vfs_syscalls.c revision 1.164.2.1 1 /* $NetBSD: vfs_syscalls.c,v 1.164.2.1 2001/03/05 22:49:48 nathanw 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/lwp.h>
56 #include <sys/proc.h>
57 #include <sys/uio.h>
58 #include <sys/malloc.h>
59 #include <sys/dirent.h>
60 #include <sys/syscallargs.h>
61
62 #include <miscfs/genfs/genfs.h>
63 #include <miscfs/syncfs/syncfs.h>
64
65 #include <uvm/uvm_extern.h>
66
67 #include <sys/sysctl.h>
68
69 static int change_dir __P((struct nameidata *, struct proc *));
70 static int change_flags __P((struct vnode *, u_long, 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
80 int dovfsusermount = 0;
81
82 /*
83 * Virtual File System System Calls
84 */
85
86 /*
87 * Mount a file system.
88 */
89
90 /*
91 * This table is used to maintain compatibility with 4.3BSD
92 * and NetBSD 0.9 mount syscalls. Note, the order is important!
93 *
94 * Also note that not all of these had actual numbers in 4.3BSD
95 * or NetBSD 0.9!
96 */
97 const char *mountcompatnames[] = {
98 NULL, /* 0 = MOUNT_NONE */
99 MOUNT_FFS, /* 1 */
100 MOUNT_NFS, /* 2 */
101 MOUNT_MFS, /* 3 */
102 MOUNT_MSDOS, /* 4 */
103 MOUNT_LFS, /* 5 */
104 NULL, /* 6 = MOUNT_LOFS */
105 MOUNT_FDESC, /* 7 */
106 MOUNT_PORTAL, /* 8 */
107 MOUNT_NULL, /* 9 */
108 MOUNT_UMAP, /* 10 */
109 MOUNT_KERNFS, /* 11 */
110 MOUNT_PROCFS, /* 12 */
111 MOUNT_AFS, /* 13 */
112 MOUNT_CD9660, /* 14 = MOUNT_ISOFS */
113 MOUNT_UNION, /* 15 */
114 MOUNT_ADOSFS, /* 16 */
115 MOUNT_EXT2FS, /* 17 */
116 MOUNT_CODA, /* 18 */
117 MOUNT_FILECORE, /* 19 */
118 MOUNT_NTFS, /* 20 */
119 };
120 const int nmountcompatnames = sizeof(mountcompatnames) /
121 sizeof(mountcompatnames[0]);
122
123 /* ARGSUSED */
124 int
125 sys_mount(l, v, retval)
126 struct lwp *l;
127 void *v;
128 register_t *retval;
129 {
130 struct sys_mount_args /* {
131 syscallarg(const char *) type;
132 syscallarg(const char *) path;
133 syscallarg(int) flags;
134 syscallarg(void *) data;
135 } */ *uap = v;
136 struct proc *p = l->l_proc;
137 struct vnode *vp;
138 struct mount *mp;
139 int error, flag = 0;
140 char fstypename[MFSNAMELEN];
141 struct vattr va;
142 struct nameidata nd;
143 struct vfsops *vfs;
144
145 if (dovfsusermount == 0 && (error = suser(p->p_ucred, &p->p_acflag)))
146 return (error);
147 /*
148 * Get vnode to be covered
149 */
150 NDINIT(&nd, LOOKUP, FOLLOW , UIO_USERSPACE,
151 SCARG(uap, path), p);
152 if ((error = namei(&nd)) != 0)
153 return (error);
154 vp = nd.ni_vp;
155 /*
156 * A lookup in VFS_MOUNT might result in an attempt to
157 * lock this vnode again, so make the lock resursive.
158 */
159 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY | LK_SETRECURSE);
160 if (SCARG(uap, flags) & MNT_UPDATE) {
161 if ((vp->v_flag & VROOT) == 0) {
162 vput(vp);
163 return (EINVAL);
164 }
165 mp = vp->v_mount;
166 flag = mp->mnt_flag;
167 vfs = mp->mnt_op;
168 /*
169 * We only allow the filesystem to be reloaded if it
170 * is currently mounted read-only.
171 */
172 if ((SCARG(uap, flags) & MNT_RELOAD) &&
173 ((mp->mnt_flag & MNT_RDONLY) == 0)) {
174 vput(vp);
175 return (EOPNOTSUPP); /* Needs translation */
176 }
177 /*
178 * In "highly secure" mode, don't let the caller do anything
179 * but downgrade a filesystem from read-write to read-only.
180 * (see also below; MNT_UPDATE is required.)
181 */
182 if (securelevel >= 2 &&
183 (SCARG(uap, flags) !=
184 (mp->mnt_flag | MNT_RDONLY |
185 MNT_RELOAD | MNT_FORCE | MNT_UPDATE))) {
186 vput(vp);
187 return (EPERM);
188 }
189 mp->mnt_flag |=
190 SCARG(uap, flags) & (MNT_RELOAD | MNT_FORCE | MNT_UPDATE);
191 /*
192 * Only root, or the user that did the original mount is
193 * permitted to update it.
194 */
195 if (mp->mnt_stat.f_owner != p->p_ucred->cr_uid &&
196 (error = suser(p->p_ucred, &p->p_acflag)) != 0) {
197 vput(vp);
198 return (error);
199 }
200 /*
201 * Do not allow NFS export by non-root users. For non-root
202 * users, silently enforce MNT_NOSUID and MNT_NODEV, and
203 * MNT_NOEXEC if mount point is already MNT_NOEXEC.
204 */
205 if (p->p_ucred->cr_uid != 0) {
206 if (SCARG(uap, flags) & MNT_EXPORTED) {
207 vput(vp);
208 return (EPERM);
209 }
210 SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
211 if (flag & MNT_NOEXEC)
212 SCARG(uap, flags) |= MNT_NOEXEC;
213 }
214 if (vfs_busy(mp, LK_NOWAIT, 0)) {
215 vput(vp);
216 return (EPERM);
217 }
218 VOP_UNLOCK(vp, 0);
219 goto update;
220 } else {
221 if (securelevel >= 2) {
222 vput(vp);
223 return (EPERM);
224 }
225 }
226 /*
227 * If the user is not root, ensure that they own the directory
228 * onto which we are attempting to mount.
229 */
230 if ((error = VOP_GETATTR(vp, &va, p->p_ucred, p)) != 0 ||
231 (va.va_uid != p->p_ucred->cr_uid &&
232 (error = suser(p->p_ucred, &p->p_acflag)) != 0)) {
233 vput(vp);
234 return (error);
235 }
236 /*
237 * Do not allow NFS export by non-root users. For non-root users,
238 * silently enforce MNT_NOSUID and MNT_NODEV, and MNT_NOEXEC if the
239 * mount point is already MNT_NOEXEC.
240 */
241 if (p->p_ucred->cr_uid != 0) {
242 if (SCARG(uap, flags) & MNT_EXPORTED) {
243 vput(vp);
244 return (EPERM);
245 }
246 SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
247 if (vp->v_mount->mnt_flag & MNT_NOEXEC)
248 SCARG(uap, flags) |= MNT_NOEXEC;
249 }
250 if ((error = vinvalbuf(vp, V_SAVE, p->p_ucred, p, 0, 0)) != 0)
251 return (error);
252 if (vp->v_type != VDIR) {
253 vput(vp);
254 return (ENOTDIR);
255 }
256 error = copyinstr(SCARG(uap, type), fstypename, MFSNAMELEN, NULL);
257 if (error) {
258 #if defined(COMPAT_09) || defined(COMPAT_43)
259 /*
260 * Historically filesystem types were identified by number.
261 * If we get an integer for the filesystem type instead of a
262 * string, we check to see if it matches one of the historic
263 * filesystem types.
264 */
265 u_long fsindex = (u_long)SCARG(uap, type);
266 if (fsindex >= nmountcompatnames ||
267 mountcompatnames[fsindex] == NULL) {
268 vput(vp);
269 return (ENODEV);
270 }
271 strncpy(fstypename, mountcompatnames[fsindex], MFSNAMELEN);
272 #else
273 vput(vp);
274 return (error);
275 #endif
276 }
277 #ifdef COMPAT_10
278 /* Accept `ufs' as an alias for `ffs'. */
279 if (!strncmp(fstypename, "ufs", MFSNAMELEN))
280 strncpy(fstypename, "ffs", MFSNAMELEN);
281 #endif
282 if ((vfs = vfs_getopsbyname(fstypename)) == NULL) {
283 vput(vp);
284 return (ENODEV);
285 }
286 if (vp->v_mountedhere != NULL) {
287 vput(vp);
288 return (EBUSY);
289 }
290
291 /*
292 * Allocate and initialize the file system.
293 */
294 mp = (struct mount *)malloc((u_long)sizeof(struct mount),
295 M_MOUNT, M_WAITOK);
296 memset((char *)mp, 0, (u_long)sizeof(struct mount));
297 lockinit(&mp->mnt_lock, PVFS, "vfslock", 0, 0);
298 (void)vfs_busy(mp, LK_NOWAIT, 0);
299 mp->mnt_op = vfs;
300 vfs->vfs_refcount++;
301 mp->mnt_vnodecovered = vp;
302 mp->mnt_stat.f_owner = p->p_ucred->cr_uid;
303 mp->mnt_unmounter = NULL;
304 update:
305 /*
306 * Set the mount level flags.
307 */
308 if (SCARG(uap, flags) & MNT_RDONLY)
309 mp->mnt_flag |= MNT_RDONLY;
310 else if (mp->mnt_flag & MNT_RDONLY)
311 mp->mnt_flag |= MNT_WANTRDWR;
312 mp->mnt_flag &=~ (MNT_NOSUID | MNT_NOEXEC | MNT_NODEV |
313 MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | MNT_NOCOREDUMP |
314 MNT_NOATIME | MNT_NODEVMTIME | MNT_SYMPERM | MNT_SOFTDEP);
315 mp->mnt_flag |= SCARG(uap, flags) & (MNT_NOSUID | MNT_NOEXEC |
316 MNT_NODEV | MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC |
317 MNT_NOCOREDUMP | MNT_NOATIME | MNT_NODEVMTIME | MNT_SYMPERM |
318 MNT_SOFTDEP);
319 /*
320 * Mount the filesystem.
321 */
322 error = VFS_MOUNT(mp, SCARG(uap, path), SCARG(uap, data), &nd, p);
323 if (mp->mnt_flag & MNT_UPDATE) {
324 vrele(vp);
325 if (mp->mnt_flag & MNT_WANTRDWR)
326 mp->mnt_flag &= ~MNT_RDONLY;
327 mp->mnt_flag &=~
328 (MNT_UPDATE | MNT_RELOAD | MNT_FORCE | MNT_WANTRDWR);
329 if (error)
330 mp->mnt_flag = flag;
331 if ((mp->mnt_flag & (MNT_RDONLY | MNT_ASYNC)) == 0) {
332 if (mp->mnt_syncer == NULL)
333 error = vfs_allocate_syncvnode(mp);
334 } else {
335 if (mp->mnt_syncer != NULL)
336 vfs_deallocate_syncvnode(mp);
337 }
338 vfs_unbusy(mp);
339 return (error);
340 }
341 /*
342 * Put the new filesystem on the mount list after root.
343 */
344 cache_purge(vp);
345 if (!error) {
346 vp->v_mountedhere = mp;
347 simple_lock(&mountlist_slock);
348 CIRCLEQ_INSERT_TAIL(&mountlist, mp, mnt_list);
349 simple_unlock(&mountlist_slock);
350 checkdirs(vp);
351 VOP_UNLOCK(vp, 0);
352 if ((mp->mnt_flag & (MNT_RDONLY | MNT_ASYNC)) == 0)
353 error = vfs_allocate_syncvnode(mp);
354 vfs_unbusy(mp);
355 (void) VFS_STATFS(mp, &mp->mnt_stat, p);
356 if ((error = VFS_START(mp, 0, p)))
357 vrele(vp);
358 } else {
359 vp->v_mountedhere = (struct mount *)0;
360 vfs->vfs_refcount--;
361 vfs_unbusy(mp);
362 free((caddr_t)mp, M_MOUNT);
363 vput(vp);
364 }
365 return (error);
366 }
367
368 /*
369 * Scan all active processes to see if any of them have a current
370 * or root directory onto which the new filesystem has just been
371 * mounted. If so, replace them with the new mount point.
