vfs_syscalls.c revision 1.214 1 /* $NetBSD: vfs_syscalls.c,v 1.214 2005/01/02 16:08:29 thorpej 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. Neither the name of the University nor the names of its contributors
21 * may be used to endorse or promote products derived from this software
22 * without specific prior written permission.
23 *
24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 * SUCH DAMAGE.
35 *
36 * @(#)vfs_syscalls.c 8.42 (Berkeley) 7/31/95
37 */
38
39 #include <sys/cdefs.h>
40 __KERNEL_RCSID(0, "$NetBSD: vfs_syscalls.c,v 1.214 2005/01/02 16:08:29 thorpej Exp $");
41
42 #include "opt_compat_netbsd.h"
43 #include "opt_compat_43.h"
44 #include "opt_ktrace.h"
45 #include "fss.h"
46
47 #include <sys/param.h>
48 #include <sys/systm.h>
49 #include <sys/namei.h>
50 #include <sys/filedesc.h>
51 #include <sys/kernel.h>
52 #include <sys/file.h>
53 #include <sys/stat.h>
54 #include <sys/vnode.h>
55 #include <sys/mount.h>
56 #include <sys/proc.h>
57 #include <sys/uio.h>
58 #include <sys/malloc.h>
59 #include <sys/dirent.h>
60 #include <sys/extattr.h>
61 #include <sys/sysctl.h>
62 #include <sys/sa.h>
63 #include <sys/syscallargs.h>
64 #ifdef KTRACE
65 #include <sys/ktrace.h>
66 #endif
67
68 #include <miscfs/genfs/genfs.h>
69 #include <miscfs/syncfs/syncfs.h>
70
71 #if NFSS > 0
72 #include <dev/fssvar.h>
73 #endif
74
75 MALLOC_DEFINE(M_MOUNT, "mount", "vfs mount struct");
76
77 static int change_dir(struct nameidata *, struct proc *);
78 static int change_flags(struct vnode *, u_long, struct proc *);
79 static int change_mode(struct vnode *, int, struct proc *p);
80 static int change_owner(struct vnode *, uid_t, gid_t, struct proc *, int);
81 static int change_utimes(struct vnode *vp, const struct timeval *,
82 struct proc *p);
83 static int rename_files(const char *, const char *, struct proc *, int);
84
85 void checkdirs(struct vnode *);
86
87 int dovfsusermount = 0;
88
89 /*
90 * Virtual File System System Calls
91 */
92
93 /*
94 * Mount a file system.
95 */
96
97 #if defined(COMPAT_09) || defined(COMPAT_43)
98 /*
99 * This table is used to maintain compatibility with 4.3BSD
100 * and NetBSD 0.9 mount syscalls. Note, the order is important!
101 *
102 * Do not modify this table. It should only contain filesystems
103 * supported by NetBSD 0.9 and 4.3BSD.
104 */
105 const char * const mountcompatnames[] = {
106 NULL, /* 0 = MOUNT_NONE */
107 MOUNT_FFS, /* 1 = MOUNT_UFS */
108 MOUNT_NFS, /* 2 */
109 MOUNT_MFS, /* 3 */
110 MOUNT_MSDOS, /* 4 */
111 MOUNT_CD9660, /* 5 = MOUNT_ISOFS */
112 MOUNT_FDESC, /* 6 */
113 MOUNT_KERNFS, /* 7 */
114 NULL, /* 8 = MOUNT_DEVFS */
115 MOUNT_AFS, /* 9 */
116 };
117 const int nmountcompatnames = sizeof(mountcompatnames) /
118 sizeof(mountcompatnames[0]);
119 #endif /* COMPAT_09 || COMPAT_43 */
120
121 /* ARGSUSED */
122 int
123 sys_mount(l, v, retval)
124 struct lwp *l;
125 void *v;
126 register_t *retval;
127 {
128 struct sys_mount_args /* {
129 syscallarg(const char *) type;
130 syscallarg(const char *) path;
131 syscallarg(int) flags;
132 syscallarg(void *) data;
133 } */ *uap = v;
134 struct proc *p = l->l_proc;
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 if (dovfsusermount == 0 && (SCARG(uap, flags) & MNT_GETARGS) == 0 &&
144 (error = suser(p->p_ucred, &p->p_acflag)))
145 return (error);
146 /*
147 * Get vnode to be covered
148 */
149 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE,
150 SCARG(uap, path), p);
151 if ((error = namei(&nd)) != 0)
152 return (error);
153 vp = nd.ni_vp;
154 /*
155 * A lookup in VFS_MOUNT might result in an attempt to
156 * lock this vnode again, so make the lock recursive.
157 */
158 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY | LK_SETRECURSE);
159 if (SCARG(uap, flags) & (MNT_UPDATE | MNT_GETARGS)) {
160 if ((vp->v_flag & VROOT) == 0) {
161 vput(vp);
162 return (EINVAL);
163 }
164 mp = vp->v_mount;
165 flag = mp->mnt_flag;
166 vfs = mp->mnt_op;
167 /*
168 * We only allow the filesystem to be reloaded if it
169 * is currently mounted read-only.
170 */
171 if ((SCARG(uap, flags) & MNT_RELOAD) &&
172 ((mp->mnt_flag & MNT_RDONLY) == 0)) {
173 vput(vp);
174 return (EOPNOTSUPP); /* Needs translation */
175 }
176 /*
177 * In "highly secure" mode, don't let the caller do anything
178 * but downgrade a filesystem from read-write to read-only.
179 * (see also below; MNT_UPDATE or MNT_GETARGS is required.)
180 */
181 if (securelevel >= 2 &&
182 SCARG(uap, flags) != MNT_GETARGS &&
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 |= SCARG(uap, flags) &
190 (MNT_RELOAD | MNT_FORCE | MNT_UPDATE | MNT_GETARGS);
191 /*
192 * Only root, or the user that did the original mount is
193 * permitted to update it.
194 */
195 if ((mp->mnt_flag & MNT_GETARGS) == 0 &&
196 mp->mnt_stat.f_owner != p->p_ucred->cr_uid &&
197 (error = suser(p->p_ucred, &p->p_acflag)) != 0) {
198 vput(vp);
199 return (error);
200 }
201 /*
202 * Do not allow NFS export by non-root users. For non-root
203 * users, silently enforce MNT_NOSUID and MNT_NODEV, and
204 * MNT_NOEXEC if mount point is already MNT_NOEXEC.
205 */
206 if (p->p_ucred->cr_uid != 0) {
207 if (SCARG(uap, flags) & MNT_EXPORTED) {
208 vput(vp);
209 return (EPERM);
210 }
211 SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
212 if (flag & MNT_NOEXEC)
213 SCARG(uap, flags) |= MNT_NOEXEC;
214 }
215 if (vfs_busy(mp, LK_NOWAIT, 0)) {
216 vput(vp);
217 return (EPERM);
218 }
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 vput(vp);
252 return (error);
253 }
254 if (vp->v_type != VDIR) {
255 vput(vp);
256 return (ENOTDIR);
257 }
258 error = copyinstr(SCARG(uap, type), fstypename, MFSNAMELEN, NULL);
259 if (error) {
260 #if defined(COMPAT_09) || defined(COMPAT_43)
261 /*
262 * Historically filesystem types were identified by number.
263 * If we get an integer for the filesystem type instead of a
264 * string, we check to see if it matches one of the historic
265 * filesystem types.
266 */
267 u_long fsindex = (u_long)SCARG(uap, type);
268 if (fsindex >= nmountcompatnames ||
269 mountcompatnames[fsindex] == NULL) {
270 vput(vp);
271 return (ENODEV);
272 }
273 strncpy(fstypename, mountcompatnames[fsindex], MFSNAMELEN);
274 #else
275 vput(vp);
276 return (error);
277 #endif
278 }
279 #ifdef COMPAT_10
280 /* Accept `ufs' as an alias for `ffs'. */
281 if (!strncmp(fstypename, "ufs", MFSNAMELEN))
282 strncpy(fstypename, "ffs", MFSNAMELEN);
283 #endif
284 if ((vfs = vfs_getopsbyname(fstypename)) == NULL) {
285 vput(vp);
286 return (ENODEV);
287 }
288 if (vp->v_mountedhere != NULL) {
289 vput(vp);
290 return (EBUSY);
291 }
292
293 /*
294 * Allocate and initialize the file system.
295 */
296 mp = (struct mount *)malloc((u_long)sizeof(struct mount),
297 M_MOUNT, M_WAITOK);
298 memset((char *)mp, 0, (u_long)sizeof(struct mount));
299 lockinit(&mp->mnt_lock, PVFS, "vfslock", 0, 0);
300 simple_lock_init(&mp->mnt_slock);
301 (void)vfs_busy(mp, LK_NOWAIT, 0);
302 mp->mnt_op = vfs;
303 vfs->vfs_refcount++;
304 mp->mnt_vnodecovered = vp;
305 mp->mnt_stat.f_owner = p->p_ucred->cr_uid;
306 mp->mnt_unmounter = NULL;
307 mp->mnt_leaf = mp;
308
309 /*
310 * The underlying file system may refuse the mount for
311 * various reasons. Allow the user to force it to happen.
312 */
313 mp->mnt_flag |= SCARG(uap, flags) & MNT_FORCE;
314 update:
315 /*
316 * Set the mount level flags.
317 */
318 if (SCARG(uap, flags) & MNT_RDONLY)
319 mp->mnt_flag |= MNT_RDONLY;
320 else if (mp->mnt_flag & MNT_RDONLY)
321 mp->mnt_iflag |= IMNT_WANTRDWR;
322 mp->mnt_flag &=
323 ~(MNT_NOSUID | MNT_NOEXEC | MNT_NODEV |
324 MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | MNT_NOCOREDUMP |
325 MNT_NOATIME | MNT_NODEVMTIME | MNT_SYMPERM | MNT_SOFTDEP);
326 mp->mnt_flag |= SCARG(uap, flags) &
327 (MNT_NOSUID | MNT_NOEXEC | MNT_NODEV |
328 MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | MNT_NOCOREDUMP |
329 MNT_NOATIME | MNT_NODEVMTIME | MNT_SYMPERM | MNT_SOFTDEP |
330 MNT_IGNORE);
331 /*
332 * Mount the filesystem.
333 */
334 error = VFS_MOUNT(mp, SCARG(uap, path), SCARG(uap, data), &nd, p);
335 if (mp->mnt_flag & (MNT_UPDATE | MNT_GETARGS)) {
336 if (mp->mnt_iflag & IMNT_WANTRDWR)
337 mp->mnt_flag &= ~MNT_RDONLY;
338 if (error || (mp->mnt_flag & MNT_GETARGS))
339 mp->mnt_flag = flag;
340 mp->mnt_flag &=~
341 (MNT_RELOAD | MNT_FORCE | MNT_UPDATE | MNT_GETARGS);
342 mp->mnt_iflag &=~ IMNT_WANTRDWR;
343 if ((mp->mnt_flag & (MNT_RDONLY | MNT_ASYNC)) == 0) {
344 if (mp->mnt_syncer == NULL)
345 error = vfs_allocate_syncvnode(mp);
346 } else {
347 if (mp->mnt_syncer != NULL)
348 vfs_deallocate_syncvnode(mp);
349 }
350 vfs_unbusy(mp);
351 VOP_UNLOCK(vp, 0);
352 vrele(vp);
353 return (error);
354 }
355 /*
356 * Put the new filesystem on the mount list after root.
