Initial import from FreeBSD RELENG_4:
[dragonfly.git] / sys / kern / vfs_syscalls.c
CommitLineData
984263bc
MD
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: src/sys/kern/vfs_syscalls.c,v 1.151.2.18 2003/04/04 20:35:58 tegge Exp $
40 */
41
42/* For 4.3 integer FS ID compatibility */
43#include "opt_compat.h"
44
45#include <sys/param.h>
46#include <sys/systm.h>
47#include <sys/buf.h>
48#include <sys/sysent.h>
49#include <sys/malloc.h>
50#include <sys/mount.h>
51#include <sys/sysproto.h>
52#include <sys/namei.h>
53#include <sys/filedesc.h>
54#include <sys/kernel.h>
55#include <sys/fcntl.h>
56#include <sys/file.h>
57#include <sys/linker.h>
58#include <sys/stat.h>
59#include <sys/unistd.h>
60#include <sys/vnode.h>
61#include <sys/proc.h>
62#include <sys/dirent.h>
63#include <sys/extattr.h>
64
65#include <machine/limits.h>
66#include <miscfs/union/union.h>
67#include <sys/sysctl.h>
68#include <vm/vm.h>
69#include <vm/vm_object.h>
70#include <vm/vm_zone.h>
71#include <vm/vm_page.h>
72
73static int change_dir __P((struct nameidata *ndp, struct proc *p));
74static void checkdirs __P((struct vnode *olddp));
75static int chroot_refuse_vdir_fds __P((struct filedesc *fdp));
76static int getutimes __P((const struct timeval *, struct timespec *));
77static int setfown __P((struct proc *, struct vnode *, uid_t, gid_t));
78static int setfmode __P((struct proc *, struct vnode *, int));
79static int setfflags __P((struct proc *, struct vnode *, int));
80static int setutimes __P((struct proc *, struct vnode *,
81 const struct timespec *, int));
82static int usermount = 0; /* if 1, non-root can mount fs. */
83
84int (*union_dircheckp) __P((struct proc *, struct vnode **, struct file *));
85
86SYSCTL_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_
96struct mount_args {
97 char *type;
98 char *path;
99 int flags;
100 caddr_t data;
101};
102#endif
103/* ARGSUSED */
104int
105mount(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 /*
167 * Only root, or the user that did the original mount is
168 * permitted to update it.
169 */
170 if (mp->mnt_stat.f_owner != p->p_ucred->cr_uid &&
171 (error = suser(p))) {
172 vput(vp);
173 return (error);
174 }
175 if (vfs_busy(mp, LK_NOWAIT, 0, p)) {
176 vput(vp);
177 return (EBUSY);
178 }
179 simple_lock(&vp->v_interlock);
180 if ((vp->v_flag & VMOUNT) != 0 ||
181 vp->v_mountedhere != NULL) {
182 simple_unlock(&vp->v_interlock);
183 vfs_unbusy(mp, p);
184 vput(vp);
185 return (EBUSY);
186 }
187 vp->v_flag |= VMOUNT;
188 simple_unlock(&vp->v_interlock);
189 mp->mnt_flag |=
190 SCARG(uap, flags) & (MNT_RELOAD | MNT_FORCE | MNT_UPDATE);
191 VOP_UNLOCK(vp, 0, p);
192 goto update;
193 }
194 /*
195 * If the user is not root, ensure that they own the directory
196 * onto which we are attempting to mount.
197 */
198 if ((error = VOP_GETATTR(vp, &va, p->p_ucred, p)) ||
199 (va.va_uid != p->p_ucred->cr_uid &&
200 (error = suser(p)))) {
201 vput(vp);
202 return (error);
203 }
204 if ((error = vinvalbuf(vp, V_SAVE, p->p_ucred, p, 0, 0)) != 0) {
205 vput(vp);
206 return (error);
207 }
208 if (vp->v_type != VDIR) {
209 vput(vp);
210 return (ENOTDIR);
211 }
212#ifdef COMPAT_43
213 /*
214 * Historically filesystem types were identified by number. If we
215 * get an integer for the filesystem type instead of a string, we
216 * check to see if it matches one of the historic filesystem types.
217 */
218 fstypenum = (uintptr_t)SCARG(uap, type);
219 if (fstypenum < maxvfsconf) {
220 for (vfsp = vfsconf; vfsp; vfsp = vfsp->vfc_next)
221 if (vfsp->vfc_typenum == fstypenum)
222 break;
223 if (vfsp == NULL) {
224 vput(vp);
225 return (ENODEV);
226 }
227 strncpy(fstypename, vfsp->vfc_name, MFSNAMELEN);
228 } else
229#endif /* COMPAT_43 */
230 if ((error = copyinstr(SCARG(uap, type), fstypename, MFSNAMELEN, NULL)) != 0) {
231 vput(vp);
232 return (error);
233 }
234 for (vfsp = vfsconf; vfsp; vfsp = vfsp->vfc_next)
235 if (!strcmp(vfsp->vfc_name, fstypename))
236 break;
237 if (vfsp == NULL) {
238 linker_file_t lf;
239
240 /* Only load modules for root (very important!) */
241 if ((error = suser(p)) != 0) {
242 vput(vp);
243 return error;
244 }
245 error = linker_load_file(fstypename, &lf);
246 if (error || lf == NULL) {
247 vput(vp);
248 if (lf == NULL)
249 error = ENODEV;
250 return error;
251 }
252 lf->userrefs++;
253 /* lookup again, see if the VFS was loaded */
254 for (vfsp = vfsconf; vfsp; vfsp = vfsp->vfc_next)
255 if (!strcmp(vfsp->vfc_name, fstypename))
256 break;
257 if (vfsp == NULL) {
258 lf->userrefs--;
259 linker_file_unload(lf);
260 vput(vp);
261 return (ENODEV);
262 }
263 }
264 simple_lock(&vp->v_interlock);
265 if ((vp->v_flag & VMOUNT) != 0 ||
266 vp->v_mountedhere != NULL) {
267 simple_unlock(&vp->v_interlock);
268 vput(vp);
269 return (EBUSY);
270 }
271 vp->v_flag |= VMOUNT;
272 simple_unlock(&vp->v_interlock);
273
274 /*
275 * Allocate and initialize the filesystem.
276 */
277 mp = malloc(sizeof(struct mount), M_MOUNT, M_WAITOK);
278 bzero((char *)mp, (u_long)sizeof(struct mount));
279 TAILQ_INIT(&mp->mnt_nvnodelist);
280 TAILQ_INIT(&mp->mnt_reservedvnlist);
281 mp->mnt_nvnodelistsize = 0;
282 lockinit(&mp->mnt_lock, PVFS, "vfslock", 0, LK_NOPAUSE);
283 (void)vfs_busy(mp, LK_NOWAIT, 0, p);
284 mp->mnt_op = vfsp->vfc_vfsops;
285 mp->mnt_vfc = vfsp;
286 vfsp->vfc_refcount++;
287 mp->mnt_stat.f_type = vfsp->vfc_typenum;
288 mp->mnt_flag |= vfsp->vfc_flags & MNT_VISFLAGMASK;
289 strncpy(mp->mnt_stat.f_fstypename, vfsp->vfc_name, MFSNAMELEN);
290 mp->mnt_vnodecovered = vp;
291 mp->mnt_stat.f_owner = p->p_ucred->cr_uid;
292 mp->mnt_iosize_max = DFLTPHYS;
293 VOP_UNLOCK(vp, 0, p);
294update:
295 /*
296 * Set the mount level flags.
297 */
298 if (SCARG(uap, flags) & MNT_RDONLY)
299 mp->mnt_flag |= MNT_RDONLY;
300 else if (mp->mnt_flag & MNT_RDONLY)
301 mp->mnt_kern_flag |= MNTK_WANTRDWR;
302 mp->mnt_flag &=~ (MNT_NOSUID | MNT_NOEXEC | MNT_NODEV |
303 MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | MNT_NOATIME |
304 MNT_NOSYMFOLLOW | MNT_IGNORE |
305 MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR);
306 mp->mnt_flag |= SCARG(uap, flags) & (MNT_NOSUID | MNT_NOEXEC |
307 MNT_NODEV | MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | MNT_FORCE |
308 MNT_NOSYMFOLLOW | MNT_IGNORE |
309 MNT_NOATIME | MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR);
310 /*
311 * Mount the filesystem.
312 * XXX The final recipients of VFS_MOUNT just overwrite the ndp they
313 * get. No freeing of cn_pnbuf.
314 */
315 error = VFS_MOUNT(mp, SCARG(uap, path), SCARG(uap, data), &nd, p);
316 if (mp->mnt_flag & MNT_UPDATE) {
317 if (mp->mnt_kern_flag & MNTK_WANTRDWR)
318 mp->mnt_flag &= ~MNT_RDONLY;
319 mp->mnt_flag &=~ (MNT_UPDATE | MNT_RELOAD | MNT_FORCE);
320 mp->mnt_kern_flag &=~ MNTK_WANTRDWR;
321 if (error) {
322 mp->mnt_flag = flag;
323 mp->mnt_kern_flag = flag2;
324 }
325 if ((mp->mnt_flag & MNT_RDONLY) == 0) {
326 if (mp->mnt_syncer == NULL)
327 error = vfs_allocate_syncvnode(mp);
328 } else {
329 if (mp->mnt_syncer != NULL)
330 vrele(mp->mnt_syncer);
331 mp->mnt_syncer = NULL;
332 }
333 vfs_unbusy(mp, p);
334 simple_lock(&vp->v_interlock);
335 vp->v_flag &= ~VMOUNT;
336 simple_unlock(&vp->v_interlock);
337 vrele(vp);
338 return (error);
339 }
340 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
341 /*
342 * Put the new filesystem on the mount list after root.
343 */
344 cache_purge(vp);
345 if (!error) {
346 simple_lock(&vp->v_interlock);
347 vp->v_flag &= ~VMOUNT;
348 vp->v_mountedhere = mp;
349 simple_unlock(&vp->v_interlock);
350 simple_lock(&mountlist_slock);
351 TAILQ_INSERT_TAIL(&mountlist, mp, mnt_list);
352 simple_unlock(&mountlist_slock);
353 checkdirs(vp);
354 VOP_UNLOCK(vp, 0, p);
355 if ((mp->mnt_flag & MNT_RDONLY) == 0)
356 error = vfs_allocate_syncvnode(mp);
357 vfs_unbusy(mp, p);
358 if ((error = VFS_START(mp, 0, p)) != 0)
359 vrele(vp);
360 } else {
361 simple_lock(&vp->v_interlock);
362 vp->v_flag &= ~VMOUNT;
363 simple_unlock(&vp->v_interlock);
364 mp->mnt_vfc->vfc_refcount--;
365 vfs_unbusy(mp, p);
366 free((caddr_t)mp, M_MOUNT);
367 vput(vp);
368 }
369 return (error);
370}
371
372/*
373 * Scan all active processes to see if any of them have a current
374 * or root directory onto which the new filesystem has just been
375 * mounted. If so, replace them with the new mount point.
376 */
377static void
378checkdirs(olddp)
379 struct vnode *olddp;
380{
381 struct filedesc *fdp;
382 struct vnode *newdp;
383 struct proc *p;
384
385 if (olddp->v_usecount == 1)
386 return;
387 if (VFS_ROOT(olddp->v_mountedhere, &newdp))
388 panic("mount: lost mount");
389 LIST_FOREACH(p, &allproc, p_list) {
390 fdp = p->p_fd;
391 if (fdp->fd_cdir == olddp) {
392 vrele(fdp->fd_cdir);
393 VREF(newdp);
394 fdp->fd_cdir = newdp;
395 }
396 if (fdp->fd_rdir == olddp) {
397 vrele(fdp->fd_rdir);
398 VREF(newdp);
399 fdp->fd_rdir = newdp;
400 }
401 }
402 if (rootvnode == olddp) {
403 vrele(rootvnode);
404 VREF(newdp);
405 rootvnode = newdp;
406 }
407 vput(newdp);
408}
409
410/*
411 * Unmount a file system.
412 *
413 * Note: unmount takes a path to the vnode mounted on as argument,
414 * not special file (as before).
415 */
416#ifndef _SYS_SYSPROTO_H_
417struct unmount_args {
418 char *path;
419 int flags;
420};
421#endif
422/* ARGSUSED */
423int
424unmount(p, uap)
425 struct proc *p;
426 register struct unmount_args /* {
427 syscallarg(char *) path;
428 syscallarg(int) flags;
429 } */ *uap;
430{
431 register struct vnode *vp;
432 struct mount *mp;
433 int error;
434 struct nameidata nd;
435
436 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
437 SCARG(uap, path), p);
438 if ((error = namei(&nd)) != 0)
439 return (error);
440 vp = nd.ni_vp;
441 NDFREE(&nd, NDF_ONLY_PNBUF);
442 mp = vp->v_mount;
443
444 /*
445 * Only root, or the user that did the original mount is
446 * permitted to unmount this filesystem.
447 */
448 if ((mp->mnt_stat.f_owner != p->p_ucred->cr_uid) &&
449 (error = suser(p))) {
450 vput(vp);
451 return (error);
452 }
453
454 /*
455 * Don't allow unmounting the root file system.
456 */
457 if (mp->mnt_flag & MNT_ROOTFS) {
458 vput(vp);
459 return (EINVAL);
460 }
461
462 /*
463 * Must be the root of the filesystem
464 */
465 if ((vp->v_flag & VROOT) == 0) {
466 vput(vp);
467 return (EINVAL);
468 }
469 vput(vp);
470 return (dounmount(mp, SCARG(uap, flags), p));
471}
472
473/*
474 * Do the actual file system unmount.