372 */
373 void
374 checkdirs(olddp)
375 struct vnode *olddp;
376 {
377 struct cwdinfo *cwdi;
378 struct vnode *newdp;
379 struct proc *p;
380
381 if (olddp->v_usecount == 1)
382 return;
383 if (VFS_ROOT(olddp->v_mountedhere, &newdp))
384 panic("mount: lost mount");
385 proclist_lock_read();
386 for (p = allproc.lh_first; p != 0; p = p->p_list.le_next) {
387 cwdi = p->p_cwdi;
388 if (cwdi->cwdi_cdir == olddp) {
389 vrele(cwdi->cwdi_cdir);
390 VREF(newdp);
391 cwdi->cwdi_cdir = newdp;
392 }
393 if (cwdi->cwdi_rdir == olddp) {
394 vrele(cwdi->cwdi_rdir);
395 VREF(newdp);
396 cwdi->cwdi_rdir = newdp;
397 }
398 }
399 proclist_unlock_read();
400 if (rootvnode == olddp) {
401 vrele(rootvnode);
402 VREF(newdp);
403 rootvnode = newdp;
404 }
405 vput(newdp);
406 }
407
408 /*
409 * Unmount a file system.
410 *
411 * Note: unmount takes a path to the vnode mounted on as argument,
412 * not special file (as before).
413 */
414 /* ARGSUSED */
415 int
416 sys_unmount(l, v, retval)
417 struct lwp *l;
418 void *v;
419 register_t *retval;
420 {
421 struct sys_unmount_args /* {
422 syscallarg(const char *) path;
423 syscallarg(int) flags;
424 } */ *uap = v;
425 struct proc *p = l->l_proc;
426 struct vnode *vp;
427 struct mount *mp;
428 int error;
429 struct nameidata nd;
430
431 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
432 SCARG(uap, path), p);
433 if ((error = namei(&nd)) != 0)
434 return (error);
435 vp = nd.ni_vp;
436 mp = vp->v_mount;
437
438 /*
439 * Only root, or the user that did the original mount is
440 * permitted to unmount this filesystem.
441 */
442 if ((mp->mnt_stat.f_owner != p->p_ucred->cr_uid) &&
443 (error = suser(p->p_ucred, &p->p_acflag)) != 0) {
444 vput(vp);
445 return (error);
446 }
447
448 /*
449 * Don't allow unmounting the root file system.
450 */
451 if (mp->mnt_flag & MNT_ROOTFS) {
452 vput(vp);
453 return (EINVAL);
454 }
455
456 /*
457 * Must be the root of the filesystem
458 */
459 if ((vp->v_flag & VROOT) == 0) {
460 vput(vp);
461 return (EINVAL);
462 }
463 vput(vp);
464
465 if (vfs_busy(mp, 0, 0))
466 return (EBUSY);
467
468 return (dounmount(mp, SCARG(uap, flags), p));
469 }
470
471 /*
472 * Do the actual file system unmount. File system is assumed to have been
473 * marked busy by the caller.
474 */
475 int
476 dounmount(mp, flags, p)
477 struct mount *mp;
478 int flags;
479 struct proc *p;
480 {
481 struct vnode *coveredvp;
482 int error;
483 int async;
484 int used_syncer;
485
486 simple_lock(&mountlist_slock);
487 vfs_unbusy(mp);
488 used_syncer = (mp->mnt_syncer != NULL);
489
490 /*
491 * XXX Freeze syncer. This should really be done on a mountpoint
492 * basis, but especially the softdep code possibly called from
493 * the syncer doesn't exactly work on a per-mountpoint basis,
494 * so the softdep code would become a maze of vfs_busy calls.
495 */
496 if (used_syncer)
497 lockmgr(&syncer_lock, LK_EXCLUSIVE, NULL);
498
499 mp->mnt_flag |= MNT_UNMOUNT;
500 mp->mnt_unmounter = p;
501 lockmgr(&mp->mnt_lock, LK_DRAIN | LK_INTERLOCK, &mountlist_slock);
502 if (mp->mnt_flag & MNT_EXPUBLIC)
503 vfs_setpublicfs(NULL, NULL, NULL);
504 async = mp->mnt_flag & MNT_ASYNC;
505 mp->mnt_flag &= ~MNT_ASYNC;
506 cache_purgevfs(mp); /* remove cache entries for this file sys */
507 if (mp->mnt_syncer != NULL)
508 vfs_deallocate_syncvnode(mp);
509 if (((mp->mnt_flag & MNT_RDONLY) ||
510 (error = VFS_SYNC(mp, MNT_WAIT, p->p_ucred, p)) == 0) ||
511 (flags & MNT_FORCE))
512 error = VFS_UNMOUNT(mp, flags, p);
513 simple_lock(&mountlist_slock);
514 if (error) {
515 if ((mp->mnt_flag & (MNT_RDONLY | MNT_ASYNC)) == 0)
516 (void) vfs_allocate_syncvnode(mp);
517 mp->mnt_flag &= ~MNT_UNMOUNT;
518 mp->mnt_unmounter = NULL;
519 mp->mnt_flag |= async;
520 lockmgr(&mp->mnt_lock, LK_RELEASE | LK_INTERLOCK | LK_REENABLE,
521 &mountlist_slock);
522 if (used_syncer)
523 lockmgr(&syncer_lock, LK_RELEASE, NULL);
524 while (mp->mnt_wcnt > 0) {
525 wakeup((caddr_t)mp);
526 tsleep(&mp->mnt_wcnt, PVFS, "mntwcnt1", 0);
527 }
528 return (error);
529 }
530 CIRCLEQ_REMOVE(&mountlist, mp, mnt_list);
531 if ((coveredvp = mp->mnt_vnodecovered) != NULLVP) {
532 coveredvp->v_mountedhere = NULL;
533 vrele(coveredvp);
534 }
535 mp->mnt_op->vfs_refcount--;
536 if (mp->mnt_vnodelist.lh_first != NULL)
537 panic("unmount: dangling vnode");
538 mp->mnt_flag |= MNT_GONE;
539 lockmgr(&mp->mnt_lock, LK_RELEASE | LK_INTERLOCK, &mountlist_slock);
540 if (used_syncer)
541 lockmgr(&syncer_lock, LK_RELEASE, NULL);
542 while(mp->mnt_wcnt > 0) {
543 wakeup((caddr_t)mp);
544 tsleep(&mp->mnt_wcnt, PVFS, "mntwcnt2", 0);
545 }
546 free((caddr_t)mp, M_MOUNT);
547 return (0);
548 }
549
550 /*
551 * Sync each mounted filesystem.
552 */
553 #ifdef DEBUG
554 int syncprt = 0;
555 struct ctldebug debug0 = { "syncprt", &syncprt };
556 #endif
557
558 /* ARGSUSED */
559 int
560 sys_sync(l, v, retval)
561 struct lwp *l;
562 void *v;
563 register_t *retval;
564 {
565 struct mount *mp, *nmp;
566 int asyncflag;
567 struct proc *p = l->l_proc;
568
569 simple_lock(&mountlist_slock);
570 for (mp = mountlist.cqh_last; mp != (void *)&mountlist; mp = nmp) {
571 if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock)) {
572 nmp = mp->mnt_list.cqe_prev;
573 continue;
574 }
575 if ((mp->mnt_flag & MNT_RDONLY) == 0) {
576 asyncflag = mp->mnt_flag & MNT_ASYNC;
577 mp->mnt_flag &= ~MNT_ASYNC;
578 VFS_SYNC(mp, MNT_NOWAIT, p->p_ucred, p);
579 if (asyncflag)
580 mp->mnt_flag |= MNT_ASYNC;
581 }
582 simple_lock(&mountlist_slock);
583 nmp = mp->mnt_list.cqe_prev;
584 vfs_unbusy(mp);
585
586 }
587 simple_unlock(&mountlist_slock);
588 #ifdef DEBUG
589 if (syncprt)
590 vfs_bufstats();
591 #endif /* DEBUG */
592 return (0);
593 }
594
595 /*
596 * Change filesystem quotas.
597 */
598 /* ARGSUSED */
599 int
600 sys_quotactl(l, v, retval)
601 struct lwp *l;
602 void *v;
603 register_t *retval;
604 {
605 struct sys_quotactl_args /* {
606 syscallarg(const char *) path;
607 syscallarg(int) cmd;
608 syscallarg(int) uid;
609 syscallarg(caddr_t) arg;
610 } */ *uap = v;
611 struct proc *p = l->l_proc;
612 struct mount *mp;
613 int error;
614 struct nameidata nd;
615
616 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
617 if ((error = namei(&nd)) != 0)
618 return (error);
619 mp = nd.ni_vp->v_mount;
620 vrele(nd.ni_vp);
621 return (VFS_QUOTACTL(mp, SCARG(uap, cmd), SCARG(uap, uid),
622 SCARG(uap, arg), p));
623 }
624
625 /*
626 * Get filesystem statistics.
627 */
628 /* ARGSUSED */
629 int
630 sys_statfs(l, v, retval)
631 struct lwp *l;
632 void *v;
633 register_t *retval;
634 {
635 struct sys_statfs_args /* {
636 syscallarg(const char *) path;
637 syscallarg(struct statfs *) buf;
638 } */ *uap = v;
639 struct proc *p = l->l_proc;
640 struct mount *mp;
641 struct statfs *sp;
642 int error;
643 struct nameidata nd;
644
645 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
646 if ((error = namei(&nd)) != 0)
647 return (error);
648 mp = nd.ni_vp->v_mount;
649 sp = &mp->mnt_stat;
650 vrele(nd.ni_vp);
651 if ((error = VFS_STATFS(mp, sp, p)) != 0)
652 return (error);
653 sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
654 sp->f_oflags = sp->f_flags & 0xffff;
655 return (copyout(sp, SCARG(uap, buf), sizeof(*sp)));
656 }
657
658 /*
659 * Get filesystem statistics.
660 */
661 /* ARGSUSED */
662 int
663 sys_fstatfs(l, v, retval)
664 struct lwp *l;
665 void *v;
666 register_t *retval;
667 {
668 struct sys_fstatfs_args /* {
669 syscallarg(int) fd;
670 syscallarg(struct statfs *) buf;
671 } */ *uap = v;
672 struct proc *p = l->l_proc;
673 struct file *fp;
674 struct mount *mp;
675 struct statfs *sp;
676 int error;
677
678 /* getvnode() will use the descriptor for us */
679 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
680 return (error);
681 mp = ((struct vnode *)fp->f_data)->v_mount;
682 sp = &mp->mnt_stat;
683 if ((error = VFS_STATFS(mp, sp, p)) != 0)
684 goto out;
685 sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
686 sp->f_oflags = sp->f_flags & 0xffff;
687 error = copyout(sp, SCARG(uap, buf), sizeof(*sp));
688 out:
689 FILE_UNUSE(fp, p);
690 return (error);
691 }
692
693 /*
694 * Get statistics on all filesystems.
695 */
696 int
697 sys_getfsstat(l, v, retval)
698 struct lwp *l;
699 void *v;
700 register_t *retval;
701 {
702 struct sys_getfsstat_args /* {
703 syscallarg(struct statfs *) buf;
704 syscallarg(long) bufsize;
705 syscallarg(int) flags;
706 } */ *uap = v;
707 struct proc *p = l->l_proc;
708 struct mount *mp, *nmp;
709 struct statfs *sp;
710 caddr_t sfsp;
711 long count, maxcount, error;
712
713 maxcount = SCARG(uap, bufsize) / sizeof(struct statfs);
714 sfsp = (caddr_t)SCARG(uap, buf);
715 simple_lock(&mountlist_slock);
716 count = 0;
717 for (mp = mountlist.cqh_first; mp != (void *)&mountlist; mp = nmp) {
718 if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock)) {
719 nmp = mp->mnt_list.cqe_next;
720 continue;
721 }
722 if (sfsp && count < maxcount) {
723 sp = &mp->mnt_stat;
724 /*
725 * If MNT_NOWAIT or MNT_LAZY is specified, do not
726 * refresh the fsstat cache. MNT_WAIT or MNT_LAXY
727 * overrides MNT_NOWAIT.