357 */
358 cache_purge(vp);
359 if (!error) {
360 mp->mnt_flag &=~
361 (MNT_RELOAD | MNT_FORCE | MNT_UPDATE | MNT_GETARGS);
362 mp->mnt_iflag &=~ IMNT_WANTRDWR;
363 vp->v_mountedhere = mp;
364 simple_lock(&mountlist_slock);
365 CIRCLEQ_INSERT_TAIL(&mountlist, mp, mnt_list);
366 simple_unlock(&mountlist_slock);
367 checkdirs(vp);
368 VOP_UNLOCK(vp, 0);
369 if ((mp->mnt_flag & (MNT_RDONLY | MNT_ASYNC)) == 0)
370 error = vfs_allocate_syncvnode(mp);
371 vfs_unbusy(mp);
372 (void) VFS_STATVFS(mp, &mp->mnt_stat, p);
373 if ((error = VFS_START(mp, 0, p)))
374 vrele(vp);
375 } else {
376 vp->v_mountedhere = (struct mount *)0;
377 vfs->vfs_refcount--;
378 vfs_unbusy(mp);
379 free(mp, M_MOUNT);
380 vput(vp);
381 }
382 return (error);
383 }
384
385 /*
386 * Scan all active processes to see if any of them have a current
387 * or root directory onto which the new filesystem has just been
388 * mounted. If so, replace them with the new mount point.
389 */
390 void
391 checkdirs(olddp)
392 struct vnode *olddp;
393 {
394 struct cwdinfo *cwdi;
395 struct vnode *newdp;
396 struct proc *p;
397
398 if (olddp->v_usecount == 1)
399 return;
400 if (VFS_ROOT(olddp->v_mountedhere, &newdp))
401 panic("mount: lost mount");
402 proclist_lock_read();
403 PROCLIST_FOREACH(p, &allproc) {
404 cwdi = p->p_cwdi;
405 if (cwdi->cwdi_cdir == olddp) {
406 vrele(cwdi->cwdi_cdir);
407 VREF(newdp);
408 cwdi->cwdi_cdir = newdp;
409 }
410 if (cwdi->cwdi_rdir == olddp) {
411 vrele(cwdi->cwdi_rdir);
412 VREF(newdp);
413 cwdi->cwdi_rdir = newdp;
414 }
415 }
416 proclist_unlock_read();
417 if (rootvnode == olddp) {
418 vrele(rootvnode);
419 VREF(newdp);
420 rootvnode = newdp;
421 }
422 vput(newdp);
423 }
424
425 /*
426 * Unmount a file system.
427 *
428 * Note: unmount takes a path to the vnode mounted on as argument,
429 * not special file (as before).
430 */
431 /* ARGSUSED */
432 int
433 sys_unmount(l, v, retval)
434 struct lwp *l;
435 void *v;
436 register_t *retval;
437 {
438 struct sys_unmount_args /* {
439 syscallarg(const char *) path;
440 syscallarg(int) flags;
441 } */ *uap = v;
442 struct proc *p = l->l_proc;
443 struct vnode *vp;
444 struct mount *mp;
445 int error;
446 struct nameidata nd;
447
448 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
449 SCARG(uap, path), p);
450 if ((error = namei(&nd)) != 0)
451 return (error);
452 vp = nd.ni_vp;
453 mp = vp->v_mount;
454
455 /*
456 * Only root, or the user that did the original mount is
457 * permitted to unmount this filesystem.
458 */
459 if ((mp->mnt_stat.f_owner != p->p_ucred->cr_uid) &&
460 (error = suser(p->p_ucred, &p->p_acflag)) != 0) {
461 vput(vp);
462 return (error);
463 }
464
465 /*
466 * Don't allow unmounting the root file system.
467 */
468 if (mp->mnt_flag & MNT_ROOTFS) {
469 vput(vp);
470 return (EINVAL);
471 }
472
473 /*
474 * Must be the root of the filesystem
475 */
476 if ((vp->v_flag & VROOT) == 0) {
477 vput(vp);
478 return (EINVAL);
479 }
480 vput(vp);
481
482 /*
483 * XXX Freeze syncer. Must do this before locking the
484 * mount point. See dounmount() for details.
485 */
486 lockmgr(&syncer_lock, LK_EXCLUSIVE, NULL);
487
488 if (vfs_busy(mp, 0, 0)) {
489 lockmgr(&syncer_lock, LK_RELEASE, NULL);
490 return (EBUSY);
491 }
492
493 return (dounmount(mp, SCARG(uap, flags), p));
494 }
495
496 /*
497 * Do the actual file system unmount. File system is assumed to have been
498 * marked busy by the caller.
499 */
500 int
501 dounmount(mp, flags, p)
502 struct mount *mp;
503 int flags;
504 struct proc *p;
505 {
506 struct vnode *coveredvp;
507 int error;
508 int async;
509 int used_syncer;
510
511 simple_lock(&mountlist_slock);
512 vfs_unbusy(mp);
513 used_syncer = (mp->mnt_syncer != NULL);
514
515 /*
516 * XXX Syncer must be frozen when we get here. This should really
517 * be done on a per-mountpoint basis, but especially the softdep
518 * code possibly called from the syncer doens't exactly work on a
519 * per-mountpoint basis, so the softdep code would become a maze
520 * of vfs_busy() calls.
521 *
522 * The caller of dounmount() must acquire syncer_lock because
523 * the syncer itself acquires locks in syncer_lock -> vfs_busy
524 * order, and we must preserve that order to avoid deadlock.
525 *
526 * So, if the file system did not use the syncer, now is
527 * the time to release the syncer_lock.
528 */
529 if (used_syncer == 0)
530 lockmgr(&syncer_lock, LK_RELEASE, NULL);
531
532 mp->mnt_iflag |= IMNT_UNMOUNT;
533 mp->mnt_unmounter = p;
534 lockmgr(&mp->mnt_lock, LK_DRAIN | LK_INTERLOCK, &mountlist_slock);
535 vn_start_write(NULL, &mp, V_WAIT);
536
537 if (mp->mnt_flag & MNT_EXPUBLIC)
538 vfs_setpublicfs(NULL, NULL, NULL);
539 async = mp->mnt_flag & MNT_ASYNC;
540 mp->mnt_flag &= ~MNT_ASYNC;
541 cache_purgevfs(mp); /* remove cache entries for this file sys */
542 if (mp->mnt_syncer != NULL)
543 vfs_deallocate_syncvnode(mp);
544 error = 0;
545 if ((mp->mnt_flag & MNT_RDONLY) == 0) {
546 #if NFSS > 0
547 error = fss_umount_hook(mp, (flags & MNT_FORCE));
548 #endif
549 if (error == 0)
550 error = VFS_SYNC(mp, MNT_WAIT, p->p_ucred, p);
551 }
552 if (error == 0 || (flags & MNT_FORCE))
553 error = VFS_UNMOUNT(mp, flags, p);
554 vn_finished_write(mp, 0);
555 simple_lock(&mountlist_slock);
556 if (error) {
557 if ((mp->mnt_flag & (MNT_RDONLY | MNT_ASYNC)) == 0)
558 (void) vfs_allocate_syncvnode(mp);
559 mp->mnt_iflag &= ~IMNT_UNMOUNT;
560 mp->mnt_unmounter = NULL;
561 mp->mnt_flag |= async;
562 lockmgr(&mp->mnt_lock, LK_RELEASE | LK_INTERLOCK | LK_REENABLE,
563 &mountlist_slock);
564 if (used_syncer)
565 lockmgr(&syncer_lock, LK_RELEASE, NULL);
566 simple_lock(&mp->mnt_slock);
567 while (mp->mnt_wcnt > 0) {
568 wakeup(mp);
569 ltsleep(&mp->mnt_wcnt, PVFS, "mntwcnt1",
570 0, &mp->mnt_slock);
571 }
572 simple_unlock(&mp->mnt_slock);
573 return (error);
574 }
575 CIRCLEQ_REMOVE(&mountlist, mp, mnt_list);
576 if ((coveredvp = mp->mnt_vnodecovered) != NULLVP) {
577 coveredvp->v_mountedhere = NULL;
578 vrele(coveredvp);
579 }
580 mp->mnt_op->vfs_refcount--;
581 if (LIST_FIRST(&mp->mnt_vnodelist) != NULL)
582 panic("unmount: dangling vnode");
583 mp->mnt_iflag |= IMNT_GONE;
584 lockmgr(&mp->mnt_lock, LK_RELEASE | LK_INTERLOCK, &mountlist_slock);
585 if (used_syncer)
586 lockmgr(&syncer_lock, LK_RELEASE, NULL);
587 simple_lock(&mp->mnt_slock);
588 while (mp->mnt_wcnt > 0) {
589 wakeup(mp);
590 ltsleep(&mp->mnt_wcnt, PVFS, "mntwcnt2", 0, &mp->mnt_slock);
591 }
592 simple_unlock(&mp->mnt_slock);
593 free(mp, M_MOUNT);
594 return (0);
595 }
596
597 /*
598 * Sync each mounted filesystem.
599 */
600 #ifdef DEBUG
601 int syncprt = 0;
602 struct ctldebug debug0 = { "syncprt", &syncprt };
603 #endif
604
605 /* ARGSUSED */
606 int
607 sys_sync(l, v, retval)
608 struct lwp *l;
609 void *v;
610 register_t *retval;
611 {
612 struct mount *mp, *nmp;
613 int asyncflag;
614 struct proc *p = l == NULL ? &proc0 : l->l_proc;
615
616 simple_lock(&mountlist_slock);
617 for (mp = mountlist.cqh_last; mp != (void *)&mountlist; mp = nmp) {
618 if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock)) {
619 nmp = mp->mnt_list.cqe_prev;
620 continue;
621 }
622 if ((mp->mnt_flag & MNT_RDONLY) == 0 &&
623 vn_start_write(NULL, &mp, V_NOWAIT) == 0) {
624 asyncflag = mp->mnt_flag & MNT_ASYNC;
625 mp->mnt_flag &= ~MNT_ASYNC;
626 VFS_SYNC(mp, MNT_NOWAIT, p->p_ucred, p);
627 if (asyncflag)
628 mp->mnt_flag |= MNT_ASYNC;
629 vn_finished_write(mp, 0);
630 }
631 simple_lock(&mountlist_slock);
632 nmp = mp->mnt_list.cqe_prev;
633 vfs_unbusy(mp);
634
635 }
636 simple_unlock(&mountlist_slock);
637 #ifdef DEBUG
638 if (syncprt)
639 vfs_bufstats();
640 #endif /* DEBUG */
641 return (0);
642 }
643
644 /*
645 * Change filesystem quotas.
646 */
647 /* ARGSUSED */
648 int
649 sys_quotactl(l, v, retval)
650 struct lwp *l;
651 void *v;
652 register_t *retval;
653 {
654 struct sys_quotactl_args /* {
655 syscallarg(const char *) path;
656 syscallarg(int) cmd;
657 syscallarg(int) uid;
658 syscallarg(caddr_t) arg;
659 } */ *uap = v;
660 struct proc *p = l->l_proc;
661 struct mount *mp;
662 int error;
663 struct nameidata nd;
664
665 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
666 if ((error = namei(&nd)) != 0)
667 return (error);
668 error = vn_start_write(nd.ni_vp, &mp, V_WAIT | V_PCATCH);
669 vrele(nd.ni_vp);
670 if (error)
671 return (error);
672 error = VFS_QUOTACTL(mp, SCARG(uap, cmd), SCARG(uap, uid),
673 SCARG(uap, arg), p);
674 vn_finished_write(mp, 0);
675 return (error);
676 }
677
678 int
679 dostatvfs(struct mount *mp, struct statvfs *sp, struct proc *p, int flags,
680 int root)
681 {
682 struct cwdinfo *cwdi = p->p_cwdi;
683 int error = 0;
684
685 /*
686 * If MNT_NOWAIT or MNT_LAZY is specified, do not
687 * refresh the fsstat cache. MNT_WAIT or MNT_LAZY
688 * overrides MNT_NOWAIT.
689 */
690 if (flags == MNT_NOWAIT || flags == MNT_LAZY ||
691 (flags != MNT_WAIT && flags != 0)) {
692 memcpy(sp, &mp->mnt_stat, sizeof(*sp));
693 goto done;
694 }
695
696 /* Get the filesystem stats now */
697 memset(sp, 0, sizeof(*sp));
698 if ((error = VFS_STATVFS(mp, sp, p)) != 0) {
699 return error;
700 }
701
702 if (cwdi->cwdi_rdir == NULL)
703 (void)memcpy(&mp->mnt_stat, sp, sizeof(mp->mnt_stat));
704 done:
705 if (cwdi->cwdi_rdir != NULL) {
706 size_t len;
707 char *bp;
708 char *path = malloc(MAXPATHLEN, M_TEMP, M_WAITOK);
709 if (!path)
710 return ENOMEM;
711
712 bp = path + MAXPATHLEN;
713 *--bp = '\0';
714 error = getcwd_common(cwdi->cwdi_rdir, rootvnode, &bp, path,
715 MAXPATHLEN / 2, 0, p);
716 if (error) {
717 free(path, M_TEMP);
718 return error;
719 }
720 len = strlen(bp);
721 /*
722 * for mount points that are below our root, we can see
723 * them, so we fix up the pathname and return them. The
724 * rest we cannot see, so we don't allow viewing the
725 * data.
