vfs_syscalls.c revision 1.147.4.5 1 /* $NetBSD: vfs_syscalls.c,v 1.147.4.5 1999/10/26 19:15:17 fvdl Exp $ */
2
3 /*
4 * Copyright (c) 1989, 1993
5 * The Regents of the University of California. All rights reserved.
6 * (c) UNIX System Laboratories, Inc.
7 * All or some portions of this file are derived from material licensed
8 * to the University of California by American Telephone and Telegraph
9 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
10 * the permission of UNIX System Laboratories, Inc.
11 *
12 * Redistribution and use in source and binary forms, with or without
13 * modification, are permitted provided that the following conditions
14 * are met:
15 * 1. Redistributions of source code must retain the above copyright
16 * notice, this list of conditions and the following disclaimer.
17 * 2. Redistributions in binary form must reproduce the above copyright
18 * notice, this list of conditions and the following disclaimer in the
19 * documentation and/or other materials provided with the distribution.
20 * 3. All advertising materials mentioning features or use of this software
21 * must display the following acknowledgement:
22 * This product includes software developed by the University of
23 * California, Berkeley and its contributors.
24 * 4. Neither the name of the University nor the names of its contributors
25 * may be used to endorse or promote products derived from this software
26 * without specific prior written permission.
27 *
28 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
29 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
30 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
31 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
32 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
33 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
34 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
35 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
36 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
37 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
38 * SUCH DAMAGE.
39 *
40 * @(#)vfs_syscalls.c 8.42 (Berkeley) 7/31/95
41 */
42
43 #include "opt_compat_netbsd.h"
44 #include "opt_compat_43.h"
45
46 #include <sys/param.h>
47 #include <sys/systm.h>
48 #include <sys/namei.h>
49 #include <sys/filedesc.h>
50 #include <sys/kernel.h>
51 #include <sys/file.h>
52 #include <sys/stat.h>
53 #include <sys/vnode.h>
54 #include <sys/mount.h>
55 #include <sys/proc.h>
56 #include <sys/uio.h>
57 #include <sys/malloc.h>
58 #include <sys/dirent.h>
59
60 #include <sys/syscallargs.h>
61
62 #include <vm/vm.h>
63 #include <sys/sysctl.h>
64
65 #include <miscfs/genfs/genfs.h>
66 #include <miscfs/syncfs/syncfs.h>
67
68 #include <uvm/uvm_extern.h>
69
70 static int change_dir __P((struct nameidata *, struct proc *));
71 static int change_mode __P((struct vnode *, int, struct proc *p));
72 static int change_owner __P((struct vnode *, uid_t, gid_t, struct proc *,
73 int));
74 static int change_utimes __P((struct vnode *vp, const struct timeval *,
75 struct proc *p));
76 static int rename_files __P((const char *, const char *, struct proc *, int));
77
78 void checkdirs __P((struct vnode *));
79 int dounmount __P((struct mount *, int, struct proc *));
80
81 /*
82 * Virtual File System System Calls
83 */
84
85 /*
86 * Mount a file system.
87 */
88
89 /*
90 * This table is used to maintain compatibility with 4.3BSD
91 * and NetBSD 0.9 mount syscalls. Note, the order is important!
92 *
93 * Also note that not all of these had actual numbers in 4.3BSD
94 * or NetBSD 0.9!
95 */
96 const char *mountcompatnames[] = {
97 NULL, /* 0 = MOUNT_NONE */
98 MOUNT_FFS, /* 1 */
99 MOUNT_NFS, /* 2 */
100 MOUNT_MFS, /* 3 */
101 MOUNT_MSDOS, /* 4 */
102 MOUNT_LFS, /* 5 */
103 NULL, /* 6 = MOUNT_LOFS */
104 MOUNT_FDESC, /* 7 */
105 MOUNT_PORTAL, /* 8 */
106 MOUNT_NULL, /* 9 */
107 MOUNT_UMAP, /* 10 */
108 MOUNT_KERNFS, /* 11 */
109 MOUNT_PROCFS, /* 12 */
110 MOUNT_AFS, /* 13 */
111 MOUNT_CD9660, /* 14 = MOUNT_ISOFS */
112 MOUNT_UNION, /* 15 */
113 MOUNT_ADOSFS, /* 16 */
114 MOUNT_EXT2FS, /* 17 */
115 MOUNT_CODA, /* 18 */
116 MOUNT_FILECORE, /* 19 */
117 MOUNT_NTFS, /* 20 */
118 };
119 const int nmountcompatnames = sizeof(mountcompatnames) /
120 sizeof(mountcompatnames[0]);
121
122 /* ARGSUSED */
123 int
124 sys_mount(p, v, retval)
125 struct proc *p;
126 void *v;
127 register_t *retval;
128 {
129 register struct sys_mount_args /* {
130 syscallarg(const char *) type;
131 syscallarg(const char *) path;
132 syscallarg(int) flags;
133 syscallarg(void *) data;
134 } */ *uap = v;
135 struct vnode *vp;
136 struct mount *mp;
137 int error, flag = 0;
138 char fstypename[MFSNAMELEN];
139 struct vattr va;
140 struct nameidata nd;
141 struct vfsops *vfs;
142
143 /*
144 * Get vnode to be covered
145 */
146 NDINIT(&nd, LOOKUP, FOLLOW , UIO_USERSPACE,
147 SCARG(uap, path), p);
148 if ((error = namei(&nd)) != 0)
149 return (error);
150 vp = nd.ni_vp;
151 /*
152 * A lookup in VFS_MOUNT might result in an attempt to
153 * lock this vnode again, so make the lock resursive.
154 */
155 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY | LK_SETRECURSE);
156 if (SCARG(uap, flags) & MNT_UPDATE) {
157 if ((vp->v_flag & VROOT) == 0) {
158 vput(vp);
159 return (EINVAL);
160 }
161 mp = vp->v_mount;
162 flag = mp->mnt_flag;
163 vfs = mp->mnt_op;
164 /*
165 * We only allow the filesystem to be reloaded if it
166 * is currently mounted read-only.
167 */
168 if ((SCARG(uap, flags) & MNT_RELOAD) &&
169 ((mp->mnt_flag & MNT_RDONLY) == 0)) {
170 vput(vp);
171 return (EOPNOTSUPP); /* Needs translation */
172 }
173 /*
174 * In "highly secure" mode, don't let the caller do anything
175 * but downgrade a filesystem from read-write to read-only.
176 * (see also below; MNT_UPDATE is required.)
177 */
178 if (securelevel >= 2 &&
179 (SCARG(uap, flags) !=
180 (mp->mnt_flag | MNT_RDONLY |
181 MNT_RELOAD | MNT_FORCE | MNT_UPDATE))) {
182 vput(vp);
183 return (EPERM);
184 }
185 mp->mnt_flag |=
186 SCARG(uap, flags) & (MNT_RELOAD | MNT_FORCE | MNT_UPDATE);
187 /*
188 * Only root, or the user that did the original mount is
189 * permitted to update it.
190 */
191 if (mp->mnt_stat.f_owner != p->p_ucred->cr_uid &&
192 (error = suser(p->p_ucred, &p->p_acflag)) != 0) {
193 vput(vp);
194 return (error);
195 }
196 /*
197 * Do not allow NFS export by non-root users. For non-root
198 * users, silently enforce MNT_NOSUID and MNT_NODEV, and
199 * MNT_NOEXEC if mount point is already MNT_NOEXEC.
200 */
201 if (p->p_ucred->cr_uid != 0) {
202 if (SCARG(uap, flags) & MNT_EXPORTED) {
203 vput(vp);
204 return (EPERM);
205 }
206 SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
207 if (flag & MNT_NOEXEC)
208 SCARG(uap, flags) |= MNT_NOEXEC;
209 }
210 if (vfs_busy(mp, LK_NOWAIT, 0)) {
211 vput(vp);
212 return (EPERM);
213 }
214 VOP_UNLOCK(vp, 0);
215 goto update;
216 } else {
217 if (securelevel >= 2)
218 return (EPERM);
219 }
220 /*
221 * If the user is not root, ensure that they own the directory
222 * onto which we are attempting to mount.
223 */
224 if ((error = VOP_GETATTR(vp, &va, p->p_ucred, p)) != 0 ||
225 (va.va_uid != p->p_ucred->cr_uid &&
226 (error = suser(p->p_ucred, &p->p_acflag)) != 0)) {
227 vput(vp);
228 return (error);
229 }
230 /*
231 * Do not allow NFS export by non-root users. For non-root users,
232 * silently enforce MNT_NOSUID and MNT_NODEV, and MNT_NOEXEC if the
233 * mount point is already MNT_NOEXEC.
234 */
235 if (p->p_ucred->cr_uid != 0) {
236 if (SCARG(uap, flags) & MNT_EXPORTED) {
237 vput(vp);
238 return (EPERM);
239 }
240 SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
241 if (vp->v_mount->mnt_flag & MNT_NOEXEC)
242 SCARG(uap, flags) |= MNT_NOEXEC;
243 }
244 if ((error = vinvalbuf(vp, V_SAVE, p->p_ucred, p, 0, 0)) != 0)
245 return (error);
246 if (vp->v_type != VDIR) {
247 vput(vp);
248 return (ENOTDIR);
249 }
250 error = copyinstr(SCARG(uap, type), fstypename, MFSNAMELEN, NULL);
251 if (error) {
252 #if defined(COMPAT_09) || defined(COMPAT_43)
253 /*
254 * Historically filesystem types were identified by number.
255 * If we get an integer for the filesystem type instead of a
256 * string, we check to see if it matches one of the historic
257 * filesystem types.
258 */
259 u_long fsindex = (u_long)SCARG(uap, type);
260 if (fsindex >= nmountcompatnames ||
261 mountcompatnames[fsindex] == NULL) {
262 vput(vp);
263 return (ENODEV);
264 }
265 strncpy(fstypename, mountcompatnames[fsindex], MFSNAMELEN);
266 #else
267 vput(vp);
268 return (error);
269 #endif
270 }
271 #ifdef COMPAT_10
272 /* Accept `ufs' as an alias for `ffs'. */
273 if (!strncmp(fstypename, "ufs", MFSNAMELEN))
274 strncpy(fstypename, "ffs", MFSNAMELEN);
275 #endif
276 if ((vfs = vfs_getopsbyname(fstypename)) == NULL) {
277 vput(vp);
278 return (ENODEV);
279 }
280 if (vp->v_mountedhere != NULL) {
281 vput(vp);
282 return (EBUSY);
283 }
284
285 /*
286 * Allocate and initialize the file system.
287 */
288 mp = (struct mount *)malloc((u_long)sizeof(struct mount),
289 M_MOUNT, M_WAITOK);
290 memset((char *)mp, 0, (u_long)sizeof(struct mount));
291 lockinit(&mp->mnt_lock, PVFS, "vfslock", 0, 0);
292 (void)vfs_busy(mp, LK_NOWAIT, 0);
293 mp->mnt_op = vfs;
294 vfs->vfs_refcount++;
295 mp->mnt_vnodecovered = vp;
296 mp->mnt_stat.f_owner = p->p_ucred->cr_uid;
297 mp->mnt_unmounter = NULL;
298 update:
299 /*
300 * Set the mount level flags.
301 */
302 if (SCARG(uap, flags) & MNT_RDONLY)
303 mp->mnt_flag |= MNT_RDONLY;
304 else if (mp->mnt_flag & MNT_RDONLY)
305 mp->mnt_flag |= MNT_WANTRDWR;
306 mp->mnt_flag &=~ (MNT_NOSUID | MNT_NOEXEC | MNT_NODEV |
307 MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | MNT_NOCOREDUMP |
308 MNT_NOATIME | MNT_NODEVMTIME | MNT_SYMPERM);
309 mp->mnt_flag |= SCARG(uap, flags) & (MNT_NOSUID | MNT_NOEXEC |
310 MNT_NODEV | MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC |
311 MNT_NOCOREDUMP | MNT_NOATIME | MNT_NODEVMTIME | MNT_SYMPERM);
312 /*
313 * Mount the filesystem.