728 */
729 if (SCARG(uap, flags) != MNT_NOWAIT &&
730 SCARG(uap, flags) != MNT_LAZY &&
731 (SCARG(uap, flags) == MNT_WAIT ||
732 SCARG(uap, flags) == 0) &&
733 (error = VFS_STATFS(mp, sp, p)) != 0) {
734 simple_lock(&mountlist_slock);
735 nmp = mp->mnt_list.cqe_next;
736 vfs_unbusy(mp);
737 continue;
738 }
739 sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
740 sp->f_oflags = sp->f_flags & 0xffff;
741 error = copyout(sp, sfsp, sizeof(*sp));
742 if (error) {
743 vfs_unbusy(mp);
744 return (error);
745 }
746 sfsp += sizeof(*sp);
747 }
748 count++;
749 simple_lock(&mountlist_slock);
750 nmp = mp->mnt_list.cqe_next;
751 vfs_unbusy(mp);
752 }
753 simple_unlock(&mountlist_slock);
754 if (sfsp && count > maxcount)
755 *retval = maxcount;
756 else
757 *retval = count;
758 return (0);
759 }
760
761 /*
762 * Change current working directory to a given file descriptor.
763 */
764 /* ARGSUSED */
765 int
766 sys_fchdir(l, v, retval)
767 struct lwp *l;
768 void *v;
769 register_t *retval;
770 {
771 struct sys_fchdir_args /* {
772 syscallarg(int) fd;
773 } */ *uap = v;
774 struct proc *p = l->l_proc;
775 struct filedesc *fdp = p->p_fd;
776 struct cwdinfo *cwdi = p->p_cwdi;
777 struct vnode *vp, *tdp;
778 struct mount *mp;
779 struct file *fp;
780 int error;
781
782 /* getvnode() will use the descriptor for us */
783 if ((error = getvnode(fdp, SCARG(uap, fd), &fp)) != 0)
784 return (error);
785 vp = (struct vnode *)fp->f_data;
786
787 VREF(vp);
788 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
789 if (vp->v_type != VDIR)
790 error = ENOTDIR;
791 else
792 error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
793 while (!error && (mp = vp->v_mountedhere) != NULL) {
794 if (vfs_busy(mp, 0, 0))
795 continue;
796 error = VFS_ROOT(mp, &tdp);
797 vfs_unbusy(mp);
798 if (error)
799 break;
800 vput(vp);
801 vp = tdp;
802 }
803 if (error) {
804 vput(vp);
805 goto out;
806 }
807 VOP_UNLOCK(vp, 0);
808
809 /*
810 * Disallow changing to a directory not under the process's
811 * current root directory (if there is one).
812 */
813 if (cwdi->cwdi_rdir && !vn_isunder(vp, NULL, p)) {
814 vrele(vp);
815 error = EPERM; /* operation not permitted */
816 goto out;
817 }
818
819 vrele(cwdi->cwdi_cdir);
820 cwdi->cwdi_cdir = vp;
821 out:
822 FILE_UNUSE(fp, p);
823 return (error);
824 }
825
826 /*
827 * Change this process's notion of the root directory to a given file descriptor.
828 */
829
830 int
831 sys_fchroot(l, v, retval)
832 struct lwp *l;
833 void *v;
834 register_t *retval;
835 {
836 struct sys_fchroot_args *uap = v;
837 struct proc *p = l->l_proc;
838 struct filedesc *fdp = p->p_fd;
839 struct cwdinfo *cwdi = p->p_cwdi;
840 struct vnode *vp;
841 struct file *fp;
842 int error;
843
844 if ((error = suser(p->p_ucred, &p->p_acflag)) != 0)
845 return error;
846 /* getvnode() will use the descriptor for us */
847 if ((error = getvnode(fdp, SCARG(uap, fd), &fp)) != 0)
848 return error;
849 vp = (struct vnode *) fp->f_data;
850 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
851 if (vp->v_type != VDIR)
852 error = ENOTDIR;
853 else
854 error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
855 VOP_UNLOCK(vp, 0);
856 if (error)
857 goto out;
858 VREF(vp);
859
860 /*
861 * Prevent escaping from chroot by putting the root under
862 * the working directory. Silently chdir to / if we aren't
863 * already there.
864 */
865 if (!vn_isunder(cwdi->cwdi_cdir, vp, p)) {
866 /*
867 * XXX would be more failsafe to change directory to a
868 * deadfs node here instead
869 */
870 vrele(cwdi->cwdi_cdir);
871 VREF(vp);
872 cwdi->cwdi_cdir = vp;
873 }
874
875 if (cwdi->cwdi_rdir != NULL)
876 vrele(cwdi->cwdi_rdir);
877 cwdi->cwdi_rdir = vp;
878 out:
879 FILE_UNUSE(fp, p);
880 return (error);
881 }
882
883
884
885 /*
886 * Change current working directory (``.'').
887 */
888 /* ARGSUSED */
889 int
890 sys_chdir(l, v, retval)
891 struct lwp *l;
892 void *v;
893 register_t *retval;
894 {
895 struct sys_chdir_args /* {
896 syscallarg(const char *) path;
897 } */ *uap = v;
898 struct proc *p = l->l_proc;
899 struct cwdinfo *cwdi = p->p_cwdi;
900 int error;
901 struct nameidata nd;
902
903 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
904 SCARG(uap, path), p);
905 if ((error = change_dir(&nd, p)) != 0)
906 return (error);
907 vrele(cwdi->cwdi_cdir);
908 cwdi->cwdi_cdir = nd.ni_vp;
909 return (0);
910 }
911
912 /*
913 * Change notion of root (``/'') directory.
914 */
915 /* ARGSUSED */
916 int
917 sys_chroot(l, v, retval)
918 struct lwp *l;
919 void *v;
920 register_t *retval;
921 {
922 struct sys_chroot_args /* {
923 syscallarg(const char *) path;
924 } */ *uap = v;
925 struct proc *p = l->l_proc;
926 struct cwdinfo *cwdi = p->p_cwdi;
927 struct vnode *vp;
928 int error;
929 struct nameidata nd;
930
931 if ((error = suser(p->p_ucred, &p->p_acflag)) != 0)
932 return (error);
933 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
934 SCARG(uap, path), p);
935 if ((error = change_dir(&nd, p)) != 0)
936 return (error);
937 if (cwdi->cwdi_rdir != NULL)
938 vrele(cwdi->cwdi_rdir);
939 vp = nd.ni_vp;
940 cwdi->cwdi_rdir = vp;
941
942 /*
943 * Prevent escaping from chroot by putting the root under
944 * the working directory. Silently chdir to / if we aren't
945 * already there.
946 */
947 if (!vn_isunder(cwdi->cwdi_cdir, vp, p)) {
948 /*
949 * XXX would be more failsafe to change directory to a
950 * deadfs node here instead
951 */
952 vrele(cwdi->cwdi_cdir);
953 VREF(vp);
954 cwdi->cwdi_cdir = vp;
955 }
956
957 return (0);
958 }
959
960 /*
961 * Common routine for chroot and chdir.
962 */
963 static int
964 change_dir(ndp, p)
965 struct nameidata *ndp;
966 struct proc *p;
967 {
968 struct vnode *vp;
969 int error;
970
971 if ((error = namei(ndp)) != 0)
972 return (error);
973 vp = ndp->ni_vp;
974 if (vp->v_type != VDIR)
975 error = ENOTDIR;
976 else
977 error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
978
979 if (error)
980 vput(vp);
981 else
982 VOP_UNLOCK(vp, 0);
983 return (error);
984 }
985
986 /*
987 * Check permissions, allocate an open file structure,
988 * and call the device open routine if any.
989 */
990 int
991 sys_open(l, v, retval)
992 struct lwp *l;
993 void *v;
994 register_t *retval;
995 {
996 struct sys_open_args /* {
997 syscallarg(const char *) path;
998 syscallarg(int) flags;
999 syscallarg(int) mode;
1000 } */ *uap = v;
1001 struct proc *p = l->l_proc;
1002 struct cwdinfo *cwdi = p->p_cwdi;
1003 struct filedesc *fdp = p->p_fd;
1004 struct file *fp;
1005 struct vnode *vp;
1006 int flags, cmode;
1007 int type, indx, error;
1008 struct flock lf;
1009 struct nameidata nd;
1010
1011 flags = FFLAGS(SCARG(uap, flags));
1012 if ((flags & (FREAD | FWRITE)) == 0)
1013 return (EINVAL);
1014 /* falloc() will use the file descriptor for us */
1015 if ((error = falloc(p, &fp, &indx)) != 0)
1016 return (error);
1017 cmode = ((SCARG(uap, mode) &~ cwdi->cwdi_cmask) & ALLPERMS) &~ S_ISTXT;
1018 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1019 p->p_dupfd = -indx - 1; /* XXX check for fdopen */
1020 if ((error = vn_open(&nd, flags, cmode)) != 0) {
1021 FILE_UNUSE(fp, p);
1022 ffree(fp);
1023 if ((error == ENODEV || error == ENXIO) &&
1024 p->p_dupfd >= 0 && /* XXX from fdopen */
1025 (error =
1026 dupfdopen(p, indx, p->p_dupfd, flags, error)) == 0) {
1027 *retval = indx;
1028 return (0);
1029 }
1030 if (error == ERESTART)
1031 error = EINTR;
1032 fdremove(fdp, indx);
1033 return (error);
1034 }
1035 p->p_dupfd = 0;
1036 vp = nd.ni_vp;
1037 fp->f_flag = flags & FMASK;
1038 fp->f_type = DTYPE_VNODE;
1039 fp->f_ops = &vnops;
1040 fp->f_data = (caddr_t)vp;
1041 if (flags & (O_EXLOCK | O_SHLOCK)) {
1042 lf.l_whence = SEEK_SET;
1043 lf.l_start = 0;
1044 lf.l_len = 0;
1045 if (flags & O_EXLOCK)
1046 lf.l_type = F_WRLCK;
1047 else
1048 lf.l_type = F_RDLCK;
1049 type = F_FLOCK;
1050 if ((flags & FNONBLOCK) == 0)
1051 type |= F_WAIT;
1052 VOP_UNLOCK(vp, 0);
1053 error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type);
1054 if (error) {
1055 (void) vn_close(vp, fp->f_flag, fp->f_cred, p);
1056 FILE_UNUSE(fp, p);
1057 ffree(fp);
1058 fdremove(fdp, indx);
1059 return (error);
1060 }
1061 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
1062 fp->f_flag |= FHASLOCK;
1063 }
1064 VOP_UNLOCK(vp, 0);
1065 *retval = indx;
1066 FILE_UNUSE(fp, p);
1067 return (0);
1068 }
1069
1070 /*
1071 * Get file handle system call
1072 */
1073 int
1074 sys_getfh(l, v, retval)
1075 struct lwp *l;
1076 void *v;
1077 register_t *retval;
1078 {
1079 struct sys_getfh_args /* {
1080 syscallarg(char *) fname;
1081 syscallarg(fhandle_t *) fhp;
1082 } */ *uap = v;
1083 struct proc *p = l->l_proc;
1084 struct vnode *vp;
1085 fhandle_t fh;
1086 int error;
1087 struct nameidata nd;
1088
1089 /*
1090 * Must be super user
1091 */
1092 error = suser(p->p_ucred, &p->p_acflag);
1093 if (error)
1094 return (error);
1095 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1096 SCARG(uap, fname), p);
1097 error = namei(&nd);
1098 if (error)
1099 return (error);
1100 vp = nd.ni_vp;
1101 memset((caddr_t)&fh, 0, sizeof(fh));
1102 fh.fh_fsid = vp->v_mount->mnt_stat.f_fsid;
1103 error = VFS_VPTOFH(vp, &fh.fh_fid);
1104 vput(vp);
1105 if (error)
1106 return (error);
1107 error = copyout((caddr_t)&fh, (caddr_t)SCARG(uap, fhp), sizeof (fh));
1108 return (error);
1109 }
1110
1111 /*
1112 * Open a file given a file handle.
1113 *
1114 * Check permissions, allocate an open file structure,
1115 * and call the device open routine if any.