726 */
727 if (strncmp(bp, sp->f_mntonname, len) == 0) {
728 strlcpy(sp->f_mntonname, &sp->f_mntonname[len],
729 sizeof(sp->f_mntonname));
730 if (sp->f_mntonname[0] == '\0')
731 (void)strlcpy(sp->f_mntonname, "/",
732 sizeof(sp->f_mntonname));
733 } else {
734 if (root)
735 (void)strlcpy(sp->f_mntonname, "/",
736 sizeof(sp->f_mntonname));
737 else
738 error = EPERM;
739 }
740 free(path, M_TEMP);
741 }
742 sp->f_flag = mp->mnt_flag & MNT_VISFLAGMASK;
743 return error;
744 }
745
746 /*
747 * Get filesystem statistics.
748 */
749 /* ARGSUSED */
750 int
751 sys_statvfs1(l, v, retval)
752 struct lwp *l;
753 void *v;
754 register_t *retval;
755 {
756 struct sys_statvfs1_args /* {
757 syscallarg(const char *) path;
758 syscallarg(struct statvfs *) buf;
759 syscallarg(int) flags;
760 } */ *uap = v;
761 struct proc *p = l->l_proc;
762 struct mount *mp;
763 struct statvfs sbuf;
764 int error;
765 struct nameidata nd;
766
767 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
768 if ((error = namei(&nd)) != 0)
769 return error;
770 mp = nd.ni_vp->v_mount;
771 vrele(nd.ni_vp);
772 if ((error = dostatvfs(mp, &sbuf, p, SCARG(uap, flags), 1)) != 0)
773 return error;
774 return copyout(&sbuf, SCARG(uap, buf), sizeof(sbuf));
775 }
776
777 /*
778 * Get filesystem statistics.
779 */
780 /* ARGSUSED */
781 int
782 sys_fstatvfs1(l, v, retval)
783 struct lwp *l;
784 void *v;
785 register_t *retval;
786 {
787 struct sys_fstatvfs1_args /* {
788 syscallarg(int) fd;
789 syscallarg(struct statvfs *) buf;
790 syscallarg(int) flags;
791 } */ *uap = v;
792 struct proc *p = l->l_proc;
793 struct file *fp;
794 struct mount *mp;
795 struct statvfs sbuf;
796 int error;
797
798 /* getvnode() will use the descriptor for us */
799 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
800 return (error);
801 mp = ((struct vnode *)fp->f_data)->v_mount;
802 if ((error = dostatvfs(mp, &sbuf, p, SCARG(uap, flags), 1)) != 0)
803 goto out;
804 error = copyout(&sbuf, SCARG(uap, buf), sizeof(sbuf));
805 out:
806 FILE_UNUSE(fp, p);
807 return error;
808 }
809
810
811 /*
812 * Get statistics on all filesystems.
813 */
814 int
815 sys_getvfsstat(l, v, retval)
816 struct lwp *l;
817 void *v;
818 register_t *retval;
819 {
820 struct sys_getvfsstat_args /* {
821 syscallarg(struct statvfs *) buf;
822 syscallarg(size_t) bufsize;
823 syscallarg(int) flags;
824 } */ *uap = v;
825 int root = 0;
826 struct proc *p = l->l_proc;
827 struct mount *mp, *nmp;
828 struct statvfs sbuf;
829 struct statvfs *sfsp;
830 size_t count, maxcount;
831 int error = 0;
832
833 maxcount = SCARG(uap, bufsize) / sizeof(struct statvfs);
834 sfsp = SCARG(uap, buf);
835 simple_lock(&mountlist_slock);
836 count = 0;
837 for (mp = CIRCLEQ_FIRST(&mountlist); mp != (void *)&mountlist;
838 mp = nmp) {
839 if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock)) {
840 nmp = CIRCLEQ_NEXT(mp, mnt_list);
841 continue;
842 }
843 if (sfsp && count < maxcount) {
844 error = dostatvfs(mp, &sbuf, p, SCARG(uap, flags), 0);
845 if (error) {
846 simple_lock(&mountlist_slock);
847 nmp = CIRCLEQ_NEXT(mp, mnt_list);
848 vfs_unbusy(mp);
849 continue;
850 }
851 error = copyout(&sbuf, sfsp, sizeof(*sfsp));
852 if (error) {
853 vfs_unbusy(mp);
854 return (error);
855 }
856 sfsp++;
857 root |= strcmp(sbuf.f_mntonname, "/") == 0;
858 }
859 count++;
860 simple_lock(&mountlist_slock);
861 nmp = CIRCLEQ_NEXT(mp, mnt_list);
862 vfs_unbusy(mp);
863 }
864 simple_unlock(&mountlist_slock);
865 if (root == 0 && p->p_cwdi->cwdi_rdir) {
866 /*
867 * fake a root entry
868 */
869 if ((error = dostatvfs(p->p_cwdi->cwdi_rdir->v_mount, &sbuf, p,
870 SCARG(uap, flags), 1)) != 0)
871 return error;
872 if (sfsp)
873 error = copyout(&sbuf, sfsp, sizeof(*sfsp));
874 count++;
875 }
876 if (sfsp && count > maxcount)
877 *retval = maxcount;
878 else
879 *retval = count;
880 return error;
881 }
882
883 /*
884 * Change current working directory to a given file descriptor.
885 */
886 /* ARGSUSED */
887 int
888 sys_fchdir(l, v, retval)
889 struct lwp *l;
890 void *v;
891 register_t *retval;
892 {
893 struct sys_fchdir_args /* {
894 syscallarg(int) fd;
895 } */ *uap = v;
896 struct proc *p = l->l_proc;
897 struct filedesc *fdp = p->p_fd;
898 struct cwdinfo *cwdi = p->p_cwdi;
899 struct vnode *vp, *tdp;
900 struct mount *mp;
901 struct file *fp;
902 int error;
903
904 /* getvnode() will use the descriptor for us */
905 if ((error = getvnode(fdp, SCARG(uap, fd), &fp)) != 0)
906 return (error);
907 vp = (struct vnode *)fp->f_data;
908
909 VREF(vp);
910 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
911 if (vp->v_type != VDIR)
912 error = ENOTDIR;
913 else
914 error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
915 while (!error && (mp = vp->v_mountedhere) != NULL) {
916 if (vfs_busy(mp, 0, 0))
917 continue;
918 error = VFS_ROOT(mp, &tdp);
919 vfs_unbusy(mp);
920 if (error)
921 break;
922 vput(vp);
923 vp = tdp;
924 }
925 if (error) {
926 vput(vp);
927 goto out;
928 }
929 VOP_UNLOCK(vp, 0);
930
931 /*
932 * Disallow changing to a directory not under the process's
933 * current root directory (if there is one).
934 */
935 if (cwdi->cwdi_rdir && !vn_isunder(vp, NULL, p)) {
936 vrele(vp);
937 error = EPERM; /* operation not permitted */
938 goto out;
939 }
940
941 vrele(cwdi->cwdi_cdir);
942 cwdi->cwdi_cdir = vp;
943 out:
944 FILE_UNUSE(fp, p);
945 return (error);
946 }
947
948 /*
949 * Change this process's notion of the root directory to a given file descriptor.
950 */
951
952 int
953 sys_fchroot(l, v, retval)
954 struct lwp *l;
955 void *v;
956 register_t *retval;
957 {
958 struct sys_fchroot_args *uap = v;
959 struct proc *p = l->l_proc;
960 struct filedesc *fdp = p->p_fd;
961 struct cwdinfo *cwdi = p->p_cwdi;
962 struct vnode *vp;
963 struct file *fp;
964 int error;
965
966 if ((error = suser(p->p_ucred, &p->p_acflag)) != 0)
967 return error;
968 /* getvnode() will use the descriptor for us */
969 if ((error = getvnode(fdp, SCARG(uap, fd), &fp)) != 0)
970 return error;
971 vp = (struct vnode *) fp->f_data;
972 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
973 if (vp->v_type != VDIR)
974 error = ENOTDIR;
975 else
976 error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
977 VOP_UNLOCK(vp, 0);
978 if (error)
979 goto out;
980 VREF(vp);
981
982 /*
983 * Prevent escaping from chroot by putting the root under
984 * the working directory. Silently chdir to / if we aren't
985 * already there.
986 */
987 if (!vn_isunder(cwdi->cwdi_cdir, vp, p)) {
988 /*
989 * XXX would be more failsafe to change directory to a
990 * deadfs node here instead
991 */
992 vrele(cwdi->cwdi_cdir);
993 VREF(vp);
994 cwdi->cwdi_cdir = vp;
995 }
996
997 if (cwdi->cwdi_rdir != NULL)
998 vrele(cwdi->cwdi_rdir);
999 cwdi->cwdi_rdir = vp;
1000 out:
1001 FILE_UNUSE(fp, p);
1002 return (error);
1003 }
1004
1005
1006
1007 /*
1008 * Change current working directory (``.'').
1009 */
1010 /* ARGSUSED */
1011 int
1012 sys_chdir(l, v, retval)
1013 struct lwp *l;
1014 void *v;
1015 register_t *retval;
1016 {
1017 struct sys_chdir_args /* {
1018 syscallarg(const char *) path;
1019 } */ *uap = v;
1020 struct proc *p = l->l_proc;
1021 struct cwdinfo *cwdi = p->p_cwdi;
1022 int error;
1023 struct nameidata nd;
1024
1025 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1026 SCARG(uap, path), p);
1027 if ((error = change_dir(&nd, p)) != 0)
1028 return (error);
1029 vrele(cwdi->cwdi_cdir);
1030 cwdi->cwdi_cdir = nd.ni_vp;
1031 return (0);
1032 }
1033
1034 /*
1035 * Change notion of root (``/'') directory.
1036 */
1037 /* ARGSUSED */
1038 int
1039 sys_chroot(l, v, retval)
1040 struct lwp *l;
1041 void *v;
1042 register_t *retval;
1043 {
1044 struct sys_chroot_args /* {
1045 syscallarg(const char *) path;
1046 } */ *uap = v;
1047 struct proc *p = l->l_proc;
1048 struct cwdinfo *cwdi = p->p_cwdi;
1049 struct vnode *vp;
1050 int error;
1051 struct nameidata nd;
1052
1053 if ((error = suser(p->p_ucred, &p->p_acflag)) != 0)
1054 return (error);
1055 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1056 SCARG(uap, path), p);
1057 if ((error = change_dir(&nd, p)) != 0)
1058 return (error);
1059 if (cwdi->cwdi_rdir != NULL)
1060 vrele(cwdi->cwdi_rdir);
1061 vp = nd.ni_vp;
1062 cwdi->cwdi_rdir = vp;
1063
1064 /*
1065 * Prevent escaping from chroot by putting the root under
1066 * the working directory. Silently chdir to / if we aren't
1067 * already there.
1068 */
1069 if (!vn_isunder(cwdi->cwdi_cdir, vp, p)) {
1070 /*
1071 * XXX would be more failsafe to change directory to a
1072 * deadfs node here instead
1073 */
1074 vrele(cwdi->cwdi_cdir);
1075 VREF(vp);
1076 cwdi->cwdi_cdir = vp;
1077 }
1078
1079 return (0);
1080 }
1081
1082 /*
1083 * Common routine for chroot and chdir.