475 */
476int
477dounmount(mp, flags, p)
478 register struct mount *mp;
479 int flags;
480 struct proc *p;
481{
482 struct vnode *coveredvp;
483 int error;
484 int async_flag;
485
486 simple_lock(&mountlist_slock);
487 if (mp->mnt_kern_flag & MNTK_UNMOUNT) {
488 simple_unlock(&mountlist_slock);
489 return (EBUSY);
490 }
491 mp->mnt_kern_flag |= MNTK_UNMOUNT;
492 /* Allow filesystems to detect that a forced unmount is in progress. */
493 if (flags & MNT_FORCE)
494 mp->mnt_kern_flag |= MNTK_UNMOUNTF;
495 error = lockmgr(&mp->mnt_lock, LK_DRAIN | LK_INTERLOCK |
496 ((flags & MNT_FORCE) ? 0 : LK_NOWAIT), &mountlist_slock, p);
497 if (error) {
498 mp->mnt_kern_flag &= ~(MNTK_UNMOUNT | MNTK_UNMOUNTF);
499 if (mp->mnt_kern_flag & MNTK_MWAIT)
500 wakeup((caddr_t)mp);
501 return (error);
502 }
503
504 if (mp->mnt_flag & MNT_EXPUBLIC)
505 vfs_setpublicfs(NULL, NULL, NULL);
506
507 vfs_msync(mp, MNT_WAIT);
508 async_flag = mp->mnt_flag & MNT_ASYNC;
509 mp->mnt_flag &=~ MNT_ASYNC;
510 cache_purgevfs(mp); /* remove cache entries for this file sys */
511 if (mp->mnt_syncer != NULL)
512 vrele(mp->mnt_syncer);
513 if (((mp->mnt_flag & MNT_RDONLY) ||
514 (error = VFS_SYNC(mp, MNT_WAIT, p->p_ucred, p)) == 0) ||
515 (flags & MNT_FORCE))
516 error = VFS_UNMOUNT(mp, flags, p);
517 simple_lock(&mountlist_slock);
518 if (error) {
519 if ((mp->mnt_flag & MNT_RDONLY) == 0 && mp->mnt_syncer == NULL)
520 (void) vfs_allocate_syncvnode(mp);
521 mp->mnt_kern_flag &= ~(MNTK_UNMOUNT | MNTK_UNMOUNTF);
522 mp->mnt_flag |= async_flag;
523 lockmgr(&mp->mnt_lock, LK_RELEASE | LK_INTERLOCK | LK_REENABLE,
524 &mountlist_slock, p);
525 if (mp->mnt_kern_flag & MNTK_MWAIT)
526 wakeup((caddr_t)mp);
527 return (error);
528 }
529 TAILQ_REMOVE(&mountlist, mp, mnt_list);
530 if ((coveredvp = mp->mnt_vnodecovered) != NULLVP) {
531 coveredvp->v_mountedhere = (struct mount *)0;
532 vrele(coveredvp);
533 }
534 mp->mnt_vfc->vfc_refcount--;
535 if (!TAILQ_EMPTY(&mp->mnt_nvnodelist))
536 panic("unmount: dangling vnode");
537 lockmgr(&mp->mnt_lock, LK_RELEASE | LK_INTERLOCK, &mountlist_slock, p);
538 if (mp->mnt_kern_flag & MNTK_MWAIT)
539 wakeup((caddr_t)mp);
540 free((caddr_t)mp, M_MOUNT);
541 return (0);
542}
543
544/*
545 * Sync each mounted filesystem.
546 */
547#ifndef _SYS_SYSPROTO_H_
548struct sync_args {
549 int dummy;
550};
551#endif
552
553#ifdef DEBUG
554static int syncprt = 0;
555SYSCTL_INT(_debug, OID_AUTO, syncprt, CTLFLAG_RW, &syncprt, 0, "");
556#endif
557
558/* ARGSUSED */
559int
560sync(p, uap)
561 struct proc *p;
562 struct sync_args *uap;
563{
564 register struct mount *mp, *nmp;
565 int asyncflag;
566
567 simple_lock(&mountlist_slock);
568 for (mp = TAILQ_FIRST(&mountlist); mp != NULL; mp = nmp) {
569 if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock, p)) {
570 nmp = TAILQ_NEXT(mp, mnt_list);
571 continue;
572 }
573 if ((mp->mnt_flag & MNT_RDONLY) == 0) {
574 asyncflag = mp->mnt_flag & MNT_ASYNC;
575 mp->mnt_flag &= ~MNT_ASYNC;
576 vfs_msync(mp, MNT_NOWAIT);
577 VFS_SYNC(mp, MNT_NOWAIT,
578 ((p != NULL) ? p->p_ucred : NOCRED), p);
579 mp->mnt_flag |= asyncflag;
580 }
581 simple_lock(&mountlist_slock);
582 nmp = TAILQ_NEXT(mp, mnt_list);
583 vfs_unbusy(mp, p);
584 }
585 simple_unlock(&mountlist_slock);
586#if 0
587/*
588 * XXX don't call vfs_bufstats() yet because that routine
589 * was not imported in the Lite2 merge.
590 */
591#ifdef DIAGNOSTIC
592 if (syncprt)
593 vfs_bufstats();
594#endif /* DIAGNOSTIC */
595#endif
596 return (0);
597}
598
599/* XXX PRISON: could be per prison flag */
600static int prison_quotas;
601#if 0
602SYSCTL_INT(_kern_prison, OID_AUTO, quotas, CTLFLAG_RW, &prison_quotas, 0, "");
603#endif
604
605/*
606 * Change filesystem quotas.
607 */
608#ifndef _SYS_SYSPROTO_H_
609struct quotactl_args {
610 char *path;
611 int cmd;
612 int uid;
613 caddr_t arg;
614};
615#endif
616/* ARGSUSED */
617int
618quotactl(p, uap)
619 struct proc *p;
620 register struct quotactl_args /* {
621 syscallarg(char *) path;
622 syscallarg(int) cmd;
623 syscallarg(int) uid;
624 syscallarg(caddr_t) arg;
625 } */ *uap;
626{
627 register struct mount *mp;
628 int error;
629 struct nameidata nd;
630
631 if (p->p_prison && !prison_quotas)
632 return (EPERM);
633 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
634 if ((error = namei(&nd)) != 0)
635 return (error);
636 mp = nd.ni_vp->v_mount;
637 NDFREE(&nd, NDF_ONLY_PNBUF);
638 vrele(nd.ni_vp);
639 return (VFS_QUOTACTL(mp, SCARG(uap, cmd), SCARG(uap, uid),
640 SCARG(uap, arg), p));
641}
642
643/*
644 * Get filesystem statistics.
645 */
646#ifndef _SYS_SYSPROTO_H_
647struct statfs_args {
648 char *path;
649 struct statfs *buf;
650};
651#endif
652/* ARGSUSED */
653int
654statfs(p, uap)
655 struct proc *p;
656 register struct statfs_args /* {
657 syscallarg(char *) path;
658 syscallarg(struct statfs *) buf;
659 } */ *uap;
660{
661 register struct mount *mp;
662 register struct statfs *sp;
663 int error;
664 struct nameidata nd;
665 struct statfs sb;
666
667 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
668 if ((error = namei(&nd)) != 0)
669 return (error);
670 mp = nd.ni_vp->v_mount;
671 sp = &mp->mnt_stat;
672 NDFREE(&nd, NDF_ONLY_PNBUF);
673 vrele(nd.ni_vp);
674 error = VFS_STATFS(mp, sp, p);
675 if (error)
676 return (error);
677 sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
678 if (suser_xxx(p->p_ucred, 0, 0)) {
679 bcopy((caddr_t)sp, (caddr_t)&sb, sizeof(sb));
680 sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0;
681 sp = &sb;
682 }
683 return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
684}
685
686/*
687 * Get filesystem statistics.
688 */
689#ifndef _SYS_SYSPROTO_H_
690struct fstatfs_args {
691 int fd;
692 struct statfs *buf;
693};
694#endif
695/* ARGSUSED */
696int
697fstatfs(p, uap)
698 struct proc *p;
699 register struct fstatfs_args /* {
700 syscallarg(int) fd;
701 syscallarg(struct statfs *) buf;
702 } */ *uap;
703{
704 struct file *fp;
705 struct mount *mp;
706 register struct statfs *sp;
707 int error;
708 struct statfs sb;
709
710 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
711 return (error);
712 mp = ((struct vnode *)fp->f_data)->v_mount;
713 if (mp == NULL)
714 return (EBADF);
715 sp = &mp->mnt_stat;
716 error = VFS_STATFS(mp, sp, p);
717 if (error)
718 return (error);
719 sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
720 if (suser_xxx(p->p_ucred, 0, 0)) {
721 bcopy((caddr_t)sp, (caddr_t)&sb, sizeof(sb));
722 sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0;
723 sp = &sb;
724 }
725 return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
726}
727
728/*
729 * Get statistics on all filesystems.
730 */
731#ifndef _SYS_SYSPROTO_H_
732struct getfsstat_args {
733 struct statfs *buf;
734 long bufsize;
735 int flags;
736};
737#endif
738int
739getfsstat(p, uap)
740 struct proc *p;
741 register struct getfsstat_args /* {
742 syscallarg(struct statfs *) buf;
743 syscallarg(long) bufsize;
744 syscallarg(int) flags;
745 } */ *uap;
746{
747 register struct mount *mp, *nmp;
748 register struct statfs *sp;
749 caddr_t sfsp;
750 long count, maxcount, error;
751
752 maxcount = SCARG(uap, bufsize) / sizeof(struct statfs);
753 sfsp = (caddr_t)SCARG(uap, buf);
754 count = 0;
755 simple_lock(&mountlist_slock);
756 for (mp = TAILQ_FIRST(&mountlist); mp != NULL; mp = nmp) {
757 if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock, p)) {
758 nmp = TAILQ_NEXT(mp, mnt_list);
759 continue;
760 }
761 if (sfsp && count < maxcount) {
762 sp = &mp->mnt_stat;
763 /*
764 * If MNT_NOWAIT or MNT_LAZY is specified, do not
765 * refresh the fsstat cache. MNT_NOWAIT or MNT_LAZY
766 * overrides MNT_WAIT.
767 */
768 if (((SCARG(uap, flags) & (MNT_LAZY|MNT_NOWAIT)) == 0 ||
769 (SCARG(uap, flags) & MNT_WAIT)) &&
770 (error = VFS_STATFS(mp, sp, p))) {
771 simple_lock(&mountlist_slock);
772 nmp = TAILQ_NEXT(mp, mnt_list);
773 vfs_unbusy(mp, p);
774 continue;
775 }
776 sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
777 error = copyout((caddr_t)sp, sfsp, sizeof(*sp));
778 if (error) {
779 vfs_unbusy(mp, p);
780 return (error);
781 }
782 sfsp += sizeof(*sp);
783 }
784 count++;
785 simple_lock(&mountlist_slock);
786 nmp = TAILQ_NEXT(mp, mnt_list);
787 vfs_unbusy(mp, p);
788 }
789 simple_unlock(&mountlist_slock);
790 if (sfsp && count > maxcount)
791 p->p_retval[0] = maxcount;
792 else
793 p->p_retval[0] = count;
794 return (0);
795}
796
797/*
798 * Change current working directory to a given file descriptor.
799 */
800#ifndef _SYS_SYSPROTO_H_
801struct fchdir_args {
802 int fd;
803};
804#endif
805/* ARGSUSED */
806int
807fchdir(p, uap)
808 struct proc *p;
809 struct fchdir_args /* {
810 syscallarg(int) fd;
811 } */ *uap;
812{
813 register struct filedesc *fdp = p->p_fd;
814 struct vnode *vp, *tdp;
815 struct mount *mp;
816 struct file *fp;
817 int error;
818
819 if ((error = getvnode(fdp, SCARG(uap, fd), &fp)) != 0)
820 return (error);
821 vp = (struct vnode *)fp->f_data;
822 VREF(vp);
823 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
824 if (vp->v_type != VDIR)
825 error = ENOTDIR;
826 else
827 error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
828 while (!error && (mp = vp->v_mountedhere) != NULL) {
829 if (vfs_busy(mp, 0, 0, p))
830 continue;
831 error = VFS_ROOT(mp, &tdp);
832 vfs_unbusy(mp, p);
833 if (error)
834 break;
835 vput(vp);
836 vp = tdp;
837 }
838 if (error) {
839 vput(vp);
840 return (error);
841 }
842 VOP_UNLOCK(vp, 0, p);
843 vrele(fdp->fd_cdir);
844 fdp->fd_cdir = vp;
845 return (0);
846}
847
848/*
849 * Change current working directory (``.'').
850 */
851#ifndef _SYS_SYSPROTO_H_
852struct chdir_args {
853 char *path;
854};
855#endif
856/* ARGSUSED */
857int
858chdir(p, uap)
859 struct proc *p;
860 struct chdir_args /* {
861 syscallarg(char *) path;
862 } */ *uap;
863{
864 register struct filedesc *fdp = p->p_fd;
865 int error;
866 struct nameidata nd;
867
868 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
869 SCARG(uap, path), p);
870 if ((error = change_dir(&nd, p)) != 0)
871 return (error);
872 NDFREE(&nd, NDF_ONLY_PNBUF);
873 vrele(fdp->fd_cdir);
874 fdp->fd_cdir = nd.ni_vp;
875 return (0);
876}
877
878/*
879 * Helper function for raised chroot(2) security function: Refuse if
880 * any filedescriptors are open directories.
881 */
882static int
883chroot_refuse_vdir_fds(fdp)
884 struct filedesc *fdp;
885{
886 struct vnode *vp;
887 struct file *fp;
888 int error;
889 int fd;
890
891 for (fd = 0; fd < fdp->fd_nfiles ; fd++) {
892 error = getvnode(fdp, fd, &fp);
893 if (error)
894 continue;
895 vp = (struct vnode *)fp->f_data;
896 if (vp->v_type != VDIR)
897 continue;
898 return(EPERM);
899 }
900 return (0);
901}
902
903/*
904 * This sysctl determines if we will allow a process to chroot(2) if it
905 * has a directory open:
906 * 0: disallowed for all processes.
907 * 1: allowed for processes that were not already chroot(2)'ed.
908 * 2: allowed for all processes.
909 */
910
911static int chroot_allow_open_directories = 1;
912
913SYSCTL_INT(_kern, OID_AUTO, chroot_allow_open_directories, CTLFLAG_RW,
914 &chroot_allow_open_directories, 0, "");
915
916/*
917 * Change notion of root (``/'') directory.
918 */
919#ifndef _SYS_SYSPROTO_H_
920struct chroot_args {
921 char *path;
922};
923#endif
924/* ARGSUSED */
925int
926chroot(p, uap)
927 struct proc *p;
928 struct chroot_args /* {
929 syscallarg(char *) path;
930 } */ *uap;
931{
932 register struct filedesc *fdp = p->p_fd;
933 int error;
934 struct nameidata nd;
935
936 error = suser_xxx(0, p, PRISON_ROOT);
937 if (error)
938 return (error);
939 if (chroot_allow_open_directories == 0 ||
940 (chroot_allow_open_directories == 1 && fdp->fd_rdir != rootvnode))
941 error = chroot_refuse_vdir_fds(fdp);
942 if (error)
943 return (error);
944 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
945 SCARG(uap, path), p);
946 if ((error = change_dir(&nd, p)) != 0)
947 return (error);
948 NDFREE(&nd, NDF_ONLY_PNBUF);
949 vrele(fdp->fd_rdir);
950 fdp->fd_rdir = nd.ni_vp;
951 if (!fdp->fd_jdir) {
952 fdp->fd_jdir = nd.ni_vp;
953 VREF(fdp->fd_jdir);
954 }
955 return (0);
956}
957
958/*
959 * Common routine for chroot and chdir.