1116 */
1117 int
1118 sys_fhopen(l, v, retval)
1119 struct lwp *l;
1120 void *v;
1121 register_t *retval;
1122 {
1123 struct sys_fhopen_args /* {
1124 syscallarg(const fhandle_t *) fhp;
1125 syscallarg(int) flags;
1126 } */ *uap = v;
1127 struct proc *p = l->l_proc;
1128 struct filedesc *fdp = p->p_fd;
1129 struct file *fp;
1130 struct vnode *vp = NULL;
1131 struct mount *mp;
1132 struct ucred *cred = p->p_ucred;
1133 int flags;
1134 struct file *nfp;
1135 int type, indx, error=0;
1136 struct flock lf;
1137 struct vattr va;
1138 fhandle_t fh;
1139
1140 /*
1141 * Must be super user
1142 */
1143 if ((error = suser(p->p_ucred, &p->p_acflag)))
1144 return (error);
1145
1146 flags = FFLAGS(SCARG(uap, flags));
1147 if ((flags & (FREAD | FWRITE)) == 0)
1148 return (EINVAL);
1149 if ((flags & O_CREAT))
1150 return (EINVAL);
1151 /* falloc() will use the file descriptor for us */
1152 if ((error = falloc(p, &nfp, &indx)) != 0)
1153 return (error);
1154 fp = nfp;
1155 if ((error = copyin(SCARG(uap, fhp), &fh, sizeof(fhandle_t))) != 0)
1156 goto bad;
1157
1158 if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL) {
1159 error = ESTALE;
1160 goto bad;
1161 }
1162
1163 if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)) != 0) {
1164 vp = NULL; /* most likely unnecessary sanity for bad: */
1165 goto bad;
1166 }
1167
1168 /* Now do an effective vn_open */
1169
1170 if (vp->v_type == VSOCK) {
1171 error = EOPNOTSUPP;
1172 goto bad;
1173 }
1174 if (flags & FREAD) {
1175 if ((error = VOP_ACCESS(vp, VREAD, cred, p)) != 0)
1176 goto bad;
1177 }
1178 if (flags & (FWRITE | O_TRUNC)) {
1179 if (vp->v_type == VDIR) {
1180 error = EISDIR;
1181 goto bad;
1182 }
1183 if ((error = vn_writechk(vp)) != 0 ||
1184 (error = VOP_ACCESS(vp, VWRITE, cred, p)) != 0)
1185 goto bad;
1186 }
1187 if (flags & O_TRUNC) {
1188 VOP_UNLOCK(vp, 0); /* XXX */
1189 VOP_LEASE(vp, p, cred, LEASE_WRITE);
1190 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); /* XXX */
1191 VATTR_NULL(&va);
1192 va.va_size = 0;
1193 if ((error = VOP_SETATTR(vp, &va, cred, p)) != 0)
1194 goto bad;
1195 }
1196 if ((error = VOP_OPEN(vp, flags, cred, p)) != 0)
1197 goto bad;
1198 if (vp->v_type == VREG &&
1199 uvn_attach(vp, flags & FWRITE ? VM_PROT_WRITE : 0) == NULL) {
1200 error = EIO;
1201 goto bad;
1202 }
1203 if (flags & FWRITE)
1204 vp->v_writecount++;
1205
1206 /* done with modified vn_open, now finish what sys_open does. */
1207
1208 fp->f_flag = flags & FMASK;
1209 fp->f_type = DTYPE_VNODE;
1210 fp->f_ops = &vnops;
1211 fp->f_data = (caddr_t)vp;
1212 if (flags & (O_EXLOCK | O_SHLOCK)) {
1213 lf.l_whence = SEEK_SET;
1214 lf.l_start = 0;
1215 lf.l_len = 0;
1216 if (flags & O_EXLOCK)
1217 lf.l_type = F_WRLCK;
1218 else
1219 lf.l_type = F_RDLCK;
1220 type = F_FLOCK;
1221 if ((flags & FNONBLOCK) == 0)
1222 type |= F_WAIT;
1223 VOP_UNLOCK(vp, 0);
1224 error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type);
1225 if (error) {
1226 (void) vn_close(vp, fp->f_flag, fp->f_cred, p);
1227 FILE_UNUSE(fp, p);
1228 ffree(fp);
1229 fdremove(fdp, indx);
1230 return (error);
1231 }
1232 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
1233 fp->f_flag |= FHASLOCK;
1234 }
1235 VOP_UNLOCK(vp, 0);
1236 *retval = indx;
1237 FILE_UNUSE(fp, p);
1238 return (0);
1239
1240 bad:
1241 FILE_UNUSE(fp, p);
1242 ffree(fp);
1243 fdremove(fdp, indx);
1244 if (vp != NULL)
1245 vput(vp);
1246 return (error);
1247 }
1248
1249 /* ARGSUSED */
1250 int
1251 sys_fhstat(l, v, retval)
1252 struct lwp *l;
1253 void *v;
1254 register_t *retval;
1255 {
1256 struct sys_fhstat_args /* {
1257 syscallarg(const fhandle_t *) fhp;
1258 syscallarg(struct stat *) sb;
1259 } */ *uap = v;
1260 struct proc *p = l->l_proc;
1261 struct stat sb;
1262 int error;
1263 fhandle_t fh;
1264 struct mount *mp;
1265 struct vnode *vp;
1266
1267 /*
1268 * Must be super user
1269 */
1270 if ((error = suser(p->p_ucred, &p->p_acflag)))
1271 return (error);
1272
1273 if ((error = copyin(SCARG(uap, fhp), &fh, sizeof(fhandle_t))) != 0)
1274 return (error);
1275
1276 if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL)
1277 return (ESTALE);
1278 if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)))
1279 return (error);
1280 error = vn_stat(vp, &sb, p);
1281 vput(vp);
1282 if (error)
1283 return (error);
1284 error = copyout(&sb, SCARG(uap, sb), sizeof(sb));
1285 return (error);
1286 }
1287
1288 /* ARGSUSED */
1289 int
1290 sys_fhstatfs(l, v, retval)
1291 struct lwp *l;
1292 void *v;
1293 register_t *retval;
1294 {
1295 struct sys_fhstatfs_args /*
1296 syscallarg(const fhandle_t *) fhp;
1297 syscallarg(struct statfs *) buf;
1298 } */ *uap = v;
1299 struct proc *p = l->l_proc;
1300 struct statfs sp;
1301 fhandle_t fh;
1302 struct mount *mp;
1303 struct vnode *vp;
1304 int error;
1305
1306 /*
1307 * Must be super user
1308 */
1309 if ((error = suser(p->p_ucred, &p->p_acflag)))
1310 return (error);
1311
1312 if ((error = copyin(SCARG(uap, fhp), &fh, sizeof(fhandle_t))) != 0)
1313 return (error);
1314
1315 if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL)
1316 return (ESTALE);
1317 if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)))
1318 return (error);
1319 mp = vp->v_mount;
1320 vput(vp);
1321 if ((error = VFS_STATFS(mp, &sp, p)) != 0)
1322 return (error);
1323 sp.f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
1324 sp.f_oflags = sp.f_flags & 0xffff;
1325 return (copyout(&sp, SCARG(uap, buf), sizeof(sp)));
1326 }
1327
1328 /*
1329 * Create a special file.
1330 */
1331 /* ARGSUSED */
1332 int
1333 sys_mknod(l, v, retval)
1334 struct lwp *l;
1335 void *v;
1336 register_t *retval;
1337 {
1338 struct sys_mknod_args /* {
1339 syscallarg(const char *) path;
1340 syscallarg(int) mode;
1341 syscallarg(int) dev;
1342 } */ *uap = v;
1343 struct proc *p = l->l_proc;
1344 struct vnode *vp;
1345 struct vattr vattr;
1346 int error;
1347 int whiteout = 0;
1348 struct nameidata nd;
1349
1350 if ((error = suser(p->p_ucred, &p->p_acflag)) != 0)
1351 return (error);
1352 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1353 if ((error = namei(&nd)) != 0)
1354 return (error);
1355 vp = nd.ni_vp;
1356 if (vp != NULL)
1357 error = EEXIST;
1358 else {
1359 VATTR_NULL(&vattr);
1360 vattr.va_mode =
1361 (SCARG(uap, mode) & ALLPERMS) &~ p->p_cwdi->cwdi_cmask;
1362 vattr.va_rdev = SCARG(uap, dev);
1363 whiteout = 0;
1364
1365 switch (SCARG(uap, mode) & S_IFMT) {
1366 case S_IFMT: /* used by badsect to flag bad sectors */
1367 vattr.va_type = VBAD;
1368 break;
1369 case S_IFCHR:
1370 vattr.va_type = VCHR;
1371 break;
1372 case S_IFBLK:
1373 vattr.va_type = VBLK;
1374 break;
1375 case S_IFWHT:
1376 whiteout = 1;
1377 break;
1378 default:
1379 error = EINVAL;
1380 break;
1381 }
1382 }
1383 if (!error) {
1384 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1385 if (whiteout) {
1386 error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, CREATE);
1387 if (error)
1388 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1389 vput(nd.ni_dvp);
1390 } else {
1391 error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp,
1392 &nd.ni_cnd, &vattr);
1393 }
1394 } else {
1395 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1396 if (nd.ni_dvp == vp)
1397 vrele(nd.ni_dvp);
1398 else
1399 vput(nd.ni_dvp);
1400 if (vp)
1401 vrele(vp);
1402 }
1403 return (error);
1404 }
1405
1406 /*
1407 * Create a named pipe.
1408 */
1409 /* ARGSUSED */
1410 int
1411 sys_mkfifo(l, v, retval)
1412 struct lwp *l;
1413 void *v;
1414 register_t *retval;
1415 {
1416 struct sys_mkfifo_args /* {
1417 syscallarg(const char *) path;
1418 syscallarg(int) mode;
1419 } */ *uap = v;
1420 struct proc *p = l->l_proc;
1421 struct vattr vattr;
1422 int error;
1423 struct nameidata nd;
1424
1425 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1426 if ((error = namei(&nd)) != 0)
1427 return (error);
1428 if (nd.ni_vp != NULL) {
1429 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1430 if (nd.ni_dvp == nd.ni_vp)
1431 vrele(nd.ni_dvp);
1432 else
1433 vput(nd.ni_dvp);
1434 vrele(nd.ni_vp);
1435 return (EEXIST);
1436 }
1437 VATTR_NULL(&vattr);
1438 vattr.va_type = VFIFO;
1439 vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_cwdi->cwdi_cmask;
1440 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1441 return (VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr));
1442 }
1443
1444 /*
1445 * Make a hard file link.
1446 */
1447 /* ARGSUSED */
1448 int
1449 sys_link(l, v, retval)
1450 struct lwp *l;
1451 void *v;
1452 register_t *retval;
1453 {
1454 struct sys_link_args /* {
1455 syscallarg(const char *) path;
1456 syscallarg(const char *) link;
1457 } */ *uap = v;
1458 struct proc *p = l->l_proc;
1459 struct vnode *vp;
1460 struct nameidata nd;
1461 int error;
1462
1463 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1464 if ((error = namei(&nd)) != 0)
1465 return (error);
1466 vp = nd.ni_vp;
1467 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, link), p);
1468 if ((error = namei(&nd)) != 0)
1469 goto out;
1470 if (nd.ni_vp) {
1471 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1472 if (nd.ni_dvp == nd.ni_vp)
1473 vrele(nd.ni_dvp);
1474 else
1475 vput(nd.ni_dvp);
1476 vrele(nd.ni_vp);
1477 error = EEXIST;
1478 goto out;
1479 }
1480 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1481 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1482 error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd);
1483 out:
1484 vrele(vp);
1485 return (error);
1486 }
1487
1488 /*
1489 * Make a symbolic link.
1490 */
1491 /* ARGSUSED */
1492 int
1493 sys_symlink(l, v, retval)
1494 struct lwp *l;
1495 void *v;
1496 register_t *retval;
1497 {
1498 struct sys_symlink_args /* {
1499 syscallarg(const char *) path;
1500 syscallarg(const char *) link;
1501 } */ *uap = v;
1502 struct proc *p = l->l_proc;
1503 struct vattr vattr;
1504 char *path;
1505 int error;
1506 struct nameidata nd;
1507
1508 path = PNBUF_GET();
1509 error = copyinstr(SCARG(uap, path), path, MAXPATHLEN, NULL);
1510 if (error)
1511 goto out;
1512 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, link), p);
1513 if ((error = namei(&nd)) != 0)
1514 goto out;
1515 if (nd.ni_vp) {
1516 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1517 if (nd.ni_dvp == nd.ni_vp)
1518 vrele(nd.ni_dvp);
1519 else
1520 vput(nd.ni_dvp);
1521 vrele(nd.ni_vp);
1522 error = EEXIST;
1523 goto out;
1524 }
1525 VATTR_NULL(&vattr);
1526 vattr.va_mode = ACCESSPERMS &~ p->p_cwdi->cwdi_cmask;
1527 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1528 error = VOP_SYMLINK(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr, path);
1529 out:
1530 PNBUF_PUT(path);
1531 return (error);
1532 }
1533
1534 /*
1535 * Delete a whiteout from the filesystem.