1084 */
1085 static int
1086 change_dir(ndp, p)
1087 struct nameidata *ndp;
1088 struct proc *p;
1089 {
1090 struct vnode *vp;
1091 int error;
1092
1093 if ((error = namei(ndp)) != 0)
1094 return (error);
1095 vp = ndp->ni_vp;
1096 if (vp->v_type != VDIR)
1097 error = ENOTDIR;
1098 else
1099 error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
1100
1101 if (error)
1102 vput(vp);
1103 else
1104 VOP_UNLOCK(vp, 0);
1105 return (error);
1106 }
1107
1108 /*
1109 * Check permissions, allocate an open file structure,
1110 * and call the device open routine if any.
1111 */
1112 int
1113 sys_open(l, v, retval)
1114 struct lwp *l;
1115 void *v;
1116 register_t *retval;
1117 {
1118 struct sys_open_args /* {
1119 syscallarg(const char *) path;
1120 syscallarg(int) flags;
1121 syscallarg(int) mode;
1122 } */ *uap = v;
1123 struct proc *p = l->l_proc;
1124 struct cwdinfo *cwdi = p->p_cwdi;
1125 struct filedesc *fdp = p->p_fd;
1126 struct file *fp;
1127 struct vnode *vp;
1128 int flags, cmode;
1129 int type, indx, error;
1130 struct flock lf;
1131 struct nameidata nd;
1132
1133 flags = FFLAGS(SCARG(uap, flags));
1134 if ((flags & (FREAD | FWRITE)) == 0)
1135 return (EINVAL);
1136 /* falloc() will use the file descriptor for us */
1137 if ((error = falloc(p, &fp, &indx)) != 0)
1138 return (error);
1139 cmode = ((SCARG(uap, mode) &~ cwdi->cwdi_cmask) & ALLPERMS) &~ S_ISTXT;
1140 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1141 l->l_dupfd = -indx - 1; /* XXX check for fdopen */
1142 if ((error = vn_open(&nd, flags, cmode)) != 0) {
1143 FILE_UNUSE(fp, p);
1144 fdp->fd_ofiles[indx] = NULL;
1145 ffree(fp);
1146 if ((error == EDUPFD || error == EMOVEFD) &&
1147 l->l_dupfd >= 0 && /* XXX from fdopen */
1148 (error =
1149 dupfdopen(p, indx, l->l_dupfd, flags, error)) == 0) {
1150 *retval = indx;
1151 return (0);
1152 }
1153 if (error == ERESTART)
1154 error = EINTR;
1155 fdremove(fdp, indx);
1156 return (error);
1157 }
1158 l->l_dupfd = 0;
1159 vp = nd.ni_vp;
1160 fp->f_flag = flags & FMASK;
1161 fp->f_type = DTYPE_VNODE;
1162 fp->f_ops = &vnops;
1163 fp->f_data = vp;
1164 if (flags & (O_EXLOCK | O_SHLOCK)) {
1165 lf.l_whence = SEEK_SET;
1166 lf.l_start = 0;
1167 lf.l_len = 0;
1168 if (flags & O_EXLOCK)
1169 lf.l_type = F_WRLCK;
1170 else
1171 lf.l_type = F_RDLCK;
1172 type = F_FLOCK;
1173 if ((flags & FNONBLOCK) == 0)
1174 type |= F_WAIT;
1175 VOP_UNLOCK(vp, 0);
1176 error = VOP_ADVLOCK(vp, fp, F_SETLK, &lf, type);
1177 if (error) {
1178 (void) vn_close(vp, fp->f_flag, fp->f_cred, p);
1179 FILE_UNUSE(fp, p);
1180 ffree(fp);
1181 fdremove(fdp, indx);
1182 return (error);
1183 }
1184 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
1185 fp->f_flag |= FHASLOCK;
1186 }
1187 VOP_UNLOCK(vp, 0);
1188 *retval = indx;
1189 FILE_SET_MATURE(fp);
1190 FILE_UNUSE(fp, p);
1191 return (0);
1192 }
1193
1194 /*
1195 * Get file handle system call
1196 */
1197 int
1198 sys_getfh(l, v, retval)
1199 struct lwp *l;
1200 void *v;
1201 register_t *retval;
1202 {
1203 struct sys_getfh_args /* {
1204 syscallarg(char *) fname;
1205 syscallarg(fhandle_t *) fhp;
1206 } */ *uap = v;
1207 struct proc *p = l->l_proc;
1208 struct vnode *vp;
1209 fhandle_t fh;
1210 int error;
1211 struct nameidata nd;
1212
1213 /*
1214 * Must be super user
1215 */
1216 error = suser(p->p_ucred, &p->p_acflag);
1217 if (error)
1218 return (error);
1219 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
1220 SCARG(uap, fname), p);
1221 error = namei(&nd);
1222 if (error)
1223 return (error);
1224 vp = nd.ni_vp;
1225 memset(&fh, 0, sizeof(fh));
1226 fh.fh_fsid = vp->v_mount->mnt_stat.f_fsidx;
1227 error = VFS_VPTOFH(vp, &fh.fh_fid);
1228 vput(vp);
1229 if (error)
1230 return (error);
1231 error = copyout(&fh, (caddr_t)SCARG(uap, fhp), sizeof (fh));
1232 return (error);
1233 }
1234
1235 /*
1236 * Open a file given a file handle.
1237 *
1238 * Check permissions, allocate an open file structure,
1239 * and call the device open routine if any.
1240 */
1241 int
1242 sys_fhopen(l, v, retval)
1243 struct lwp *l;
1244 void *v;
1245 register_t *retval;
1246 {
1247 struct sys_fhopen_args /* {
1248 syscallarg(const fhandle_t *) fhp;
1249 syscallarg(int) flags;
1250 } */ *uap = v;
1251 struct proc *p = l->l_proc;
1252 struct filedesc *fdp = p->p_fd;
1253 struct file *fp;
1254 struct vnode *vp = NULL;
1255 struct mount *mp;
1256 struct ucred *cred = p->p_ucred;
1257 int flags;
1258 struct file *nfp;
1259 int type, indx, error=0;
1260 struct flock lf;
1261 struct vattr va;
1262 fhandle_t fh;
1263
1264 /*
1265 * Must be super user
1266 */
1267 if ((error = suser(p->p_ucred, &p->p_acflag)))
1268 return (error);
1269
1270 flags = FFLAGS(SCARG(uap, flags));
1271 if ((flags & (FREAD | FWRITE)) == 0)
1272 return (EINVAL);
1273 if ((flags & O_CREAT))
1274 return (EINVAL);
1275 /* falloc() will use the file descriptor for us */
1276 if ((error = falloc(p, &nfp, &indx)) != 0)
1277 return (error);
1278 fp = nfp;
1279 if ((error = copyin(SCARG(uap, fhp), &fh, sizeof(fhandle_t))) != 0)
1280 goto bad;
1281
1282 if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL) {
1283 error = ESTALE;
1284 goto bad;
1285 }
1286
1287 if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)) != 0) {
1288 vp = NULL; /* most likely unnecessary sanity for bad: */
1289 goto bad;
1290 }
1291
1292 /* Now do an effective vn_open */
1293
1294 if (vp->v_type == VSOCK) {
1295 error = EOPNOTSUPP;
1296 goto bad;
1297 }
1298 if (flags & FREAD) {
1299 if ((error = VOP_ACCESS(vp, VREAD, cred, p)) != 0)
1300 goto bad;
1301 }
1302 if (flags & (FWRITE | O_TRUNC)) {
1303 if (vp->v_type == VDIR) {
1304 error = EISDIR;
1305 goto bad;
1306 }
1307 if ((error = vn_writechk(vp)) != 0 ||
1308 (error = VOP_ACCESS(vp, VWRITE, cred, p)) != 0)
1309 goto bad;
1310 }
1311 if (flags & O_TRUNC) {
1312 if ((error = vn_start_write(vp, &mp, V_WAIT | V_PCATCH)) != 0)
1313 goto bad;
1314 VOP_UNLOCK(vp, 0); /* XXX */
1315 VOP_LEASE(vp, p, cred, LEASE_WRITE);
1316 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); /* XXX */
1317 VATTR_NULL(&va);
1318 va.va_size = 0;
1319 error = VOP_SETATTR(vp, &va, cred, p);
1320 vn_finished_write(mp, 0);
1321 if (error)
1322 goto bad;
1323 }
1324 if ((error = VOP_OPEN(vp, flags, cred, p)) != 0)
1325 goto bad;
1326 if (vp->v_type == VREG &&
1327 uvn_attach(vp, flags & FWRITE ? VM_PROT_WRITE : 0) == NULL) {
1328 error = EIO;
1329 goto bad;
1330 }
1331 if (flags & FWRITE)
1332 vp->v_writecount++;
1333
1334 /* done with modified vn_open, now finish what sys_open does. */
1335
1336 fp->f_flag = flags & FMASK;
1337 fp->f_type = DTYPE_VNODE;
1338 fp->f_ops = &vnops;
1339 fp->f_data = vp;
1340 if (flags & (O_EXLOCK | O_SHLOCK)) {
1341 lf.l_whence = SEEK_SET;
1342 lf.l_start = 0;
1343 lf.l_len = 0;
1344 if (flags & O_EXLOCK)
1345 lf.l_type = F_WRLCK;
1346 else
1347 lf.l_type = F_RDLCK;
1348 type = F_FLOCK;
1349 if ((flags & FNONBLOCK) == 0)
1350 type |= F_WAIT;
1351 VOP_UNLOCK(vp, 0);
1352 error = VOP_ADVLOCK(vp, fp, F_SETLK, &lf, type);
1353 if (error) {
1354 (void) vn_close(vp, fp->f_flag, fp->f_cred, p);
1355 FILE_UNUSE(fp, p);
1356 ffree(fp);
1357 fdremove(fdp, indx);
1358 return (error);
1359 }
1360 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
1361 fp->f_flag |= FHASLOCK;
1362 }
1363 VOP_UNLOCK(vp, 0);
1364 *retval = indx;
1365 FILE_SET_MATURE(fp);
1366 FILE_UNUSE(fp, p);
1367 return (0);
1368
1369 bad:
1370 FILE_UNUSE(fp, p);
1371 ffree(fp);
1372 fdremove(fdp, indx);
1373 if (vp != NULL)
1374 vput(vp);
1375 return (error);
1376 }
1377
1378 /* ARGSUSED */
1379 int
1380 sys_fhstat(l, v, retval)
1381 struct lwp *l;
1382 void *v;
1383 register_t *retval;
1384 {
1385 struct sys_fhstat_args /* {
1386 syscallarg(const fhandle_t *) fhp;
1387 syscallarg(struct stat *) sb;
1388 } */ *uap = v;
1389 struct proc *p = l->l_proc;
1390 struct stat sb;
1391 int error;
1392 fhandle_t fh;
1393 struct mount *mp;
1394 struct vnode *vp;
1395
1396 /*
1397 * Must be super user
1398 */
1399 if ((error = suser(p->p_ucred, &p->p_acflag)))
1400 return (error);
1401
1402 if ((error = copyin(SCARG(uap, fhp), &fh, sizeof(fhandle_t))) != 0)
1403 return (error);
1404
1405 if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL)
1406 return (ESTALE);
1407 if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)))
1408 return (error);
1409 error = vn_stat(vp, &sb, p);
1410 vput(vp);
1411 if (error)
1412 return (error);
1413 error = copyout(&sb, SCARG(uap, sb), sizeof(sb));
1414 return (error);
1415 }
1416
1417 /* ARGSUSED */
1418 int
1419 sys_fhstatvfs1(l, v, retval)
1420 struct lwp *l;
1421 void *v;
1422 register_t *retval;
1423 {
1424 struct sys_fhstatvfs1_args /*
1425 syscallarg(const fhandle_t *) fhp;
1426 syscallarg(struct statvfs *) buf;
1427 syscallarg(int) flags;
1428 } */ *uap = v;
1429 struct proc *p = l->l_proc;
1430 struct statvfs sbuf;
1431 fhandle_t fh;
1432 struct mount *mp;
1433 struct vnode *vp;
1434 int error;
1435
1436 /*
1437 * Must be super user
1438 */
1439 if ((error = suser(p->p_ucred, &p->p_acflag)) != 0)
1440 return error;
1441
1442 if ((error = copyin(SCARG(uap, fhp), &fh, sizeof(fhandle_t))) != 0)
1443 return error;
1444
1445 if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL)
1446 return ESTALE;
1447 if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)))
1448 return error;
1449
1450 mp = vp->v_mount;
1451 if ((error = dostatvfs(mp, &sbuf, p, SCARG(uap, flags), 1)) != 0) {
1452 vput(vp);
1453 return error;
1454 }
1455 vput(vp);
1456 return copyout(&sbuf, SCARG(uap, buf), sizeof(sbuf));
1457 }
1458
1459 /*
1460 * Create a special file.