960 */
961static int
962change_dir(ndp, p)
963 register struct nameidata *ndp;
964 struct proc *p;
965{
966 struct vnode *vp;
967 int error;
968
969 error = namei(ndp);
970 if (error)
971 return (error);
972 vp = ndp->ni_vp;
973 if (vp->v_type != VDIR)
974 error = ENOTDIR;
975 else
976 error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
977 if (error)
978 vput(vp);
979 else
980 VOP_UNLOCK(vp, 0, p);
981 return (error);
982}
983
984/*
985 * Check permissions, allocate an open file structure,
986 * and call the device open routine if any.
987 */
988#ifndef _SYS_SYSPROTO_H_
989struct open_args {
990 char *path;
991 int flags;
992 int mode;
993};
994#endif
995int
996open(p, uap)
997 struct proc *p;
998 register struct open_args /* {
999 syscallarg(char *) path;
1000 syscallarg(int) flags;
1001 syscallarg(int) mode;
1002 } */ *uap;
1003{
1004 register struct filedesc *fdp = p->p_fd;
1005 register struct file *fp;
1006 register struct vnode *vp;
1007 int cmode, flags, oflags;
1008 struct file *nfp;
1009 int type, indx, error;
1010 struct flock lf;
1011 struct nameidata nd;
1012
1013 oflags = SCARG(uap, flags);
1014 if ((oflags & O_ACCMODE) == O_ACCMODE)
1015 return (EINVAL);
1016 flags = FFLAGS(oflags);
1017 error = falloc(p, &nfp, &indx);
1018 if (error)
1019 return (error);
1020 fp = nfp;
1021 cmode = ((SCARG(uap, mode) &~ fdp->fd_cmask) & ALLPERMS) &~ S_ISTXT;
1022 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1023 p->p_dupfd = -indx - 1; /* XXX check for fdopen */
1024 /*
1025 * Bump the ref count to prevent another process from closing
1026 * the descriptor while we are blocked in vn_open()
1027 */
1028 fhold(fp);
1029 error = vn_open(&nd, flags, cmode);
1030 if (error) {
1031 /*
1032 * release our own reference
1033 */
1034 fdrop(fp, p);
1035
1036 /*
1037 * handle special fdopen() case. bleh. dupfdopen() is
1038 * responsible for dropping the old contents of ofiles[indx]
1039 * if it succeeds.
1040 */
1041 if ((error == ENODEV || error == ENXIO) &&
1042 p->p_dupfd >= 0 && /* XXX from fdopen */
1043 (error =
1044 dupfdopen(p, fdp, indx, p->p_dupfd, flags, error)) == 0) {
1045 p->p_retval[0] = indx;
1046 return (0);
1047 }
1048 /*
1049 * Clean up the descriptor, but only if another thread hadn't
1050 * replaced or closed it.
1051 */
1052 if (fdp->fd_ofiles[indx] == fp) {
1053 fdp->fd_ofiles[indx] = NULL;
1054 fdrop(fp, p);
1055 }
1056
1057 if (error == ERESTART)
1058 error = EINTR;
1059 return (error);
1060 }
1061 p->p_dupfd = 0;
1062 NDFREE(&nd, NDF_ONLY_PNBUF);
1063 vp = nd.ni_vp;
1064
1065 /*
1066 * There should be 2 references on the file, one from the descriptor
1067 * table, and one for us.
1068 *
1069 * Handle the case where someone closed the file (via its file
1070 * descriptor) while we were blocked. The end result should look
1071 * like opening the file succeeded but it was immediately closed.
1072 */
1073 if (fp->f_count == 1) {
1074 KASSERT(fdp->fd_ofiles[indx] != fp,
1075 ("Open file descriptor lost all refs"));
1076 VOP_UNLOCK(vp, 0, p);
1077 vn_close(vp, flags & FMASK, fp->f_cred, p);
1078 fdrop(fp, p);
1079 p->p_retval[0] = indx;
1080 return 0;
1081 }
1082
1083 fp->f_data = (caddr_t)vp;
1084 fp->f_flag = flags & FMASK;
1085 fp->f_ops = &vnops;
1086 fp->f_type = (vp->v_type == VFIFO ? DTYPE_FIFO : DTYPE_VNODE);
1087 if (flags & (O_EXLOCK | O_SHLOCK)) {
1088 lf.l_whence = SEEK_SET;
1089 lf.l_start = 0;
1090 lf.l_len = 0;
1091 if (flags & O_EXLOCK)
1092 lf.l_type = F_WRLCK;
1093 else
1094 lf.l_type = F_RDLCK;
1095 type = F_FLOCK;
1096 if ((flags & FNONBLOCK) == 0)
1097 type |= F_WAIT;
1098 VOP_UNLOCK(vp, 0, p);
1099 if ((error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type)) != 0) {
1100 /*
1101 * lock request failed. Normally close the descriptor
1102 * but handle the case where someone might have dup()d
1103 * it when we weren't looking. One reference is
1104 * owned by the descriptor array, the other by us.
1105 */
1106 if (fdp->fd_ofiles[indx] == fp) {
1107 fdp->fd_ofiles[indx] = NULL;
1108 fdrop(fp, p);
1109 }
1110 fdrop(fp, p);
1111 return (error);
1112 }
1113 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
1114 fp->f_flag |= FHASLOCK;
1115 }
1116 /* assert that vn_open created a backing object if one is needed */
1117 KASSERT(!vn_canvmio(vp) || VOP_GETVOBJECT(vp, NULL) == 0,
1118 ("open: vmio vnode has no backing object after vn_open"));
1119 VOP_UNLOCK(vp, 0, p);
1120
1121 /*
1122 * release our private reference, leaving the one associated with the
1123 * descriptor table intact.
1124 */
1125 fdrop(fp, p);
1126 p->p_retval[0] = indx;
1127 return (0);
1128}
1129
1130#ifdef COMPAT_43
1131/*
1132 * Create a file.
1133 */
1134#ifndef _SYS_SYSPROTO_H_
1135struct ocreat_args {
1136 char *path;
1137 int mode;
1138};
1139#endif
1140int
1141ocreat(p, uap)
1142 struct proc *p;
1143 register struct ocreat_args /* {
1144 syscallarg(char *) path;
1145 syscallarg(int) mode;
1146 } */ *uap;
1147{
1148 struct open_args /* {
1149 syscallarg(char *) path;
1150 syscallarg(int) flags;
1151 syscallarg(int) mode;
1152 } */ nuap;
1153
1154 SCARG(&nuap, path) = SCARG(uap, path);
1155 SCARG(&nuap, mode) = SCARG(uap, mode);
1156 SCARG(&nuap, flags) = O_WRONLY | O_CREAT | O_TRUNC;
1157 return (open(p, &nuap));
1158}
1159#endif /* COMPAT_43 */
1160
1161/*
1162 * Create a special file.
1163 */
1164#ifndef _SYS_SYSPROTO_H_
1165struct mknod_args {
1166 char *path;
1167 int mode;
1168 int dev;
1169};
1170#endif
1171/* ARGSUSED */
1172int
1173mknod(p, uap)
1174 struct proc *p;
1175 register struct mknod_args /* {
1176 syscallarg(char *) path;
1177 syscallarg(int) mode;
1178 syscallarg(int) dev;
1179 } */ *uap;
1180{
1181 register struct vnode *vp;
1182 struct vattr vattr;
1183 int error;
1184 int whiteout = 0;
1185 struct nameidata nd;
1186
1187 switch (SCARG(uap, mode) & S_IFMT) {
1188 case S_IFCHR:
1189 case S_IFBLK:
1190 error = suser(p);
1191 break;
1192 default:
1193 error = suser_xxx(0, p, PRISON_ROOT);
1194 break;
1195 }
1196 if (error)
1197 return (error);
1198 bwillwrite();
1199 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1200 if ((error = namei(&nd)) != 0)
1201 return (error);
1202 vp = nd.ni_vp;
1203 if (vp != NULL)
1204 error = EEXIST;
1205 else {
1206 VATTR_NULL(&vattr);
1207 vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
1208 vattr.va_rdev = SCARG(uap, dev);
1209 whiteout = 0;
1210
1211 switch (SCARG(uap, mode) & S_IFMT) {
1212 case S_IFMT: /* used by badsect to flag bad sectors */
1213 vattr.va_type = VBAD;
1214 break;
1215 case S_IFCHR:
1216 vattr.va_type = VCHR;
1217 break;
1218 case S_IFBLK:
1219 vattr.va_type = VBLK;
1220 break;
1221 case S_IFWHT:
1222 whiteout = 1;
1223 break;
1224 default:
1225 error = EINVAL;
1226 break;
1227 }
1228 }
1229 if (!error) {
1230 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1231 if (whiteout)
1232 error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, CREATE);
1233 else {
1234 error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp,
1235 &nd.ni_cnd, &vattr);
1236 if (error == 0)
1237 vput(nd.ni_vp);
1238 }
1239 NDFREE(&nd, NDF_ONLY_PNBUF);
1240 vput(nd.ni_dvp);
1241 } else {
1242 NDFREE(&nd, NDF_ONLY_PNBUF);
1243 if (nd.ni_dvp == vp)
1244 vrele(nd.ni_dvp);
1245 else
1246 vput(nd.ni_dvp);
1247 if (vp)
1248 vrele(vp);
1249 }
1250 ASSERT_VOP_UNLOCKED(nd.ni_dvp, "mknod");
1251 ASSERT_VOP_UNLOCKED(nd.ni_vp, "mknod");
1252 return (error);
1253}
1254
1255/*
1256 * Create a named pipe.
1257 */
1258#ifndef _SYS_SYSPROTO_H_
1259struct mkfifo_args {
1260 char *path;
1261 int mode;
1262};
1263#endif
1264/* ARGSUSED */
1265int
1266mkfifo(p, uap)
1267 struct proc *p;
1268 register struct mkfifo_args /* {
1269 syscallarg(char *) path;
1270 syscallarg(int) mode;
1271 } */ *uap;
1272{
1273 struct vattr vattr;
1274 int error;
1275 struct nameidata nd;
1276
1277 bwillwrite();
1278 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1279 if ((error = namei(&nd)) != 0)
1280 return (error);
1281 if (nd.ni_vp != NULL) {
1282 NDFREE(&nd, NDF_ONLY_PNBUF);
1283 if (nd.ni_dvp == nd.ni_vp)
1284 vrele(nd.ni_dvp);
1285 else
1286 vput(nd.ni_dvp);
1287 vrele(nd.ni_vp);
1288 return (EEXIST);
1289 }
1290 VATTR_NULL(&vattr);
1291 vattr.va_type = VFIFO;
1292 vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
1293 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1294 error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
1295 if (error == 0)
1296 vput(nd.ni_vp);
1297 NDFREE(&nd, NDF_ONLY_PNBUF);
1298 vput(nd.ni_dvp);
1299 return (error);
1300}
1301
1302/*
1303 * Make a hard file link.
1304 */
1305#ifndef _SYS_SYSPROTO_H_
1306struct link_args {
1307 char *path;
1308 char *link;
1309};
1310#endif
1311/* ARGSUSED */
1312int
1313link(p, uap)
1314 struct proc *p;
1315 register struct link_args /* {
1316 syscallarg(char *) path;
1317 syscallarg(char *) link;
1318 } */ *uap;
1319{
1320 register struct vnode *vp;
1321 struct nameidata nd;
1322 int error;
1323
1324 bwillwrite();
1325 NDINIT(&nd, LOOKUP, FOLLOW|NOOBJ, UIO_USERSPACE, SCARG(uap, path), p);
1326 if ((error = namei(&nd)) != 0)
1327 return (error);
1328 NDFREE(&nd, NDF_ONLY_PNBUF);
1329 vp = nd.ni_vp;
1330 if (vp->v_type == VDIR)
1331 error = EPERM; /* POSIX */
1332 else {
1333 NDINIT(&nd, CREATE, LOCKPARENT|NOOBJ, UIO_USERSPACE, SCARG(uap, link), p);
1334 error = namei(&nd);
1335 if (!error) {
1336 if (nd.ni_vp != NULL) {
1337 if (nd.ni_vp)
1338 vrele(nd.ni_vp);
1339 error = EEXIST;
1340 } else {
1341 VOP_LEASE(nd.ni_dvp, p, p->p_ucred,
1342 LEASE_WRITE);
1343 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1344 error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd);
1345 }
1346 NDFREE(&nd, NDF_ONLY_PNBUF);
1347 if (nd.ni_dvp == nd.ni_vp)
1348 vrele(nd.ni_dvp);
1349 else
1350 vput(nd.ni_dvp);
1351 }
1352 }
1353 vrele(vp);
1354 ASSERT_VOP_UNLOCKED(nd.ni_dvp, "link");
1355 ASSERT_VOP_UNLOCKED(nd.ni_vp, "link");
1356 return (error);
1357}
1358
1359/*
1360 * Make a symbolic link.
1361 */
1362#ifndef _SYS_SYSPROTO_H_
1363struct symlink_args {
1364 char *path;
1365 char *link;
1366};
1367#endif
1368/* ARGSUSED */
1369int
1370symlink(p, uap)
1371 struct proc *p;
1372 register struct symlink_args /* {
1373 syscallarg(char *) path;
1374 syscallarg(char *) link;
1375 } */ *uap;
1376{
1377 struct vattr vattr;
1378 char *path;
1379 int error;
1380 struct nameidata nd;
1381
1382 path = zalloc(namei_zone);
1383 if ((error = copyinstr(SCARG(uap, path), path, MAXPATHLEN, NULL)) != 0)
1384 goto out;
1385 bwillwrite();
1386 NDINIT(&nd, CREATE, LOCKPARENT|NOOBJ, UIO_USERSPACE, SCARG(uap, link), p);
1387 if ((error = namei(&nd)) != 0)
1388 goto out;
1389 if (nd.ni_vp) {
1390 NDFREE(&nd, NDF_ONLY_PNBUF);
1391 if (nd.ni_dvp == nd.ni_vp)
1392 vrele(nd.ni_dvp);
1393 else
1394 vput(nd.ni_dvp);
1395 vrele(nd.ni_vp);
1396 error = EEXIST;
1397 goto out;
1398 }
1399 VATTR_NULL(&vattr);
1400 vattr.va_mode = ACCESSPERMS &~ p->p_fd->fd_cmask;
1401 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1402 error = VOP_SYMLINK(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr, path);
1403 NDFREE(&nd, NDF_ONLY_PNBUF);
1404 if (error == 0)
1405 vput(nd.ni_vp);
1406 vput(nd.ni_dvp);
1407 ASSERT_VOP_UNLOCKED(nd.ni_dvp, "symlink");
1408 ASSERT_VOP_UNLOCKED(nd.ni_vp, "symlink");
1409out:
1410 zfree(namei_zone, path);
1411 return (error);
1412}
1413
1414/*
1415 * Delete a whiteout from the filesystem.