1536 */
1537 /* ARGSUSED */
1538 int
1539 sys_undelete(l, v, retval)
1540 struct lwp *l;
1541 void *v;
1542 register_t *retval;
1543 {
1544 struct sys_undelete_args /* {
1545 syscallarg(const char *) path;
1546 } */ *uap = v;
1547 struct proc *p = l->l_proc;
1548 int error;
1549 struct nameidata nd;
1550
1551 NDINIT(&nd, DELETE, LOCKPARENT|DOWHITEOUT, UIO_USERSPACE,
1552 SCARG(uap, path), p);
1553 error = namei(&nd);
1554 if (error)
1555 return (error);
1556
1557 if (nd.ni_vp != NULLVP || !(nd.ni_cnd.cn_flags & ISWHITEOUT)) {
1558 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1559 if (nd.ni_dvp == nd.ni_vp)
1560 vrele(nd.ni_dvp);
1561 else
1562 vput(nd.ni_dvp);
1563 if (nd.ni_vp)
1564 vrele(nd.ni_vp);
1565 return (EEXIST);
1566 }
1567
1568 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1569 if ((error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, DELETE)) != 0)
1570 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1571 vput(nd.ni_dvp);
1572 return (error);
1573 }
1574
1575 /*
1576 * Delete a name from the filesystem.
1577 */
1578 /* ARGSUSED */
1579 int
1580 sys_unlink(l, v, retval)
1581 struct lwp *l;
1582 void *v;
1583 register_t *retval;
1584 {
1585 struct sys_unlink_args /* {
1586 syscallarg(const char *) path;
1587 } */ *uap = v;
1588 struct proc *p = l->l_proc;
1589 struct vnode *vp;
1590 int error;
1591 struct nameidata nd;
1592
1593 NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
1594 SCARG(uap, path), p);
1595 if ((error = namei(&nd)) != 0)
1596 return (error);
1597 vp = nd.ni_vp;
1598
1599 /*
1600 * The root of a mounted filesystem cannot be deleted.
1601 */
1602 if (vp->v_flag & VROOT) {
1603 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1604 if (nd.ni_dvp == vp)
1605 vrele(nd.ni_dvp);
1606 else
1607 vput(nd.ni_dvp);
1608 vput(vp);
1609 error = EBUSY;
1610 goto out;
1611 }
1612
1613 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1614 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1615 error = VOP_REMOVE(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
1616 out:
1617 return (error);
1618 }
1619
1620 /*
1621 * Reposition read/write file offset.
1622 */
1623 int
1624 sys_lseek(l, v, retval)
1625 struct lwp *l;
1626 void *v;
1627 register_t *retval;
1628 {
1629 struct sys_lseek_args /* {
1630 syscallarg(int) fd;
1631 syscallarg(int) pad;
1632 syscallarg(off_t) offset;
1633 syscallarg(int) whence;
1634 } */ *uap = v;
1635 struct proc *p = l->l_proc;
1636 struct ucred *cred = p->p_ucred;
1637 struct filedesc *fdp = p->p_fd;
1638 struct file *fp;
1639 struct vnode *vp;
1640 struct vattr vattr;
1641 off_t newoff;
1642 int error;
1643
1644 if ((u_int)SCARG(uap, fd) >= fdp->fd_nfiles ||
1645 (fp = fdp->fd_ofiles[SCARG(uap, fd)]) == NULL ||
1646 (fp->f_iflags & FIF_WANTCLOSE) != 0)
1647 return (EBADF);
1648
1649 FILE_USE(fp);
1650
1651 vp = (struct vnode *)fp->f_data;
1652 if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
1653 error = ESPIPE;
1654 goto out;
1655 }
1656
1657 switch (SCARG(uap, whence)) {
1658 case SEEK_CUR:
1659 newoff = fp->f_offset + SCARG(uap, offset);
1660 break;
1661 case SEEK_END:
1662 error = VOP_GETATTR(vp, &vattr, cred, p);
1663 if (error)
1664 goto out;
1665 newoff = SCARG(uap, offset) + vattr.va_size;
1666 break;
1667 case SEEK_SET:
1668 newoff = SCARG(uap, offset);
1669 break;
1670 default:
1671 error = EINVAL;
1672 goto out;
1673 }
1674 if ((error = VOP_SEEK(vp, fp->f_offset, newoff, cred)) != 0)
1675 goto out;
1676
1677 *(off_t *)retval = fp->f_offset = newoff;
1678 out:
1679 FILE_UNUSE(fp, p);
1680 return (error);
1681 }
1682
1683 /*
1684 * Positional read system call.
1685 */
1686 int
1687 sys_pread(l, v, retval)
1688 struct lwp *l;
1689 void *v;
1690 register_t *retval;
1691 {
1692 struct sys_pread_args /* {
1693 syscallarg(int) fd;
1694 syscallarg(void *) buf;
1695 syscallarg(size_t) nbyte;
1696 syscallarg(off_t) offset;
1697 } */ *uap = v;
1698 struct proc *p = l->l_proc;
1699 struct filedesc *fdp = p->p_fd;
1700 struct file *fp;
1701 struct vnode *vp;
1702 off_t offset;
1703 int error, fd = SCARG(uap, fd);
1704
1705 if ((u_int)fd >= fdp->fd_nfiles ||
1706 (fp = fdp->fd_ofiles[fd]) == NULL ||
1707 (fp->f_iflags & FIF_WANTCLOSE) != 0 ||
1708 (fp->f_flag & FREAD) == 0)
1709 return (EBADF);
1710
1711 FILE_USE(fp);
1712
1713 vp = (struct vnode *)fp->f_data;
1714 if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
1715 error = ESPIPE;
1716 goto out;
1717 }
1718
1719 offset = SCARG(uap, offset);
1720
1721 /*
1722 * XXX This works because no file systems actually
1723 * XXX take any action on the seek operation.
1724 */
1725 if ((error = VOP_SEEK(vp, fp->f_offset, offset, fp->f_cred)) != 0)
1726 goto out;
1727
1728 /* dofileread() will unuse the descriptor for us */
1729 return (dofileread(p, fd, fp, SCARG(uap, buf), SCARG(uap, nbyte),
1730 &offset, 0, retval));
1731
1732 out:
1733 FILE_UNUSE(fp, p);
1734 return (error);
1735 }
1736
1737 /*
1738 * Positional scatter read system call.
1739 */
1740 int
1741 sys_preadv(l, v, retval)
1742 struct lwp *l;
1743 void *v;
1744 register_t *retval;
1745 {
1746 struct sys_preadv_args /* {
1747 syscallarg(int) fd;
1748 syscallarg(const struct iovec *) iovp;
1749 syscallarg(int) iovcnt;
1750 syscallarg(off_t) offset;
1751 } */ *uap = v;
1752 struct proc *p = l->l_proc;
1753 struct filedesc *fdp = p->p_fd;
1754 struct file *fp;
1755 struct vnode *vp;
1756 off_t offset;
1757 int error, fd = SCARG(uap, fd);
1758
1759 if ((u_int)fd >= fdp->fd_nfiles ||
1760 (fp = fdp->fd_ofiles[fd]) == NULL ||
1761 (fp->f_iflags & FIF_WANTCLOSE) != 0 ||
1762 (fp->f_flag & FREAD) == 0)
1763 return (EBADF);
1764
1765 FILE_USE(fp);
1766
1767 vp = (struct vnode *)fp->f_data;
1768 if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
1769 error = ESPIPE;
1770 goto out;
1771 }
1772
1773 offset = SCARG(uap, offset);
1774
1775 /*
1776 * XXX This works because no file systems actually
1777 * XXX take any action on the seek operation.
1778 */
1779 if ((error = VOP_SEEK(vp, fp->f_offset, offset, fp->f_cred)) != 0)
1780 goto out;
1781
1782 /* dofilereadv() will unuse the descriptor for us */
1783 return (dofilereadv(p, fd, fp, SCARG(uap, iovp), SCARG(uap, iovcnt),
1784 &offset, 0, retval));
1785
1786 out:
1787 FILE_UNUSE(fp, p);
1788 return (error);
1789 }
1790
1791 /*
1792 * Positional write system call.
1793 */
1794 int
1795 sys_pwrite(l, v, retval)
1796 struct lwp *l;
1797 void *v;
1798 register_t *retval;
1799 {
1800 struct sys_pwrite_args /* {
1801 syscallarg(int) fd;
1802 syscallarg(const void *) buf;
1803 syscallarg(size_t) nbyte;
1804 syscallarg(off_t) offset;
1805 } */ *uap = v;
1806 struct proc *p = l->l_proc;
1807 struct filedesc *fdp = p->p_fd;
1808 struct file *fp;
1809 struct vnode *vp;
1810 off_t offset;
1811 int error, fd = SCARG(uap, fd);
1812
1813 if ((u_int)fd >= fdp->fd_nfiles ||
1814 (fp = fdp->fd_ofiles[fd]) == NULL ||
1815 (fp->f_iflags & FIF_WANTCLOSE) != 0 ||
1816 (fp->f_flag & FWRITE) == 0)
1817 return (EBADF);
1818
1819 FILE_USE(fp);
1820
1821 vp = (struct vnode *)fp->f_data;
1822 if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
1823 error = ESPIPE;
1824 goto out;
1825 }
1826
1827 offset = SCARG(uap, offset);
1828
1829 /*
1830 * XXX This works because no file systems actually
1831 * XXX take any action on the seek operation.
1832 */
1833 if ((error = VOP_SEEK(vp, fp->f_offset, offset, fp->f_cred)) != 0)
1834 goto out;
1835
1836 /* dofilewrite() will unuse the descriptor for us */
1837 return (dofilewrite(p, fd, fp, SCARG(uap, buf), SCARG(uap, nbyte),
1838 &offset, 0, retval));
1839
1840 out:
1841 FILE_UNUSE(fp, p);
1842 return (error);
1843 }
1844
1845 /*
1846 * Positional gather write system call.
1847 */
1848 int
1849 sys_pwritev(l, v, retval)
1850 struct lwp *l;
1851 void *v;
1852 register_t *retval;
1853 {
1854 struct sys_pwritev_args /* {
1855 syscallarg(int) fd;
1856 syscallarg(const struct iovec *) iovp;
1857 syscallarg(int) iovcnt;
1858 syscallarg(off_t) offset;
1859 } */ *uap = v;
1860 struct proc *p = l->l_proc;
1861 struct filedesc *fdp = p->p_fd;
1862 struct file *fp;
1863 struct vnode *vp;
1864 off_t offset;
1865 int error, fd = SCARG(uap, fd);
1866
1867 if ((u_int)fd >= fdp->fd_nfiles ||
1868 (fp = fdp->fd_ofiles[fd]) == NULL ||
1869 (fp->f_iflags & FIF_WANTCLOSE) != 0 ||
1870 (fp->f_flag & FWRITE) == 0)
1871 return (EBADF);
1872
1873 FILE_USE(fp);
1874
1875 vp = (struct vnode *)fp->f_data;
1876 if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
1877 error = ESPIPE;
1878 goto out;
1879 }
1880
1881 offset = SCARG(uap, offset);
1882
1883 /*
1884 * XXX This works because no file systems actually
1885 * XXX take any action on the seek operation.
1886 */
1887 if ((error = VOP_SEEK(vp, fp->f_offset, offset, fp->f_cred)) != 0)
1888 goto out;
1889
1890 /* dofilewritev() will unuse the descriptor for us */
1891 return (dofilewritev(p, fd, fp, SCARG(uap, iovp), SCARG(uap, iovcnt),
1892 &offset, 0, retval));
1893
1894 out:
1895 FILE_UNUSE(fp, p);
1896 return (error);
1897 }
1898
1899 /*
1900 * Check access permissions.
1901 */
1902 int
1903 sys_access(l, v, retval)
1904 struct lwp *l;
1905 void *v;
1906 register_t *retval;
1907 {
1908 struct sys_access_args /* {
1909 syscallarg(const char *) path;
1910 syscallarg(int) flags;
1911 } */ *uap = v;
1912 struct proc *p = l->l_proc;
1913 struct ucred *cred = p->p_ucred;
1914 struct vnode *vp;
1915 int error, flags, t_gid, t_uid;
1916 struct nameidata nd;
1917
1918 t_uid = cred->cr_uid;
1919 t_gid = cred->cr_gid;
1920 cred->cr_uid = p->p_cred->p_ruid;
1921 cred->cr_gid = p->p_cred->p_rgid;
1922 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1923 SCARG(uap, path), p);
1924 if ((error = namei(&nd)) != 0)
1925 goto out1;
1926 vp = nd.ni_vp;
1927
1928 /* Flags == 0 means only check for existence. */
1929 if (SCARG(uap, flags)) {
1930 flags = 0;
1931 if (SCARG(uap, flags) & R_OK)
1932 flags |= VREAD;
1933 if (SCARG(uap, flags) & W_OK)
1934 flags |= VWRITE;
1935 if (SCARG(uap, flags) & X_OK)
1936 flags |= VEXEC;
1937
1938 error = VOP_ACCESS(vp, flags, cred, p);
1939 if (!error && (flags & VWRITE))
1940 error = vn_writechk(vp);
1941 }
1942 vput(vp);
1943 out1:
1944 cred->cr_uid = t_uid;
1945 cred->cr_gid = t_gid;
1946 return (error);
1947 }
1948
1949 /*
1950 * Get file status; this version follows links.
