
1751		else
1752			vput(tdvp);
1753		if (tvp)
1754			vput(tvp);
1755		VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
1756		vrele(fromnd.ni_dvp);
1757		vrele(fvp);
1758	}
1759	vrele(tond.ni_startdir);
1760	FREE(tond.ni_cnd.cn_pnbuf, M_NAMEI);
1761out1:
1762	if (fromnd.ni_startdir)
1763		vrele(fromnd.ni_startdir);
1764	FREE(fromnd.ni_cnd.cn_pnbuf, M_NAMEI);
1765	if (error == -1)
1766		return (0);
1767	return (error);
1768}
1769
1770/*
1771 * Make a directory file.
1772 */
1773/* ARGSUSED */
1774sys_mkdir(p, v, retval)
1775	struct proc *p;
1776	void *v;
1777	register_t *retval;
1778{
1779	register struct sys_mkdir_args /* {
1780		syscallarg(char *) path;
1781		syscallarg(int) mode;
1782	} */ *uap = v;
1783	register struct vnode *vp;
1784	struct vattr vattr;
1785	int error;
1786	struct nameidata nd;
1787
1788	NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1789	if (error = namei(&nd))
1790		return (error);
1791	vp = nd.ni_vp;
1792	if (vp != NULL) {
1793		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1794		if (nd.ni_dvp == vp)
1795			vrele(nd.ni_dvp);
1796		else
1797			vput(nd.ni_dvp);
1798		vrele(vp);
1799		return (EEXIST);
1800	}
1801	VATTR_NULL(&vattr);
1802	vattr.va_type = VDIR;
1803	vattr.va_mode = (SCARG(uap, mode) & ACCESSPERMS) &~ p->p_fd->fd_cmask;
1804	VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1805	error = VOP_MKDIR(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
1806	if (!error)
1807		vput(nd.ni_vp);
1808	return (error);
1809}
1810
1811/*
1812 * Remove a directory file.
1813 */
1814/* ARGSUSED */
1815sys_rmdir(p, v, retval)
1816	struct proc *p;
1817	void *v;
1818	register_t *retval;
1819{
1820	struct sys_rmdir_args /* {
1821		syscallarg(char *) path;
1822	} */ *uap = v;
1823	register struct vnode *vp;
1824	int error;
1825	struct nameidata nd;
1826
1827	NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
1828	    SCARG(uap, path), p);
1829	if (error = namei(&nd))
1830		return (error);
1831	vp = nd.ni_vp;
1832	if (vp->v_type != VDIR) {
1833		error = ENOTDIR;
1834		goto out;
1835	}
1836	/*
1837	 * No rmdir "." please.
1838	 */
1839	if (nd.ni_dvp == vp) {
1840		error = EINVAL;
1841		goto out;
1842	}
1843	/*
1844	 * The root of a mounted filesystem cannot be deleted.
1845	 */
1846	if (vp->v_flag & VROOT)
1847		error = EBUSY;
1848out:
1849	if (!error) {
1850		VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1851		VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1852		error = VOP_RMDIR(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
1853	} else {
1854		VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1855		if (nd.ni_dvp == vp)
1856			vrele(nd.ni_dvp);
1857		else
1858			vput(nd.ni_dvp);
1859		vput(vp);
1860	}
1861	return (error);
1862}
1863
1864/*
1865 * Read a block of directory entries in a file system independent format.