1416 */
1417/* ARGSUSED */
1418int
1419undelete(p, uap)
1420 struct proc *p;
1421 register struct undelete_args /* {
1422 syscallarg(char *) path;
1423 } */ *uap;
1424{
1425 int error;
1426 struct nameidata nd;
1427
1428 bwillwrite();
1429 NDINIT(&nd, DELETE, LOCKPARENT|DOWHITEOUT, UIO_USERSPACE,
1430 SCARG(uap, path), p);
1431 error = namei(&nd);
1432 if (error)
1433 return (error);
1434
1435 if (nd.ni_vp != NULLVP || !(nd.ni_cnd.cn_flags & ISWHITEOUT)) {
1436 NDFREE(&nd, NDF_ONLY_PNBUF);
1437 if (nd.ni_dvp == nd.ni_vp)
1438 vrele(nd.ni_dvp);
1439 else
1440 vput(nd.ni_dvp);
1441 if (nd.ni_vp)
1442 vrele(nd.ni_vp);
1443 return (EEXIST);
1444 }
1445
1446 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1447 error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, DELETE);
1448 NDFREE(&nd, NDF_ONLY_PNBUF);
1449 vput(nd.ni_dvp);
1450 ASSERT_VOP_UNLOCKED(nd.ni_dvp, "undelete");
1451 ASSERT_VOP_UNLOCKED(nd.ni_vp, "undelete");
1452 return (error);
1453}
1454
1455/*
1456 * Delete a name from the filesystem.
1457 */
1458#ifndef _SYS_SYSPROTO_H_
1459struct unlink_args {
1460 char *path;
1461};
1462#endif
1463/* ARGSUSED */
1464int
1465unlink(p, uap)
1466 struct proc *p;
1467 struct unlink_args /* {
1468 syscallarg(char *) path;
1469 } */ *uap;
1470{
1471 register struct vnode *vp;
1472 int error;
1473 struct nameidata nd;
1474
1475 bwillwrite();
1476 NDINIT(&nd, DELETE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1477 if ((error = namei(&nd)) != 0)
1478 return (error);
1479 vp = nd.ni_vp;
1480 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1481 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
1482
1483 if (vp->v_type == VDIR)
1484 error = EPERM; /* POSIX */
1485 else {
1486 /*
1487 * The root of a mounted filesystem cannot be deleted.
1488 *
1489 * XXX: can this only be a VDIR case?
1490 */
1491 if (vp->v_flag & VROOT)
1492 error = EBUSY;
1493 }
1494
1495 if (!error) {
1496 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1497 error = VOP_REMOVE(nd.ni_dvp, vp, &nd.ni_cnd);
1498 }
1499 NDFREE(&nd, NDF_ONLY_PNBUF);
1500 if (nd.ni_dvp == vp)
1501 vrele(nd.ni_dvp);
1502 else
1503 vput(nd.ni_dvp);
1504 if (vp != NULLVP)
1505 vput(vp);
1506 ASSERT_VOP_UNLOCKED(nd.ni_dvp, "unlink");
1507 ASSERT_VOP_UNLOCKED(nd.ni_vp, "unlink");
1508 return (error);
1509}
1510
1511/*
1512 * Reposition read/write file offset.
1513 */
1514#ifndef _SYS_SYSPROTO_H_
1515struct lseek_args {
1516 int fd;
1517 int pad;
1518 off_t offset;
1519 int whence;
1520};
1521#endif
1522int
1523lseek(p, uap)
1524 struct proc *p;
1525 register struct lseek_args /* {
1526 syscallarg(int) fd;
1527 syscallarg(int) pad;
1528 syscallarg(off_t) offset;
1529 syscallarg(int) whence;
1530 } */ *uap;
1531{
1532 struct ucred *cred = p->p_ucred;
1533 register struct filedesc *fdp = p->p_fd;
1534 register struct file *fp;
1535 struct vattr vattr;
1536 int error;
1537
1538 if ((u_int)SCARG(uap, fd) >= fdp->fd_nfiles ||
1539 (fp = fdp->fd_ofiles[SCARG(uap, fd)]) == NULL)
1540 return (EBADF);
1541 if (fp->f_type != DTYPE_VNODE)
1542 return (ESPIPE);
1543 switch (SCARG(uap, whence)) {
1544 case L_INCR:
1545 fp->f_offset += SCARG(uap, offset);
1546 break;
1547 case L_XTND:
1548 error=VOP_GETATTR((struct vnode *)fp->f_data, &vattr, cred, p);
1549 if (error)
1550 return (error);
1551 fp->f_offset = SCARG(uap, offset) + vattr.va_size;
1552 break;
1553 case L_SET:
1554 fp->f_offset = SCARG(uap, offset);
1555 break;
1556 default:
1557 return (EINVAL);
1558 }
1559 *(off_t *)(p->p_retval) = fp->f_offset;
1560 return (0);
1561}
1562
1563#if defined(COMPAT_43) || defined(COMPAT_SUNOS)
1564/*
1565 * Reposition read/write file offset.
1566 */
1567#ifndef _SYS_SYSPROTO_H_
1568struct olseek_args {
1569 int fd;
1570 long offset;
1571 int whence;
1572};
1573#endif
1574int
1575olseek(p, uap)
1576 struct proc *p;
1577 register struct olseek_args /* {
1578 syscallarg(int) fd;
1579 syscallarg(long) offset;
1580 syscallarg(int) whence;
1581 } */ *uap;
1582{
1583 struct lseek_args /* {
1584 syscallarg(int) fd;
1585 syscallarg(int) pad;
1586 syscallarg(off_t) offset;
1587 syscallarg(int) whence;
1588 } */ nuap;
1589 int error;
1590
1591 SCARG(&nuap, fd) = SCARG(uap, fd);
1592 SCARG(&nuap, offset) = SCARG(uap, offset);
1593 SCARG(&nuap, whence) = SCARG(uap, whence);
1594 error = lseek(p, &nuap);
1595 return (error);
1596}
1597#endif /* COMPAT_43 */
1598
1599/*
1600 * Check access permissions.
1601 */
1602#ifndef _SYS_SYSPROTO_H_
1603struct access_args {
1604 char *path;
1605 int flags;
1606};
1607#endif
1608int
1609access(p, uap)
1610 struct proc *p;
1611 register struct access_args /* {
1612 syscallarg(char *) path;
1613 syscallarg(int) flags;
1614 } */ *uap;
1615{
1616 struct ucred *cred, *tmpcred;
1617 register struct vnode *vp;
1618 int error, flags;
1619 struct nameidata nd;
1620
1621 cred = p->p_ucred;
1622 /*
1623 * Create and modify a temporary credential instead of one that
1624 * is potentially shared. This could also mess up socket
1625 * buffer accounting which can run in an interrupt context.
1626 */
1627 tmpcred = crdup(cred);
1628 tmpcred->cr_uid = p->p_cred->p_ruid;
1629 tmpcred->cr_groups[0] = p->p_cred->p_rgid;
1630 p->p_ucred = tmpcred;
1631 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1632 SCARG(uap, path), p);
1633 if ((error = namei(&nd)) != 0)
1634 goto out1;
1635 vp = nd.ni_vp;
1636
1637 /* Flags == 0 means only check for existence. */
1638 if (SCARG(uap, flags)) {
1639 flags = 0;
1640 if (SCARG(uap, flags) & R_OK)
1641 flags |= VREAD;
1642 if (SCARG(uap, flags) & W_OK)
1643 flags |= VWRITE;
1644 if (SCARG(uap, flags) & X_OK)
1645 flags |= VEXEC;
1646 if ((flags & VWRITE) == 0 || (error = vn_writechk(vp)) == 0)
1647 error = VOP_ACCESS(vp, flags, tmpcred, p);
1648 }
1649 NDFREE(&nd, NDF_ONLY_PNBUF);
1650 vput(vp);
1651out1:
1652 p->p_ucred = cred;
1653 crfree(tmpcred);
1654 return (error);
1655}
1656
1657#if defined(COMPAT_43) || defined(COMPAT_SUNOS)
1658/*
1659 * Get file status; this version follows links.
1660 */
1661#ifndef _SYS_SYSPROTO_H_
1662struct ostat_args {
1663 char *path;
1664 struct ostat *ub;
1665};
1666#endif
1667/* ARGSUSED */
1668int
1669ostat(p, uap)
1670 struct proc *p;
1671 register struct ostat_args /* {
1672 syscallarg(char *) path;
1673 syscallarg(struct ostat *) ub;
1674 } */ *uap;
1675{
1676 struct stat sb;
1677 struct ostat osb;
1678 int error;
1679 struct nameidata nd;
1680
1681 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1682 SCARG(uap, path), p);
1683 if ((error = namei(&nd)) != 0)
1684 return (error);
1685 NDFREE(&nd, NDF_ONLY_PNBUF);
1686 error = vn_stat(nd.ni_vp, &sb, p);
1687 vput(nd.ni_vp);
1688 if (error)
1689 return (error);
1690 cvtstat(&sb, &osb);
1691 error = copyout((caddr_t)&osb, (caddr_t)SCARG(uap, ub), sizeof (osb));
1692 return (error);
1693}
1694
1695/*
1696 * Get file status; this version does not follow links.
1697 */
1698#ifndef _SYS_SYSPROTO_H_
1699struct olstat_args {
1700 char *path;
1701 struct ostat *ub;
1702};
1703#endif
1704/* ARGSUSED */
1705int
1706olstat(p, uap)
1707 struct proc *p;
1708 register struct olstat_args /* {
1709 syscallarg(char *) path;
1710 syscallarg(struct ostat *) ub;
1711 } */ *uap;
1712{
1713 struct vnode *vp;
1714 struct stat sb;
1715 struct ostat osb;
1716 int error;
1717 struct nameidata nd;
1718
1719 NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1720 SCARG(uap, path), p);
1721 if ((error = namei(&nd)) != 0)
1722 return (error);
1723 vp = nd.ni_vp;
1724 error = vn_stat(vp, &sb, p);
1725 NDFREE(&nd, NDF_ONLY_PNBUF);
1726 vput(vp);
1727 if (error)
1728 return (error);
1729 cvtstat(&sb, &osb);
1730 error = copyout((caddr_t)&osb, (caddr_t)SCARG(uap, ub), sizeof (osb));
1731 return (error);
1732}
1733
1734/*
1735 * Convert from an old to a new stat structure.
1736 */
1737void
1738cvtstat(st, ost)
1739 struct stat *st;
1740 struct ostat *ost;
1741{
1742
1743 ost->st_dev = st->st_dev;
1744 ost->st_ino = st->st_ino;
1745 ost->st_mode = st->st_mode;
1746 ost->st_nlink = st->st_nlink;
1747 ost->st_uid = st->st_uid;
1748 ost->st_gid = st->st_gid;
1749 ost->st_rdev = st->st_rdev;
1750 if (st->st_size < (quad_t)1 << 32)
1751 ost->st_size = st->st_size;
1752 else
1753 ost->st_size = -2;
1754 ost->st_atime = st->st_atime;
1755 ost->st_mtime = st->st_mtime;
1756 ost->st_ctime = st->st_ctime;
1757 ost->st_blksize = st->st_blksize;
1758 ost->st_blocks = st->st_blocks;
1759 ost->st_flags = st->st_flags;
1760 ost->st_gen = st->st_gen;
1761}
1762#endif /* COMPAT_43 || COMPAT_SUNOS */
1763
1764/*
1765 * Get file status; this version follows links.
1766 */
1767#ifndef _SYS_SYSPROTO_H_
1768struct stat_args {
1769 char *path;
1770 struct stat *ub;
1771};
1772#endif
1773/* ARGSUSED */
1774int
1775stat(p, uap)
1776 struct proc *p;
1777 register struct stat_args /* {
1778 syscallarg(char *) path;
1779 syscallarg(struct stat *) ub;
1780 } */ *uap;
1781{
1782 struct stat sb;
1783 int error;
1784 struct nameidata nd;
1785
1786 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1787 SCARG(uap, path), p);
1788 if ((error = namei(&nd)) != 0)
1789 return (error);
1790 error = vn_stat(nd.ni_vp, &sb, p);
1791 NDFREE(&nd, NDF_ONLY_PNBUF);
1792 vput(nd.ni_vp);
1793 if (error)
1794 return (error);
1795 error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
1796 return (error);
1797}
1798
1799/*
1800 * Get file status; this version does not follow links.
1801 */
1802#ifndef _SYS_SYSPROTO_H_
1803struct lstat_args {
1804 char *path;
1805 struct stat *ub;
1806};
1807#endif
1808/* ARGSUSED */
1809int
1810lstat(p, uap)
1811 struct proc *p;
1812 register struct lstat_args /* {
1813 syscallarg(char *) path;
1814 syscallarg(struct stat *) ub;
1815 } */ *uap;
1816{
1817 int error;
1818 struct vnode *vp;
1819 struct stat sb;
1820 struct nameidata nd;
1821
1822 NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1823 SCARG(uap, path), p);
1824 if ((error = namei(&nd)) != 0)
1825 return (error);
1826 vp = nd.ni_vp;
1827 error = vn_stat(vp, &sb, p);
1828 NDFREE(&nd, NDF_ONLY_PNBUF);
1829 vput(vp);
1830 if (error)
1831 return (error);
1832 error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
1833 return (error);
1834}
1835
1836void
1837cvtnstat(sb, nsb)
1838 struct stat *sb;
1839 struct nstat *nsb;
1840{
1841 nsb->st_dev = sb->st_dev;
1842 nsb->st_ino = sb->st_ino;
1843 nsb->st_mode = sb->st_mode;
1844 nsb->st_nlink = sb->st_nlink;
1845 nsb->st_uid = sb->st_uid;
1846 nsb->st_gid = sb->st_gid;
1847 nsb->st_rdev = sb->st_rdev;
1848 nsb->st_atimespec = sb->st_atimespec;
1849 nsb->st_mtimespec = sb->st_mtimespec;
1850 nsb->st_ctimespec = sb->st_ctimespec;
1851 nsb->st_size = sb->st_size;
1852 nsb->st_blocks = sb->st_blocks;
1853 nsb->st_blksize = sb->st_blksize;
1854 nsb->st_flags = sb->st_flags;
1855 nsb->st_gen = sb->st_gen;
1856 nsb->st_qspare[0] = sb->st_qspare[0];
1857 nsb->st_qspare[1] = sb->st_qspare[1];
1858}
1859
1860#ifndef _SYS_SYSPROTO_H_
1861struct nstat_args {
1862 char *path;
1863 struct nstat *ub;
1864};
1865#endif
1866/* ARGSUSED */
1867int
1868nstat(p, uap)
1869 struct proc *p;
1870 register struct nstat_args /* {
1871 syscallarg(char *) path;
1872 syscallarg(struct nstat *) ub;
1873 } */ *uap;
1874{
1875 struct stat sb;
1876 struct nstat nsb;
1877 int error;
1878 struct nameidata nd;
1879
1880 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1881 SCARG(uap, path), p);
1882 if ((error = namei(&nd)) != 0)
1883 return (error);
1884 NDFREE(&nd, NDF_ONLY_PNBUF);
1885 error = vn_stat(nd.ni_vp, &sb, p);
1886 vput(nd.ni_vp);
1887 if (error)
1888 return (error);
1889 cvtnstat(&sb, &nsb);
1890 error = copyout((caddr_t)&nsb, (caddr_t)SCARG(uap, ub), sizeof (nsb));
1891 return (error);
1892}
1893
1894/*
1895 * Get file status; this version does not follow links.