1951 */
1952 /* ARGSUSED */
1953 int
1954 sys___stat13(l, v, retval)
1955 struct lwp *l;
1956 void *v;
1957 register_t *retval;
1958 {
1959 struct sys___stat13_args /* {
1960 syscallarg(const char *) path;
1961 syscallarg(struct stat *) ub;
1962 } */ *uap = v;
1963 struct proc *p = l->l_proc;
1964 struct stat sb;
1965 int error;
1966 struct nameidata nd;
1967
1968 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1969 SCARG(uap, path), p);
1970 if ((error = namei(&nd)) != 0)
1971 return (error);
1972 error = vn_stat(nd.ni_vp, &sb, p);
1973 vput(nd.ni_vp);
1974 if (error)
1975 return (error);
1976 error = copyout(&sb, SCARG(uap, ub), sizeof(sb));
1977 return (error);
1978 }
1979
1980 /*
1981 * Get file status; this version does not follow links.
1982 */
1983 /* ARGSUSED */
1984 int
1985 sys___lstat13(l, v, retval)
1986 struct lwp *l;
1987 void *v;
1988 register_t *retval;
1989 {
1990 struct sys___lstat13_args /* {
1991 syscallarg(const char *) path;
1992 syscallarg(struct stat *) ub;
1993 } */ *uap = v;
1994 struct proc *p = l->l_proc;
1995 struct stat sb;
1996 int error;
1997 struct nameidata nd;
1998
1999 NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF, UIO_USERSPACE,
2000 SCARG(uap, path), p);
2001 if ((error = namei(&nd)) != 0)
2002 return (error);
2003 error = vn_stat(nd.ni_vp, &sb, p);
2004 vput(nd.ni_vp);
2005 if (error)
2006 return (error);
2007 error = copyout(&sb, SCARG(uap, ub), sizeof(sb));
2008 return (error);
2009 }
2010
2011 /*
2012 * Get configurable pathname variables.
2013 */
2014 /* ARGSUSED */
2015 int
2016 sys_pathconf(l, v, retval)
2017 struct lwp *l;
2018 void *v;
2019 register_t *retval;
2020 {
2021 struct sys_pathconf_args /* {
2022 syscallarg(const char *) path;
2023 syscallarg(int) name;
2024 } */ *uap = v;
2025 struct proc *p = l->l_proc;
2026 int error;
2027 struct nameidata nd;
2028
2029 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
2030 SCARG(uap, path), p);
2031 if ((error = namei(&nd)) != 0)
2032 return (error);
2033 error = VOP_PATHCONF(nd.ni_vp, SCARG(uap, name), retval);
2034 vput(nd.ni_vp);
2035 return (error);
2036 }
2037
2038 /*
2039 * Return target name of a symbolic link.
2040 */
2041 /* ARGSUSED */
2042 int
2043 sys_readlink(l, v, retval)
2044 struct lwp *l;
2045 void *v;
2046 register_t *retval;
2047 {
2048 struct sys_readlink_args /* {
2049 syscallarg(const char *) path;
2050 syscallarg(char *) buf;
2051 syscallarg(size_t) count;
2052 } */ *uap = v;
2053 struct proc *p = l->l_proc;
2054 struct vnode *vp;
2055 struct iovec aiov;
2056 struct uio auio;
2057 int error;
2058 struct nameidata nd;
2059
2060 NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF, UIO_USERSPACE,
2061 SCARG(uap, path), p);
2062 if ((error = namei(&nd)) != 0)
2063 return (error);
2064 vp = nd.ni_vp;
2065 if (vp->v_type != VLNK)
2066 error = EINVAL;
2067 else if (!(vp->v_mount->mnt_flag & MNT_SYMPERM) ||
2068 (error = VOP_ACCESS(vp, VREAD, p->p_ucred, p)) == 0) {
2069 aiov.iov_base = SCARG(uap, buf);
2070 aiov.iov_len = SCARG(uap, count);
2071 auio.uio_iov = &aiov;
2072 auio.uio_iovcnt = 1;
2073 auio.uio_offset = 0;
2074 auio.uio_rw = UIO_READ;
2075 auio.uio_segflg = UIO_USERSPACE;
2076 auio.uio_procp = p;
2077 auio.uio_resid = SCARG(uap, count);
2078 error = VOP_READLINK(vp, &auio, p->p_ucred);
2079 }
2080 vput(vp);
2081 *retval = SCARG(uap, count) - auio.uio_resid;
2082 return (error);
2083 }
2084
2085 /*
2086 * Change flags of a file given a path name.
2087 */
2088 /* ARGSUSED */
2089 int
2090 sys_chflags(l, v, retval)
2091 struct lwp *l;
2092 void *v;
2093 register_t *retval;
2094 {
2095 struct sys_chflags_args /* {
2096 syscallarg(const char *) path;
2097 syscallarg(u_long) flags;
2098 } */ *uap = v;
2099 struct proc *p = l->l_proc;
2100 struct vnode *vp;
2101 int error;
2102 struct nameidata nd;
2103
2104 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2105 if ((error = namei(&nd)) != 0)
2106 return (error);
2107 vp = nd.ni_vp;
2108 error = change_flags(vp, SCARG(uap, flags), p);
2109 vput(vp);
2110 return (error);
2111 }
2112
2113 /*
2114 * Change flags of a file given a file descriptor.
2115 */
2116 /* ARGSUSED */
2117 int
2118 sys_fchflags(l, v, retval)
2119 struct lwp *l;
2120 void *v;
2121 register_t *retval;
2122 {
2123 struct sys_fchflags_args /* {
2124 syscallarg(int) fd;
2125 syscallarg(u_long) flags;
2126 } */ *uap = v;
2127 struct proc *p = l->l_proc;
2128 struct vnode *vp;
2129 struct file *fp;
2130 int error;
2131
2132 /* getvnode() will use the descriptor for us */
2133 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2134 return (error);
2135 vp = (struct vnode *)fp->f_data;
2136 error = change_flags(vp, SCARG(uap, flags), p);
2137 VOP_UNLOCK(vp, 0);
2138 FILE_UNUSE(fp, p);
2139 return (error);
2140 }
2141
2142 /*
2143 * Change flags of a file given a path name; this version does
2144 * not follow links.
2145 */
2146 int
2147 sys_lchflags(l, v, retval)
2148 struct lwp *l;
2149 void *v;
2150 register_t *retval;
2151 {
2152 struct sys_lchflags_args /* {
2153 syscallarg(const char *) path;
2154 syscallarg(u_long) flags;
2155 } */ *uap = v;
2156 struct proc *p = l->l_proc;
2157 struct vnode *vp;
2158 int error;
2159 struct nameidata nd;
2160
2161 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2162 if ((error = namei(&nd)) != 0)
2163 return (error);
2164 vp = nd.ni_vp;
2165 error = change_flags(vp, SCARG(uap, flags), p);
2166 vput(vp);
2167 return (error);
2168 }
2169
2170 /*
2171 * Common routine to change flags of a file.
2172 */
2173 int
2174 change_flags(vp, flags, p)
2175 struct vnode *vp;
2176 u_long flags;
2177 struct proc *p;
2178 {
2179 struct vattr vattr;
2180 int error;
2181
2182 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2183 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2184 /*
2185 * Non-superusers cannot change the flags on devices, even if they
2186 * own them.
2187 */
2188 if (suser(p->p_ucred, &p->p_acflag) != 0) {
2189 if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)) != 0)
2190 goto out;
2191 if (vattr.va_type == VCHR || vattr.va_type == VBLK) {
2192 error = EINVAL;
2193 goto out;
2194 }
2195 }
2196 VATTR_NULL(&vattr);
2197 vattr.va_flags = flags;
2198 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2199 out:
2200 return (error);
2201 }
2202
2203 /*
2204 * Change mode of a file given path name; this version follows links.
2205 */
2206 /* ARGSUSED */
2207 int
2208 sys_chmod(l, v, retval)
2209 struct lwp *l;
2210 void *v;
2211 register_t *retval;
2212 {
2213 struct sys_chmod_args /* {
2214 syscallarg(const char *) path;
2215 syscallarg(int) mode;
2216 } */ *uap = v;
2217 struct proc *p = l->l_proc;
2218 int error;
2219 struct nameidata nd;
2220
2221 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2222 if ((error = namei(&nd)) != 0)
2223 return (error);
2224
2225 error = change_mode(nd.ni_vp, SCARG(uap, mode), p);
2226
2227 vrele(nd.ni_vp);
2228 return (error);
2229 }
2230
2231 /*
2232 * Change mode of a file given a file descriptor.
2233 */
2234 /* ARGSUSED */
2235 int
2236 sys_fchmod(l, v, retval)
2237 struct lwp *l;
2238 void *v;
2239 register_t *retval;
2240 {
2241 struct sys_fchmod_args /* {
2242 syscallarg(int) fd;
2243 syscallarg(int) mode;
2244 } */ *uap = v;
2245 struct proc *p = l->l_proc;
2246 struct file *fp;
2247 int error;
2248
2249 /* getvnode() will use the descriptor for us */
2250 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2251 return (error);
2252
2253 error = change_mode((struct vnode *)fp->f_data, SCARG(uap, mode), p);
2254 FILE_UNUSE(fp, p);
2255 return (error);
2256 }
2257
2258 /*
2259 * Change mode of a file given path name; this version does not follow links.
2260 */
2261 /* ARGSUSED */
2262 int
2263 sys_lchmod(l, v, retval)
2264 struct lwp *l;
2265 void *v;
2266 register_t *retval;
2267 {
2268 struct sys_lchmod_args /* {
2269 syscallarg(const char *) path;
2270 syscallarg(int) mode;
2271 } */ *uap = v;
2272 struct proc *p = l->l_proc;
2273 int error;
2274 struct nameidata nd;
2275
2276 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2277 if ((error = namei(&nd)) != 0)
2278 return (error);
2279
2280 error = change_mode(nd.ni_vp, SCARG(uap, mode), p);
2281
2282 vrele(nd.ni_vp);
2283 return (error);
2284 }
2285
2286 /*
2287 * Common routine to set mode given a vnode.
2288 */
2289 static int
2290 change_mode(vp, mode, p)
2291 struct vnode *vp;
2292 int mode;
2293 struct proc *p;
2294 {
2295 struct vattr vattr;
2296 int error;
2297
2298 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2299 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2300 VATTR_NULL(&vattr);
2301 vattr.va_mode = mode & ALLPERMS;
2302 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2303 VOP_UNLOCK(vp, 0);
2304 return (error);
2305 }
2306
2307 /*
2308 * Set ownership given a path name; this version follows links.
2309 */
2310 /* ARGSUSED */
2311 int
2312 sys_chown(l, v, retval)
2313 struct lwp *l;
2314 void *v;
2315 register_t *retval;
2316 {
2317 struct sys_chown_args /* {
2318 syscallarg(const char *) path;
2319 syscallarg(uid_t) uid;
2320 syscallarg(gid_t) gid;
2321 } */ *uap = v;
2322 struct proc *p = l->l_proc;
2323 int error;
2324 struct nameidata nd;
2325
2326 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2327 if ((error = namei(&nd)) != 0)
2328 return (error);
2329
2330 error = change_owner(nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid), p, 0);
2331
2332 vrele(nd.ni_vp);
2333 return (error);
2334 }
2335
2336 /*
2337 * Set ownership given a path name; this version follows links.
2338 * Provides POSIX semantics.
2339 */
2340 /* ARGSUSED */
2341 int
2342 sys___posix_chown(l, v, retval)
2343 struct lwp *l;
2344 void *v;
2345 register_t *retval;
2346 {
2347 struct sys_chown_args /* {
2348 syscallarg(const char *) path;
2349 syscallarg(uid_t) uid;
2350 syscallarg(gid_t) gid;
2351 } */ *uap = v;
2352 struct proc *p = l->l_proc;
2353 int error;
2354 struct nameidata nd;
2355
2356 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2357 if ((error = namei(&nd)) != 0)
2358 return (error);
2359
2360 error = change_owner(nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid), p, 1);
2361
2362 vrele(nd.ni_vp);
2363 return (error);
2364 }
2365
2366 /*
2367 * Set ownership given a file descriptor.