314 */
315 error = VFS_MOUNT(mp, SCARG(uap, path), SCARG(uap, data), &nd, p);
316 if (mp->mnt_flag & MNT_UPDATE) {
317 vrele(vp);
318 if (mp->mnt_flag & MNT_WANTRDWR)
319 mp->mnt_flag &= ~MNT_RDONLY;
320 mp->mnt_flag &=~
321 (MNT_UPDATE | MNT_RELOAD | MNT_FORCE | MNT_WANTRDWR);
322 if (error)
323 mp->mnt_flag = flag;
324 if ((mp->mnt_flag & MNT_RDONLY) == 0) {
325 if (mp->mnt_syncer == NULL)
326 error = vfs_allocate_syncvnode(mp);
327 } else {
328 if (mp->mnt_syncer != NULL)
329 vgone(mp->mnt_syncer);
330 mp->mnt_syncer = NULL;
331 }
332 vfs_unbusy(mp);
333 return (error);
334 }
335 /*
336 * Put the new filesystem on the mount list after root.
337 */
338 cache_purge(vp);
339 if (!error) {
340 vp->v_mountedhere = mp;
341 simple_lock(&mountlist_slock);
342 CIRCLEQ_INSERT_TAIL(&mountlist, mp, mnt_list);
343 simple_unlock(&mountlist_slock);
344 checkdirs(vp);
345 VOP_UNLOCK(vp, 0);
346 if ((mp->mnt_flag & MNT_RDONLY) == 0)
347 error = vfs_allocate_syncvnode(mp);
348 vfs_unbusy(mp);
349 (void) VFS_STATFS(mp, &mp->mnt_stat, p);
350 if ((error = VFS_START(mp, 0, p)))
351 vrele(vp);
352 } else {
353 vp->v_mountedhere = (struct mount *)0;
354 vfs->vfs_refcount--;
355 vfs_unbusy(mp);
356 free((caddr_t)mp, M_MOUNT);
357 vput(vp);
358 }
359 return (error);
360 }
361
362 /*
363 * Scan all active processes to see if any of them have a current
364 * or root directory onto which the new filesystem has just been
365 * mounted. If so, replace them with the new mount point.
366 */
367 void
368 checkdirs(olddp)
369 struct vnode *olddp;
370 {
371 struct cwdinfo *cwdi;
372 struct vnode *newdp;
373 struct proc *p;
374
375 if (olddp->v_usecount == 1)
376 return;
377 if (VFS_ROOT(olddp->v_mountedhere, &newdp))
378 panic("mount: lost mount");
379 proclist_lock_read();
380 for (p = allproc.lh_first; p != 0; p = p->p_list.le_next) {
381 cwdi = p->p_cwdi;
382 if (cwdi->cwdi_cdir == olddp) {
383 vrele(cwdi->cwdi_cdir);
384 VREF(newdp);
385 cwdi->cwdi_cdir = newdp;
386 }
387 if (cwdi->cwdi_rdir == olddp) {
388 vrele(cwdi->cwdi_rdir);
389 VREF(newdp);
390 cwdi->cwdi_rdir = newdp;
391 }
392 }
393 proclist_unlock_read();
394 if (rootvnode == olddp) {
395 vrele(rootvnode);
396 VREF(newdp);
397 rootvnode = newdp;
398 }
399 vput(newdp);
400 }
401
402 /*
403 * Unmount a file system.
404 *
405 * Note: unmount takes a path to the vnode mounted on as argument,
406 * not special file (as before).
407 */
408 /* ARGSUSED */
409 int
410 sys_unmount(p, v, retval)
411 struct proc *p;
412 void *v;
413 register_t *retval;
414 {
415 register struct sys_unmount_args /* {
416 syscallarg(const char *) path;
417 syscallarg(int) flags;
418 } */ *uap = v;
419 register struct vnode *vp;
420 struct mount *mp;
421 int error;
422 struct nameidata nd;
423
424 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
425 SCARG(uap, path), p);
426 if ((error = namei(&nd)) != 0)
427 return (error);
428 vp = nd.ni_vp;
429 mp = vp->v_mount;
430
431 /*
432 * Only root, or the user that did the original mount is
433 * permitted to unmount this filesystem.
434 */
435 if ((mp->mnt_stat.f_owner != p->p_ucred->cr_uid) &&
436 (error = suser(p->p_ucred, &p->p_acflag)) != 0) {
437 vput(vp);
438 return (error);
439 }
440
441 /*
442 * Don't allow unmounting the root file system.
443 */
444 if (mp->mnt_flag & MNT_ROOTFS) {
445 vput(vp);
446 return (EINVAL);
447 }
448
449 /*
450 * Must be the root of the filesystem
451 */
452 if ((vp->v_flag & VROOT) == 0) {
453 vput(vp);
454 return (EINVAL);
455 }
456 vput(vp);
457
458 if (vfs_busy(mp, 0, 0))
459 return (EBUSY);
460
461 return (dounmount(mp, SCARG(uap, flags), p));
462 }
463
464 /*
465 * Do the actual file system unmount. File system is assumed to have been
466 * marked busy by the caller.
467 */
468 int
469 dounmount(mp, flags, p)
470 register struct mount *mp;
471 int flags;
472 struct proc *p;
473 {
474 struct vnode *coveredvp;
475 int error;
476 int async;
477
478 simple_lock(&mountlist_slock);
479 vfs_unbusy(mp);
480 /*
481 * XXX Freeze syncer. This should really be done on a mountpoint
482 * basis, but especially the softdep code possibly called from
483 * the syncer doesn't exactly work on a per-mountpoint basis,
484 * so the softdep code would become a maze of vfs_busy calls.
485 */
486 lockmgr(&syncer_lock, LK_EXCLUSIVE, NULL);
487
488 mp->mnt_flag |= MNT_UNMOUNT;
489 mp->mnt_unmounter = p;
490 lockmgr(&mp->mnt_lock, LK_DRAIN | LK_INTERLOCK, &mountlist_slock);
491 if (mp->mnt_flag & MNT_EXPUBLIC)
492 vfs_setpublicfs(NULL, NULL, NULL);
493 async = mp->mnt_flag & MNT_ASYNC;
494 mp->mnt_flag &=~ MNT_ASYNC;
495 cache_purgevfs(mp); /* remove cache entries for this file sys */
496 if (mp->mnt_syncer != NULL)
497 vgone(mp->mnt_syncer);
498 if (((mp->mnt_flag & MNT_RDONLY) ||
499 (error = VFS_SYNC(mp, MNT_WAIT, p->p_ucred, p)) == 0) ||
500 (flags & MNT_FORCE))
501 error = VFS_UNMOUNT(mp, flags, p);
502 simple_lock(&mountlist_slock);
503 if (error) {
504 if ((mp->mnt_flag & MNT_RDONLY) == 0 && mp->mnt_syncer == NULL)
505 (void) vfs_allocate_syncvnode(mp);
506 mp->mnt_flag &= ~MNT_UNMOUNT;
507 mp->mnt_unmounter = NULL;
508 mp->mnt_flag |= async;
509 lockmgr(&mp->mnt_lock, LK_RELEASE | LK_INTERLOCK | LK_REENABLE,
510 &mountlist_slock);
511 lockmgr(&syncer_lock, LK_RELEASE, NULL);
512 while(mp->mnt_wcnt > 0) {
513 wakeup((caddr_t)mp);
514 tsleep(&mp->mnt_wcnt, PVFS, "mntwcnt1", 0);
515 }
516 return (error);
517 }
518 CIRCLEQ_REMOVE(&mountlist, mp, mnt_list);
519 if ((coveredvp = mp->mnt_vnodecovered) != NULLVP) {
520 coveredvp->v_mountedhere = NULL;
521 vrele(coveredvp);
522 }
523 mp->mnt_op->vfs_refcount--;
524 if (mp->mnt_vnodelist.lh_first != NULL)
525 panic("unmount: dangling vnode");
526 mp->mnt_flag |= MNT_GONE;
527 lockmgr(&mp->mnt_lock, LK_RELEASE | LK_INTERLOCK, &mountlist_slock);
528 lockmgr(&syncer_lock, LK_RELEASE, NULL);
529 while(mp->mnt_wcnt > 0) {
530 wakeup((caddr_t)mp);
531 tsleep(&mp->mnt_wcnt, PVFS, "mntwcnt2", 0);
532 }
533 free((caddr_t)mp, M_MOUNT);
534 return (0);
535 }
536
537 /*
538 * Sync each mounted filesystem.
539 */
540 #ifdef DEBUG
541 int syncprt = 0;
542 struct ctldebug debug0 = { "syncprt", &syncprt };
543 #endif
544
545 /* ARGSUSED */
546 int
547 sys_sync(p, v, retval)
548 struct proc *p;
549 void *v;
550 register_t *retval;
551 {
552 register struct mount *mp, *nmp;
553 int asyncflag;
554
555 simple_lock(&mountlist_slock);
556 for (mp = mountlist.cqh_last; mp != (void *)&mountlist; mp = nmp) {
557 if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock)) {
558 nmp = mp->mnt_list.cqe_prev;
559 continue;
560 }
561 if ((mp->mnt_flag & MNT_RDONLY) == 0) {
562 asyncflag = mp->mnt_flag & MNT_ASYNC;
563 mp->mnt_flag &= ~MNT_ASYNC;
564 uvm_vnp_sync(mp);
565 VFS_SYNC(mp, MNT_NOWAIT, p->p_ucred, p);
566 if (asyncflag)
567 mp->mnt_flag |= MNT_ASYNC;
568 }
569 simple_lock(&mountlist_slock);
570 nmp = mp->mnt_list.cqe_prev;
571 vfs_unbusy(mp);
572
573 }
574 simple_unlock(&mountlist_slock);
575 #ifdef DEBUG
576 if (syncprt)
577 vfs_bufstats();
578 #endif /* DEBUG */
579 return (0);
580 }
581
582 /*
583 * Change filesystem quotas.
584 */
585 /* ARGSUSED */
586 int
587 sys_quotactl(p, v, retval)
588 struct proc *p;
589 void *v;
590 register_t *retval;
591 {
592 register struct sys_quotactl_args /* {
593 syscallarg(const char *) path;
594 syscallarg(int) cmd;
595 syscallarg(int) uid;
596 syscallarg(caddr_t) arg;
597 } */ *uap = v;
598 register struct mount *mp;
599 int error;
600 struct nameidata nd;
601
602 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
603 if ((error = namei(&nd)) != 0)
604 return (error);
605 mp = nd.ni_vp->v_mount;
606 vrele(nd.ni_vp);
607 return (VFS_QUOTACTL(mp, SCARG(uap, cmd), SCARG(uap, uid),
608 SCARG(uap, arg), p));
609 }
610
611 /*
612 * Get filesystem statistics.
613 */
614 /* ARGSUSED */
615 int
616 sys_statfs(p, v, retval)
617 struct proc *p;
618 void *v;
619 register_t *retval;
620 {
621 register struct sys_statfs_args /* {
622 syscallarg(const char *) path;
623 syscallarg(struct statfs *) buf;
624 } */ *uap = v;
625 register struct mount *mp;
626 register struct statfs *sp;
627 int error;
628 struct nameidata nd;
629
630 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
631 if ((error = namei(&nd)) != 0)
632 return (error);
633 mp = nd.ni_vp->v_mount;
634 sp = &mp->mnt_stat;
635 vrele(nd.ni_vp);
636 if ((error = VFS_STATFS(mp, sp, p)) != 0)
637 return (error);
638 sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
639 return (copyout(sp, SCARG(uap, buf), sizeof(*sp)));
640 }
641
642 /*
643 * Get filesystem statistics.
644 */
645 /* ARGSUSED */
646 int
647 sys_fstatfs(p, v, retval)
648 struct proc *p;
649 void *v;
650 register_t *retval;
651 {
652 register struct sys_fstatfs_args /* {
653 syscallarg(int) fd;
654 syscallarg(struct statfs *) buf;
655 } */ *uap = v;
656 struct file *fp;
657 struct mount *mp;
658 register struct statfs *sp;
659 int error;
660
661 /* getvnode() will use the descriptor for us */
662 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
663 return (error);
664 mp = ((struct vnode *)fp->f_data)->v_mount;
665 sp = &mp->mnt_stat;
666 if ((error = VFS_STATFS(mp, sp, p)) != 0)
667 goto out;
668 sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
669 error = copyout(sp, SCARG(uap, buf), sizeof(*sp));
670 out:
671 FILE_UNUSE(fp, p);
672 return (error);
673 }
674
675 /*
676 * Get statistics on all filesystems.