1896 */
1897#ifndef _SYS_SYSPROTO_H_
1898struct lstat_args {
1899 char *path;
1900 struct stat *ub;
1901};
1902#endif
1903/* ARGSUSED */
1904int
1905nlstat(p, uap)
1906 struct proc *p;
1907 register struct nlstat_args /* {
1908 syscallarg(char *) path;
1909 syscallarg(struct nstat *) ub;
1910 } */ *uap;
1911{
1912 int error;
1913 struct vnode *vp;
1914 struct stat sb;
1915 struct nstat nsb;
1916 struct nameidata nd;
1917
1918 NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1919 SCARG(uap, path), p);
1920 if ((error = namei(&nd)) != 0)
1921 return (error);
1922 vp = nd.ni_vp;
1923 NDFREE(&nd, NDF_ONLY_PNBUF);
1924 error = vn_stat(vp, &sb, p);
1925 vput(vp);
1926 if (error)
1927 return (error);
1928 cvtnstat(&sb, &nsb);
1929 error = copyout((caddr_t)&nsb, (caddr_t)SCARG(uap, ub), sizeof (nsb));
1930 return (error);
1931}
1932
1933/*
1934 * Get configurable pathname variables.
1935 */
1936#ifndef _SYS_SYSPROTO_H_
1937struct pathconf_args {
1938 char *path;
1939 int name;
1940};
1941#endif
1942/* ARGSUSED */
1943int
1944pathconf(p, uap)
1945 struct proc *p;
1946 register struct pathconf_args /* {
1947 syscallarg(char *) path;
1948 syscallarg(int) name;
1949 } */ *uap;
1950{
1951 int error;
1952 struct nameidata nd;
1953
1954 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1955 SCARG(uap, path), p);
1956 if ((error = namei(&nd)) != 0)
1957 return (error);
1958 NDFREE(&nd, NDF_ONLY_PNBUF);
1959 error = VOP_PATHCONF(nd.ni_vp, SCARG(uap, name), p->p_retval);
1960 vput(nd.ni_vp);
1961 return (error);
1962}
1963
1964/*
1965 * Return target name of a symbolic link.
1966 */
1967#ifndef _SYS_SYSPROTO_H_
1968struct readlink_args {
1969 char *path;
1970 char *buf;
1971 int count;
1972};
1973#endif
1974/* ARGSUSED */
1975int
1976readlink(p, uap)
1977 struct proc *p;
1978 register struct readlink_args /* {
1979 syscallarg(char *) path;
1980 syscallarg(char *) buf;
1981 syscallarg(int) count;
1982 } */ *uap;
1983{
1984 register struct vnode *vp;
1985 struct iovec aiov;
1986 struct uio auio;
1987 int error;
1988 struct nameidata nd;
1989
1990 NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
1991 SCARG(uap, path), p);
1992 if ((error = namei(&nd)) != 0)
1993 return (error);
1994 NDFREE(&nd, NDF_ONLY_PNBUF);
1995 vp = nd.ni_vp;
1996 if (vp->v_type != VLNK)
1997 error = EINVAL;
1998 else {
1999 aiov.iov_base = SCARG(uap, buf);
2000 aiov.iov_len = SCARG(uap, count);
2001 auio.uio_iov = &aiov;
2002 auio.uio_iovcnt = 1;
2003 auio.uio_offset = 0;
2004 auio.uio_rw = UIO_READ;
2005 auio.uio_segflg = UIO_USERSPACE;
2006 auio.uio_procp = p;
2007 auio.uio_resid = SCARG(uap, count);
2008 error = VOP_READLINK(vp, &auio, p->p_ucred);
2009 }
2010 vput(vp);
2011 p->p_retval[0] = SCARG(uap, count) - auio.uio_resid;
2012 return (error);
2013}
2014
2015static int
2016setfflags(p, vp, flags)
2017 struct proc *p;
2018 struct vnode *vp;
2019 int flags;
2020{
2021 int error;
2022 struct vattr vattr;
2023
2024 /*
2025 * Prevent non-root users from setting flags on devices. When
2026 * a device is reused, users can retain ownership of the device
2027 * if they are allowed to set flags and programs assume that
2028 * chown can't fail when done as root.
2029 */
2030 if ((vp->v_type == VCHR || vp->v_type == VBLK) &&
2031 ((error = suser_xxx(p->p_ucred, p, PRISON_ROOT)) != 0))
2032 return (error);
2033
2034 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2035 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2036 VATTR_NULL(&vattr);
2037 vattr.va_flags = flags;
2038 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2039 VOP_UNLOCK(vp, 0, p);
2040 return (error);
2041}
2042
2043/*
2044 * Change flags of a file given a path name.
2045 */
2046#ifndef _SYS_SYSPROTO_H_
2047struct chflags_args {
2048 char *path;
2049 int flags;
2050};
2051#endif
2052/* ARGSUSED */
2053int
2054chflags(p, uap)
2055 struct proc *p;
2056 register struct chflags_args /* {
2057 syscallarg(char *) path;
2058 syscallarg(int) flags;
2059 } */ *uap;
2060{
2061 int error;
2062 struct nameidata nd;
2063
2064 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2065 if ((error = namei(&nd)) != 0)
2066 return (error);
2067 NDFREE(&nd, NDF_ONLY_PNBUF);
2068 error = setfflags(p, nd.ni_vp, SCARG(uap, flags));
2069 vrele(nd.ni_vp);
2070 return error;
2071}
2072
2073/*
2074 * Change flags of a file given a file descriptor.
2075 */
2076#ifndef _SYS_SYSPROTO_H_
2077struct fchflags_args {
2078 int fd;
2079 int flags;
2080};
2081#endif
2082/* ARGSUSED */
2083int
2084fchflags(p, uap)
2085 struct proc *p;
2086 register struct fchflags_args /* {
2087 syscallarg(int) fd;
2088 syscallarg(int) flags;
2089 } */ *uap;
2090{
2091 struct file *fp;
2092 int error;
2093
2094 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2095 return (error);
2096 return setfflags(p, (struct vnode *) fp->f_data, SCARG(uap, flags));
2097}
2098
2099static int
2100setfmode(p, vp, mode)
2101 struct proc *p;
2102 struct vnode *vp;
2103 int mode;
2104{
2105 int error;
2106 struct vattr vattr;
2107
2108 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2109 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2110 VATTR_NULL(&vattr);
2111 vattr.va_mode = mode & ALLPERMS;
2112 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2113 VOP_UNLOCK(vp, 0, p);
2114 return error;
2115}
2116
2117/*
2118 * Change mode of a file given path name.
2119 */
2120#ifndef _SYS_SYSPROTO_H_
2121struct chmod_args {
2122 char *path;
2123 int mode;
2124};
2125#endif
2126/* ARGSUSED */
2127int
2128chmod(p, uap)
2129 struct proc *p;
2130 register struct chmod_args /* {
2131 syscallarg(char *) path;
2132 syscallarg(int) mode;
2133 } */ *uap;
2134{
2135 int error;
2136 struct nameidata nd;
2137
2138 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2139 if ((error = namei(&nd)) != 0)
2140 return (error);
2141 NDFREE(&nd, NDF_ONLY_PNBUF);
2142 error = setfmode(p, nd.ni_vp, SCARG(uap, mode));
2143 vrele(nd.ni_vp);
2144 return error;
2145}
2146
2147/*
2148 * Change mode of a file given path name (don't follow links.)
2149 */
2150#ifndef _SYS_SYSPROTO_H_
2151struct lchmod_args {
2152 char *path;
2153 int mode;
2154};
2155#endif
2156/* ARGSUSED */
2157int
2158lchmod(p, uap)
2159 struct proc *p;
2160 register struct lchmod_args /* {
2161 syscallarg(char *) path;
2162 syscallarg(int) mode;
2163 } */ *uap;
2164{
2165 int error;
2166 struct nameidata nd;
2167
2168 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2169 if ((error = namei(&nd)) != 0)
2170 return (error);
2171 NDFREE(&nd, NDF_ONLY_PNBUF);
2172 error = setfmode(p, nd.ni_vp, SCARG(uap, mode));
2173 vrele(nd.ni_vp);
2174 return error;
2175}
2176
2177/*
2178 * Change mode of a file given a file descriptor.
2179 */
2180#ifndef _SYS_SYSPROTO_H_
2181struct fchmod_args {
2182 int fd;
2183 int mode;
2184};
2185#endif
2186/* ARGSUSED */
2187int
2188fchmod(p, uap)
2189 struct proc *p;
2190 register struct fchmod_args /* {
2191 syscallarg(int) fd;
2192 syscallarg(int) mode;
2193 } */ *uap;
2194{
2195 struct file *fp;
2196 int error;
2197
2198 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2199 return (error);
2200 return setfmode(p, (struct vnode *)fp->f_data, SCARG(uap, mode));
2201}
2202
2203static int
2204setfown(p, vp, uid, gid)
2205 struct proc *p;
2206 struct vnode *vp;
2207 uid_t uid;
2208 gid_t gid;
2209{
2210 int error;
2211 struct vattr vattr;
2212
2213 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2214 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2215 VATTR_NULL(&vattr);
2216 vattr.va_uid = uid;
2217 vattr.va_gid = gid;
2218 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2219 VOP_UNLOCK(vp, 0, p);
2220 return error;
2221}
2222
2223/*
2224 * Set ownership given a path name.
2225 */
2226#ifndef _SYS_SYSPROTO_H_
2227struct chown_args {
2228 char *path;
2229 int uid;
2230 int gid;
2231};
2232#endif
2233/* ARGSUSED */
2234int
2235chown(p, uap)
2236 struct proc *p;
2237 register struct chown_args /* {
2238 syscallarg(char *) path;
2239 syscallarg(int) uid;
2240 syscallarg(int) gid;
2241 } */ *uap;
2242{
2243 int error;
2244 struct nameidata nd;
2245
2246 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2247 if ((error = namei(&nd)) != 0)
2248 return (error);
2249 NDFREE(&nd, NDF_ONLY_PNBUF);
2250 error = setfown(p, nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid));
2251 vrele(nd.ni_vp);
2252 return (error);
2253}
2254
2255/*
2256 * Set ownership given a path name, do not cross symlinks.
2257 */
2258#ifndef _SYS_SYSPROTO_H_
2259struct lchown_args {
2260 char *path;
2261 int uid;
2262 int gid;
2263};
2264#endif
2265/* ARGSUSED */
2266int
2267lchown(p, uap)
2268 struct proc *p;
2269 register struct lchown_args /* {
2270 syscallarg(char *) path;
2271 syscallarg(int) uid;
2272 syscallarg(int) gid;
2273 } */ *uap;
2274{
2275 int error;
2276 struct nameidata nd;
2277
2278 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2279 if ((error = namei(&nd)) != 0)
2280 return (error);
2281 NDFREE(&nd, NDF_ONLY_PNBUF);
2282 error = setfown(p, nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid));
2283 vrele(nd.ni_vp);
2284 return (error);
2285}
2286
2287/*
2288 * Set ownership given a file descriptor.
2289 */
2290#ifndef _SYS_SYSPROTO_H_
2291struct fchown_args {
2292 int fd;
2293 int uid;
2294 int gid;
2295};
2296#endif
2297/* ARGSUSED */
2298int
2299fchown(p, uap)
2300 struct proc *p;
2301 register struct fchown_args /* {
2302 syscallarg(int) fd;
2303 syscallarg(int) uid;
2304 syscallarg(int) gid;
2305 } */ *uap;
2306{
2307 struct file *fp;
2308 int error;
2309
2310 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2311 return (error);
2312 return setfown(p, (struct vnode *)fp->f_data,
2313 SCARG(uap, uid), SCARG(uap, gid));
2314}
2315
2316static int
2317getutimes(usrtvp, tsp)
2318 const struct timeval *usrtvp;
2319 struct timespec *tsp;
2320{
2321 struct timeval tv[2];
2322 int error;
2323
2324 if (usrtvp == NULL) {
2325 microtime(&tv[0]);
2326 TIMEVAL_TO_TIMESPEC(&tv[0], &tsp[0]);
2327 tsp[1] = tsp[0];
2328 } else {
2329 if ((error = copyin(usrtvp, tv, sizeof (tv))) != 0)
2330 return (error);
2331 TIMEVAL_TO_TIMESPEC(&tv[0], &tsp[0]);
2332 TIMEVAL_TO_TIMESPEC(&tv[1], &tsp[1]);
2333 }
2334 return 0;
2335}
2336
2337static int
2338setutimes(p, vp, ts, nullflag)
2339 struct proc *p;
2340 struct vnode *vp;
2341 const struct timespec *ts;
2342 int nullflag;
2343{
2344 int error;
2345 struct vattr vattr;
2346
2347 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2348 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2349 VATTR_NULL(&vattr);
2350 vattr.va_atime = ts[0];
2351 vattr.va_mtime = ts[1];
2352 if (nullflag)
2353 vattr.va_vaflags |= VA_UTIMES_NULL;
2354 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2355 VOP_UNLOCK(vp, 0, p);
2356 return error;
2357}
2358
2359/*
2360 * Set the access and modification times of a file.