1461 */
1462 /* ARGSUSED */
1463 int
1464 sys_mknod(l, v, retval)
1465 struct lwp *l;
1466 void *v;
1467 register_t *retval;
1468 {
1469 struct sys_mknod_args /* {
1470 syscallarg(const char *) path;
1471 syscallarg(int) mode;
1472 syscallarg(int) dev;
1473 } */ *uap = v;
1474 struct proc *p = l->l_proc;
1475 struct vnode *vp;
1476 struct mount *mp;
1477 struct vattr vattr;
1478 int error;
1479 int whiteout = 0;
1480 struct nameidata nd;
1481
1482 if ((error = suser(p->p_ucred, &p->p_acflag)) != 0)
1483 return (error);
1484 restart:
1485 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1486 if ((error = namei(&nd)) != 0)
1487 return (error);
1488 vp = nd.ni_vp;
1489 if (vp != NULL)
1490 error = EEXIST;
1491 else {
1492 VATTR_NULL(&vattr);
1493 vattr.va_mode =
1494 (SCARG(uap, mode) & ALLPERMS) &~ p->p_cwdi->cwdi_cmask;
1495 vattr.va_rdev = SCARG(uap, dev);
1496 whiteout = 0;
1497
1498 switch (SCARG(uap, mode) & S_IFMT) {
1499 case S_IFMT: /* used by badsect to flag bad sectors */
1500 vattr.va_type = VBAD;
1501 break;
1502 case S_IFCHR:
1503 vattr.va_type = VCHR;
1504 break;
1505 case S_IFBLK:
1506 vattr.va_type = VBLK;
1507 break;
1508 case S_IFWHT:
1509 whiteout = 1;
1510 break;
1511 default:
1512 error = EINVAL;
1513 break;
1514 }
1515 }
1516 if (vn_start_write(nd.ni_dvp, &mp, V_NOWAIT) != 0) {
1517 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1518 if (nd.ni_dvp == vp)
1519 vrele(nd.ni_dvp);
1520 else
1521 vput(nd.ni_dvp);
1522 if (vp)
1523 vrele(vp);
1524 if ((error = vn_start_write(NULL, &mp,
1525 V_WAIT | V_SLEEPONLY | V_PCATCH)) != 0)
1526 return (error);
1527 goto restart;
1528 }
1529 if (!error) {
1530 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1531 if (whiteout) {
1532 error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, CREATE);
1533 if (error)
1534 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1535 vput(nd.ni_dvp);
1536 } else {
1537 error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp,
1538 &nd.ni_cnd, &vattr);
1539 if (error == 0)
1540 vput(nd.ni_vp);
1541 }
1542 } else {
1543 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1544 if (nd.ni_dvp == vp)
1545 vrele(nd.ni_dvp);
1546 else
1547 vput(nd.ni_dvp);
1548 if (vp)
1549 vrele(vp);
1550 }
1551 vn_finished_write(mp, 0);
1552 return (error);
1553 }
1554
1555 /*
1556 * Create a named pipe.
1557 */
1558 /* ARGSUSED */
1559 int
1560 sys_mkfifo(l, v, retval)
1561 struct lwp *l;
1562 void *v;
1563 register_t *retval;
1564 {
1565 struct sys_mkfifo_args /* {
1566 syscallarg(const char *) path;
1567 syscallarg(int) mode;
1568 } */ *uap = v;
1569 struct proc *p = l->l_proc;
1570 struct mount *mp;
1571 struct vattr vattr;
1572 int error;
1573 struct nameidata nd;
1574
1575 restart:
1576 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
1577 if ((error = namei(&nd)) != 0)
1578 return (error);
1579 if (nd.ni_vp != NULL) {
1580 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1581 if (nd.ni_dvp == nd.ni_vp)
1582 vrele(nd.ni_dvp);
1583 else
1584 vput(nd.ni_dvp);
1585 vrele(nd.ni_vp);
1586 return (EEXIST);
1587 }
1588 if (vn_start_write(nd.ni_dvp, &mp, V_NOWAIT) != 0) {
1589 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1590 if (nd.ni_dvp == nd.ni_vp)
1591 vrele(nd.ni_dvp);
1592 else
1593 vput(nd.ni_dvp);
1594 if (nd.ni_vp)
1595 vrele(nd.ni_vp);
1596 if ((error = vn_start_write(NULL, &mp,
1597 V_WAIT | V_SLEEPONLY | V_PCATCH)) != 0)
1598 return (error);
1599 goto restart;
1600 }
1601 VATTR_NULL(&vattr);
1602 vattr.va_type = VFIFO;
1603 vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_cwdi->cwdi_cmask;
1604 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1605 error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
1606 if (error == 0)
1607 vput(nd.ni_vp);
1608 vn_finished_write(mp, 0);
1609 return (error);
1610 }
1611
1612 /*
1613 * Make a hard file link.
1614 */
1615 /* ARGSUSED */
1616 int
1617 sys_link(l, v, retval)
1618 struct lwp *l;
1619 void *v;
1620 register_t *retval;
1621 {
1622 struct sys_link_args /* {
1623 syscallarg(const char *) path;
1624 syscallarg(const char *) link;
1625 } */ *uap = v;
1626 struct proc *p = l->l_proc;
1627 struct vnode *vp;
1628 struct mount *mp;
1629 struct nameidata nd;
1630 int error;
1631
1632 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
1633 if ((error = namei(&nd)) != 0)
1634 return (error);
1635 vp = nd.ni_vp;
1636 if ((error = vn_start_write(vp, &mp, V_WAIT | V_PCATCH)) != 0) {
1637 vrele(vp);
1638 return (error);
1639 }
1640 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, link), p);
1641 if ((error = namei(&nd)) != 0)
1642 goto out;
1643 if (nd.ni_vp) {
1644 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1645 if (nd.ni_dvp == nd.ni_vp)
1646 vrele(nd.ni_dvp);
1647 else
1648 vput(nd.ni_dvp);
1649 vrele(nd.ni_vp);
1650 error = EEXIST;
1651 goto out;
1652 }
1653 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1654 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1655 error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd);
1656 out:
1657 vrele(vp);
1658 vn_finished_write(mp, 0);
1659 return (error);
1660 }
1661
1662 /*
1663 * Make a symbolic link.
1664 */
1665 /* ARGSUSED */
1666 int
1667 sys_symlink(l, v, retval)
1668 struct lwp *l;
1669 void *v;
1670 register_t *retval;
1671 {
1672 struct sys_symlink_args /* {
1673 syscallarg(const char *) path;
1674 syscallarg(const char *) link;
1675 } */ *uap = v;
1676 struct proc *p = l->l_proc;
1677 struct mount *mp;
1678 struct vattr vattr;
1679 char *path;
1680 int error;
1681 struct nameidata nd;
1682
1683 path = PNBUF_GET();
1684 error = copyinstr(SCARG(uap, path), path, MAXPATHLEN, NULL);
1685 if (error)
1686 goto out;
1687 restart:
1688 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, link), p);
1689 if ((error = namei(&nd)) != 0)
1690 goto out;
1691 if (nd.ni_vp) {
1692 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1693 if (nd.ni_dvp == nd.ni_vp)
1694 vrele(nd.ni_dvp);
1695 else
1696 vput(nd.ni_dvp);
1697 vrele(nd.ni_vp);
1698 error = EEXIST;
1699 goto out;
1700 }
1701 if (vn_start_write(nd.ni_dvp, &mp, V_NOWAIT) != 0) {
1702 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1703 if (nd.ni_dvp == nd.ni_vp)
1704 vrele(nd.ni_dvp);
1705 else
1706 vput(nd.ni_dvp);
1707 if ((error = vn_start_write(NULL, &mp,
1708 V_WAIT | V_SLEEPONLY | V_PCATCH)) != 0)
1709 return (error);
1710 goto restart;
1711 }
1712 VATTR_NULL(&vattr);
1713 vattr.va_mode = ACCESSPERMS &~ p->p_cwdi->cwdi_cmask;
1714 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1715 error = VOP_SYMLINK(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr, path);
1716 if (error == 0)
1717 vput(nd.ni_vp);
1718 vn_finished_write(mp, 0);
1719 out:
1720 PNBUF_PUT(path);
1721 return (error);
1722 }
1723
1724 /*
1725 * Delete a whiteout from the filesystem.
1726 */
1727 /* ARGSUSED */
1728 int
1729 sys_undelete(l, v, retval)
1730 struct lwp *l;
1731 void *v;
1732 register_t *retval;
1733 {
1734 struct sys_undelete_args /* {
1735 syscallarg(const char *) path;
1736 } */ *uap = v;
1737 struct proc *p = l->l_proc;
1738 int error;
1739 struct mount *mp;
1740 struct nameidata nd;
1741
1742 restart:
1743 NDINIT(&nd, DELETE, LOCKPARENT|DOWHITEOUT, UIO_USERSPACE,
1744 SCARG(uap, path), p);
1745 error = namei(&nd);
1746 if (error)
1747 return (error);
1748
1749 if (nd.ni_vp != NULLVP || !(nd.ni_cnd.cn_flags & ISWHITEOUT)) {
1750 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1751 if (nd.ni_dvp == nd.ni_vp)
1752 vrele(nd.ni_dvp);
1753 else
1754 vput(nd.ni_dvp);
1755 if (nd.ni_vp)
1756 vrele(nd.ni_vp);
1757 return (EEXIST);
1758 }
1759 if (vn_start_write(nd.ni_dvp, &mp, V_NOWAIT) != 0) {
1760 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1761 if (nd.ni_dvp == nd.ni_vp)
1762 vrele(nd.ni_dvp);
1763 else
1764 vput(nd.ni_dvp);
1765 if ((error = vn_start_write(NULL, &mp,
1766 V_WAIT | V_SLEEPONLY | V_PCATCH)) != 0)
1767 return (error);
1768 goto restart;
1769 }
1770 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1771 if ((error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, DELETE)) != 0)
1772 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1773 vput(nd.ni_dvp);
1774 vn_finished_write(mp, 0);
1775 return (error);
1776 }
1777
1778 /*
1779 * Delete a name from the filesystem.
1780 */
1781 /* ARGSUSED */
1782 int
1783 sys_unlink(l, v, retval)
1784 struct lwp *l;
1785 void *v;
1786 register_t *retval;
1787 {
1788 struct sys_unlink_args /* {
1789 syscallarg(const char *) path;
1790 } */ *uap = v;
1791 struct proc *p = l->l_proc;
1792 struct mount *mp;
1793 struct vnode *vp;
1794 int error;
1795 struct nameidata nd;
1796
1797 restart:
1798 NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
1799 SCARG(uap, path), p);
1800 if ((error = namei(&nd)) != 0)
1801 return (error);
1802 vp = nd.ni_vp;
1803
1804 /*
1805 * The root of a mounted filesystem cannot be deleted.
1806 */
1807 if (vp->v_flag & VROOT) {
1808 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1809 if (nd.ni_dvp == vp)
1810 vrele(nd.ni_dvp);
1811 else
1812 vput(nd.ni_dvp);
1813 vput(vp);
1814 error = EBUSY;
1815 goto out;
1816 }
1817 if (vn_start_write(nd.ni_dvp, &mp, V_NOWAIT) != 0) {
1818 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
1819 if (nd.ni_dvp == vp)
1820 vrele(nd.ni_dvp);
1821 else
1822 vput(nd.ni_dvp);
1823 vput(vp);
1824 if ((error = vn_start_write(NULL, &mp,
1825 V_WAIT | V_SLEEPONLY | V_PCATCH)) != 0)
1826 return (error);
1827 goto restart;
1828 }
1829 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
1830 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
1831 error = VOP_REMOVE(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
1832 vn_finished_write(mp, 0);
1833 out:
1834 return (error);
1835 }
1836
1837 /*
1838 * Reposition read/write file offset.