677 */
678 int
679 sys_getfsstat(p, v, retval)
680 struct proc *p;
681 void *v;
682 register_t *retval;
683 {
684 register struct sys_getfsstat_args /* {
685 syscallarg(struct statfs *) buf;
686 syscallarg(long) bufsize;
687 syscallarg(int) flags;
688 } */ *uap = v;
689 register struct mount *mp, *nmp;
690 register struct statfs *sp;
691 caddr_t sfsp;
692 long count, maxcount, error;
693
694 maxcount = SCARG(uap, bufsize) / sizeof(struct statfs);
695 sfsp = (caddr_t)SCARG(uap, buf);
696 simple_lock(&mountlist_slock);
697 count = 0;
698 for (mp = mountlist.cqh_first; mp != (void *)&mountlist; mp = nmp) {
699 if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock)) {
700 nmp = mp->mnt_list.cqe_next;
701 continue;
702 }
703 if (sfsp && count < maxcount) {
704 sp = &mp->mnt_stat;
705 /*
706 * If MNT_NOWAIT or MNT_LAZY is specified, do not
707 * refresh the fsstat cache. MNT_WAIT or MNT_LAXY
708 * overrides MNT_NOWAIT.
709 */
710 if (SCARG(uap, flags) != MNT_NOWAIT &&
711 SCARG(uap, flags) != MNT_LAZY &&
712 (SCARG(uap, flags) == MNT_WAIT ||
713 SCARG(uap, flags) == 0) &&
714 (error = VFS_STATFS(mp, sp, p)) != 0) {
715 simple_lock(&mountlist_slock);
716 nmp = mp->mnt_list.cqe_next;
717 vfs_unbusy(mp);
718 continue;
719 }
720 sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
721 error = copyout(sp, sfsp, sizeof(*sp));
722 if (error) {
723 vfs_unbusy(mp);
724 return (error);
725 }
726 sfsp += sizeof(*sp);
727 }
728 count++;
729 simple_lock(&mountlist_slock);
730 nmp = mp->mnt_list.cqe_next;
731 vfs_unbusy(mp);
732 }
733 simple_unlock(&mountlist_slock);
734 if (sfsp && count > maxcount)
735 *retval = maxcount;
736 else
737 *retval = count;
738 return (0);
739 }
740
741 /*
742 * Change current working directory to a given file descriptor.
743 */
744 /* ARGSUSED */
745 int
746 sys_fchdir(p, v, retval)
747 struct proc *p;
748 void *v;
749 register_t *retval;
750 {
751 struct sys_fchdir_args /* {
752 syscallarg(int) fd;
753 } */ *uap = v;
754 struct filedesc *fdp = p->p_fd;
755 struct cwdinfo *cwdi = p->p_cwdi;
756 struct vnode *vp, *tdp;
757 struct mount *mp;
758 struct file *fp;
759 int error;
760
761 /* getvnode() will use the descriptor for us */
762 if ((error = getvnode(fdp, SCARG(uap, fd), &fp)) != 0)
763 return (error);
764 vp = (struct vnode *)fp->f_data;
765
766 VREF(vp);
767 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
768 if (vp->v_type != VDIR)
769 error = ENOTDIR;
770 else
771 error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
772 while (!error && (mp = vp->v_mountedhere) != NULL) {
773 if (vfs_busy(mp, 0, 0))
774 continue;
775 error = VFS_ROOT(mp, &tdp);
776 vfs_unbusy(mp);
777 if (error)
778 break;
779 vput(vp);
780 vp = tdp;
781 }
782 if (error) {
783 vput(vp);
784 goto out;
785 }
786 VOP_UNLOCK(vp, 0);
787
788 /*
789 * Disallow changing to a directory not under the process's
790 * current root directory (if there is one).
791 */
792 if (cwdi->cwdi_rdir && !vn_isunder(vp, NULL, p)) {
793 vrele(vp);
794 error = EPERM; /* operation not permitted */
795 goto out;
796 }
797
798 vrele(cwdi->cwdi_cdir);
799 cwdi->cwdi_cdir = vp;
800 out:
801 FILE_UNUSE(fp, p);
802 return (error);
803 }
804
805 /*
806 * Change this process's notion of the root directory to a given file descriptor.
807 */
808
809 int
810 sys_fchroot(p, v, retval)
811 struct proc *p;
812 void *v;
813 register_t *retval;
814 {
815 struct sys_fchroot_args *uap = v;
816 struct filedesc *fdp = p->p_fd;
817 struct cwdinfo *cwdi = p->p_cwdi;
818 struct vnode *vp;
819 struct file *fp;
820 int error;
821
822 if ((error = suser(p->p_ucred, &p->p_acflag)) != 0)
823 return error;
824 /* getvnode() will use the descriptor for us */
825 if ((error = getvnode(fdp, SCARG(uap, fd), &fp)) != 0)
826 return error;
827 vp = (struct vnode *) fp->f_data;
828 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
829 if (vp->v_type != VDIR)
830 error = ENOTDIR;
831 else
832 error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
833 VOP_UNLOCK(vp, 0);
834 if (error)
835 goto out;
836 VREF(vp);
837
838 /*
839 * Prevent escaping from chroot by putting the root under
840 * the working directory. Silently chdir to / if we aren't
841 * already there.
842 */
843 if (!vn_isunder(cwdi->cwdi_cdir, vp, p)) {
844 /*
845 * XXX would be more failsafe to change directory to a
846 * deadfs node here instead
847 */
848 vrele(cwdi->cwdi_cdir);
849 VREF(vp);
850 cwdi->cwdi_cdir = vp;
851 }
852
853 if (cwdi->cwdi_rdir != NULL)
854 vrele(cwdi->cwdi_rdir);
855 cwdi->cwdi_rdir = vp;
856 out:
857 FILE_UNUSE(fp, p);
858 return (error);
859 }
860
861
862
863 /*
864 * Change current working directory (``.'').
865 */
866 /* ARGSUSED */
867 int
868 sys_chdir(p, v, retval)
869 struct proc *p;
870 void *v;
871 register_t *retval;
872 {
873 struct sys_chdir_args /* {
874 syscallarg(const char *) path;
875 } */ *uap = v;
876 struct cwdinfo *cwdi = p->p_cwdi;
877 int error;
878 struct nameidata nd;
879
880 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
881 SCARG(uap, path), p);
882 if ((error = change_dir(&nd, p)) != 0)
883 return (error);
884 vrele(cwdi->cwdi_cdir);
885 cwdi->cwdi_cdir = nd.ni_vp;
886 return (0);
887 }
888
889 /*
890 * Change notion of root (``/'') directory.
891 */
892 /* ARGSUSED */
893 int
894 sys_chroot(p, v, retval)
895 struct proc *p;
896 void *v;
897 register_t *retval;
898 {
899 struct sys_chroot_args /* {
900 syscallarg(const char *) path;
901 } */ *uap = v;
902 struct cwdinfo *cwdi = p->p_cwdi;
903 struct vnode *vp;
904 int error;
905 struct nameidata nd;
906
907 if ((error = suser(p->p_ucred, &p->p_acflag)) != 0)
908 return (error);
909 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
910 SCARG(uap, path), p);
911 if ((error = change_dir(&nd, p)) != 0)
912 return (error);
913 if (cwdi->cwdi_rdir != NULL)
914 vrele(cwdi->cwdi_rdir);
915 vp = nd.ni_vp;
916 cwdi->cwdi_rdir = vp;
917
918 /*
919 * Prevent escaping from chroot by putting the root under
920 * the working directory. Silently chdir to / if we aren't
921 * already there.
922 */
923 if (!vn_isunder(cwdi->cwdi_cdir, vp, p)) {
924 /*
925 * XXX would be more failsafe to change directory to a
926 * deadfs node here instead
927 */
928 vrele(cwdi->cwdi_cdir);
929 VREF(vp);
930 cwdi->cwdi_cdir = vp;
931 }
932
933 return (0);
934 }
935
936 /*
937 * Common routine for chroot and chdir.
938 */
939 static int
940 change_dir(ndp, p)
941 register struct nameidata *ndp;
942 struct proc *p;
943 {
944 struct vnode *vp;
945 int error;
946
947 if ((error = namei(ndp)) != 0)
948 return (error);
949 vp = ndp->ni_vp;
950 if (vp->v_type != VDIR)
951 error = ENOTDIR;
952 else
953 error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
954
955 if (error)
956 vput(vp);
957 else
958 VOP_UNLOCK(vp, 0);
959 return (error);
960 }
961
962 /*
963 * Check permissions, allocate an open file structure,
964 * and call the device open routine if any.
965 */
966 int
967 sys_open(p, v, retval)
968 struct proc *p;
969 void *v;
970 register_t *retval;
971 {
972 register struct sys_open_args /* {
973 syscallarg(const char *) path;
974 syscallarg(int) flags;
975 syscallarg(int) mode;
976 } */ *uap = v;
977 struct cwdinfo *cwdi = p->p_cwdi;
978 struct filedesc *fdp = p->p_fd;
979 struct file *fp;
980 struct vnode *vp;
981 int flags, cmode;
982 int type, indx, error;
983 struct flock lf;
984 struct nameidata nd;
985 extern struct fileops vnops;
986
987 flags = FFLAGS(SCARG(uap, flags));
988 if ((flags & (FREAD | FWRITE)) == 0)
989 return (EINVAL);
990 /* falloc() will use the file descriptor for us */
991 if ((error = falloc(p, &fp, &indx)) != 0)
992 return (error);
993 cmode = ((SCARG(uap, mode) &~ cwdi->cwdi_cmask) & ALLPERMS) &~ S_ISTXT;
994 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
995 p->p_dupfd = -indx - 1; /* XXX check for fdopen */
996 if ((error = vn_open(&nd, flags, cmode)) != 0) {
997 FILE_UNUSE(fp, p);
998 ffree(fp);
999 if ((error == ENODEV || error == ENXIO) &&
1000 p->p_dupfd >= 0 && /* XXX from fdopen */
1001 (error =
1002 dupfdopen(p, indx, p->p_dupfd, flags, error)) == 0) {
1003 *retval = indx;
1004 return (0);
1005 }
1006 if (error == ERESTART)
1007 error = EINTR;
1008 fdp->fd_ofiles[indx] = NULL;
1009 return (error);
1010 }
1011 p->p_dupfd = 0;
1012 vp = nd.ni_vp;
1013 fp->f_flag = flags & FMASK;
1014 fp->f_type = DTYPE_VNODE;
1015 fp->f_ops = &vnops;
1016 fp->f_data = (caddr_t)vp;
1017 if (flags & (O_EXLOCK | O_SHLOCK)) {
1018 lf.l_whence = SEEK_SET;
1019 lf.l_start = 0;
1020 lf.l_len = 0;
1021 if (flags & O_EXLOCK)
1022 lf.l_type = F_WRLCK;
1023 else
1024 lf.l_type = F_RDLCK;
1025 type = F_FLOCK;
1026 if ((flags & FNONBLOCK) == 0)
1027 type |= F_WAIT;
1028 VOP_UNLOCK(vp, 0);
1029 error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type);
1030 if (error) {
1031 (void) vn_close(vp, fp->f_flag, fp->f_cred, p);
1032 FILE_UNUSE(fp, p);
1033 ffree(fp);
1034 fdp->fd_ofiles[indx] = NULL;
1035 return (error);
1036 }
1037 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
1038 fp->f_flag |= FHASLOCK;
1039 }
1040 VOP_UNLOCK(vp, 0);
1041 *retval = indx;
1042 FILE_UNUSE(fp, p);
1043 return (0);
1044 }
1045
1046 /*
1047 * Get file handle system call
1048 */
1049 int
1050 sys_getfh(p, v, retval)
1051 struct proc *p;
1052 register void *v;
1053 register_t *retval;
1054 {
1055 register struct sys_getfh_args /* {
1056 syscallarg(char *) fname;
1057 syscallarg(fhandle_t *) fhp;
1058 } */ *uap = v;
1059 register struct vnode *vp;
1060 fhandle_t fh;
1061 int error;
1062 struct nameidata nd;
1063
1064 /*
1065 * Must be super user
1066 */
1067 error = suser(p->p_ucred, &p->p_acflag);
1068 if (error)
1069 return (error);
1070 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1071 SCARG(uap, fname), p);
1072 error = namei(&nd);
1073 if (error)
1074 return (error);
1075 vp = nd.ni_vp;
1076 memset((caddr_t)&fh, 0, sizeof(fh));
1077 fh.fh_fsid = vp->v_mount->mnt_stat.f_fsid;
1078 error = VFS_VPTOFH(vp, &fh.fh_fid);
1079 vput(vp);
1080 if (error)
1081 return (error);
1082 error = copyout((caddr_t)&fh, (caddr_t)SCARG(uap, fhp), sizeof (fh));
1083 return (error);
1084 }
1085
1086 /*
1087 * Open a file given a file handle.