2361 */
2362#ifndef _SYS_SYSPROTO_H_
2363struct utimes_args {
2364 char *path;
2365 struct timeval *tptr;
2366};
2367#endif
2368/* ARGSUSED */
2369int
2370utimes(p, uap)
2371 struct proc *p;
2372 register struct utimes_args /* {
2373 syscallarg(char *) path;
2374 syscallarg(struct timeval *) tptr;
2375 } */ *uap;
2376{
2377 struct timespec ts[2];
2378 struct timeval *usrtvp;
2379 int error;
2380 struct nameidata nd;
2381
2382 usrtvp = SCARG(uap, tptr);
2383 if ((error = getutimes(usrtvp, ts)) != 0)
2384 return (error);
2385 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2386 if ((error = namei(&nd)) != 0)
2387 return (error);
2388 NDFREE(&nd, NDF_ONLY_PNBUF);
2389 error = setutimes(p, nd.ni_vp, ts, usrtvp == NULL);
2390 vrele(nd.ni_vp);
2391 return (error);
2392}
2393
2394/*
2395 * Set the access and modification times of a file.
2396 */
2397#ifndef _SYS_SYSPROTO_H_
2398struct lutimes_args {
2399 char *path;
2400 struct timeval *tptr;
2401};
2402#endif
2403/* ARGSUSED */
2404int
2405lutimes(p, uap)
2406 struct proc *p;
2407 register struct lutimes_args /* {
2408 syscallarg(char *) path;
2409 syscallarg(struct timeval *) tptr;
2410 } */ *uap;
2411{
2412 struct timespec ts[2];
2413 struct timeval *usrtvp;
2414 int error;
2415 struct nameidata nd;
2416
2417 usrtvp = SCARG(uap, tptr);
2418 if ((error = getutimes(usrtvp, ts)) != 0)
2419 return (error);
2420 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2421 if ((error = namei(&nd)) != 0)
2422 return (error);
2423 NDFREE(&nd, NDF_ONLY_PNBUF);
2424 error = setutimes(p, nd.ni_vp, ts, usrtvp == NULL);
2425 vrele(nd.ni_vp);
2426 return (error);
2427}
2428
2429/*
2430 * Set the access and modification times of a file.
2431 */
2432#ifndef _SYS_SYSPROTO_H_
2433struct futimes_args {
2434 int fd;
2435 struct timeval *tptr;
2436};
2437#endif
2438/* ARGSUSED */
2439int
2440futimes(p, uap)
2441 struct proc *p;
2442 register struct futimes_args /* {
2443 syscallarg(int ) fd;
2444 syscallarg(struct timeval *) tptr;
2445 } */ *uap;
2446{
2447 struct timespec ts[2];
2448 struct file *fp;
2449 struct timeval *usrtvp;
2450 int error;
2451
2452 usrtvp = SCARG(uap, tptr);
2453 if ((error = getutimes(usrtvp, ts)) != 0)
2454 return (error);
2455 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2456 return (error);
2457 return setutimes(p, (struct vnode *)fp->f_data, ts, usrtvp == NULL);
2458}
2459
2460/*
2461 * Truncate a file given its path name.
2462 */
2463#ifndef _SYS_SYSPROTO_H_
2464struct truncate_args {
2465 char *path;
2466 int pad;
2467 off_t length;
2468};
2469#endif
2470/* ARGSUSED */
2471int
2472truncate(p, uap)
2473 struct proc *p;
2474 register struct truncate_args /* {
2475 syscallarg(char *) path;
2476 syscallarg(int) pad;
2477 syscallarg(off_t) length;
2478 } */ *uap;
2479{
2480 register struct vnode *vp;
2481 struct vattr vattr;
2482 int error;
2483 struct nameidata nd;
2484
2485 if (uap->length < 0)
2486 return(EINVAL);
2487 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2488 if ((error = namei(&nd)) != 0)
2489 return (error);
2490 vp = nd.ni_vp;
2491 NDFREE(&nd, NDF_ONLY_PNBUF);
2492 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2493 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2494 if (vp->v_type == VDIR)
2495 error = EISDIR;
2496 else if ((error = vn_writechk(vp)) == 0 &&
2497 (error = VOP_ACCESS(vp, VWRITE, p->p_ucred, p)) == 0) {
2498 VATTR_NULL(&vattr);
2499 vattr.va_size = SCARG(uap, length);
2500 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2501 }
2502 vput(vp);
2503 return (error);
2504}
2505
2506/*
2507 * Truncate a file given a file descriptor.
2508 */
2509#ifndef _SYS_SYSPROTO_H_
2510struct ftruncate_args {
2511 int fd;
2512 int pad;
2513 off_t length;
2514};
2515#endif
2516/* ARGSUSED */
2517int
2518ftruncate(p, uap)
2519 struct proc *p;
2520 register struct ftruncate_args /* {
2521 syscallarg(int) fd;
2522 syscallarg(int) pad;
2523 syscallarg(off_t) length;
2524 } */ *uap;
2525{
2526 struct vattr vattr;
2527 struct vnode *vp;
2528 struct file *fp;
2529 int error;
2530
2531 if (uap->length < 0)
2532 return(EINVAL);
2533 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2534 return (error);
2535 if ((fp->f_flag & FWRITE) == 0)
2536 return (EINVAL);
2537 vp = (struct vnode *)fp->f_data;
2538 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2539 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2540 if (vp->v_type == VDIR)
2541 error = EISDIR;
2542 else if ((error = vn_writechk(vp)) == 0) {
2543 VATTR_NULL(&vattr);
2544 vattr.va_size = SCARG(uap, length);
2545 error = VOP_SETATTR(vp, &vattr, fp->f_cred, p);
2546 }
2547 VOP_UNLOCK(vp, 0, p);
2548 return (error);
2549}
2550
2551#if defined(COMPAT_43) || defined(COMPAT_SUNOS)
2552/*
2553 * Truncate a file given its path name.
2554 */
2555#ifndef _SYS_SYSPROTO_H_
2556struct otruncate_args {
2557 char *path;
2558 long length;
2559};
2560#endif
2561/* ARGSUSED */
2562int
2563otruncate(p, uap)
2564 struct proc *p;
2565 register struct otruncate_args /* {
2566 syscallarg(char *) path;
2567 syscallarg(long) length;
2568 } */ *uap;
2569{
2570 struct truncate_args /* {
2571 syscallarg(char *) path;
2572 syscallarg(int) pad;
2573 syscallarg(off_t) length;
2574 } */ nuap;
2575
2576 SCARG(&nuap, path) = SCARG(uap, path);
2577 SCARG(&nuap, length) = SCARG(uap, length);
2578 return (truncate(p, &nuap));
2579}
2580
2581/*
2582 * Truncate a file given a file descriptor.
2583 */
2584#ifndef _SYS_SYSPROTO_H_
2585struct oftruncate_args {
2586 int fd;
2587 long length;
2588};
2589#endif
2590/* ARGSUSED */
2591int
2592oftruncate(p, uap)
2593 struct proc *p;
2594 register struct oftruncate_args /* {
2595 syscallarg(int) fd;
2596 syscallarg(long) length;
2597 } */ *uap;
2598{
2599 struct ftruncate_args /* {
2600 syscallarg(int) fd;
2601 syscallarg(int) pad;
2602 syscallarg(off_t) length;
2603 } */ nuap;
2604
2605 SCARG(&nuap, fd) = SCARG(uap, fd);
2606 SCARG(&nuap, length) = SCARG(uap, length);
2607 return (ftruncate(p, &nuap));
2608}
2609#endif /* COMPAT_43 || COMPAT_SUNOS */
2610
2611/*
2612 * Sync an open file.
2613 */
2614#ifndef _SYS_SYSPROTO_H_
2615struct fsync_args {
2616 int fd;
2617};
2618#endif
2619/* ARGSUSED */
2620int
2621fsync(p, uap)
2622 struct proc *p;
2623 struct fsync_args /* {
2624 syscallarg(int) fd;
2625 } */ *uap;
2626{
2627 register struct vnode *vp;
2628 struct file *fp;
2629 vm_object_t obj;
2630 int error;
2631
2632 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2633 return (error);
2634 vp = (struct vnode *)fp->f_data;
2635 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2636 if (VOP_GETVOBJECT(vp, &obj) == 0)
2637 vm_object_page_clean(obj, 0, 0, 0);
2638 if ((error = VOP_FSYNC(vp, fp->f_cred, MNT_WAIT, p)) == 0 &&
2639 vp->v_mount && (vp->v_mount->mnt_flag & MNT_SOFTDEP) &&
2640 bioops.io_fsync)
2641 error = (*bioops.io_fsync)(vp);
2642 VOP_UNLOCK(vp, 0, p);
2643 return (error);
2644}
2645
2646/*
2647 * Rename files. Source and destination must either both be directories,
2648 * or both not be directories. If target is a directory, it must be empty.
2649 */
2650#ifndef _SYS_SYSPROTO_H_
2651struct rename_args {
2652 char *from;
2653 char *to;
2654};
2655#endif
2656/* ARGSUSED */
2657int
2658rename(p, uap)
2659 struct proc *p;
2660 register struct rename_args /* {
2661 syscallarg(char *) from;
2662 syscallarg(char *) to;
2663 } */ *uap;
2664{
2665 register struct vnode *tvp, *fvp, *tdvp;
2666 struct nameidata fromnd, tond;
2667 int error;
2668
2669 bwillwrite();
2670 NDINIT(&fromnd, DELETE, WANTPARENT | SAVESTART, UIO_USERSPACE,
2671 SCARG(uap, from), p);
2672 if ((error = namei(&fromnd)) != 0)
2673 return (error);
2674 fvp = fromnd.ni_vp;
2675 NDINIT(&tond, RENAME, LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART | NOOBJ,
2676 UIO_USERSPACE, SCARG(uap, to), p);
2677 if (fromnd.ni_vp->v_type == VDIR)
2678 tond.ni_cnd.cn_flags |= WILLBEDIR;
2679 if ((error = namei(&tond)) != 0) {
2680 /* Translate error code for rename("dir1", "dir2/."). */
2681 if (error == EISDIR && fvp->v_type == VDIR)
2682 error = EINVAL;
2683 NDFREE(&fromnd, NDF_ONLY_PNBUF);
2684 vrele(fromnd.ni_dvp);
2685 vrele(fvp);
2686 goto out1;
2687 }
2688 tdvp = tond.ni_dvp;
2689 tvp = tond.ni_vp;
2690 if (tvp != NULL) {
2691 if (fvp->v_type == VDIR && tvp->v_type != VDIR) {
2692 error = ENOTDIR;
2693 goto out;
2694 } else if (fvp->v_type != VDIR && tvp->v_type == VDIR) {
2695 error = EISDIR;
2696 goto out;
2697 }
2698 }
2699 if (fvp == tdvp)
2700 error = EINVAL;
2701 /*
2702 * If the source is the same as the destination (that is, if they
2703 * are links to the same vnode), then there is nothing to do.
2704 */
2705 if (fvp == tvp)
2706 error = -1;
2707out:
2708 if (!error) {
2709 VOP_LEASE(tdvp, p, p->p_ucred, LEASE_WRITE);
2710 if (fromnd.ni_dvp != tdvp) {
2711 VOP_LEASE(fromnd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
2712 }
2713 if (tvp) {
2714 VOP_LEASE(tvp, p, p->p_ucred, LEASE_WRITE);
2715 }
2716 error = VOP_RENAME(fromnd.ni_dvp, fromnd.ni_vp, &fromnd.ni_cnd,
2717 tond.ni_dvp, tond.ni_vp, &tond.ni_cnd);
2718 NDFREE(&fromnd, NDF_ONLY_PNBUF);
2719 NDFREE(&tond, NDF_ONLY_PNBUF);
2720 } else {
2721 NDFREE(&fromnd, NDF_ONLY_PNBUF);
2722 NDFREE(&tond, NDF_ONLY_PNBUF);
2723 if (tdvp == tvp)
2724 vrele(tdvp);
2725 else
2726 vput(tdvp);
2727 if (tvp)
2728 vput(tvp);
2729 vrele(fromnd.ni_dvp);
2730 vrele(fvp);
2731 }
2732 vrele(tond.ni_startdir);
2733 ASSERT_VOP_UNLOCKED(fromnd.ni_dvp, "rename");
2734 ASSERT_VOP_UNLOCKED(fromnd.ni_vp, "rename");
2735 ASSERT_VOP_UNLOCKED(tond.ni_dvp, "rename");
2736 ASSERT_VOP_UNLOCKED(tond.ni_vp, "rename");
2737out1:
2738 if (fromnd.ni_startdir)
2739 vrele(fromnd.ni_startdir);
2740 if (error == -1)
2741 return (0);
2742 return (error);
2743}
2744
2745/*
2746 * Make a directory file.
2747 */
2748#ifndef _SYS_SYSPROTO_H_
2749struct mkdir_args {
2750 char *path;
2751 int mode;
2752};
2753#endif
2754/* ARGSUSED */
2755int
2756mkdir(p, uap)
2757 struct proc *p;
2758 register struct mkdir_args /* {
2759 syscallarg(char *) path;
2760 syscallarg(int) mode;
2761 } */ *uap;
2762{
2763 register struct vnode *vp;
2764 struct vattr vattr;
2765 int error;
2766 struct nameidata nd;
2767
2768 bwillwrite();
2769 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
2770 nd.ni_cnd.cn_flags |= WILLBEDIR;
2771 if ((error = namei(&nd)) != 0)
2772 return (error);
2773 vp = nd.ni_vp;
2774 if (vp != NULL) {
2775 NDFREE(&nd, NDF_ONLY_PNBUF);
2776 if (nd.ni_dvp == vp)
2777 vrele(nd.ni_dvp);
2778 else
2779 vput(nd.ni_dvp);
2780 vrele(vp);
2781 return (EEXIST);
2782 }
2783 VATTR_NULL(&vattr);
2784 vattr.va_type = VDIR;
2785 vattr.va_mode = (SCARG(uap, mode) & ACCESSPERMS) &~ p->p_fd->fd_cmask;
2786 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
2787 error = VOP_MKDIR(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
2788 NDFREE(&nd, NDF_ONLY_PNBUF);
2789 vput(nd.ni_dvp);
2790 if (!error)
2791 vput(nd.ni_vp);
2792 ASSERT_VOP_UNLOCKED(nd.ni_dvp, "mkdir");
2793 ASSERT_VOP_UNLOCKED(nd.ni_vp, "mkdir");
2794 return (error);
2795}
2796
2797/*
2798 * Remove a directory file.