2368 */
2369 /* ARGSUSED */
2370 int
2371 sys_fchown(l, v, retval)
2372 struct lwp *l;
2373 void *v;
2374 register_t *retval;
2375 {
2376 struct sys_fchown_args /* {
2377 syscallarg(int) fd;
2378 syscallarg(uid_t) uid;
2379 syscallarg(gid_t) gid;
2380 } */ *uap = v;
2381 struct proc *p = l->l_proc;
2382 int error;
2383 struct file *fp;
2384
2385 /* getvnode() will use the descriptor for us */
2386 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2387 return (error);
2388
2389 error = change_owner((struct vnode *)fp->f_data, SCARG(uap, uid),
2390 SCARG(uap, gid), p, 0);
2391 FILE_UNUSE(fp, p);
2392 return (error);
2393 }
2394
2395 /*
2396 * Set ownership given a file descriptor, providing POSIX/XPG semantics.
2397 */
2398 /* ARGSUSED */
2399 int
2400 sys___posix_fchown(l, v, retval)
2401 struct lwp *l;
2402 void *v;
2403 register_t *retval;
2404 {
2405 struct sys_fchown_args /* {
2406 syscallarg(int) fd;
2407 syscallarg(uid_t) uid;
2408 syscallarg(gid_t) gid;
2409 } */ *uap = v;
2410 struct proc *p = l->l_proc;
2411 int error;
2412 struct file *fp;
2413
2414 /* getvnode() will use the descriptor for us */
2415 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2416 return (error);
2417
2418 error = change_owner((struct vnode *)fp->f_data, SCARG(uap, uid),
2419 SCARG(uap, gid), p, 1);
2420 FILE_UNUSE(fp, p);
2421 return (error);
2422 }
2423
2424 /*
2425 * Set ownership given a path name; this version does not follow links.
2426 */
2427 /* ARGSUSED */
2428 int
2429 sys_lchown(l, v, retval)
2430 struct lwp *l;
2431 void *v;
2432 register_t *retval;
2433 {
2434 struct sys_lchown_args /* {
2435 syscallarg(const char *) path;
2436 syscallarg(uid_t) uid;
2437 syscallarg(gid_t) gid;
2438 } */ *uap = v;
2439 struct proc *p = l->l_proc;
2440 int error;
2441 struct nameidata nd;
2442
2443 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2444 if ((error = namei(&nd)) != 0)
2445 return (error);
2446
2447 error = change_owner(nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid), p, 0);
2448
2449 vrele(nd.ni_vp);
2450 return (error);
2451 }
2452
2453 /*
2454 * Set ownership given a path name; this version does not follow links.
2455 * Provides POSIX/XPG semantics.
2456 */
2457 /* ARGSUSED */
2458 int
2459 sys___posix_lchown(l, v, retval)
2460 struct lwp *l;
2461 void *v;
2462 register_t *retval;
2463 {
2464 struct sys_lchown_args /* {
2465 syscallarg(const char *) path;
2466 syscallarg(uid_t) uid;
2467 syscallarg(gid_t) gid;
2468 } */ *uap = v;
2469 struct proc *p = l->l_proc;
2470 int error;
2471 struct nameidata nd;
2472
2473 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2474 if ((error = namei(&nd)) != 0)
2475 return (error);
2476
2477 error = change_owner(nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid), p, 1);
2478
2479 vrele(nd.ni_vp);
2480 return (error);
2481 }
2482
2483 /*
2484 * Common routine to set ownership given a vnode.
2485 */
2486 static int
2487 change_owner(vp, uid, gid, p, posix_semantics)
2488 struct vnode *vp;
2489 uid_t uid;
2490 gid_t gid;
2491 struct proc *p;
2492 int posix_semantics;
2493 {
2494 struct vattr vattr;
2495 mode_t newmode;
2496 int error;
2497
2498 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2499 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2500 if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)) != 0)
2501 goto out;
2502
2503 #define CHANGED(x) ((x) != -1)
2504 newmode = vattr.va_mode;
2505 if (posix_semantics) {
2506 /*
2507 * POSIX/XPG semantics: if the caller is not the super-user,
2508 * clear set-user-id and set-group-id bits. Both POSIX and
2509 * the XPG consider the behaviour for calls by the super-user
2510 * implementation-defined; we leave the set-user-id and set-
2511 * group-id settings intact in that case.
2512 */
2513 if (suser(p->p_ucred, NULL) != 0)
2514 newmode &= ~(S_ISUID | S_ISGID);
2515 } else {
2516 /*
2517 * NetBSD semantics: when changing owner and/or group,
2518 * clear the respective bit(s).
2519 */
2520 if (CHANGED(uid))
2521 newmode &= ~S_ISUID;
2522 if (CHANGED(gid))
2523 newmode &= ~S_ISGID;
2524 }
2525 /* Update va_mode iff altered. */
2526 if (vattr.va_mode == newmode)
2527 newmode = VNOVAL;
2528
2529 VATTR_NULL(&vattr);
2530 vattr.va_uid = CHANGED(uid) ? uid : VNOVAL;
2531 vattr.va_gid = CHANGED(gid) ? gid : VNOVAL;
2532 vattr.va_mode = newmode;
2533 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2534 #undef CHANGED
2535
2536 out:
2537 VOP_UNLOCK(vp, 0);
2538 return (error);
2539 }
2540
2541 /*
2542 * Set the access and modification times given a path name; this
2543 * version follows links.
2544 */
2545 /* ARGSUSED */
2546 int
2547 sys_utimes(l, v, retval)
2548 struct lwp *l;
2549 void *v;
2550 register_t *retval;
2551 {
2552 struct sys_utimes_args /* {
2553 syscallarg(const char *) path;
2554 syscallarg(const struct timeval *) tptr;
2555 } */ *uap = v;
2556 struct proc *p = l->l_proc;
2557 int error;
2558 struct nameidata nd;
2559
2560 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2561 if ((error = namei(&nd)) != 0)
2562 return (error);
2563
2564 error = change_utimes(nd.ni_vp, SCARG(uap, tptr), p);
2565
2566 vrele(nd.ni_vp);
2567 return (error);
2568 }
2569
2570 /*
2571 * Set the access and modification times given a file descriptor.
2572 */
2573 /* ARGSUSED */
2574 int
2575 sys_futimes(l, v, retval)
2576 struct lwp *l;
2577 void *v;
2578 register_t *retval;
2579 {
2580 struct sys_futimes_args /* {
2581 syscallarg(int) fd;
2582 syscallarg(const struct timeval *) tptr;
2583 } */ *uap = v;
2584 struct proc *p = l->l_proc;
2585 int error;
2586 struct file *fp;
2587
2588 /* getvnode() will use the descriptor for us */
2589 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2590 return (error);
2591
2592 error = change_utimes((struct vnode *)fp->f_data, SCARG(uap, tptr), p);
2593 FILE_UNUSE(fp, p);
2594 return (error);
2595 }
2596
2597 /*
2598 * Set the access and modification times given a path name; this
2599 * version does not follow links.
2600 */
2601 /* ARGSUSED */
2602 int
2603 sys_lutimes(l, v, retval)
2604 struct lwp *l;
2605 void *v;
2606 register_t *retval;
2607 {
2608 struct sys_lutimes_args /* {
2609 syscallarg(const char *) path;
2610 syscallarg(const struct timeval *) tptr;
2611 } */ *uap = v;
2612 struct proc *p = l->l_proc;
2613 int error;
2614 struct nameidata nd;
2615
2616 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2617 if ((error = namei(&nd)) != 0)
2618 return (error);
2619
2620 error = change_utimes(nd.ni_vp, SCARG(uap, tptr), p);
2621
2622 vrele(nd.ni_vp);
2623 return (error);
2624 }
2625
2626 /*
2627 * Common routine to set access and modification times given a vnode.
2628 */
2629 static int
2630 change_utimes(vp, tptr, p)
2631 struct vnode *vp;
2632 const struct timeval *tptr;
2633 struct proc *p;
2634 {
2635 struct timeval tv[2];
2636 struct vattr vattr;
2637 int error;
2638
2639 VATTR_NULL(&vattr);
2640 if (tptr == NULL) {
2641 microtime(&tv[0]);
2642 tv[1] = tv[0];
2643 vattr.va_vaflags |= VA_UTIMES_NULL;
2644 } else {
2645 error = copyin(tptr, tv, sizeof(tv));
2646 if (error)
2647 return (error);
2648 }
2649 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2650 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2651 vattr.va_atime.tv_sec = tv[0].tv_sec;
2652 vattr.va_atime.tv_nsec = tv[0].tv_usec * 1000;
2653 vattr.va_mtime.tv_sec = tv[1].tv_sec;
2654 vattr.va_mtime.tv_nsec = tv[1].tv_usec * 1000;
2655 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2656 VOP_UNLOCK(vp, 0);
2657 return (error);
2658 }
2659
2660 /*
2661 * Truncate a file given its path name.
2662 */
2663 /* ARGSUSED */
2664 int
2665 sys_truncate(l, v, retval)
2666 struct lwp *l;
2667 void *v;
2668 register_t *retval;
2669 {
2670 struct sys_truncate_args /* {
2671 syscallarg(const char *) path;
2672 syscallarg(int) pad;
2673 syscallarg(off_t) length;
2674 } */ *uap = v;
2675 struct proc *p = l->l_proc;
2676 struct vnode *vp;
2677 struct vattr vattr;
2678 int error;
2679 struct nameidata nd;
2680
2681 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2682 if ((error = namei(&nd)) != 0)
2683 return (error);
2684 vp = nd.ni_vp;
2685 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2686 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2687 if (vp->v_type == VDIR)
2688 error = EISDIR;
2689 else if ((error = vn_writechk(vp)) == 0 &&
2690 (error = VOP_ACCESS(vp, VWRITE, p->p_ucred, p)) == 0) {
2691 VATTR_NULL(&vattr);
2692 vattr.va_size = SCARG(uap, length);
2693 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2694 }
2695 vput(vp);
2696 return (error);
2697 }
2698
2699 /*
2700 * Truncate a file given a file descriptor.
2701 */
2702 /* ARGSUSED */
2703 int
2704 sys_ftruncate(l, v, retval)
2705 struct lwp *l;
2706 void *v;
2707 register_t *retval;
2708 {
2709 struct sys_ftruncate_args /* {
2710 syscallarg(int) fd;
2711 syscallarg(int) pad;
2712 syscallarg(off_t) length;
2713 } */ *uap = v;
2714 struct proc *p = l->l_proc;
2715 struct vattr vattr;
2716 struct vnode *vp;
2717 struct file *fp;
2718 int error;
2719
2720 /* getvnode() will use the descriptor for us */
2721 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2722 return (error);
2723 if ((fp->f_flag & FWRITE) == 0) {
2724 error = EINVAL;
2725 goto out;
2726 }
2727 vp = (struct vnode *)fp->f_data;
2728 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2729 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2730 if (vp->v_type == VDIR)
2731 error = EISDIR;
2732 else if ((error = vn_writechk(vp)) == 0) {
2733 VATTR_NULL(&vattr);
2734 vattr.va_size = SCARG(uap, length);
2735 error = VOP_SETATTR(vp, &vattr, fp->f_cred, p);
2736 }
2737 VOP_UNLOCK(vp, 0);
2738 out:
2739 FILE_UNUSE(fp, p);
2740 return (error);
2741 }
2742
2743 /*
2744 * Sync an open file.
2745 */
2746 /* ARGSUSED */
2747 int
2748 sys_fsync(l, v, retval)
2749 struct lwp *l;
2750 void *v;
2751 register_t *retval;
2752 {
2753 struct sys_fsync_args /* {
2754 syscallarg(int) fd;
2755 } */ *uap = v;
2756 struct proc *p = l->l_proc;
2757 struct vnode *vp;
2758 struct file *fp;
2759 int error;
2760
2761 /* getvnode() will use the descriptor for us */
2762 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2763 return (error);
2764 vp = (struct vnode *)fp->f_data;
2765 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2766 error = VOP_FSYNC(vp, fp->f_cred, FSYNC_WAIT, 0, 0, p);
2767 if (error == 0 && bioops.io_fsync != NULL &&
2768 vp->v_mount && (vp->v_mount->mnt_flag & MNT_SOFTDEP))
2769 (*bioops.io_fsync)(vp);
2770 VOP_UNLOCK(vp, 0);
2771 FILE_UNUSE(fp, p);
2772 return (error);
2773 }
2774
2775 /*
2776 * Sync the data of an open file.