1088 *
1089 * Check permissions, allocate an open file structure,
1090 * and call the device open routine if any.
1091 */
1092 int
1093 sys_fhopen(p, v, retval)
1094 struct proc *p;
1095 void *v;
1096 register_t *retval;
1097 {
1098 register struct sys_fhopen_args /* {
1099 syscallarg(const fhandle_t *) fhp;
1100 syscallarg(int) flags;
1101 } */ *uap = v;
1102 struct filedesc *fdp = p->p_fd;
1103 struct file *fp;
1104 struct vnode *vp = NULL;
1105 struct mount *mp;
1106 struct ucred *cred = p->p_ucred;
1107 int flags;
1108 struct file *nfp;
1109 int type, indx, error=0;
1110 struct flock lf;
1111 struct vattr va;
1112 fhandle_t fh;
1113 extern struct fileops vnops;
1114
1115 /*
1116 * Must be super user
1117 */
1118 if ((error = suser(p->p_ucred, &p->p_acflag)))
1119 return (error);
1120
1121 flags = FFLAGS(SCARG(uap, flags));
1122 if ((flags & (FREAD | FWRITE)) == 0)
1123 return (EINVAL);
1124 if ((flags & O_CREAT))
1125 return (EINVAL);
1126 /* falloc() will use the file descriptor for us */
1127 if ((error = falloc(p, &nfp, &indx)) != 0)
1128 return (error);
1129 fp = nfp;
1130 if ((error = copyin(SCARG(uap, fhp), &fh, sizeof(fhandle_t))) != 0)
1131 goto bad;
1132
1133 if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL) {
1134 error = ESTALE;
1135 goto bad;
1136 }
1137
1138 if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)) != 0) {
1139 vp = NULL; /* most likely unnecessary sanity for bad: */
1140 goto bad;
1141 }
1142
1143 /* Now do an effective vn_open */
1144
1145 if (vp->v_type == VSOCK) {
1146 error = EOPNOTSUPP;
1147 goto bad;
1148 }
1149 if (flags & FREAD) {
1150 if ((error = VOP_ACCESS(vp, VREAD, cred, p)) != 0)
1151 goto bad;
1152 }
1153 if (flags & (FWRITE | O_TRUNC)) {
1154 if (vp->v_type == VDIR) {
1155 error = EISDIR;
1156 goto bad;
1157 }
1158 if ((error = vn_writechk(vp)) != 0 ||
1159 (error = VOP_ACCESS(vp, VWRITE, cred, p)) != 0)
1160 goto bad;
1161 }
1162 if (flags & O_TRUNC) {
1163 VOP_UNLOCK(vp, 0); /* XXX */
1164 VOP_LEASE(vp, p, cred, LEASE_WRITE);
1165 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); /* XXX */
1166 VATTR_NULL(&va);
1167 va.va_size = 0;
1168 if ((error = VOP_SETATTR(vp, &va, cred, p)) != 0)
1169 goto bad;
1170 }
1171 if ((error = VOP_OPEN(vp, flags, cred, p)) != 0)
1172 goto bad;
1173 if (flags & FWRITE)
1174 vp->v_writecount++;
1175
1176 /* done with modified vn_open, now finish what sys_open does. */
1177
1178 fp->f_flag = flags & FMASK;
1179 fp->f_type = DTYPE_VNODE;
1180 fp->f_ops = &vnops;
1181 fp->f_data = (caddr_t)vp;
1182 if (flags & (O_EXLOCK | O_SHLOCK)) {
1183 lf.l_whence = SEEK_SET;
1184 lf.l_start = 0;
1185 lf.l_len = 0;
1186 if (flags & O_EXLOCK)
1187 lf.l_type = F_WRLCK;
1188 else
1189 lf.l_type = F_RDLCK;
1190 type = F_FLOCK;
1191 if ((flags & FNONBLOCK) == 0)
1192 type |= F_WAIT;
1193 VOP_UNLOCK(vp, 0);
1194 error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type);
1195 if (error) {
1196 (void) vn_close(vp, fp->f_flag, fp->f_cred, p);
1197 FILE_UNUSE(fp, p);
1198 ffree(fp);
1199 fdp->fd_ofiles[indx] = NULL;
1200 return (error);
1201 }
1202 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
1203 fp->f_flag |= FHASLOCK;
1204 }
1205 VOP_UNLOCK(vp, 0);
1206 *retval = indx;
1207 FILE_UNUSE(fp, p);
1208 return (0);
1209
1210 bad:
1211 FILE_UNUSE(fp, p);
1212 ffree(fp);
1213 fdp->fd_ofiles[indx] = NULL;
1214 if (vp != NULL)
1215 vput(vp);
1216 return (error);
1217 }
1218
1219 /* ARGSUSED */
1220 int
1221 sys_fhstat(p, v, retval)
1222 struct proc *p;
1223 void *v;
1224 register_t *retval;
1225 {
1226 register struct sys_fhstat_args /* {
1227 syscallarg(const fhandle_t *) fhp;
1228 syscallarg(struct stat *) sb;
1229 } */ *uap = v;
1230 struct stat sb;
1231 int error;
1232 fhandle_t fh;
1233 struct mount *mp;
1234 struct vnode *vp;
1235
1236 /*
1237 * Must be super user
1238 */
1239 if ((error = suser(p->p_ucred, &p->p_acflag)))
1240 return (error);
1241
1242 if ((error = copyin(SCARG(uap, fhp), &fh, sizeof(fhandle_t))) != 0)
1243 return (error);
1244
1245 if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL)
1246 return (ESTALE);
1247 if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)))
1248 return (error);
1249 error = vn_stat(vp, &sb, p);
1250 vput(vp);
1251 if (error)
1252 return (error);
1253 error = copyout(&sb, SCARG(uap, sb), sizeof(sb));
1254 return (error);
1255 }
1256
1257 /* ARGSUSED */
1258 int
1259 sys_fhstatfs(p, v, retval)
1260 struct proc *p;
1261 void *v;
1262 register_t *retval;
1263 {
1264 register struct sys_fhstatfs_args /*
1265 syscallarg(const fhandle_t *) fhp;
1266 syscallarg(struct statfs *) buf;
1267 } */ *uap = v;
1268 struct statfs sp;
1269 fhandle_t fh;
1270 struct mount *mp;
1271 struct vnode *vp;
1272 int error;
1273
1274 /*
1275 * Must be super user
1276 */
1277 if ((error = suser(p->p_ucred, &p->p_acflag)))
1278 return (error);
1279
1280 if ((error = copyin(SCARG(uap, fhp), &fh, sizeof(fhandle_t))) != 0)
1281 return (error);
1282
1283 if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL)
1284 return (ESTALE);
1285 if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)))
1286 return (error);
1287 mp = vp->v_mount;
1288 vput(vp);
1289 if ((error = VFS_STATFS(mp, &sp, p)) != 0)
1290 return (error);
1291 sp.f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
1292 return (copyout(&sp, SCARG(uap, buf), sizeof(sp)));
1293 }
1294
1295 /*
1296 * Create a special file.
1297 */
1298 /* ARGSUSED */
1299 int
1300 sys_mknod(p, v, retval)
1301 struct proc *p;
1302 void *v;
1303 register_t *retval;
1304 {
1305 register struct sys_mknod_args /* {
1306 syscallarg(const char *) path;
1307 syscallarg(int) mode;
1308 syscallarg(int) dev;
1309 } */ *uap = v;
1310 register struct vnode *vp;
1311 struct vattr vattr;
1312 int error;
1313 int whiteout = 0;
1314 struct nameidata nd;
1315
1316 if ((error = suser(p->p_ucred, &p->p_acflag)) != 0)
1317 return (error);
1318 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1319 if ((error = namei(&nd)) != 0)
1320 return (error);
1321 vp = nd.ni_vp;
1322 if (vp != NULL)
1323 error = EEXIST;
1324 else {
1325 VATTR_NULL(&vattr);
1326 vattr.va_mode =
1327 (SCARG(uap, mode) & ALLPERMS) &~ p->p_cwdi->cwdi_cmask;
1328 vattr.va_rdev = SCARG(uap, dev);
1329 whiteout = 0;
1330
1331 switch (SCARG(uap, mode) & S_IFMT) {
1332 case S_IFMT: /* used by badsect to flag bad sectors */
1333 vattr.va_type = VBAD;
1334 break;
1335 case S_IFCHR:
1336 vattr.va_type = VCHR;
1337 break;
1338 case S_IFBLK:
1339 vattr.va_type = VBLK;
1340 break;
1341 case S_IFWHT:
1342 whiteout = 1;
1343 break;
1344 default:
1345 error = EINVAL;
1346 break;
1347 }
1348 }
1349 if (!error) {
1350 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1351 if (whiteout) {
1352 error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, CREATE);
1353 if (error)
1354 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1355 vput(nd.ni_dvp);
1356 } else {
1357 error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp,
1358 &nd.ni_cnd, &vattr);
1359 }
1360 } else {
1361 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1362 if (nd.ni_dvp == vp)
1363 vrele(nd.ni_dvp);
1364 else
1365 vput(nd.ni_dvp);
1366 if (vp)
1367 vrele(vp);
1368 }
1369 return (error);
1370 }
1371
1372 /*
1373 * Create a named pipe.
1374 */
1375 /* ARGSUSED */
1376 int
1377 sys_mkfifo(p, v, retval)
1378 struct proc *p;
1379 void *v;
1380 register_t *retval;
1381 {
1382 register struct sys_mkfifo_args /* {
1383 syscallarg(const char *) path;
1384 syscallarg(int) mode;
1385 } */ *uap = v;
1386 struct vattr vattr;
1387 int error;
1388 struct nameidata nd;
1389
1390 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1391 if ((error = namei(&nd)) != 0)
1392 return (error);
1393 if (nd.ni_vp != NULL) {
1394 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1395 if (nd.ni_dvp == nd.ni_vp)
1396 vrele(nd.ni_dvp);
1397 else
1398 vput(nd.ni_dvp);
1399 vrele(nd.ni_vp);
1400 return (EEXIST);
1401 }
1402 VATTR_NULL(&vattr);
1403 vattr.va_type = VFIFO;
1404 vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_cwdi->cwdi_cmask;
1405 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1406 return (VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr));
1407 }
1408
1409 /*
1410 * Make a hard file link.
1411 */
1412 /* ARGSUSED */
1413 int
1414 sys_link(p, v, retval)
1415 struct proc *p;
1416 void *v;
1417 register_t *retval;
1418 {
1419 register struct sys_link_args /* {
1420 syscallarg(const char *) path;
1421 syscallarg(const char *) link;
1422 } */ *uap = v;
1423 register struct vnode *vp;
1424 struct nameidata nd;
1425 int error;
1426
1427 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1428 if ((error = namei(&nd)) != 0)
1429 return (error);
1430 vp = nd.ni_vp;
1431 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, link), p);
1432 if ((error = namei(&nd)) != 0)
1433 goto out;
1434 if (nd.ni_vp) {
1435 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1436 if (nd.ni_dvp == nd.ni_vp)
1437 vrele(nd.ni_dvp);
1438 else
1439 vput(nd.ni_dvp);
1440 vrele(nd.ni_vp);
1441 error = EEXIST;
1442 goto out;
1443 }
1444 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1445 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1446 error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd);
1447 out:
1448 vrele(vp);
1449 return (error);
1450 }
1451
1452 /*
1453 * Make a symbolic link.