1839 */
1840 int
1841 sys_lseek(l, v, retval)
1842 struct lwp *l;
1843 void *v;
1844 register_t *retval;
1845 {
1846 struct sys_lseek_args /* {
1847 syscallarg(int) fd;
1848 syscallarg(int) pad;
1849 syscallarg(off_t) offset;
1850 syscallarg(int) whence;
1851 } */ *uap = v;
1852 struct proc *p = l->l_proc;
1853 struct ucred *cred = p->p_ucred;
1854 struct filedesc *fdp = p->p_fd;
1855 struct file *fp;
1856 struct vnode *vp;
1857 struct vattr vattr;
1858 off_t newoff;
1859 int error;
1860
1861 if ((fp = fd_getfile(fdp, SCARG(uap, fd))) == NULL)
1862 return (EBADF);
1863
1864 FILE_USE(fp);
1865
1866 vp = (struct vnode *)fp->f_data;
1867 if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
1868 error = ESPIPE;
1869 goto out;
1870 }
1871
1872 switch (SCARG(uap, whence)) {
1873 case SEEK_CUR:
1874 newoff = fp->f_offset + SCARG(uap, offset);
1875 break;
1876 case SEEK_END:
1877 error = VOP_GETATTR(vp, &vattr, cred, p);
1878 if (error)
1879 goto out;
1880 newoff = SCARG(uap, offset) + vattr.va_size;
1881 break;
1882 case SEEK_SET:
1883 newoff = SCARG(uap, offset);
1884 break;
1885 default:
1886 error = EINVAL;
1887 goto out;
1888 }
1889 if ((error = VOP_SEEK(vp, fp->f_offset, newoff, cred)) != 0)
1890 goto out;
1891
1892 *(off_t *)retval = fp->f_offset = newoff;
1893 out:
1894 FILE_UNUSE(fp, p);
1895 return (error);
1896 }
1897
1898 /*
1899 * Positional read system call.
1900 */
1901 int
1902 sys_pread(l, v, retval)
1903 struct lwp *l;
1904 void *v;
1905 register_t *retval;
1906 {
1907 struct sys_pread_args /* {
1908 syscallarg(int) fd;
1909 syscallarg(void *) buf;
1910 syscallarg(size_t) nbyte;
1911 syscallarg(off_t) offset;
1912 } */ *uap = v;
1913 struct proc *p = l->l_proc;
1914 struct filedesc *fdp = p->p_fd;
1915 struct file *fp;
1916 struct vnode *vp;
1917 off_t offset;
1918 int error, fd = SCARG(uap, fd);
1919
1920 if ((fp = fd_getfile(fdp, fd)) == NULL)
1921 return (EBADF);
1922
1923 if ((fp->f_flag & FREAD) == 0) {
1924 simple_unlock(&fp->f_slock);
1925 return (EBADF);
1926 }
1927
1928 FILE_USE(fp);
1929
1930 vp = (struct vnode *)fp->f_data;
1931 if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
1932 error = ESPIPE;
1933 goto out;
1934 }
1935
1936 offset = SCARG(uap, offset);
1937
1938 /*
1939 * XXX This works because no file systems actually
1940 * XXX take any action on the seek operation.
1941 */
1942 if ((error = VOP_SEEK(vp, fp->f_offset, offset, fp->f_cred)) != 0)
1943 goto out;
1944
1945 /* dofileread() will unuse the descriptor for us */
1946 return (dofileread(p, fd, fp, SCARG(uap, buf), SCARG(uap, nbyte),
1947 &offset, 0, retval));
1948
1949 out:
1950 FILE_UNUSE(fp, p);
1951 return (error);
1952 }
1953
1954 /*
1955 * Positional scatter read system call.
1956 */
1957 int
1958 sys_preadv(l, v, retval)
1959 struct lwp *l;
1960 void *v;
1961 register_t *retval;
1962 {
1963 struct sys_preadv_args /* {
1964 syscallarg(int) fd;
1965 syscallarg(const struct iovec *) iovp;
1966 syscallarg(int) iovcnt;
1967 syscallarg(off_t) offset;
1968 } */ *uap = v;
1969 struct proc *p = l->l_proc;
1970 struct filedesc *fdp = p->p_fd;
1971 struct file *fp;
1972 struct vnode *vp;
1973 off_t offset;
1974 int error, fd = SCARG(uap, fd);
1975
1976 if ((fp = fd_getfile(fdp, fd)) == NULL)
1977 return (EBADF);
1978
1979 if ((fp->f_flag & FREAD) == 0) {
1980 simple_unlock(&fp->f_slock);
1981 return (EBADF);
1982 }
1983
1984 FILE_USE(fp);
1985
1986 vp = (struct vnode *)fp->f_data;
1987 if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
1988 error = ESPIPE;
1989 goto out;
1990 }
1991
1992 offset = SCARG(uap, offset);
1993
1994 /*
1995 * XXX This works because no file systems actually
1996 * XXX take any action on the seek operation.
1997 */
1998 if ((error = VOP_SEEK(vp, fp->f_offset, offset, fp->f_cred)) != 0)
1999 goto out;
2000
2001 /* dofilereadv() will unuse the descriptor for us */
2002 return (dofilereadv(p, fd, fp, SCARG(uap, iovp), SCARG(uap, iovcnt),
2003 &offset, 0, retval));
2004
2005 out:
2006 FILE_UNUSE(fp, p);
2007 return (error);
2008 }
2009
2010 /*
2011 * Positional write system call.
2012 */
2013 int
2014 sys_pwrite(l, v, retval)
2015 struct lwp *l;
2016 void *v;
2017 register_t *retval;
2018 {
2019 struct sys_pwrite_args /* {
2020 syscallarg(int) fd;
2021 syscallarg(const void *) buf;
2022 syscallarg(size_t) nbyte;
2023 syscallarg(off_t) offset;
2024 } */ *uap = v;
2025 struct proc *p = l->l_proc;
2026 struct filedesc *fdp = p->p_fd;
2027 struct file *fp;
2028 struct vnode *vp;
2029 off_t offset;
2030 int error, fd = SCARG(uap, fd);
2031
2032 if ((fp = fd_getfile(fdp, fd)) == NULL)
2033 return (EBADF);
2034
2035 if ((fp->f_flag & FWRITE) == 0) {
2036 simple_unlock(&fp->f_slock);
2037 return (EBADF);
2038 }
2039
2040 FILE_USE(fp);
2041
2042 vp = (struct vnode *)fp->f_data;
2043 if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
2044 error = ESPIPE;
2045 goto out;
2046 }
2047
2048 offset = SCARG(uap, offset);
2049
2050 /*
2051 * XXX This works because no file systems actually
2052 * XXX take any action on the seek operation.
2053 */
2054 if ((error = VOP_SEEK(vp, fp->f_offset, offset, fp->f_cred)) != 0)
2055 goto out;
2056
2057 /* dofilewrite() will unuse the descriptor for us */
2058 return (dofilewrite(p, fd, fp, SCARG(uap, buf), SCARG(uap, nbyte),
2059 &offset, 0, retval));
2060
2061 out:
2062 FILE_UNUSE(fp, p);
2063 return (error);
2064 }
2065
2066 /*
2067 * Positional gather write system call.
2068 */
2069 int
2070 sys_pwritev(l, v, retval)
2071 struct lwp *l;
2072 void *v;
2073 register_t *retval;
2074 {
2075 struct sys_pwritev_args /* {
2076 syscallarg(int) fd;
2077 syscallarg(const struct iovec *) iovp;
2078 syscallarg(int) iovcnt;
2079 syscallarg(off_t) offset;
2080 } */ *uap = v;
2081 struct proc *p = l->l_proc;
2082 struct filedesc *fdp = p->p_fd;
2083 struct file *fp;
2084 struct vnode *vp;
2085 off_t offset;
2086 int error, fd = SCARG(uap, fd);
2087
2088 if ((fp = fd_getfile(fdp, fd)) == NULL)
2089 return (EBADF);
2090
2091 if ((fp->f_flag & FWRITE) == 0) {
2092 simple_unlock(&fp->f_slock);
2093 return (EBADF);
2094 }
2095
2096 FILE_USE(fp);
2097
2098 vp = (struct vnode *)fp->f_data;
2099 if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
2100 error = ESPIPE;
2101 goto out;
2102 }
2103
2104 offset = SCARG(uap, offset);
2105
2106 /*
2107 * XXX This works because no file systems actually
2108 * XXX take any action on the seek operation.
2109 */
2110 if ((error = VOP_SEEK(vp, fp->f_offset, offset, fp->f_cred)) != 0)
2111 goto out;
2112
2113 /* dofilewritev() will unuse the descriptor for us */
2114 return (dofilewritev(p, fd, fp, SCARG(uap, iovp), SCARG(uap, iovcnt),
2115 &offset, 0, retval));
2116
2117 out:
2118 FILE_UNUSE(fp, p);
2119 return (error);
2120 }
2121
2122 /*
2123 * Check access permissions.
2124 */
2125 int
2126 sys_access(l, v, retval)
2127 struct lwp *l;
2128 void *v;
2129 register_t *retval;
2130 {
2131 struct sys_access_args /* {
2132 syscallarg(const char *) path;
2133 syscallarg(int) flags;
2134 } */ *uap = v;
2135 struct proc *p = l->l_proc;
2136 struct ucred *cred;
2137 struct vnode *vp;
2138 int error, flags;
2139 struct nameidata nd;
2140
2141 cred = crdup(p->p_ucred);
2142 cred->cr_uid = p->p_cred->p_ruid;
2143 cred->cr_gid = p->p_cred->p_rgid;
2144 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
2145 SCARG(uap, path), p);
2146 /* Override default credentials */
2147 nd.ni_cnd.cn_cred = cred;
2148 if ((error = namei(&nd)) != 0)
2149 goto out;
2150 vp = nd.ni_vp;
2151
2152 /* Flags == 0 means only check for existence. */
2153 if (SCARG(uap, flags)) {
2154 flags = 0;
2155 if (SCARG(uap, flags) & R_OK)
2156 flags |= VREAD;
2157 if (SCARG(uap, flags) & W_OK)
2158 flags |= VWRITE;
2159 if (SCARG(uap, flags) & X_OK)
2160 flags |= VEXEC;
2161
2162 error = VOP_ACCESS(vp, flags, cred, p);
2163 if (!error && (flags & VWRITE))
2164 error = vn_writechk(vp);
2165 }
2166 vput(vp);
2167 out:
2168 crfree(cred);
2169 return (error);
2170 }
2171
2172 /*
2173 * Get file status; this version follows links.
2174 */
2175 /* ARGSUSED */
2176 int
2177 sys___stat13(l, v, retval)
2178 struct lwp *l;
2179 void *v;
2180 register_t *retval;
2181 {
2182 struct sys___stat13_args /* {
2183 syscallarg(const char *) path;
2184 syscallarg(struct stat *) ub;
2185 } */ *uap = v;
2186 struct proc *p = l->l_proc;
2187 struct stat sb;
2188 int error;
2189 struct nameidata nd;
2190
2191 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
2192 SCARG(uap, path), p);
2193 if ((error = namei(&nd)) != 0)
2194 return (error);
2195 error = vn_stat(nd.ni_vp, &sb, p);
2196 vput(nd.ni_vp);
2197 if (error)
2198 return (error);
2199 error = copyout(&sb, SCARG(uap, ub), sizeof(sb));
2200 return (error);
2201 }
2202
2203 /*
2204 * Get file status; this version does not follow links.