1866 */
1867sys_getdirentries(p, v, retval)
1868	struct proc *p;
1869	void *v;
1870	register_t *retval;
1871{
1872	register struct sys_getdirentries_args /* {
1873		syscallarg(int) fd;
1874		syscallarg(char *) buf;
1875		syscallarg(u_int) count;
1876		syscallarg(long *) basep;
1877	} */ *uap = v;
1878	register struct vnode *vp;
1879	struct file *fp;
1880	struct uio auio;
1881	struct iovec aiov;
1882	long loff;
1883	int error, eofflag;
1884
1885	if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
1886		return (error);
1887	if ((fp->f_flag & FREAD) == 0)
1888		return (EBADF);
1889	vp = (struct vnode *)fp->f_data;
1890unionread:
1891	if (vp->v_type != VDIR)
1892		return (EINVAL);
1893	aiov.iov_base = SCARG(uap, buf);
1894	aiov.iov_len = SCARG(uap, count);
1895	auio.uio_iov = &aiov;
1896	auio.uio_iovcnt = 1;
1897	auio.uio_rw = UIO_READ;
1898	auio.uio_segflg = UIO_USERSPACE;
1899	auio.uio_procp = p;
1900	auio.uio_resid = SCARG(uap, count);
1901	VOP_LOCK(vp);
1902	loff = auio.uio_offset = fp->f_offset;
1903	error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, (u_long *)0, 0);
1904	fp->f_offset = auio.uio_offset;
1905	VOP_UNLOCK(vp);
1906	if (error)
1907		return (error);
1908
1909#ifdef UNION
1910{
1911	extern int (**union_vnodeop_p)();
1912	extern struct vnode *union_dircache __P((struct vnode *));
1913
1914	if ((SCARG(uap, count) == auio.uio_resid) &&
1915	    (vp->v_op == union_vnodeop_p)) {
1916		struct vnode *lvp;
1917
1918		lvp = union_dircache(vp);
1919		if (lvp != NULLVP) {
1920			struct vattr va;
1921
1922			/*
1923			 * If the directory is opaque,
1924			 * then don't show lower entries
1925			 */
1926			error = VOP_GETATTR(vp, &va, fp->f_cred, p);
1927			if (va.va_flags & OPAQUE) {
1928				vput(lvp);
1929				lvp = NULL;
1930			}
1931		}
1932
1933		if (lvp != NULLVP) {
1934			error = VOP_OPEN(lvp, FREAD, fp->f_cred, p);
1935			VOP_UNLOCK(lvp);
1936
1937			if (error) {
1938				vrele(lvp);
1939				return (error);
1940			}
1941			fp->f_data = (caddr_t) lvp;
1942			fp->f_offset = 0;
1943			error = vn_close(vp, FREAD, fp->f_cred, p);
1944			if (error)
1945				return (error);
1946			vp = lvp;
1947			goto unionread;
1948		}
1949	}
1950}
1951#endif /* UNION */
1952
1953	if ((SCARG(uap, count) == auio.uio_resid) &&
1954	    (vp->v_flag & VROOT) &&
1955	    (vp->v_mount->mnt_flag & MNT_UNION)) {
1956		struct vnode *tvp = vp;
1957		vp = vp->v_mount->mnt_vnodecovered;
1958		VREF(vp);
1959		fp->f_data = (caddr_t) vp;
1960		fp->f_offset = 0;
1961		vrele(tvp);
1962		goto unionread;
1963	}
1964	error = copyout((caddr_t)&loff, (caddr_t)SCARG(uap, basep),
1965	    sizeof(long));
1966	*retval = SCARG(uap, count) - auio.uio_resid;
1967	return (error);
1968}
1969
1970/*
1971 * Set the mode mask for creation of filesystem nodes.
1972 */
1973int
1974sys_umask(p, v, retval)
1975	struct proc *p;
1976	void *v;
1977	register_t *retval;
1978{
1979	struct sys_umask_args /* {
1980		syscallarg(int) newmask;
1981	} */ *uap = v;
1982	register struct filedesc *fdp;
1983
1984	fdp = p->p_fd;
1985	*retval = fdp->fd_cmask;
1986	fdp->fd_cmask = SCARG(uap, newmask) & ALLPERMS;
1987	return (0);
1988}
1989
1990/*
1991 * Void all references to file by ripping underlying filesystem
1992 * away from vnode.
1993 */
1994/* ARGSUSED */
1995sys_revoke(p, v, retval)
1996	struct proc *p;
1997	void *v;
1998	register_t *retval;
1999{
2000	register struct sys_revoke_args /* {
2001		syscallarg(char *) path;
2002	} */ *uap = v;
2003	register struct vnode *vp;
2004	struct vattr vattr;
2005	int error;
2006	struct nameidata nd;
2007
2008	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2009	if (error = namei(&nd))
2010		return (error);
2011	vp = nd.ni_vp;
2012	if (vp->v_type != VCHR && vp->v_type != VBLK) {
2013		error = EINVAL;
2014		goto out;
2015	}
2016	if (error = VOP_GETATTR(vp, &vattr, p->p_ucred, p))
2017		goto out;
2018	if (p->p_ucred->cr_uid != vattr.va_uid &&
2019	    (error = suser(p->p_ucred, &p->p_acflag)))
2020		goto out;
2021	if (vp->v_usecount > 1 || (vp->v_flag & VALIASED))
2022		vgoneall(vp);
2023out:
2024	vrele(vp);
2025	return (error);
2026}
2027
2028/*
2029 * Convert a user file descriptor to a kernel file entry.
2030 */
2031getvnode(fdp, fd, fpp)
2032	struct filedesc *fdp;
2033	struct file **fpp;
2034	int fd;
2035{
2036	struct file *fp;
2037
2038	if ((u_int)fd >= fdp->fd_nfiles ||
2039	    (fp = fdp->fd_ofiles[fd]) == NULL)
2040		return (EBADF);
2041	if (fp->f_type != DTYPE_VNODE)
2042		return (EINVAL);
2043	*fpp = fp;
2044	return (0);
2045}
2046