1454 */
1455 /* ARGSUSED */
1456 int
1457 sys_symlink(p, v, retval)
1458 struct proc *p;
1459 void *v;
1460 register_t *retval;
1461 {
1462 register struct sys_symlink_args /* {
1463 syscallarg(const char *) path;
1464 syscallarg(const char *) link;
1465 } */ *uap = v;
1466 struct vattr vattr;
1467 char *path;
1468 int error;
1469 struct nameidata nd;
1470
1471 MALLOC(path, char *, MAXPATHLEN, M_NAMEI, M_WAITOK);
1472 error = copyinstr(SCARG(uap, path), path, MAXPATHLEN, NULL);
1473 if (error)
1474 goto out;
1475 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, link), p);
1476 if ((error = namei(&nd)) != 0)
1477 goto out;
1478 if (nd.ni_vp) {
1479 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1480 if (nd.ni_dvp == nd.ni_vp)
1481 vrele(nd.ni_dvp);
1482 else
1483 vput(nd.ni_dvp);
1484 vrele(nd.ni_vp);
1485 error = EEXIST;
1486 goto out;
1487 }
1488 VATTR_NULL(&vattr);
1489 vattr.va_mode = ACCESSPERMS &~ p->p_cwdi->cwdi_cmask;
1490 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1491 error = VOP_SYMLINK(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr, path);
1492 out:
1493 FREE(path, M_NAMEI);
1494 return (error);
1495 }
1496
1497 /*
1498 * Delete a whiteout from the filesystem.
1499 */
1500 /* ARGSUSED */
1501 int
1502 sys_undelete(p, v, retval)
1503 struct proc *p;
1504 void *v;
1505 register_t *retval;
1506 {
1507 register struct sys_undelete_args /* {
1508 syscallarg(const char *) path;
1509 } */ *uap = v;
1510 int error;
1511 struct nameidata nd;
1512
1513 NDINIT(&nd, DELETE, LOCKPARENT|DOWHITEOUT, UIO_USERSPACE,
1514 SCARG(uap, path), p);
1515 error = namei(&nd);
1516 if (error)
1517 return (error);
1518
1519 if (nd.ni_vp != NULLVP || !(nd.ni_cnd.cn_flags & ISWHITEOUT)) {
1520 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1521 if (nd.ni_dvp == nd.ni_vp)
1522 vrele(nd.ni_dvp);
1523 else
1524 vput(nd.ni_dvp);
1525 if (nd.ni_vp)
1526 vrele(nd.ni_vp);
1527 return (EEXIST);
1528 }
1529
1530 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1531 if ((error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, DELETE)) != 0)
1532 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1533 vput(nd.ni_dvp);
1534 return (error);
1535 }
1536
1537 /*
1538 * Delete a name from the filesystem.
1539 */
1540 /* ARGSUSED */
1541 int
1542 sys_unlink(p, v, retval)
1543 struct proc *p;
1544 void *v;
1545 register_t *retval;
1546 {
1547 struct sys_unlink_args /* {
1548 syscallarg(const char *) path;
1549 } */ *uap = v;
1550 register struct vnode *vp;
1551 int error;
1552 struct nameidata nd;
1553
1554 NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
1555 SCARG(uap, path), p);
1556 if ((error = namei(&nd)) != 0)
1557 return (error);
1558 vp = nd.ni_vp;
1559
1560 /*
1561 * The root of a mounted filesystem cannot be deleted.
1562 */
1563 if (vp->v_flag & VROOT) {
1564 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1565 if (nd.ni_dvp == vp)
1566 vrele(nd.ni_dvp);
1567 else
1568 vput(nd.ni_dvp);
1569 vput(vp);
1570 error = EBUSY;
1571 goto out;
1572 }
1573
1574 (void)uvm_vnp_uncache(vp);
1575
1576 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1577 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1578 error = VOP_REMOVE(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
1579 out:
1580 return (error);
1581 }
1582
1583 /*
1584 * Reposition read/write file offset.
1585 */
1586 int
1587 sys_lseek(p, v, retval)
1588 struct proc *p;
1589 void *v;
1590 register_t *retval;
1591 {
1592 register struct sys_lseek_args /* {
1593 syscallarg(int) fd;
1594 syscallarg(int) pad;
1595 syscallarg(off_t) offset;
1596 syscallarg(int) whence;
1597 } */ *uap = v;
1598 struct ucred *cred = p->p_ucred;
1599 register struct filedesc *fdp = p->p_fd;
1600 register struct file *fp;
1601 struct vnode *vp;
1602 struct vattr vattr;
1603 register off_t newoff;
1604 int error;
1605
1606 if ((u_int)SCARG(uap, fd) >= fdp->fd_nfiles ||
1607 (fp = fdp->fd_ofiles[SCARG(uap, fd)]) == NULL ||
1608 (fp->f_iflags & FIF_WANTCLOSE) != 0)
1609 return (EBADF);
1610
1611 FILE_USE(fp);
1612
1613 vp = (struct vnode *)fp->f_data;
1614 if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
1615 error = ESPIPE;
1616 goto out;
1617 }
1618
1619 switch (SCARG(uap, whence)) {
1620 case SEEK_CUR:
1621 newoff = fp->f_offset + SCARG(uap, offset);
1622 break;
1623 case SEEK_END:
1624 error = VOP_GETATTR(vp, &vattr, cred, p);
1625 if (error)
1626 goto out;
1627 newoff = SCARG(uap, offset) + vattr.va_size;
1628 break;
1629 case SEEK_SET:
1630 newoff = SCARG(uap, offset);
1631 break;
1632 default:
1633 error = EINVAL;
1634 goto out;
1635 }
1636 if ((error = VOP_SEEK(vp, fp->f_offset, newoff, cred)) != 0)
1637 goto out;
1638
1639 *(off_t *)retval = fp->f_offset = newoff;
1640 out:
1641 FILE_UNUSE(fp, p);
1642 return (error);
1643 }
1644
1645 /*
1646 * Positional read system call.
1647 */
1648 int
1649 sys_pread(p, v, retval)
1650 struct proc *p;
1651 void *v;
1652 register_t *retval;
1653 {
1654 struct sys_pread_args /* {
1655 syscallarg(int) fd;
1656 syscallarg(void *) buf;
1657 syscallarg(size_t) nbyte;
1658 syscallarg(off_t) offset;
1659 } */ *uap = v;
1660 struct filedesc *fdp = p->p_fd;
1661 struct file *fp;
1662 struct vnode *vp;
1663 off_t offset;
1664 int error, fd = SCARG(uap, fd);
1665
1666 if ((u_int)fd >= fdp->fd_nfiles ||
1667 (fp = fdp->fd_ofiles[fd]) == NULL ||
1668 (fp->f_iflags & FIF_WANTCLOSE) != 0 ||
1669 (fp->f_flag & FREAD) == 0)
1670 return (EBADF);
1671
1672 FILE_USE(fp);
1673
1674 vp = (struct vnode *)fp->f_data;
1675 if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
1676 error = ESPIPE;
1677 goto out;
1678 }
1679
1680 offset = SCARG(uap, offset);
1681
1682 /*
1683 * XXX This works because no file systems actually
1684 * XXX take any action on the seek operation.
1685 */
1686 if ((error = VOP_SEEK(vp, fp->f_offset, offset, fp->f_cred)) != 0)
1687 goto out;
1688
1689 /* dofileread() will unuse the descriptor for us */
1690 return (dofileread(p, fd, fp, SCARG(uap, buf), SCARG(uap, nbyte),
1691 &offset, 0, retval));
1692
1693 out:
1694 FILE_UNUSE(fp, p);
1695 return (error);
1696 }
1697
1698 /*
1699 * Positional scatter read system call.
1700 */
1701 int
1702 sys_preadv(p, v, retval)
1703 struct proc *p;
1704 void *v;
1705 register_t *retval;
1706 {
1707 struct sys_preadv_args /* {
1708 syscallarg(int) fd;
1709 syscallarg(const struct iovec *) iovp;
1710 syscallarg(int) iovcnt;
1711 syscallarg(off_t) offset;
1712 } */ *uap = v;
1713 struct filedesc *fdp = p->p_fd;
1714 struct file *fp;
1715 struct vnode *vp;
1716 off_t offset;
1717 int error, fd = SCARG(uap, fd);
1718
1719 if ((u_int)fd >= fdp->fd_nfiles ||
1720 (fp = fdp->fd_ofiles[fd]) == NULL ||
1721 (fp->f_iflags & FIF_WANTCLOSE) != 0 ||
1722 (fp->f_flag & FREAD) == 0)
1723 return (EBADF);
1724
1725 FILE_USE(fp);
1726
1727 vp = (struct vnode *)fp->f_data;
1728 if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
1729 error = ESPIPE;
1730 goto out;
1731 }
1732
1733 offset = SCARG(uap, offset);
1734
1735 /*
1736 * XXX This works because no file systems actually
1737 * XXX take any action on the seek operation.
1738 */
1739 if ((error = VOP_SEEK(vp, fp->f_offset, offset, fp->f_cred)) != 0)
1740 goto out;
1741
1742 /* dofilereadv() will unuse the descriptor for us */
1743 return (dofilereadv(p, fd, fp, SCARG(uap, iovp), SCARG(uap, iovcnt),
1744 &offset, 0, retval));
1745
1746 out:
1747 FILE_UNUSE(fp, p);
1748 return (error);
1749 }
1750
1751 /*
1752 * Positional write system call.
1753 */
1754 int
1755 sys_pwrite(p, v, retval)
1756 struct proc *p;
1757 void *v;
1758 register_t *retval;
1759 {
1760 struct sys_pwrite_args /* {
1761 syscallarg(int) fd;
1762 syscallarg(const void *) buf;
1763 syscallarg(size_t) nbyte;
1764 syscallarg(off_t) offset;
1765 } */ *uap = v;
1766 struct filedesc *fdp = p->p_fd;
1767 struct file *fp;
1768 struct vnode *vp;
1769 off_t offset;
1770 int error, fd = SCARG(uap, fd);
1771
1772 if ((u_int)fd >= fdp->fd_nfiles ||
1773 (fp = fdp->fd_ofiles[fd]) == NULL ||
1774 (fp->f_iflags & FIF_WANTCLOSE) != 0 ||
1775 (fp->f_flag & FWRITE) == 0)
1776 return (EBADF);
1777
1778 FILE_USE(fp);
1779
1780 vp = (struct vnode *)fp->f_data;
1781 if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
1782 error = ESPIPE;
1783 goto out;
1784 }
1785
1786 offset = SCARG(uap, offset);
1787
1788 /*
1789 * XXX This works because no file systems actually
1790 * XXX take any action on the seek operation.
1791 */
1792 if ((error = VOP_SEEK(vp, fp->f_offset, offset, fp->f_cred)) != 0)
1793 goto out;
1794
1795 /* dofilewrite() will unuse the descriptor for us */
1796 return (dofilewrite(p, fd, fp, SCARG(uap, buf), SCARG(uap, nbyte),
1797 &offset, 0, retval));
1798
1799 out:
1800 FILE_UNUSE(fp, p);
1801 return (error);
1802 }
1803
1804 /*
1805 * Positional gather write system call.
1806 */
1807 int
1808 sys_pwritev(p, v, retval)
1809 struct proc *p;
1810 void *v;
1811 register_t *retval;
1812 {
1813 struct sys_pwritev_args /* {
1814 syscallarg(int) fd;
1815 syscallarg(const struct iovec *) iovp;
1816 syscallarg(int) iovcnt;
1817 syscallarg(off_t) offset;
1818 } */ *uap = v;
1819 struct filedesc *fdp = p->p_fd;
1820 struct file *fp;
1821 struct vnode *vp;
1822 off_t offset;
1823 int error, fd = SCARG(uap, fd);
1824
1825 if ((u_int)fd >= fdp->fd_nfiles ||
1826 (fp = fdp->fd_ofiles[fd]) == NULL ||
1827 (fp->f_iflags & FIF_WANTCLOSE) != 0 ||
1828 (fp->f_flag & FWRITE) == 0)
1829 return (EBADF);
1830
1831 FILE_USE(fp);
1832
1833 vp = (struct vnode *)fp->f_data;
1834 if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
1835 error = ESPIPE;
1836 goto out;
1837 }
1838
1839 offset = SCARG(uap, offset);
1840
1841 /*
1842 * XXX This works because no file systems actually
1843 * XXX take any action on the seek operation.
1844 */
1845 if ((error = VOP_SEEK(vp, fp->f_offset, offset, fp->f_cred)) != 0)
1846 goto out;
1847
1848 /* dofilewritev() will unuse the descriptor for us */
1849 return (dofilewritev(p, fd, fp, SCARG(uap, iovp), SCARG(uap, iovcnt),
1850 &offset, 0, retval));
1851
1852 out:
1853 FILE_UNUSE(fp, p);
1854 return (error);
1855 }
1856
1857 /*
1858 * Check access permissions.