2777 */
2778 /* ARGSUSED */
2779 int
2780 sys_fdatasync(l, v, retval)
2781 struct lwp *l;
2782 void *v;
2783 register_t *retval;
2784 {
2785 struct sys_fdatasync_args /* {
2786 syscallarg(int) fd;
2787 } */ *uap = v;
2788 struct proc *p = l->l_proc;
2789 struct vnode *vp;
2790 struct file *fp;
2791 int error;
2792
2793 /* getvnode() will use the descriptor for us */
2794 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2795 return (error);
2796 vp = (struct vnode *)fp->f_data;
2797 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2798 error = VOP_FSYNC(vp, fp->f_cred, FSYNC_WAIT|FSYNC_DATAONLY, 0, 0, p);
2799 VOP_UNLOCK(vp, 0);
2800 FILE_UNUSE(fp, p);
2801 return (error);
2802 }
2803
2804 /*
2805 * Rename files, (standard) BSD semantics frontend.
2806 */
2807 /* ARGSUSED */
2808 int
2809 sys_rename(l, v, retval)
2810 struct lwp *l;
2811 void *v;
2812 register_t *retval;
2813 {
2814 struct sys_rename_args /* {
2815 syscallarg(const char *) from;
2816 syscallarg(const char *) to;
2817 } */ *uap = v;
2818 struct proc *p = l->l_proc;
2819
2820 return (rename_files(SCARG(uap, from), SCARG(uap, to), p, 0));
2821 }
2822
2823 /*
2824 * Rename files, POSIX semantics frontend.
2825 */
2826 /* ARGSUSED */
2827 int
2828 sys___posix_rename(l, v, retval)
2829 struct lwp *l;
2830 void *v;
2831 register_t *retval;
2832 {
2833 struct sys___posix_rename_args /* {
2834 syscallarg(const char *) from;
2835 syscallarg(const char *) to;
2836 } */ *uap = v;
2837 struct proc *p = l->l_proc;
2838
2839 return (rename_files(SCARG(uap, from), SCARG(uap, to), p, 1));
2840 }
2841
2842 /*
2843 * Rename files. Source and destination must either both be directories,
2844 * or both not be directories. If target is a directory, it must be empty.
2845 * If `from' and `to' refer to the same object, the value of the `retain'
2846 * argument is used to determine whether `from' will be
2847 *
2848 * (retain == 0) deleted unless `from' and `to' refer to the same
2849 * object in the file system's name space (BSD).
2850 * (retain == 1) always retained (POSIX).
2851 */
2852 static int
2853 rename_files(from, to, p, retain)
2854 const char *from, *to;
2855 struct proc *p;
2856 int retain;
2857 {
2858 struct vnode *tvp, *fvp, *tdvp;
2859 struct nameidata fromnd, tond;
2860 int error;
2861
2862 NDINIT(&fromnd, DELETE, WANTPARENT | SAVESTART, UIO_USERSPACE,
2863 from, p);
2864 if ((error = namei(&fromnd)) != 0)
2865 return (error);
2866 fvp = fromnd.ni_vp;
2867 NDINIT(&tond, RENAME, LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART,
2868 UIO_USERSPACE, to, p);
2869 if ((error = namei(&tond)) != 0) {
2870 VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
2871 vrele(fromnd.ni_dvp);
2872 vrele(fvp);
2873 goto out1;
2874 }
2875 tdvp = tond.ni_dvp;
2876 tvp = tond.ni_vp;
2877
2878 if (tvp != NULL) {
2879 if (fvp->v_type == VDIR && tvp->v_type != VDIR) {
2880 error = ENOTDIR;
2881 goto out;
2882 } else if (fvp->v_type != VDIR && tvp->v_type == VDIR) {
2883 error = EISDIR;
2884 goto out;
2885 }
2886 }
2887
2888 if (fvp == tdvp)
2889 error = EINVAL;
2890
2891 /*
2892 * Source and destination refer to the same object.
2893 */
2894 if (fvp == tvp) {
2895 if (retain)
2896 error = -1;
2897 else if (fromnd.ni_dvp == tdvp &&
2898 fromnd.ni_cnd.cn_namelen == tond.ni_cnd.cn_namelen &&
2899 !memcmp(fromnd.ni_cnd.cn_nameptr,
2900 tond.ni_cnd.cn_nameptr,
2901 fromnd.ni_cnd.cn_namelen))
2902 error = -1;
2903 }
2904
2905 out:
2906 if (!error) {
2907 VOP_LEASE(tdvp, p, p->p_ucred, LEASE_WRITE);
2908 if (fromnd.ni_dvp != tdvp)
2909 VOP_LEASE(fromnd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
2910 if (tvp) {
2911 VOP_LEASE(tvp, p, p->p_ucred, LEASE_WRITE);
2912 }
2913 error = VOP_RENAME(fromnd.ni_dvp, fromnd.ni_vp, &fromnd.ni_cnd,
2914 tond.ni_dvp, tond.ni_vp, &tond.ni_cnd);
2915 } else {
2916 VOP_ABORTOP(tond.ni_dvp, &tond.ni_cnd);
2917 if (tdvp == tvp)
2918 vrele(tdvp);
2919 else
2920 vput(tdvp);
2921 if (tvp)
2922 vput(tvp);
2923 VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
2924 vrele(fromnd.ni_dvp);
2925 vrele(fvp);
2926 }
2927 vrele(tond.ni_startdir);
2928 PNBUF_PUT(tond.ni_cnd.cn_pnbuf);
2929 out1:
2930 if (fromnd.ni_startdir)
2931 vrele(fromnd.ni_startdir);
2932 PNBUF_PUT(fromnd.ni_cnd.cn_pnbuf);
2933 return (error == -1 ? 0 : error);
2934 }
2935
2936 /*
2937 * Make a directory file.
2938 */
2939 /* ARGSUSED */
2940 int
2941 sys_mkdir(l, v, retval)
2942 struct lwp *l;
2943 void *v;
2944 register_t *retval;
2945 {
2946 struct sys_mkdir_args /* {
2947 syscallarg(const char *) path;
2948 syscallarg(int) mode;
2949 } */ *uap = v;
2950 struct proc *p = l->l_proc;
2951 struct vnode *vp;
2952 struct vattr vattr;
2953 int error;
2954 struct nameidata nd;
2955
2956 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
2957 if ((error = namei(&nd)) != 0)
2958 return (error);
2959 vp = nd.ni_vp;
2960 if (vp != NULL) {
2961 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
2962 if (nd.ni_dvp == vp)
2963 vrele(nd.ni_dvp);
2964 else
2965 vput(nd.ni_dvp);
2966 vrele(vp);
2967 return (EEXIST);
2968 }
2969 VATTR_NULL(&vattr);
2970 vattr.va_type = VDIR;
2971 vattr.va_mode =
2972 (SCARG(uap, mode) & ACCESSPERMS) &~ p->p_cwdi->cwdi_cmask;
2973 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
2974 error = VOP_MKDIR(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
2975 if (!error)
2976 vput(nd.ni_vp);
2977 return (error);
2978 }
2979
2980 /*
2981 * Remove a directory file.
2982 */
2983 /* ARGSUSED */
2984 int
2985 sys_rmdir(l, v, retval)
2986 struct lwp *l;
2987 void *v;
2988 register_t *retval;
2989 {
2990 struct sys_rmdir_args /* {
2991 syscallarg(const char *) path;
2992 } */ *uap = v;
2993 struct proc *p = l->l_proc;
2994 struct vnode *vp;
2995 int error;
2996 struct nameidata nd;
2997
2998 NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
2999 SCARG(uap, path), p);
3000 if ((error = namei(&nd)) != 0)
3001 return (error);
3002 vp = nd.ni_vp;
3003 if (vp->v_type != VDIR) {
3004 error = ENOTDIR;
3005 goto out;
3006 }
3007 /*
3008 * No rmdir "." please.
3009 */
3010 if (nd.ni_dvp == vp) {
3011 error = EINVAL;
3012 goto out;
3013 }
3014 /*
3015 * The root of a mounted filesystem cannot be deleted.
3016 */
3017 if (vp->v_flag & VROOT)
3018 error = EBUSY;
3019 out:
3020 if (!error) {
3021 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
3022 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
3023 error = VOP_RMDIR(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
3024 } else {
3025 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
3026 if (nd.ni_dvp == vp)
3027 vrele(nd.ni_dvp);
3028 else
3029 vput(nd.ni_dvp);
3030 vput(vp);
3031 }
3032 return (error);
3033 }
3034
3035 /*
3036 * Read a block of directory entries in a file system independent format.
3037 */
3038 int
3039 sys_getdents(l, v, retval)
3040 struct lwp *l;
3041 void *v;
3042 register_t *retval;
3043 {
3044 struct sys_getdents_args /* {
3045 syscallarg(int) fd;
3046 syscallarg(char *) buf;
3047 syscallarg(size_t) count;
3048 } */ *uap = v;
3049 struct proc *p = l->l_proc;
3050 struct file *fp;
3051 int error, done;
3052
3053 /* getvnode() will use the descriptor for us */
3054 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
3055 return (error);
3056 if ((fp->f_flag & FREAD) == 0) {
3057 error = EBADF;
3058 goto out;
3059 }
3060 error = vn_readdir(fp, SCARG(uap, buf), UIO_USERSPACE,
3061 SCARG(uap, count), &done, p, 0, 0);
3062 *retval = done;
3063 out:
3064 FILE_UNUSE(fp, p);
3065 return (error);
3066 }
3067
3068 /*
3069 * Set the mode mask for creation of filesystem nodes.
3070 */
3071 int
3072 sys_umask(l, v, retval)
3073 struct lwp *l;
3074 void *v;
3075 register_t *retval;
3076 {
3077 struct sys_umask_args /* {
3078 syscallarg(mode_t) newmask;
3079 } */ *uap = v;
3080 struct proc *p = l->l_proc;
3081 struct cwdinfo *cwdi;
3082
3083 cwdi = p->p_cwdi;
3084 *retval = cwdi->cwdi_cmask;
3085 cwdi->cwdi_cmask = SCARG(uap, newmask) & ALLPERMS;
3086 return (0);
3087 }
3088
3089 /*
3090 * Void all references to file by ripping underlying filesystem
3091 * away from vnode.
3092 */
3093 /* ARGSUSED */
3094 int
3095 sys_revoke(l, v, retval)
3096 struct lwp *l;
3097 void *v;
3098 register_t *retval;
3099 {
3100 struct sys_revoke_args /* {
3101 syscallarg(const char *) path;
3102 } */ *uap = v;
3103 struct proc *p = l->l_proc;
3104 struct vnode *vp;
3105 struct vattr vattr;
3106 int error;
3107 struct nameidata nd;
3108
3109 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
3110 if ((error = namei(&nd)) != 0)
3111 return (error);
3112 vp = nd.ni_vp;
3113 if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)) != 0)
3114 goto out;
3115 if (p->p_ucred->cr_uid != vattr.va_uid &&
3116 (error = suser(p->p_ucred, &p->p_acflag)) != 0)
3117 goto out;
3118 if (vp->v_usecount > 1 || (vp->v_flag & (VALIASED | VLAYER)))
3119 VOP_REVOKE(vp, REVOKEALL);
3120 out:
3121 vrele(vp);
3122 return (error);
3123 }
3124
3125 /*
3126 * Convert a user file descriptor to a kernel file entry.
3127 */
3128 int
3129 getvnode(fdp, fd, fpp)
3130 struct filedesc *fdp;
3131 int fd;
3132 struct file **fpp;
3133 {
3134 struct vnode *vp;
3135 struct file *fp;
3136
3137 if ((u_int)fd >= fdp->fd_nfiles ||
3138 (fp = fdp->fd_ofiles[fd]) == NULL ||
3139 (fp->f_iflags & FIF_WANTCLOSE) != 0)
3140 return (EBADF);
3141
3142 FILE_USE(fp);
3143
3144 if (fp->f_type != DTYPE_VNODE) {
3145 FILE_UNUSE(fp, NULL);
3146 return (EINVAL);
3147 }
3148
3149 vp = (struct vnode *)fp->f_data;
3150 if (vp->v_type == VBAD) {
3151 FILE_UNUSE(fp, NULL);
3152 return (EBADF);
3153 }
3154
3155 *fpp = fp;
3156 return (0);
3157 }
3158