2799 */
2800#ifndef _SYS_SYSPROTO_H_
2801struct rmdir_args {
2802 char *path;
2803};
2804#endif
2805/* ARGSUSED */
2806int
2807rmdir(p, uap)
2808 struct proc *p;
2809 struct rmdir_args /* {
2810 syscallarg(char *) path;
2811 } */ *uap;
2812{
2813 register struct vnode *vp;
2814 int error;
2815 struct nameidata nd;
2816
2817 bwillwrite();
2818 NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
2819 SCARG(uap, path), p);
2820 if ((error = namei(&nd)) != 0)
2821 return (error);
2822 vp = nd.ni_vp;
2823 if (vp->v_type != VDIR) {
2824 error = ENOTDIR;
2825 goto out;
2826 }
2827 /*
2828 * No rmdir "." please.
2829 */
2830 if (nd.ni_dvp == vp) {
2831 error = EINVAL;
2832 goto out;
2833 }
2834 /*
2835 * The root of a mounted filesystem cannot be deleted.
2836 */
2837 if (vp->v_flag & VROOT)
2838 error = EBUSY;
2839 else {
2840 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
2841 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2842 error = VOP_RMDIR(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
2843 }
2844out:
2845 NDFREE(&nd, NDF_ONLY_PNBUF);
2846 if (nd.ni_dvp == vp)
2847 vrele(nd.ni_dvp);
2848 else
2849 vput(nd.ni_dvp);
2850 if (vp != NULLVP)
2851 vput(vp);
2852 ASSERT_VOP_UNLOCKED(nd.ni_dvp, "rmdir");
2853 ASSERT_VOP_UNLOCKED(nd.ni_vp, "rmdir");
2854 return (error);
2855}
2856
2857#ifdef COMPAT_43
2858/*
2859 * Read a block of directory entries in a file system independent format.
2860 */
2861#ifndef _SYS_SYSPROTO_H_
2862struct ogetdirentries_args {
2863 int fd;
2864 char *buf;
2865 u_int count;
2866 long *basep;
2867};
2868#endif
2869int
2870ogetdirentries(p, uap)
2871 struct proc *p;
2872 register struct ogetdirentries_args /* {
2873 syscallarg(int) fd;
2874 syscallarg(char *) buf;
2875 syscallarg(u_int) count;
2876 syscallarg(long *) basep;
2877 } */ *uap;
2878{
2879 struct vnode *vp;
2880 struct file *fp;
2881 struct uio auio, kuio;
2882 struct iovec aiov, kiov;
2883 struct dirent *dp, *edp;
2884 caddr_t dirbuf;
2885 int error, eofflag, readcnt;
2886 long loff;
2887
2888 /* XXX arbitrary sanity limit on `count'. */
2889 if (SCARG(uap, count) > 64 * 1024)
2890 return (EINVAL);
2891 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2892 return (error);
2893 if ((fp->f_flag & FREAD) == 0)
2894 return (EBADF);
2895 vp = (struct vnode *)fp->f_data;
2896unionread:
2897 if (vp->v_type != VDIR)
2898 return (EINVAL);
2899 aiov.iov_base = SCARG(uap, buf);
2900 aiov.iov_len = SCARG(uap, count);
2901 auio.uio_iov = &aiov;
2902 auio.uio_iovcnt = 1;
2903 auio.uio_rw = UIO_READ;
2904 auio.uio_segflg = UIO_USERSPACE;
2905 auio.uio_procp = p;
2906 auio.uio_resid = SCARG(uap, count);
2907 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
2908 loff = auio.uio_offset = fp->f_offset;
2909# if (BYTE_ORDER != LITTLE_ENDIAN)
2910 if (vp->v_mount->mnt_maxsymlinklen <= 0) {
2911 error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag,
2912 NULL, NULL);
2913 fp->f_offset = auio.uio_offset;
2914 } else
2915# endif
2916 {
2917 kuio = auio;
2918 kuio.uio_iov = &kiov;
2919 kuio.uio_segflg = UIO_SYSSPACE;
2920 kiov.iov_len = SCARG(uap, count);
2921 MALLOC(dirbuf, caddr_t, SCARG(uap, count), M_TEMP, M_WAITOK);
2922 kiov.iov_base = dirbuf;
2923 error = VOP_READDIR(vp, &kuio, fp->f_cred, &eofflag,
2924 NULL, NULL);
2925 fp->f_offset = kuio.uio_offset;
2926 if (error == 0) {
2927 readcnt = SCARG(uap, count) - kuio.uio_resid;
2928 edp = (struct dirent *)&dirbuf[readcnt];
2929 for (dp = (struct dirent *)dirbuf; dp < edp; ) {
2930# if (BYTE_ORDER == LITTLE_ENDIAN)
2931 /*
2932 * The expected low byte of
2933 * dp->d_namlen is our dp->d_type.
2934 * The high MBZ byte of dp->d_namlen
2935 * is our dp->d_namlen.
2936 */
2937 dp->d_type = dp->d_namlen;
2938 dp->d_namlen = 0;
2939# else
2940 /*
2941 * The dp->d_type is the high byte
2942 * of the expected dp->d_namlen,
2943 * so must be zero'ed.
2944 */
2945 dp->d_type = 0;
2946# endif
2947 if (dp->d_reclen > 0) {
2948 dp = (struct dirent *)
2949 ((char *)dp + dp->d_reclen);
2950 } else {
2951 error = EIO;
2952 break;
2953 }
2954 }
2955 if (dp >= edp)
2956 error = uiomove(dirbuf, readcnt, &auio);
2957 }
2958 FREE(dirbuf, M_TEMP);
2959 }
2960 VOP_UNLOCK(vp, 0, p);
2961 if (error)
2962 return (error);
2963 if (SCARG(uap, count) == auio.uio_resid) {
2964 if (union_dircheckp) {
2965 error = union_dircheckp(p, &vp, fp);
2966 if (error == -1)
2967 goto unionread;
2968 if (error)
2969 return (error);
2970 }
2971 if ((vp->v_flag & VROOT) &&
2972 (vp->v_mount->mnt_flag & MNT_UNION)) {
2973 struct vnode *tvp = vp;
2974 vp = vp->v_mount->mnt_vnodecovered;
2975 VREF(vp);
2976 fp->f_data = (caddr_t) vp;
2977 fp->f_offset = 0;
2978 vrele(tvp);
2979 goto unionread;
2980 }
2981 }
2982 error = copyout((caddr_t)&loff, (caddr_t)SCARG(uap, basep),
2983 sizeof(long));
2984 p->p_retval[0] = SCARG(uap, count) - auio.uio_resid;
2985 return (error);
2986}
2987#endif /* COMPAT_43 */
2988
2989/*
2990 * Read a block of directory entries in a file system independent format.
2991 */
2992#ifndef _SYS_SYSPROTO_H_
2993struct getdirentries_args {
2994 int fd;
2995 char *buf;
2996 u_int count;
2997 long *basep;
2998};
2999#endif
3000int
3001getdirentries(p, uap)
3002 struct proc *p;
3003 register struct getdirentries_args /* {
3004 syscallarg(int) fd;
3005 syscallarg(char *) buf;
3006 syscallarg(u_int) count;
3007 syscallarg(long *) basep;
3008 } */ *uap;
3009{
3010 struct vnode *vp;
3011 struct file *fp;
3012 struct uio auio;
3013 struct iovec aiov;
3014 long loff;
3015 int error, eofflag;
3016
3017 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
3018 return (error);
3019 if ((fp->f_flag & FREAD) == 0)
3020 return (EBADF);
3021 vp = (struct vnode *)fp->f_data;
3022unionread:
3023 if (vp->v_type != VDIR)
3024 return (EINVAL);
3025 aiov.iov_base = SCARG(uap, buf);
3026 aiov.iov_len = SCARG(uap, count);
3027 auio.uio_iov = &aiov;
3028 auio.uio_iovcnt = 1;
3029 auio.uio_rw = UIO_READ;
3030 auio.uio_segflg = UIO_USERSPACE;
3031 auio.uio_procp = p;
3032 auio.uio_resid = SCARG(uap, count);
3033 /* vn_lock(vp, LK_SHARED | LK_RETRY, p); */
3034 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
3035 loff = auio.uio_offset = fp->f_offset;
3036 error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, NULL, NULL);
3037 fp->f_offset = auio.uio_offset;
3038 VOP_UNLOCK(vp, 0, p);
3039 if (error)
3040 return (error);
3041 if (SCARG(uap, count) == auio.uio_resid) {
3042 if (union_dircheckp) {
3043 error = union_dircheckp(p, &vp, fp);
3044 if (error == -1)
3045 goto unionread;
3046 if (error)
3047 return (error);
3048 }
3049 if ((vp->v_flag & VROOT) &&
3050 (vp->v_mount->mnt_flag & MNT_UNION)) {
3051 struct vnode *tvp = vp;
3052 vp = vp->v_mount->mnt_vnodecovered;
3053 VREF(vp);
3054 fp->f_data = (caddr_t) vp;
3055 fp->f_offset = 0;
3056 vrele(tvp);
3057 goto unionread;
3058 }
3059 }
3060 if (SCARG(uap, basep) != NULL) {
3061 error = copyout((caddr_t)&loff, (caddr_t)SCARG(uap, basep),
3062 sizeof(long));
3063 }
3064 p->p_retval[0] = SCARG(uap, count) - auio.uio_resid;
3065 return (error);
3066}
3067#ifndef _SYS_SYSPROTO_H_
3068struct getdents_args {
3069 int fd;
3070 char *buf;
3071 size_t count;
3072};
3073#endif
3074int
3075getdents(p, uap)
3076 struct proc *p;
3077 register struct getdents_args /* {
3078 syscallarg(int) fd;
3079 syscallarg(char *) buf;
3080 syscallarg(u_int) count;
3081 } */ *uap;
3082{
3083 struct getdirentries_args ap;
3084 ap.fd = uap->fd;
3085 ap.buf = uap->buf;
3086 ap.count = uap->count;
3087 ap.basep = NULL;
3088 return getdirentries(p, &ap);
3089}
3090
3091/*
3092 * Set the mode mask for creation of filesystem nodes.
3093 *
3094 * MP SAFE
3095 */
3096#ifndef _SYS_SYSPROTO_H_
3097struct umask_args {
3098 int newmask;
3099};
3100#endif
3101int
3102umask(p, uap)
3103 struct proc *p;
3104 struct umask_args /* {
3105 syscallarg(int) newmask;
3106 } */ *uap;
3107{
3108 register struct filedesc *fdp;
3109
3110 fdp = p->p_fd;
3111 p->p_retval[0] = fdp->fd_cmask;
3112 fdp->fd_cmask = SCARG(uap, newmask) & ALLPERMS;
3113 return (0);
3114}
3115
3116/*
3117 * Void all references to file by ripping underlying filesystem
3118 * away from vnode.
3119 */
3120#ifndef _SYS_SYSPROTO_H_
3121struct revoke_args {
3122 char *path;
3123};
3124#endif
3125/* ARGSUSED */
3126int
3127revoke(p, uap)
3128 struct proc *p;
3129 register struct revoke_args /* {
3130 syscallarg(char *) path;
3131 } */ *uap;
3132{
3133 register struct vnode *vp;
3134 struct vattr vattr;
3135 int error;
3136 struct nameidata nd;
3137
3138 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
3139 if ((error = namei(&nd)) != 0)
3140 return (error);
3141 vp = nd.ni_vp;
3142 NDFREE(&nd, NDF_ONLY_PNBUF);
3143 if (vp->v_type != VCHR && vp->v_type != VBLK) {
3144 error = EINVAL;
3145 goto out;
3146 }
3147 if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)) != 0)
3148 goto out;
3149 if (p->p_ucred->cr_uid != vattr.va_uid &&
3150 (error = suser_xxx(0, p, PRISON_ROOT)))
3151 goto out;
3152 if (vcount(vp) > 1)
3153 VOP_REVOKE(vp, REVOKEALL);
3154out:
3155 vrele(vp);
3156 return (error);
3157}
3158
3159/*
3160 * Convert a user file descriptor to a kernel file entry.
3161 */
3162int
3163getvnode(fdp, fd, fpp)
3164 struct filedesc *fdp;
3165 int fd;
3166 struct file **fpp;
3167{
3168 struct file *fp;
3169
3170 if ((u_int)fd >= fdp->fd_nfiles ||
3171 (fp = fdp->fd_ofiles[fd]) == NULL)
3172 return (EBADF);
3173 if (fp->f_type != DTYPE_VNODE && fp->f_type != DTYPE_FIFO)
3174 return (EINVAL);
3175 *fpp = fp;
3176 return (0);
3177}
3178/*
3179 * Get (NFS) file handle
3180 */
3181#ifndef _SYS_SYSPROTO_H_
3182struct getfh_args {
3183 char *fname;
3184 fhandle_t *fhp;
3185};
3186#endif
3187int
3188getfh(p, uap)
3189 struct proc *p;
3190 register struct getfh_args *uap;
3191{
3192 struct nameidata nd;
3193 fhandle_t fh;
3194 register struct vnode *vp;
3195 int error;
3196
3197 /*
3198 * Must be super user
3199 */
3200 error = suser(p);
3201 if (error)
3202 return (error);
3203 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE, uap->fname, p);
3204 error = namei(&nd);
3205 if (error)
3206 return (error);
3207 NDFREE(&nd, NDF_ONLY_PNBUF);
3208 vp = nd.ni_vp;
3209 bzero(&fh, sizeof(fh));
3210 fh.fh_fsid = vp->v_mount->mnt_stat.f_fsid;
3211 error = VFS_VPTOFH(vp, &fh.fh_fid);
3212 vput(vp);
3213 if (error)
3214 return (error);
3215 error = copyout(&fh, uap->fhp, sizeof (fh));
3216 return (error);
3217}
3218
3219/*
3220 * syscall for the rpc.lockd to use to translate a NFS file handle into
3221 * an open descriptor.
3222 *
3223 * warning: do not remove the suser() call or this becomes one giant
3224 * security hole.