1859 */
1860 int
1861 sys_access(p, v, retval)
1862 struct proc *p;
1863 void *v;
1864 register_t *retval;
1865 {
1866 register struct sys_access_args /* {
1867 syscallarg(const char *) path;
1868 syscallarg(int) flags;
1869 } */ *uap = v;
1870 register struct ucred *cred = p->p_ucred;
1871 register struct vnode *vp;
1872 int error, flags, t_gid, t_uid;
1873 struct nameidata nd;
1874
1875 t_uid = cred->cr_uid;
1876 t_gid = cred->cr_gid;
1877 cred->cr_uid = p->p_cred->p_ruid;
1878 cred->cr_gid = p->p_cred->p_rgid;
1879 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1880 SCARG(uap, path), p);
1881 if ((error = namei(&nd)) != 0)
1882 goto out1;
1883 vp = nd.ni_vp;
1884
1885 /* Flags == 0 means only check for existence. */
1886 if (SCARG(uap, flags)) {
1887 flags = 0;
1888 if (SCARG(uap, flags) & R_OK)
1889 flags |= VREAD;
1890 if (SCARG(uap, flags) & W_OK)
1891 flags |= VWRITE;
1892 if (SCARG(uap, flags) & X_OK)
1893 flags |= VEXEC;
1894
1895 error = VOP_ACCESS(vp, flags, cred, p);
1896 if (!error && (flags & VWRITE))
1897 error = vn_writechk(vp);
1898 }
1899 vput(vp);
1900 out1:
1901 cred->cr_uid = t_uid;
1902 cred->cr_gid = t_gid;
1903 return (error);
1904 }
1905
1906 /*
1907 * Get file status; this version follows links.
1908 */
1909 /* ARGSUSED */
1910 int
1911 sys___stat13(p, v, retval)
1912 struct proc *p;
1913 void *v;
1914 register_t *retval;
1915 {
1916 register struct sys___stat13_args /* {
1917 syscallarg(const char *) path;
1918 syscallarg(struct stat *) ub;
1919 } */ *uap = v;
1920 struct stat sb;
1921 int error;
1922 struct nameidata nd;
1923
1924 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1925 SCARG(uap, path), p);
1926 if ((error = namei(&nd)) != 0)
1927 return (error);
1928 error = vn_stat(nd.ni_vp, &sb, p);
1929 vput(nd.ni_vp);
1930 if (error)
1931 return (error);
1932 error = copyout(&sb, SCARG(uap, ub), sizeof(sb));
1933 return (error);
1934 }
1935
1936 /*
1937 * Get file status; this version does not follow links.
1938 */
1939 /* ARGSUSED */
1940 int
1941 sys___lstat13(p, v, retval)
1942 struct proc *p;
1943 void *v;
1944 register_t *retval;
1945 {
1946 register struct sys___lstat13_args /* {
1947 syscallarg(const char *) path;
1948 syscallarg(struct stat *) ub;
1949 } */ *uap = v;
1950 struct stat sb;
1951 int error;
1952 struct nameidata nd;
1953
1954 NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF, UIO_USERSPACE,
1955 SCARG(uap, path), p);
1956 if ((error = namei(&nd)) != 0)
1957 return (error);
1958 error = vn_stat(nd.ni_vp, &sb, p);
1959 vput(nd.ni_vp);
1960 if (error)
1961 return (error);
1962 error = copyout(&sb, SCARG(uap, ub), sizeof(sb));
1963 return (error);
1964 }
1965
1966 /*
1967 * Get configurable pathname variables.
1968 */
1969 /* ARGSUSED */
1970 int
1971 sys_pathconf(p, v, retval)
1972 struct proc *p;
1973 void *v;
1974 register_t *retval;
1975 {
1976 register struct sys_pathconf_args /* {
1977 syscallarg(const char *) path;
1978 syscallarg(int) name;
1979 } */ *uap = v;
1980 int error;
1981 struct nameidata nd;
1982
1983 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1984 SCARG(uap, path), p);
1985 if ((error = namei(&nd)) != 0)
1986 return (error);
1987 error = VOP_PATHCONF(nd.ni_vp, SCARG(uap, name), retval);
1988 vput(nd.ni_vp);
1989 return (error);
1990 }
1991
1992 /*
1993 * Return target name of a symbolic link.
1994 */
1995 /* ARGSUSED */
1996 int
1997 sys_readlink(p, v, retval)
1998 struct proc *p;
1999 void *v;
2000 register_t *retval;
2001 {
2002 register struct sys_readlink_args /* {
2003 syscallarg(const char *) path;
2004 syscallarg(char *) buf;
2005 syscallarg(size_t) count;
2006 } */ *uap = v;
2007 register struct vnode *vp;
2008 struct iovec aiov;
2009 struct uio auio;
2010 int error;
2011 struct nameidata nd;
2012
2013 NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF, UIO_USERSPACE,
2014 SCARG(uap, path), p);
2015 if ((error = namei(&nd)) != 0)
2016 return (error);
2017 vp = nd.ni_vp;
2018 if (vp->v_type != VLNK)
2019 error = EINVAL;
2020 else if (!(vp->v_mount->mnt_flag & MNT_SYMPERM) ||
2021 (error = VOP_ACCESS(vp, VREAD, p->p_ucred, p)) == 0) {
2022 aiov.iov_base = SCARG(uap, buf);
2023 aiov.iov_len = SCARG(uap, count);
2024 auio.uio_iov = &aiov;
2025 auio.uio_iovcnt = 1;
2026 auio.uio_offset = 0;
2027 auio.uio_rw = UIO_READ;
2028 auio.uio_segflg = UIO_USERSPACE;
2029 auio.uio_procp = p;
2030 auio.uio_resid = SCARG(uap, count);
2031 error = VOP_READLINK(vp, &auio, p->p_ucred);
2032 }
2033 vput(vp);
2034 *retval = SCARG(uap, count) - auio.uio_resid;
2035 return (error);
2036 }
2037
2038 /*
2039 * Change flags of a file given a path name.
2040 */
2041 /* ARGSUSED */
2042 int
2043 sys_chflags(p, v, retval)
2044 struct proc *p;
2045 void *v;
2046 register_t *retval;
2047 {
2048 register struct sys_chflags_args /* {
2049 syscallarg(const char *) path;
2050 syscallarg(u_long) flags;
2051 } */ *uap = v;
2052 register struct vnode *vp;
2053 struct vattr vattr;
2054 int error;
2055 struct nameidata nd;
2056
2057 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2058 if ((error = namei(&nd)) != 0)
2059 return (error);
2060 vp = nd.ni_vp;
2061 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2062 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2063 /* Non-superusers cannot change the flags on devices, even if they
2064 own them. */
2065 if (suser(p->p_ucred, &p->p_acflag)) {
2066 if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)) != 0)
2067 goto out;
2068 if (vattr.va_type == VCHR || vattr.va_type == VBLK) {
2069 error = EINVAL;
2070 goto out;
2071 }
2072 }
2073 VATTR_NULL(&vattr);
2074 vattr.va_flags = SCARG(uap, flags);
2075 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2076 out:
2077 vput(vp);
2078 return (error);
2079 }
2080
2081 /*
2082 * Change flags of a file given a file descriptor.
2083 */
2084 /* ARGSUSED */
2085 int
2086 sys_fchflags(p, v, retval)
2087 struct proc *p;
2088 void *v;
2089 register_t *retval;
2090 {
2091 register struct sys_fchflags_args /* {
2092 syscallarg(int) fd;
2093 syscallarg(u_long) flags;
2094 } */ *uap = v;
2095 struct vattr vattr;
2096 struct vnode *vp;
2097 struct file *fp;
2098 int error;
2099
2100 /* getvnode() will use the descriptor for us */
2101 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2102 return (error);
2103 vp = (struct vnode *)fp->f_data;
2104 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2105 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2106 /* Non-superusers cannot change the flags on devices, even if they
2107 own them. */
2108 if (suser(p->p_ucred, &p->p_acflag)) {
2109 if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p))
2110 != 0)
2111 goto out;
2112 if (vattr.va_type == VCHR || vattr.va_type == VBLK) {
2113 error = EINVAL;
2114 goto out;
2115 }
2116 }
2117 VATTR_NULL(&vattr);
2118 vattr.va_flags = SCARG(uap, flags);
2119 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2120 out:
2121 VOP_UNLOCK(vp, 0);
2122 FILE_UNUSE(fp, p);
2123 return (error);
2124 }
2125
2126 /*
2127 * Change mode of a file given path name; this version follows links.
2128 */
2129 /* ARGSUSED */
2130 int
2131 sys_chmod(p, v, retval)
2132 struct proc *p;
2133 void *v;
2134 register_t *retval;
2135 {
2136 register struct sys_chmod_args /* {
2137 syscallarg(const char *) path;
2138 syscallarg(int) mode;
2139 } */ *uap = v;
2140 int error;
2141 struct nameidata nd;
2142
2143 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2144 if ((error = namei(&nd)) != 0)
2145 return (error);
2146
2147 error = change_mode(nd.ni_vp, SCARG(uap, mode), p);
2148
2149 vrele(nd.ni_vp);
2150 return (error);
2151 }
2152
2153 /*
2154 * Change mode of a file given a file descriptor.
2155 */
2156 /* ARGSUSED */
2157 int
2158 sys_fchmod(p, v, retval)
2159 struct proc *p;
2160 void *v;
2161 register_t *retval;
2162 {
2163 register struct sys_fchmod_args /* {
2164 syscallarg(int) fd;
2165 syscallarg(int) mode;
2166 } */ *uap = v;
2167 struct file *fp;
2168 int error;
2169
2170 /* getvnode() will use the descriptor for us */
2171 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2172 return (error);
2173
2174 error = change_mode((struct vnode *)fp->f_data, SCARG(uap, mode), p);
2175 FILE_UNUSE(fp, p);
2176 return (error);
2177 }
2178
2179 /*
2180 * Change mode of a file given path name; this version does not follow links.
2181 */
2182 /* ARGSUSED */
2183 int
2184 sys_lchmod(p, v, retval)
2185 struct proc *p;
2186 void *v;
2187 register_t *retval;
2188 {
2189 register struct sys_lchmod_args /* {
2190 syscallarg(const char *) path;
2191 syscallarg(int) mode;
2192 } */ *uap = v;
2193 int error;
2194 struct nameidata nd;
2195
2196 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2197 if ((error = namei(&nd)) != 0)
2198 return (error);
2199
2200 error = change_mode(nd.ni_vp, SCARG(uap, mode), p);
2201
2202 vrele(nd.ni_vp);
2203 return (error);
2204 }
2205
2206 /*
2207 * Common routine to set mode given a vnode.
2208 */
2209 static int
2210 change_mode(vp, mode, p)
2211 struct vnode *vp;
2212 int mode;
2213 struct proc *p;
2214 {
2215 struct vattr vattr;
2216 int error;
2217
2218 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2219 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2220 VATTR_NULL(&vattr);
2221 vattr.va_mode = mode & ALLPERMS;
2222 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2223 VOP_UNLOCK(vp, 0);
2224 return (error);
2225 }
2226
2227 /*
2228 * Set ownership given a path name; this version follows links.
2229 */
2230 /* ARGSUSED */
2231 int
2232 sys_chown(p, v, retval)
2233 struct proc *p;
2234 void *v;
2235 register_t *retval;
2236 {
2237 register struct sys_chown_args /* {
2238 syscallarg(const char *) path;
2239 syscallarg(uid_t) uid;
2240 syscallarg(gid_t) gid;
2241 } */ *uap = v;
2242 int error;
2243 struct nameidata nd;
2244
2245 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2246 if ((error = namei(&nd)) != 0)
2247 return (error);
2248
2249 error = change_owner(nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid), p, 0);
2250
2251 vrele(nd.ni_vp);
2252 return (error);
2253 }
2254
2255 /*
2256 * Set ownership given a path name; this version follows links.
2257 * Provides POSIX semantics.