2205 */
2206 /* ARGSUSED */
2207 int
2208 sys___lstat13(l, v, retval)
2209 struct lwp *l;
2210 void *v;
2211 register_t *retval;
2212 {
2213 struct sys___lstat13_args /* {
2214 syscallarg(const char *) path;
2215 syscallarg(struct stat *) ub;
2216 } */ *uap = v;
2217 struct proc *p = l->l_proc;
2218 struct stat sb;
2219 int error;
2220 struct nameidata nd;
2221
2222 NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF, UIO_USERSPACE,
2223 SCARG(uap, path), p);
2224 if ((error = namei(&nd)) != 0)
2225 return (error);
2226 error = vn_stat(nd.ni_vp, &sb, p);
2227 vput(nd.ni_vp);
2228 if (error)
2229 return (error);
2230 error = copyout(&sb, SCARG(uap, ub), sizeof(sb));
2231 return (error);
2232 }
2233
2234 /*
2235 * Get configurable pathname variables.
2236 */
2237 /* ARGSUSED */
2238 int
2239 sys_pathconf(l, v, retval)
2240 struct lwp *l;
2241 void *v;
2242 register_t *retval;
2243 {
2244 struct sys_pathconf_args /* {
2245 syscallarg(const char *) path;
2246 syscallarg(int) name;
2247 } */ *uap = v;
2248 struct proc *p = l->l_proc;
2249 int error;
2250 struct nameidata nd;
2251
2252 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
2253 SCARG(uap, path), p);
2254 if ((error = namei(&nd)) != 0)
2255 return (error);
2256 error = VOP_PATHCONF(nd.ni_vp, SCARG(uap, name), retval);
2257 vput(nd.ni_vp);
2258 return (error);
2259 }
2260
2261 /*
2262 * Return target name of a symbolic link.
2263 */
2264 /* ARGSUSED */
2265 int
2266 sys_readlink(l, v, retval)
2267 struct lwp *l;
2268 void *v;
2269 register_t *retval;
2270 {
2271 struct sys_readlink_args /* {
2272 syscallarg(const char *) path;
2273 syscallarg(char *) buf;
2274 syscallarg(size_t) count;
2275 } */ *uap = v;
2276 struct proc *p = l->l_proc;
2277 struct vnode *vp;
2278 struct iovec aiov;
2279 struct uio auio;
2280 int error;
2281 struct nameidata nd;
2282
2283 NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF, UIO_USERSPACE,
2284 SCARG(uap, path), p);
2285 if ((error = namei(&nd)) != 0)
2286 return (error);
2287 vp = nd.ni_vp;
2288 if (vp->v_type != VLNK)
2289 error = EINVAL;
2290 else if (!(vp->v_mount->mnt_flag & MNT_SYMPERM) ||
2291 (error = VOP_ACCESS(vp, VREAD, p->p_ucred, p)) == 0) {
2292 aiov.iov_base = SCARG(uap, buf);
2293 aiov.iov_len = SCARG(uap, count);
2294 auio.uio_iov = &aiov;
2295 auio.uio_iovcnt = 1;
2296 auio.uio_offset = 0;
2297 auio.uio_rw = UIO_READ;
2298 auio.uio_segflg = UIO_USERSPACE;
2299 auio.uio_procp = p;
2300 auio.uio_resid = SCARG(uap, count);
2301 error = VOP_READLINK(vp, &auio, p->p_ucred);
2302 }
2303 vput(vp);
2304 *retval = SCARG(uap, count) - auio.uio_resid;
2305 return (error);
2306 }
2307
2308 /*
2309 * Change flags of a file given a path name.
2310 */
2311 /* ARGSUSED */
2312 int
2313 sys_chflags(l, v, retval)
2314 struct lwp *l;
2315 void *v;
2316 register_t *retval;
2317 {
2318 struct sys_chflags_args /* {
2319 syscallarg(const char *) path;
2320 syscallarg(u_long) flags;
2321 } */ *uap = v;
2322 struct proc *p = l->l_proc;
2323 struct vnode *vp;
2324 int error;
2325 struct nameidata nd;
2326
2327 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2328 if ((error = namei(&nd)) != 0)
2329 return (error);
2330 vp = nd.ni_vp;
2331 error = change_flags(vp, SCARG(uap, flags), p);
2332 vput(vp);
2333 return (error);
2334 }
2335
2336 /*
2337 * Change flags of a file given a file descriptor.
2338 */
2339 /* ARGSUSED */
2340 int
2341 sys_fchflags(l, v, retval)
2342 struct lwp *l;
2343 void *v;
2344 register_t *retval;
2345 {
2346 struct sys_fchflags_args /* {
2347 syscallarg(int) fd;
2348 syscallarg(u_long) flags;
2349 } */ *uap = v;
2350 struct proc *p = l->l_proc;
2351 struct vnode *vp;
2352 struct file *fp;
2353 int error;
2354
2355 /* getvnode() will use the descriptor for us */
2356 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2357 return (error);
2358 vp = (struct vnode *)fp->f_data;
2359 error = change_flags(vp, SCARG(uap, flags), p);
2360 VOP_UNLOCK(vp, 0);
2361 FILE_UNUSE(fp, p);
2362 return (error);
2363 }
2364
2365 /*
2366 * Change flags of a file given a path name; this version does
2367 * not follow links.
2368 */
2369 int
2370 sys_lchflags(l, v, retval)
2371 struct lwp *l;
2372 void *v;
2373 register_t *retval;
2374 {
2375 struct sys_lchflags_args /* {
2376 syscallarg(const char *) path;
2377 syscallarg(u_long) flags;
2378 } */ *uap = v;
2379 struct proc *p = l->l_proc;
2380 struct vnode *vp;
2381 int error;
2382 struct nameidata nd;
2383
2384 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2385 if ((error = namei(&nd)) != 0)
2386 return (error);
2387 vp = nd.ni_vp;
2388 error = change_flags(vp, SCARG(uap, flags), p);
2389 vput(vp);
2390 return (error);
2391 }
2392
2393 /*
2394 * Common routine to change flags of a file.
2395 */
2396 int
2397 change_flags(vp, flags, p)
2398 struct vnode *vp;
2399 u_long flags;
2400 struct proc *p;
2401 {
2402 struct mount *mp;
2403 struct vattr vattr;
2404 int error;
2405
2406 if ((error = vn_start_write(vp, &mp, V_WAIT | V_PCATCH)) != 0)
2407 return (error);
2408 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2409 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2410 /*
2411 * Non-superusers cannot change the flags on devices, even if they
2412 * own them.
2413 */
2414 if (suser(p->p_ucred, &p->p_acflag) != 0) {
2415 if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)) != 0)
2416 goto out;
2417 if (vattr.va_type == VCHR || vattr.va_type == VBLK) {
2418 error = EINVAL;
2419 goto out;
2420 }
2421 }
2422 VATTR_NULL(&vattr);
2423 vattr.va_flags = flags;
2424 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2425 out:
2426 vn_finished_write(mp, 0);
2427 return (error);
2428 }
2429
2430 /*
2431 * Change mode of a file given path name; this version follows links.
2432 */
2433 /* ARGSUSED */
2434 int
2435 sys_chmod(l, v, retval)
2436 struct lwp *l;
2437 void *v;
2438 register_t *retval;
2439 {
2440 struct sys_chmod_args /* {
2441 syscallarg(const char *) path;
2442 syscallarg(int) mode;
2443 } */ *uap = v;
2444 struct proc *p = l->l_proc;
2445 int error;
2446 struct nameidata nd;
2447
2448 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2449 if ((error = namei(&nd)) != 0)
2450 return (error);
2451
2452 error = change_mode(nd.ni_vp, SCARG(uap, mode), p);
2453
2454 vrele(nd.ni_vp);
2455 return (error);
2456 }
2457
2458 /*
2459 * Change mode of a file given a file descriptor.
2460 */
2461 /* ARGSUSED */
2462 int
2463 sys_fchmod(l, v, retval)
2464 struct lwp *l;
2465 void *v;
2466 register_t *retval;
2467 {
2468 struct sys_fchmod_args /* {
2469 syscallarg(int) fd;
2470 syscallarg(int) mode;
2471 } */ *uap = v;
2472 struct proc *p = l->l_proc;
2473 struct file *fp;
2474 int error;
2475
2476 /* getvnode() will use the descriptor for us */
2477 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2478 return (error);
2479
2480 error = change_mode((struct vnode *)fp->f_data, SCARG(uap, mode), p);
2481 FILE_UNUSE(fp, p);
2482 return (error);
2483 }
2484
2485 /*
2486 * Change mode of a file given path name; this version does not follow links.
2487 */
2488 /* ARGSUSED */
2489 int
2490 sys_lchmod(l, v, retval)
2491 struct lwp *l;
2492 void *v;
2493 register_t *retval;
2494 {
2495 struct sys_lchmod_args /* {
2496 syscallarg(const char *) path;
2497 syscallarg(int) mode;
2498 } */ *uap = v;
2499 struct proc *p = l->l_proc;
2500 int error;
2501 struct nameidata nd;
2502
2503 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2504 if ((error = namei(&nd)) != 0)
2505 return (error);
2506
2507 error = change_mode(nd.ni_vp, SCARG(uap, mode), p);
2508
2509 vrele(nd.ni_vp);
2510 return (error);
2511 }
2512
2513 /*
2514 * Common routine to set mode given a vnode.
2515 */
2516 static int
2517 change_mode(vp, mode, p)
2518 struct vnode *vp;
2519 int mode;
2520 struct proc *p;
2521 {
2522 struct mount *mp;
2523 struct vattr vattr;
2524 int error;
2525
2526 if ((error = vn_start_write(vp, &mp, V_WAIT | V_PCATCH)) != 0)
2527 return (error);
2528 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2529 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2530 VATTR_NULL(&vattr);
2531 vattr.va_mode = mode & ALLPERMS;
2532 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2533 VOP_UNLOCK(vp, 0);
2534 vn_finished_write(mp, 0);
2535 return (error);
2536 }
2537
2538 /*
2539 * Set ownership given a path name; this version follows links.
2540 */
2541 /* ARGSUSED */
2542 int
2543 sys_chown(l, v, retval)
2544 struct lwp *l;
2545 void *v;
2546 register_t *retval;
2547 {
2548 struct sys_chown_args /* {
2549 syscallarg(const char *) path;
2550 syscallarg(uid_t) uid;
2551 syscallarg(gid_t) gid;
2552 } */ *uap = v;
2553 struct proc *p = l->l_proc;
2554 int error;
2555 struct nameidata nd;
2556
2557 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2558 if ((error = namei(&nd)) != 0)
2559 return (error);
2560
2561 error = change_owner(nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid), p, 0);
2562
2563 vrele(nd.ni_vp);
2564 return (error);
2565 }
2566
2567 /*
2568 * Set ownership given a path name; this version follows links.
2569 * Provides POSIX semantics.
2570 */
2571 /* ARGSUSED */
2572 int
2573 sys___posix_chown(l, v, retval)
2574 struct lwp *l;
2575 void *v;
2576 register_t *retval;
2577 {
2578 struct sys_chown_args /* {
2579 syscallarg(const char *) path;
2580 syscallarg(uid_t) uid;
2581 syscallarg(gid_t) gid;
2582 } */ *uap = v;
2583 struct proc *p = l->l_proc;
2584 int error;
2585 struct nameidata nd;
2586
2587 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2588 if ((error = namei(&nd)) != 0)
2589 return (error);
2590
2591 error = change_owner(nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid), p, 1);
2592
2593 vrele(nd.ni_vp);
2594 return (error);
2595 }
2596
2597 /*
2598 * Set ownership given a file descriptor.
2599 */
2600 /* ARGSUSED */
2601 int
2602 sys_fchown(l, v, retval)
2603 struct lwp *l;
2604 void *v;
2605 register_t *retval;
2606 {
2607 struct sys_fchown_args /* {
2608 syscallarg(int) fd;
2609 syscallarg(uid_t) uid;
2610 syscallarg(gid_t) gid;
2611 } */ *uap = v;
2612 struct proc *p = l->l_proc;
2613 int error;
2614 struct file *fp;
2615
2616 /* getvnode() will use the descriptor for us */
2617 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2618 return (error);
2619
2620 error = change_owner((struct vnode *)fp->f_data, SCARG(uap, uid),
2621 SCARG(uap, gid), p, 0);
2622 FILE_UNUSE(fp, p);
2623 return (error);
2624 }
2625
2626 /*
2627 * Set ownership given a file descriptor, providing POSIX/XPG semantics.