3225 */
3226#ifndef _SYS_SYSPROTO_H_
3227struct fhopen_args {
3228 const struct fhandle *u_fhp;
3229 int flags;
3230};
3231#endif
3232int
3233fhopen(p, uap)
3234 struct proc *p;
3235 struct fhopen_args /* {
3236 syscallarg(const struct fhandle *) u_fhp;
3237 syscallarg(int) flags;
3238 } */ *uap;
3239{
3240 struct mount *mp;
3241 struct vnode *vp;
3242 struct fhandle fhp;
3243 struct vattr vat;
3244 struct vattr *vap = &vat;
3245 struct flock lf;
3246 struct file *fp;
3247 register struct filedesc *fdp = p->p_fd;
3248 int fmode, mode, error, type;
3249 struct file *nfp;
3250 int indx;
3251
3252 /*
3253 * Must be super user
3254 */
3255 error = suser(p);
3256 if (error)
3257 return (error);
3258
3259 fmode = FFLAGS(SCARG(uap, flags));
3260 /* why not allow a non-read/write open for our lockd? */
3261 if (((fmode & (FREAD | FWRITE)) == 0) || (fmode & O_CREAT))
3262 return (EINVAL);
3263 error = copyin(SCARG(uap,u_fhp), &fhp, sizeof(fhp));
3264 if (error)
3265 return(error);
3266 /* find the mount point */
3267 mp = vfs_getvfs(&fhp.fh_fsid);
3268 if (mp == NULL)
3269 return (ESTALE);
3270 /* now give me my vnode, it gets returned to me locked */
3271 error = VFS_FHTOVP(mp, &fhp.fh_fid, &vp);
3272 if (error)
3273 return (error);
3274 /*
3275 * from now on we have to make sure not
3276 * to forget about the vnode
3277 * any error that causes an abort must vput(vp)
3278 * just set error = err and 'goto bad;'.
3279 */
3280
3281 /*
3282 * from vn_open
3283 */
3284 if (vp->v_type == VLNK) {
3285 error = EMLINK;
3286 goto bad;
3287 }
3288 if (vp->v_type == VSOCK) {
3289 error = EOPNOTSUPP;
3290 goto bad;
3291 }
3292 mode = 0;
3293 if (fmode & (FWRITE | O_TRUNC)) {
3294 if (vp->v_type == VDIR) {
3295 error = EISDIR;
3296 goto bad;
3297 }
3298 error = vn_writechk(vp);
3299 if (error)
3300 goto bad;
3301 mode |= VWRITE;
3302 }
3303 if (fmode & FREAD)
3304 mode |= VREAD;
3305 if (mode) {
3306 error = VOP_ACCESS(vp, mode, p->p_ucred, p);
3307 if (error)
3308 goto bad;
3309 }
3310 if (fmode & O_TRUNC) {
3311 VOP_UNLOCK(vp, 0, p); /* XXX */
3312 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
3313 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p); /* XXX */
3314 VATTR_NULL(vap);
3315 vap->va_size = 0;
3316 error = VOP_SETATTR(vp, vap, p->p_ucred, p);
3317 if (error)
3318 goto bad;
3319 }
3320 error = VOP_OPEN(vp, fmode, p->p_ucred, p);
3321 if (error)
3322 goto bad;
3323 /*
3324 * Make sure that a VM object is created for VMIO support.
3325 */
3326 if (vn_canvmio(vp) == TRUE) {
3327 if ((error = vfs_object_create(vp, p, p->p_ucred)) != 0)
3328 goto bad;
3329 }
3330 if (fmode & FWRITE)
3331 vp->v_writecount++;
3332
3333 /*
3334 * end of vn_open code
3335 */
3336
3337 if ((error = falloc(p, &nfp, &indx)) != 0) {
3338 if (fmode & FWRITE)
3339 vp->v_writecount--;
3340 goto bad;
3341 }
3342 fp = nfp;
3343
3344 /*
3345 * hold an extra reference to avoid having fp ripped out
3346 * from under us while we block in the lock op.
3347 */
3348 fhold(fp);
3349 nfp->f_data = (caddr_t)vp;
3350 nfp->f_flag = fmode & FMASK;
3351 nfp->f_ops = &vnops;
3352 nfp->f_type = DTYPE_VNODE;
3353 if (fmode & (O_EXLOCK | O_SHLOCK)) {
3354 lf.l_whence = SEEK_SET;
3355 lf.l_start = 0;
3356 lf.l_len = 0;
3357 if (fmode & O_EXLOCK)
3358 lf.l_type = F_WRLCK;
3359 else
3360 lf.l_type = F_RDLCK;
3361 type = F_FLOCK;
3362 if ((fmode & FNONBLOCK) == 0)
3363 type |= F_WAIT;
3364 VOP_UNLOCK(vp, 0, p);
3365 if ((error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type)) != 0) {
3366 /*
3367 * lock request failed. Normally close the descriptor
3368 * but handle the case where someone might have dup()d
3369 * or close()d it when we weren't looking.
3370 */
3371 if (fdp->fd_ofiles[indx] == fp) {
3372 fdp->fd_ofiles[indx] = NULL;
3373 fdrop(fp, p);
3374 }
3375
3376 /*
3377 * release our private reference.
3378 */
3379 fdrop(fp, p);
3380 return (error);
3381 }
3382 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
3383 fp->f_flag |= FHASLOCK;
3384 }
3385 if ((vp->v_type == VREG) && (VOP_GETVOBJECT(vp, NULL) != 0))
3386 vfs_object_create(vp, p, p->p_ucred);
3387
3388 VOP_UNLOCK(vp, 0, p);
3389 fdrop(fp, p);
3390 p->p_retval[0] = indx;
3391 return (0);
3392
3393bad:
3394 vput(vp);
3395 return (error);
3396}
3397
3398#ifndef _SYS_SYSPROTO_H_
3399struct fhstat_args {
3400 struct fhandle *u_fhp;
3401 struct stat *sb;
3402};
3403#endif
3404int
3405fhstat(p, uap)
3406 struct proc *p;
3407 register struct fhstat_args /* {
3408 syscallarg(struct fhandle *) u_fhp;
3409 syscallarg(struct stat *) sb;
3410 } */ *uap;
3411{
3412 struct stat sb;
3413 fhandle_t fh;
3414 struct mount *mp;
3415 struct vnode *vp;
3416 int error;
3417
3418 /*
3419 * Must be super user
3420 */
3421 error = suser(p);
3422 if (error)
3423 return (error);
3424
3425 error = copyin(SCARG(uap, u_fhp), &fh, sizeof(fhandle_t));
3426 if (error)
3427 return (error);
3428
3429 if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL)
3430 return (ESTALE);
3431 if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)))
3432 return (error);
3433 error = vn_stat(vp, &sb, p);
3434 vput(vp);
3435 if (error)
3436 return (error);
3437 error = copyout(&sb, SCARG(uap, sb), sizeof(sb));
3438 return (error);
3439}
3440
3441#ifndef _SYS_SYSPROTO_H_
3442struct fhstatfs_args {
3443 struct fhandle *u_fhp;
3444 struct statfs *buf;
3445};
3446#endif
3447int
3448fhstatfs(p, uap)
3449 struct proc *p;
3450 struct fhstatfs_args /* {
3451 syscallarg(struct fhandle) *u_fhp;
3452 syscallarg(struct statfs) *buf;
3453 } */ *uap;
3454{
3455 struct statfs *sp;
3456 struct mount *mp;
3457 struct vnode *vp;
3458 struct statfs sb;
3459 fhandle_t fh;
3460 int error;
3461
3462 /*
3463 * Must be super user
3464 */
3465 if ((error = suser(p)))
3466 return (error);
3467
3468 if ((error = copyin(SCARG(uap, u_fhp), &fh, sizeof(fhandle_t))) != 0)
3469 return (error);
3470
3471 if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL)
3472 return (ESTALE);
3473 if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)))
3474 return (error);
3475 mp = vp->v_mount;
3476 sp = &mp->mnt_stat;
3477 vput(vp);
3478 if ((error = VFS_STATFS(mp, sp, p)) != 0)
3479 return (error);
3480 sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
3481 if (suser_xxx(p->p_ucred, 0, 0)) {
3482 bcopy((caddr_t)sp, (caddr_t)&sb, sizeof(sb));
3483 sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0;
3484 sp = &sb;
3485 }
3486 return (copyout(sp, SCARG(uap, buf), sizeof(*sp)));
3487}
3488
3489/*
3490 * Syscall to push extended attribute configuration information into the
3491 * VFS. Accepts a path, which it converts to a mountpoint, as well as
3492 * a command (int cmd), and attribute name and misc data. For now, the
3493 * attribute name is left in userspace for consumption by the VFS_op.
3494 * It will probably be changed to be copied into sysspace by the
3495 * syscall in the future, once issues with various consumers of the
3496 * attribute code have raised their hands.
3497 *
3498 * Currently this is used only by UFS Extended Attributes.
3499 */
3500int
3501extattrctl(p, uap)
3502 struct proc *p;
3503 struct extattrctl_args *uap;
3504{
3505 struct nameidata nd;
3506 struct mount *mp;
3507 int error;
3508
3509 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
3510 if ((error = namei(&nd)) != 0)
3511 return (error);
3512 mp = nd.ni_vp->v_mount;
3513 NDFREE(&nd, 0);
3514 return (VFS_EXTATTRCTL(mp, SCARG(uap, cmd), SCARG(uap, attrname),
3515 SCARG(uap, arg), p));
3516}
3517
3518/*
3519 * Syscall to set a named extended attribute on a file or directory.
3520 * Accepts attribute name, and a uio structure pointing to the data to set.
3521 * The uio is consumed in the style of writev(). The real work happens
3522 * in VOP_SETEXTATTR().
3523 */
3524int
3525extattr_set_file(p, uap)
3526 struct proc *p;
3527 struct extattr_set_file_args *uap;
3528{
3529 struct nameidata nd;
3530 struct uio auio;
3531 struct iovec *iov, *needfree = NULL, aiov[UIO_SMALLIOV];
3532 char attrname[EXTATTR_MAXNAMELEN];
3533 u_int iovlen, cnt;
3534 int error, i;
3535
3536 error = copyin(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN);
3537 if (error)
3538 return (error);
3539 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
3540 SCARG(uap, path), p);
3541 if ((error = namei(&nd)) != 0)
3542 return(error);
3543 iovlen = uap->iovcnt * sizeof(struct iovec);
3544 if (uap->iovcnt > UIO_SMALLIOV) {
3545 if (uap->iovcnt > UIO_MAXIOV) {
3546 error = EINVAL;
3547 goto done;
3548 }
3549 MALLOC(iov, struct iovec *, iovlen, M_IOV, M_WAITOK);
3550 needfree = iov;
3551 } else
3552 iov = aiov;
3553 auio.uio_iov = iov;
3554 auio.uio_iovcnt = uap->iovcnt;
3555 auio.uio_rw = UIO_WRITE;
3556 auio.uio_segflg = UIO_USERSPACE;
3557 auio.uio_procp = p;
3558 auio.uio_offset = 0;
3559 if ((error = copyin((caddr_t)uap->iovp, (caddr_t)iov, iovlen)))
3560 goto done;
3561 auio.uio_resid = 0;
3562 for (i = 0; i < uap->iovcnt; i++) {
3563 if (iov->iov_len > INT_MAX - auio.uio_resid) {
3564 error = EINVAL;
3565 goto done;
3566 }
3567 auio.uio_resid += iov->iov_len;
3568 iov++;
3569 }
3570 cnt = auio.uio_resid;
3571 error = VOP_SETEXTATTR(nd.ni_vp, attrname, &auio, p->p_cred->pc_ucred,
3572 p);
3573 cnt -= auio.uio_resid;
3574 p->p_retval[0] = cnt;
3575done:
3576 if (needfree)
3577 FREE(needfree, M_IOV);
3578 NDFREE(&nd, 0);
3579 return (error);
3580}
3581
3582/*
3583 * Syscall to get a named extended attribute on a file or directory.
3584 * Accepts attribute name, and a uio structure pointing to a buffer for the
3585 * data. The uio is consumed in the style of readv(). The real work
3586 * happens in VOP_GETEXTATTR();
3587 */
3588int
3589extattr_get_file(p, uap)
3590 struct proc *p;
3591 struct extattr_get_file_args *uap;
3592{
3593 struct nameidata nd;
3594 struct uio auio;
3595 struct iovec *iov, *needfree, aiov[UIO_SMALLIOV];
3596 char attrname[EXTATTR_MAXNAMELEN];
3597 u_int iovlen, cnt;
3598 int error, i;
3599
3600 error = copyin(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN);
3601 if (error)
3602 return (error);
3603 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
3604 SCARG(uap, path), p);
3605 if ((error = namei(&nd)) != 0)
3606 return (error);
3607 iovlen = uap->iovcnt * sizeof (struct iovec);
3608 if (uap->iovcnt > UIO_SMALLIOV) {
3609 if (uap->iovcnt > UIO_MAXIOV) {
3610 NDFREE(&nd, 0);
3611 return (EINVAL);
3612 }
3613 MALLOC(iov, struct iovec *, iovlen, M_IOV, M_WAITOK);
3614 needfree = iov;
3615 } else {
3616 iov = aiov;
3617 needfree = NULL;
3618 }
3619 auio.uio_iov = iov;
3620 auio.uio_iovcnt = uap->iovcnt;
3621 auio.uio_rw = UIO_READ;
3622 auio.uio_segflg = UIO_USERSPACE;
3623 auio.uio_procp = p;
3624 auio.uio_offset = 0;
3625 if ((error = copyin((caddr_t)uap->iovp, (caddr_t)iov, iovlen)))
3626 goto done;
3627 auio.uio_resid = 0;
3628 for (i = 0; i < uap->iovcnt; i++) {
3629 if (iov->iov_len > INT_MAX - auio.uio_resid) {
3630 error = EINVAL;
3631 goto done;
3632 }
3633 auio.uio_resid += iov->iov_len;
3634 iov++;
3635 }
3636 cnt = auio.uio_resid;
3637 error = VOP_GETEXTATTR(nd.ni_vp, attrname, &auio, p->p_cred->pc_ucred,
3638 p);
3639 cnt -= auio.uio_resid;
3640 p->p_retval[0] = cnt;
3641done:
3642 if (needfree)
3643 FREE(needfree, M_IOV);
3644 NDFREE(&nd, 0);
3645 return(error);
3646}
3647
3648/*
3649 * Syscall to delete a named extended attribute from a file or directory.
3650 * Accepts attribute name. The real work happens in VOP_SETEXTATTR().
3651 */
3652int
3653extattr_delete_file(p, uap)
3654 struct proc *p;
3655 struct extattr_delete_file_args *uap;
3656{
3657 struct nameidata nd;
3658 char attrname[EXTATTR_MAXNAMELEN];
3659 int error;
3660
3661 error = copyin(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN);
3662 if (error)
3663 return(error);
3664 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
3665 SCARG(uap, path), p);
3666 if ((error = namei(&nd)) != 0)
3667 return(error);
3668 error = VOP_SETEXTATTR(nd.ni_vp, attrname, NULL, p->p_cred->pc_ucred,
3669 p);
3670 NDFREE(&nd, 0);
3671 return(error);
3672}