2258 */
2259 /* ARGSUSED */
2260 int
2261 sys___posix_chown(p, v, retval)
2262 struct proc *p;
2263 void *v;
2264 register_t *retval;
2265 {
2266 register struct sys_chown_args /* {
2267 syscallarg(const char *) path;
2268 syscallarg(uid_t) uid;
2269 syscallarg(gid_t) gid;
2270 } */ *uap = v;
2271 int error;
2272 struct nameidata nd;
2273
2274 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2275 if ((error = namei(&nd)) != 0)
2276 return (error);
2277
2278 error = change_owner(nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid), p, 1);
2279
2280 vrele(nd.ni_vp);
2281 return (error);
2282 }
2283
2284 /*
2285 * Set ownership given a file descriptor.
2286 */
2287 /* ARGSUSED */
2288 int
2289 sys_fchown(p, v, retval)
2290 struct proc *p;
2291 void *v;
2292 register_t *retval;
2293 {
2294 register struct sys_fchown_args /* {
2295 syscallarg(int) fd;
2296 syscallarg(uid_t) uid;
2297 syscallarg(gid_t) gid;
2298 } */ *uap = v;
2299 int error;
2300 struct file *fp;
2301
2302 /* getvnode() will use the descriptor for us */
2303 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2304 return (error);
2305
2306 error = change_owner((struct vnode *)fp->f_data, SCARG(uap, uid),
2307 SCARG(uap, gid), p, 0);
2308 FILE_UNUSE(fp, p);
2309 return (error);
2310 }
2311
2312 /*
2313 * Set ownership given a file descriptor, providing POSIX/XPG semantics.
2314 */
2315 /* ARGSUSED */
2316 int
2317 sys___posix_fchown(p, v, retval)
2318 struct proc *p;
2319 void *v;
2320 register_t *retval;
2321 {
2322 register struct sys_fchown_args /* {
2323 syscallarg(int) fd;
2324 syscallarg(uid_t) uid;
2325 syscallarg(gid_t) gid;
2326 } */ *uap = v;
2327 int error;
2328 struct file *fp;
2329
2330 /* getvnode() will use the descriptor for us */
2331 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2332 return (error);
2333
2334 error = change_owner((struct vnode *)fp->f_data, SCARG(uap, uid),
2335 SCARG(uap, gid), p, 1);
2336 FILE_UNUSE(fp, p);
2337 return (error);
2338 }
2339
2340 /*
2341 * Set ownership given a path name; this version does not follow links.
2342 */
2343 /* ARGSUSED */
2344 int
2345 sys_lchown(p, v, retval)
2346 struct proc *p;
2347 void *v;
2348 register_t *retval;
2349 {
2350 register struct sys_lchown_args /* {
2351 syscallarg(const char *) path;
2352 syscallarg(uid_t) uid;
2353 syscallarg(gid_t) gid;
2354 } */ *uap = v;
2355 int error;
2356 struct nameidata nd;
2357
2358 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2359 if ((error = namei(&nd)) != 0)
2360 return (error);
2361
2362 error = change_owner(nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid), p, 0);
2363
2364 vrele(nd.ni_vp);
2365 return (error);
2366 }
2367
2368 /*
2369 * Set ownership given a path name; this version does not follow links.
2370 * Provides POSIX/XPG semantics.
2371 */
2372 /* ARGSUSED */
2373 int
2374 sys___posix_lchown(p, v, retval)
2375 struct proc *p;
2376 void *v;
2377 register_t *retval;
2378 {
2379 register struct sys_lchown_args /* {
2380 syscallarg(const char *) path;
2381 syscallarg(uid_t) uid;
2382 syscallarg(gid_t) gid;
2383 } */ *uap = v;
2384 int error;
2385 struct nameidata nd;
2386
2387 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2388 if ((error = namei(&nd)) != 0)
2389 return (error);
2390
2391 error = change_owner(nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid), p, 1);
2392
2393 vrele(nd.ni_vp);
2394 return (error);
2395 }
2396
2397 /*
2398 * Common routine to set ownership given a vnode.
2399 */
2400 static int
2401 change_owner(vp, uid, gid, p, posix_semantics)
2402 register struct vnode *vp;
2403 uid_t uid;
2404 gid_t gid;
2405 struct proc *p;
2406 int posix_semantics;
2407 {
2408 struct vattr vattr;
2409 mode_t newmode;
2410 int error;
2411
2412 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2413 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2414 if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)) != 0)
2415 goto out;
2416
2417 #define CHANGED(x) ((x) != -1)
2418 newmode = vattr.va_mode;
2419 if (posix_semantics) {
2420 /*
2421 * POSIX/XPG semantics: if the caller is not the super-user,
2422 * clear set-user-id and set-group-id bits. Both POSIX and
2423 * the XPG consider the behaviour for calls by the super-user
2424 * implementation-defined; we leave the set-user-id and set-
2425 * group-id settings intact in that case.
2426 */
2427 if (suser(p->p_ucred, NULL) != 0)
2428 newmode &= ~(S_ISUID | S_ISGID);
2429 } else {
2430 /*
2431 * NetBSD semantics: when changing owner and/or group,
2432 * clear the respective bit(s).
2433 */
2434 if (CHANGED(uid))
2435 newmode &= ~S_ISUID;
2436 if (CHANGED(gid))
2437 newmode &= ~S_ISGID;
2438 }
2439 /* Update va_mode iff altered. */
2440 if (vattr.va_mode == newmode)
2441 newmode = VNOVAL;
2442
2443 VATTR_NULL(&vattr);
2444 vattr.va_uid = CHANGED(uid) ? uid : VNOVAL;
2445 vattr.va_gid = CHANGED(gid) ? gid : VNOVAL;
2446 vattr.va_mode = newmode;
2447 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2448 #undef CHANGED
2449
2450 out:
2451 VOP_UNLOCK(vp, 0);
2452 return (error);
2453 }
2454
2455 /*
2456 * Set the access and modification times given a path name; this
2457 * version follows links.
2458 */
2459 /* ARGSUSED */
2460 int
2461 sys_utimes(p, v, retval)
2462 struct proc *p;
2463 void *v;
2464 register_t *retval;
2465 {
2466 register struct sys_utimes_args /* {
2467 syscallarg(const char *) path;
2468 syscallarg(const struct timeval *) tptr;
2469 } */ *uap = v;
2470 int error;
2471 struct nameidata nd;
2472
2473 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2474 if ((error = namei(&nd)) != 0)
2475 return (error);
2476
2477 error = change_utimes(nd.ni_vp, SCARG(uap, tptr), p);
2478
2479 vrele(nd.ni_vp);
2480 return (error);
2481 }
2482
2483 /*
2484 * Set the access and modification times given a file descriptor.
2485 */
2486 /* ARGSUSED */
2487 int
2488 sys_futimes(p, v, retval)
2489 struct proc *p;
2490 void *v;
2491 register_t *retval;
2492 {
2493 register struct sys_futimes_args /* {
2494 syscallarg(int) fd;
2495 syscallarg(const struct timeval *) tptr;
2496 } */ *uap = v;
2497 int error;
2498 struct file *fp;
2499
2500 /* getvnode() will use the descriptor for us */
2501 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2502 return (error);
2503
2504 error = change_utimes((struct vnode *)fp->f_data, SCARG(uap, tptr), p);
2505 FILE_UNUSE(fp, p);
2506 return (error);
2507 }
2508
2509 /*
2510 * Set the access and modification times given a path name; this
2511 * version does not follow links.
2512 */
2513 /* ARGSUSED */
2514 int
2515 sys_lutimes(p, v, retval)
2516 struct proc *p;
2517 void *v;
2518 register_t *retval;
2519 {
2520 register struct sys_lutimes_args /* {
2521 syscallarg(const char *) path;
2522 syscallarg(const struct timeval *) tptr;
2523 } */ *uap = v;
2524 int error;
2525 struct nameidata nd;
2526
2527 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2528 if ((error = namei(&nd)) != 0)
2529 return (error);
2530
2531 error = change_utimes(nd.ni_vp, SCARG(uap, tptr), p);
2532
2533 vrele(nd.ni_vp);
2534 return (error);
2535 }
2536
2537 /*
2538 * Common routine to set access and modification times given a vnode.
2539 */
2540 static int
2541 change_utimes(vp, tptr, p)
2542 struct vnode *vp;
2543 const struct timeval *tptr;
2544 struct proc *p;
2545 {
2546 struct timeval tv[2];
2547 struct vattr vattr;
2548 int error;
2549
2550 VATTR_NULL(&vattr);
2551 if (tptr == NULL) {
2552 microtime(&tv[0]);
2553 tv[1] = tv[0];
2554 vattr.va_vaflags |= VA_UTIMES_NULL;
2555 } else {
2556 error = copyin(tptr, tv, sizeof(tv));
2557 if (error)
2558 return (error);
2559 }
2560 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2561 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2562 vattr.va_atime.tv_sec = tv[0].tv_sec;
2563 vattr.va_atime.tv_nsec = tv[0].tv_usec * 1000;
2564 vattr.va_mtime.tv_sec = tv[1].tv_sec;
2565 vattr.va_mtime.tv_nsec = tv[1].tv_usec * 1000;
2566 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2567 VOP_UNLOCK(vp, 0);
2568 return (error);
2569 }
2570
2571 /*
2572 * Truncate a file given its path name.
2573 */
2574 /* ARGSUSED */
2575 int
2576 sys_truncate(p, v, retval)
2577 struct proc *p;
2578 void *v;
2579 register_t *retval;
2580 {
2581 register struct sys_truncate_args /* {
2582 syscallarg(const char *) path;
2583 syscallarg(int) pad;
2584 syscallarg(off_t) length;
2585 } */ *uap = v;
2586 register struct vnode *vp;
2587 struct vattr vattr;
2588 int error;
2589 struct nameidata nd;
2590
2591 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2592 if ((error = namei(&nd)) != 0)
2593 return (error);
2594 vp = nd.ni_vp;
2595 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2596 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2597 if (vp->v_type == VDIR)
2598 error = EISDIR;
2599 else if ((error = vn_writechk(vp)) == 0 &&
2600 (error = VOP_ACCESS(vp, VWRITE, p->p_ucred, p)) == 0) {
2601 VATTR_NULL(&vattr);
2602 vattr.va_size = SCARG(uap, length);
2603 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2604 }
2605 vput(vp);
2606 return (error);
2607 }
2608
2609 /*
2610 * Truncate a file given a file descriptor.
2611 */
2612 /* ARGSUSED */
2613 int
2614 sys_ftruncate(p, v, retval)
2615 struct proc *p;
2616 void *v;
2617 register_t *retval;
2618 {
2619 register struct sys_ftruncate_args /* {
2620 syscallarg(int) fd;
2621 syscallarg(int) pad;
2622 syscallarg(off_t) length;
2623 } */ *uap = v;
2624 struct vattr vattr;
2625 struct vnode *vp;
2626 struct file *fp;
2627 int error;
2628
2629 /* getvnode() will use the descriptor for us */
2630 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2631 return (error);
2632 if ((fp->f_flag & FWRITE) == 0) {
2633 error = EINVAL;
2634 goto out;
2635 }
2636 vp = (struct vnode *)fp->f_data;
2637 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2638 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2639 if (vp->v_type == VDIR)
2640 error = EISDIR;
2641 else if ((error = vn_writechk(vp)) == 0) {
2642 VATTR_NULL(&vattr);
2643 vattr.va_size = SCARG(uap, length);
2644 error = VOP_SETATTR(vp, &vattr, fp->f_cred, p);
2645 }
2646 VOP_UNLOCK(vp, 0);
2647 out:
2648 FILE_UNUSE(fp, p);
2649 return (error);
2650 }
2651
2652 /*
2653 * Sync an open file.
2654 */
2655 /* ARGSUSED */
2656 int
2657 sys_fsync(p, v, retval)
2658 struct proc *p;
2659 void *v;
2660 register_t *retval;
2661 {
2662 struct sys_fsync_args /* {
2663 syscallarg(int) fd;
2664 } */ *uap = v;
2665 register struct vnode *vp;
2666 struct file *fp;
2667 int error;
2668
2669 /* getvnode() will use the descriptor for us */
2670 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2671 return (error);
2672 vp = (struct vnode *)fp->f_data;
2673 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2674 error = VOP_FSYNC(vp, fp->f_cred, FSYNC_WAIT, p);
2675 if (error == 0 && bioops.io_fsync != NULL &&
2676 vp->v_mount && (vp->v_mount->mnt_flag & MNT_SOFTDEP))
2677 (*bioops.io_fsync)(vp);
2678 VOP_UNLOCK(vp, 0);
2679 FILE_UNUSE(fp, p);
2680 return (error);
2681 }
2682
2683 /*
2684 * Sync the data of an open file.
2685 */
2686 /* ARGSUSED */
2687 int
2688 sys_fdatasync(p, v, retval)
2689 struct proc *p;
2690 void *v;
2691 register_t *retval;
2692 {
2693 struct sys_fdatasync_args /* {
2694 syscallarg(int) fd;
2695 } */ *uap = v;
2696 struct vnode *vp;
2697 struct file *fp;
2698 int error;
2699
2700 /* getvnode() will use the descriptor for us */
2701 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2702 return (error);
2703 vp = (struct vnode *)fp->f_data;
2704 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2705 error = VOP_FSYNC(vp, fp->f_cred, FSYNC_WAIT|FSYNC_DATAONLY, p);
2706 if (error == 0 && bioops.io_fsync != NULL)
2707 (*bioops.io_fsync)(vp);
2708 VOP_UNLOCK(vp, 0);
2709 FILE_UNUSE(fp, p);
2710 return (error);
2711 }
2712
2713 /*
2714 * Rename files, (standard) BSD semantics frontend.
2715 */
2716 /* ARGSUSED */
2717 int
2718 sys_rename(p, v, retval)
2719 struct proc *p;
2720 void *v;
2721 register_t *retval;
2722 {
2723 register struct sys_rename_args /* {
2724 syscallarg(const char *) from;
2725 syscallarg(const char *) to;
2726 } */ *uap = v;
2727
2728 return (rename_files(SCARG(uap, from), SCARG(uap, to), p, 0));
2729 }
2730
2731 /*
2732 * Rename files, POSIX semantics frontend.
2733 */
2734 /* ARGSUSED */
2735 int
2736 sys___posix_rename(p, v, retval)
2737 struct proc *p;
2738 void *v;
2739 register_t *retval;
2740 {
2741 register struct sys___posix_rename_args /* {
2742 syscallarg(const char *) from;
2743 syscallarg(const char *) to;
2744 } */ *uap = v;
2745
2746 return (rename_files(SCARG(uap, from), SCARG(uap, to), p, 1));
2747 }
2748
2749 /*
2750 * Rename files. Source and destination must either both be directories,
2751 * or both not be directories. If target is a directory, it must be empty.
2752 * If `from' and `to' refer to the same object, the value of the `retain'
2753 * argument is used to determine whether `from' will be
2754 *
2755 * (retain == 0) deleted unless `from' and `to' refer to the same
2756 * object in the file system's name space (BSD).
2757 * (retain == 1) always retained (POSIX).
2758 */
2759 static int
2760 rename_files(from, to, p, retain)
2761 const char *from, *to;
2762 struct proc *p;
2763 int retain;
2764 {
2765 register struct vnode *tvp, *fvp, *tdvp;
2766 struct nameidata fromnd, tond;
2767 int error;
2768
2769 NDINIT(&fromnd, DELETE, WANTPARENT | SAVESTART, UIO_USERSPACE,
2770 from, p);
2771 if ((error = namei(&fromnd)) != 0)
2772 return (error);
2773 fvp = fromnd.ni_vp;
2774 NDINIT(&tond, RENAME, LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART,
2775 UIO_USERSPACE, to, p);
2776 if ((error = namei(&tond)) != 0) {
2777 VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
2778 vrele(fromnd.ni_dvp);
2779 vrele(fvp);
2780 goto out1;
2781 }
2782 tdvp = tond.ni_dvp;
2783 tvp = tond.ni_vp;
2784
2785 if (tvp != NULL) {
2786 if (fvp->v_type == VDIR && tvp->v_type != VDIR) {
2787 error = ENOTDIR;
2788 goto out;
2789 } else if (fvp->v_type != VDIR && tvp->v_type == VDIR) {
2790 error = EISDIR;
2791 goto out;
2792 }
2793 }
2794
2795 if (fvp == tdvp)
2796 error = EINVAL;
2797
2798 /*
2799 * Source and destination refer to the same object.
2800 */
2801 if (fvp == tvp) {
2802 if (retain)
2803 error = -1;
2804 else if (fromnd.ni_dvp == tdvp &&
2805 fromnd.ni_cnd.cn_namelen == tond.ni_cnd.cn_namelen &&
2806 !memcmp(fromnd.ni_cnd.cn_nameptr,
2807 tond.ni_cnd.cn_nameptr,
2808 fromnd.ni_cnd.cn_namelen))
2809 error = -1;
2810 }
2811
2812 out:
2813 if (!error) {
2814 VOP_LEASE(tdvp, p, p->p_ucred, LEASE_WRITE);
2815 if (fromnd.ni_dvp != tdvp)
2816 VOP_LEASE(fromnd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
2817 if (tvp) {
2818 (void)uvm_vnp_uncache(tvp);
2819 VOP_LEASE(tvp, p, p->p_ucred, LEASE_WRITE);
2820 }
2821 error = VOP_RENAME(fromnd.ni_dvp, fromnd.ni_vp, &fromnd.ni_cnd,
2822 tond.ni_dvp, tond.ni_vp, &tond.ni_cnd);
2823 } else {
2824 VOP_ABORTOP(tond.ni_dvp, &tond.ni_cnd);
2825 if (tdvp == tvp)
2826 vrele(tdvp);
2827 else
2828 vput(tdvp);
2829 if (tvp)
2830 vput(tvp);
2831 VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
2832 vrele(fromnd.ni_dvp);
2833 vrele(fvp);
2834 }
2835 vrele(tond.ni_startdir);
2836 FREE(tond.ni_cnd.cn_pnbuf, M_NAMEI);
2837 out1:
2838 if (fromnd.ni_startdir)
2839 vrele(fromnd.ni_startdir);
2840 FREE(fromnd.ni_cnd.cn_pnbuf, M_NAMEI);
2841 return (error == -1 ? 0 : error);
2842 }
2843
2844 /*
2845 * Make a directory file.
2846 */
2847 /* ARGSUSED */
2848 int
2849 sys_mkdir(p, v, retval)
2850 struct proc *p;
2851 void *v;
2852 register_t *retval;
2853 {
2854 register struct sys_mkdir_args /* {
2855 syscallarg(const char *) path;
2856 syscallarg(int) mode;
2857 } */ *uap = v;
2858 register struct vnode *vp;
2859 struct vattr vattr;
2860 int error;
2861 struct nameidata nd;
2862
2863 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
2864 if ((error = namei(&nd)) != 0)
2865 return (error);
2866 vp = nd.ni_vp;
2867 if (vp != NULL) {
2868 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
2869 if (nd.ni_dvp == vp)
2870 vrele(nd.ni_dvp);
2871 else
2872 vput(nd.ni_dvp);
2873 vrele(vp);
2874 return (EEXIST);
2875 }
2876 VATTR_NULL(&vattr);
2877 vattr.va_type = VDIR;
2878 vattr.va_mode =
2879 (SCARG(uap, mode) & ACCESSPERMS) &~ p->p_cwdi->cwdi_cmask;
2880 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
2881 error = VOP_MKDIR(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
2882 if (!error)
2883 vput(nd.ni_vp);
2884 return (error);
2885 }
2886
2887 /*
2888 * Remove a directory file.
2889 */
2890 /* ARGSUSED */
2891 int
2892 sys_rmdir(p, v, retval)
2893 struct proc *p;
2894 void *v;
2895 register_t *retval;
2896 {
2897 struct sys_rmdir_args /* {
2898 syscallarg(const char *) path;
2899 } */ *uap = v;
2900 register struct vnode *vp;
2901 int error;
2902 struct nameidata nd;
2903
2904 NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
2905 SCARG(uap, path), p);
2906 if ((error = namei(&nd)) != 0)
2907 return (error);
2908 vp = nd.ni_vp;
2909 if (vp->v_type != VDIR) {
2910 error = ENOTDIR;
2911 goto out;
2912 }
2913 /*
2914 * No rmdir "." please.
2915 */
2916 if (nd.ni_dvp == vp) {
2917 error = EINVAL;
2918 goto out;
2919 }
2920 /*
2921 * The root of a mounted filesystem cannot be deleted.
2922 */
2923 if (vp->v_flag & VROOT)
2924 error = EBUSY;
2925 out:
2926 if (!error) {
2927 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
2928 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2929 error = VOP_RMDIR(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
2930 } else {
2931 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
2932 if (nd.ni_dvp == vp)
2933 vrele(nd.ni_dvp);
2934 else
2935 vput(nd.ni_dvp);
2936 vput(vp);
2937 }
2938 return (error);
2939 }
2940
2941 /*
2942 * Read a block of directory entries in a file system independent format.
2943 */
2944 int
2945 sys_getdents(p, v, retval)
2946 struct proc *p;
2947 void *v;
2948 register_t *retval;
2949 {
2950 register struct sys_getdents_args /* {
2951 syscallarg(int) fd;
2952 syscallarg(char *) buf;
2953 syscallarg(size_t) count;
2954 } */ *uap = v;
2955 struct file *fp;
2956 int error, done;
2957
2958 /* getvnode() will use the descriptor for us */
2959 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2960 return (error);
2961 if ((fp->f_flag & FREAD) == 0) {
2962 error = EBADF;
2963 goto out;
2964 }
2965 error = vn_readdir(fp, SCARG(uap, buf), UIO_USERSPACE,
2966 SCARG(uap, count), &done, p, 0, 0);
2967 *retval = done;
2968 out:
2969 FILE_UNUSE(fp, p);
2970 return (error);
2971 }
2972
2973 /*
2974 * Set the mode mask for creation of filesystem nodes.
2975 */
2976 int
2977 sys_umask(p, v, retval)
2978 struct proc *p;
2979 void *v;
2980 register_t *retval;
2981 {
2982 struct sys_umask_args /* {
2983 syscallarg(mode_t) newmask;
2984 } */ *uap = v;
2985 struct cwdinfo *cwdi;
2986
2987 cwdi = p->p_cwdi;
2988 *retval = cwdi->cwdi_cmask;
2989 cwdi->cwdi_cmask = SCARG(uap, newmask) & ALLPERMS;
2990 return (0);
2991 }
2992
2993 /*
2994 * Void all references to file by ripping underlying filesystem
2995 * away from vnode.
2996 */
2997 /* ARGSUSED */
2998 int
2999 sys_revoke(p, v, retval)
3000 struct proc *p;
3001 void *v;
3002 register_t *retval;
3003 {
3004 register struct sys_revoke_args /* {
3005 syscallarg(const char *) path;
3006 } */ *uap = v;
3007 register struct vnode *vp;
3008 struct vattr vattr;
3009 int error;
3010 struct nameidata nd;
3011
3012 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
3013 if ((error = namei(&nd)) != 0)
3014 return (error);
3015 vp = nd.ni_vp;
3016 if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)) != 0)
3017 goto out;
3018 if (p->p_ucred->cr_uid != vattr.va_uid &&
3019 (error = suser(p->p_ucred, &p->p_acflag)) != 0)
3020 goto out;
3021 if (vp->v_usecount > 1 || (vp->v_flag & (VALIASED | VLAYER)))
3022 VOP_REVOKE(vp, REVOKEALL);
3023 out:
3024 vrele(vp);
3025 return (error);
3026 }
3027
3028 /*
3029 * Convert a user file descriptor to a kernel file entry.
3030 */
3031 int
3032 getvnode(fdp, fd, fpp)
3033 struct filedesc *fdp;
3034 int fd;
3035 struct file **fpp;
3036 {
3037 struct vnode *vp;
3038 struct file *fp;
3039
3040 if ((u_int)fd >= fdp->fd_nfiles ||
3041 (fp = fdp->fd_ofiles[fd]) == NULL ||
3042 (fp->f_iflags & FIF_WANTCLOSE) != 0)
3043 return (EBADF);
3044
3045 FILE_USE(fp);
3046
3047 if (fp->f_type != DTYPE_VNODE) {
3048 FILE_UNUSE(fp, NULL);
3049 return (EINVAL);
3050 }
3051
3052 vp = (struct vnode *)fp->f_data;
3053 if (vp->v_type == VBAD) {
3054 FILE_UNUSE(fp, NULL);
3055 return (EBADF);
3056 }
3057
3058 *fpp = fp;
3059 return (0);
3060 }
3061