2628 */
2629 /* ARGSUSED */
2630 int
2631 sys___posix_fchown(l, v, retval)
2632 struct lwp *l;
2633 void *v;
2634 register_t *retval;
2635 {
2636 struct sys_fchown_args /* {
2637 syscallarg(int) fd;
2638 syscallarg(uid_t) uid;
2639 syscallarg(gid_t) gid;
2640 } */ *uap = v;
2641 struct proc *p = l->l_proc;
2642 int error;
2643 struct file *fp;
2644
2645 /* getvnode() will use the descriptor for us */
2646 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2647 return (error);
2648
2649 error = change_owner((struct vnode *)fp->f_data, SCARG(uap, uid),
2650 SCARG(uap, gid), p, 1);
2651 FILE_UNUSE(fp, p);
2652 return (error);
2653 }
2654
2655 /*
2656 * Set ownership given a path name; this version does not follow links.
2657 */
2658 /* ARGSUSED */
2659 int
2660 sys_lchown(l, v, retval)
2661 struct lwp *l;
2662 void *v;
2663 register_t *retval;
2664 {
2665 struct sys_lchown_args /* {
2666 syscallarg(const char *) path;
2667 syscallarg(uid_t) uid;
2668 syscallarg(gid_t) gid;
2669 } */ *uap = v;
2670 struct proc *p = l->l_proc;
2671 int error;
2672 struct nameidata nd;
2673
2674 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2675 if ((error = namei(&nd)) != 0)
2676 return (error);
2677
2678 error = change_owner(nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid), p, 0);
2679
2680 vrele(nd.ni_vp);
2681 return (error);
2682 }
2683
2684 /*
2685 * Set ownership given a path name; this version does not follow links.
2686 * Provides POSIX/XPG semantics.
2687 */
2688 /* ARGSUSED */
2689 int
2690 sys___posix_lchown(l, v, retval)
2691 struct lwp *l;
2692 void *v;
2693 register_t *retval;
2694 {
2695 struct sys_lchown_args /* {
2696 syscallarg(const char *) path;
2697 syscallarg(uid_t) uid;
2698 syscallarg(gid_t) gid;
2699 } */ *uap = v;
2700 struct proc *p = l->l_proc;
2701 int error;
2702 struct nameidata nd;
2703
2704 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2705 if ((error = namei(&nd)) != 0)
2706 return (error);
2707
2708 error = change_owner(nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid), p, 1);
2709
2710 vrele(nd.ni_vp);
2711 return (error);
2712 }
2713
2714 /*
2715 * Common routine to set ownership given a vnode.
2716 */
2717 static int
2718 change_owner(vp, uid, gid, p, posix_semantics)
2719 struct vnode *vp;
2720 uid_t uid;
2721 gid_t gid;
2722 struct proc *p;
2723 int posix_semantics;
2724 {
2725 struct mount *mp;
2726 struct vattr vattr;
2727 mode_t newmode;
2728 int error;
2729
2730 if ((error = vn_start_write(vp, &mp, V_WAIT | V_PCATCH)) != 0)
2731 return (error);
2732 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2733 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2734 if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)) != 0)
2735 goto out;
2736
2737 #define CHANGED(x) ((int)(x) != -1)
2738 newmode = vattr.va_mode;
2739 if (posix_semantics) {
2740 /*
2741 * POSIX/XPG semantics: if the caller is not the super-user,
2742 * clear set-user-id and set-group-id bits. Both POSIX and
2743 * the XPG consider the behaviour for calls by the super-user
2744 * implementation-defined; we leave the set-user-id and set-
2745 * group-id settings intact in that case.
2746 */
2747 if (suser(p->p_ucred, NULL) != 0)
2748 newmode &= ~(S_ISUID | S_ISGID);
2749 } else {
2750 /*
2751 * NetBSD semantics: when changing owner and/or group,
2752 * clear the respective bit(s).
2753 */
2754 if (CHANGED(uid))
2755 newmode &= ~S_ISUID;
2756 if (CHANGED(gid))
2757 newmode &= ~S_ISGID;
2758 }
2759 /* Update va_mode iff altered. */
2760 if (vattr.va_mode == newmode)
2761 newmode = VNOVAL;
2762
2763 VATTR_NULL(&vattr);
2764 vattr.va_uid = CHANGED(uid) ? uid : (uid_t)VNOVAL;
2765 vattr.va_gid = CHANGED(gid) ? gid : (gid_t)VNOVAL;
2766 vattr.va_mode = newmode;
2767 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2768 #undef CHANGED
2769
2770 out:
2771 VOP_UNLOCK(vp, 0);
2772 vn_finished_write(mp, 0);
2773 return (error);
2774 }
2775
2776 /*
2777 * Set the access and modification times given a path name; this
2778 * version follows links.
2779 */
2780 /* ARGSUSED */
2781 int
2782 sys_utimes(l, v, retval)
2783 struct lwp *l;
2784 void *v;
2785 register_t *retval;
2786 {
2787 struct sys_utimes_args /* {
2788 syscallarg(const char *) path;
2789 syscallarg(const struct timeval *) tptr;
2790 } */ *uap = v;
2791 struct proc *p = l->l_proc;
2792 int error;
2793 struct nameidata nd;
2794
2795 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2796 if ((error = namei(&nd)) != 0)
2797 return (error);
2798
2799 error = change_utimes(nd.ni_vp, SCARG(uap, tptr), p);
2800
2801 vrele(nd.ni_vp);
2802 return (error);
2803 }
2804
2805 /*
2806 * Set the access and modification times given a file descriptor.
2807 */
2808 /* ARGSUSED */
2809 int
2810 sys_futimes(l, v, retval)
2811 struct lwp *l;
2812 void *v;
2813 register_t *retval;
2814 {
2815 struct sys_futimes_args /* {
2816 syscallarg(int) fd;
2817 syscallarg(const struct timeval *) tptr;
2818 } */ *uap = v;
2819 struct proc *p = l->l_proc;
2820 int error;
2821 struct file *fp;
2822
2823 /* getvnode() will use the descriptor for us */
2824 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2825 return (error);
2826
2827 error = change_utimes((struct vnode *)fp->f_data, SCARG(uap, tptr), p);
2828 FILE_UNUSE(fp, p);
2829 return (error);
2830 }
2831
2832 /*
2833 * Set the access and modification times given a path name; this
2834 * version does not follow links.
2835 */
2836 /* ARGSUSED */
2837 int
2838 sys_lutimes(l, v, retval)
2839 struct lwp *l;
2840 void *v;
2841 register_t *retval;
2842 {
2843 struct sys_lutimes_args /* {
2844 syscallarg(const char *) path;
2845 syscallarg(const struct timeval *) tptr;
2846 } */ *uap = v;
2847 struct proc *p = l->l_proc;
2848 int error;
2849 struct nameidata nd;
2850
2851 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2852 if ((error = namei(&nd)) != 0)
2853 return (error);
2854
2855 error = change_utimes(nd.ni_vp, SCARG(uap, tptr), p);
2856
2857 vrele(nd.ni_vp);
2858 return (error);
2859 }
2860
2861 /*
2862 * Common routine to set access and modification times given a vnode.
2863 */
2864 static int
2865 change_utimes(vp, tptr, p)
2866 struct vnode *vp;
2867 const struct timeval *tptr;
2868 struct proc *p;
2869 {
2870 struct timeval tv[2];
2871 struct mount *mp;
2872 struct vattr vattr;
2873 int error;
2874
2875 if ((error = vn_start_write(vp, &mp, V_WAIT | V_PCATCH)) != 0)
2876 return (error);
2877 VATTR_NULL(&vattr);
2878 if (tptr == NULL) {
2879 microtime(&tv[0]);
2880 tv[1] = tv[0];
2881 vattr.va_vaflags |= VA_UTIMES_NULL;
2882 } else {
2883 error = copyin(tptr, tv, sizeof(tv));
2884 if (error)
2885 goto out;
2886 }
2887 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2888 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2889 vattr.va_atime.tv_sec = tv[0].tv_sec;
2890 vattr.va_atime.tv_nsec = tv[0].tv_usec * 1000;
2891 vattr.va_mtime.tv_sec = tv[1].tv_sec;
2892 vattr.va_mtime.tv_nsec = tv[1].tv_usec * 1000;
2893 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2894 VOP_UNLOCK(vp, 0);
2895 out:
2896 vn_finished_write(mp, 0);
2897 return (error);
2898 }
2899
2900 /*
2901 * Truncate a file given its path name.
2902 */
2903 /* ARGSUSED */
2904 int
2905 sys_truncate(l, v, retval)
2906 struct lwp *l;
2907 void *v;
2908 register_t *retval;
2909 {
2910 struct sys_truncate_args /* {
2911 syscallarg(const char *) path;
2912 syscallarg(int) pad;
2913 syscallarg(off_t) length;
2914 } */ *uap = v;
2915 struct proc *p = l->l_proc;
2916 struct vnode *vp;
2917 struct mount *mp;
2918 struct vattr vattr;
2919 int error;
2920 struct nameidata nd;
2921
2922 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
2923 if ((error = namei(&nd)) != 0)
2924 return (error);
2925 vp = nd.ni_vp;
2926 if ((error = vn_start_write(vp, &mp, V_WAIT | V_PCATCH)) != 0) {
2927 vrele(vp);
2928 return (error);
2929 }
2930 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2931 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2932 if (vp->v_type == VDIR)
2933 error = EISDIR;
2934 else if ((error = vn_writechk(vp)) == 0 &&
2935 (error = VOP_ACCESS(vp, VWRITE, p->p_ucred, p)) == 0) {
2936 VATTR_NULL(&vattr);
2937 vattr.va_size = SCARG(uap, length);
2938 error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
2939 }
2940 vput(vp);
2941 vn_finished_write(mp, 0);
2942 return (error);
2943 }
2944
2945 /*
2946 * Truncate a file given a file descriptor.
2947 */
2948 /* ARGSUSED */
2949 int
2950 sys_ftruncate(l, v, retval)
2951 struct lwp *l;
2952 void *v;
2953 register_t *retval;
2954 {
2955 struct sys_ftruncate_args /* {
2956 syscallarg(int) fd;
2957 syscallarg(int) pad;
2958 syscallarg(off_t) length;
2959 } */ *uap = v;
2960 struct proc *p = l->l_proc;
2961 struct mount *mp;
2962 struct vattr vattr;
2963 struct vnode *vp;
2964 struct file *fp;
2965 int error;
2966
2967 /* getvnode() will use the descriptor for us */
2968 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2969 return (error);
2970 if ((fp->f_flag & FWRITE) == 0) {
2971 error = EINVAL;
2972 goto out;
2973 }
2974 vp = (struct vnode *)fp->f_data;
2975 if ((error = vn_start_write(vp, &mp, V_WAIT | V_PCATCH)) != 0) {
2976 FILE_UNUSE(fp, p);
2977 return (error);
2978 }
2979 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
2980 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
2981 if (vp->v_type == VDIR)
2982 error = EISDIR;
2983 else if ((error = vn_writechk(vp)) == 0) {
2984 VATTR_NULL(&vattr);
2985 vattr.va_size = SCARG(uap, length);
2986 error = VOP_SETATTR(vp, &vattr, fp->f_cred, p);
2987 }
2988 VOP_UNLOCK(vp, 0);
2989 vn_finished_write(mp, 0);
2990 out:
2991 FILE_UNUSE(fp, p);
2992 return (error);
2993 }
2994
2995 /*
2996 * Sync an open file.
2997 */
2998 /* ARGSUSED */
2999 int
3000 sys_fsync(l, v, retval)
3001 struct lwp *l;
3002 void *v;
3003 register_t *retval;
3004 {
3005 struct sys_fsync_args /* {
3006 syscallarg(int) fd;
3007 } */ *uap = v;
3008 struct proc *p = l->l_proc;
3009 struct vnode *vp;
3010 struct mount *mp;
3011 struct file *fp;
3012 int error;
3013
3014 /* getvnode() will use the descriptor for us */
3015 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
3016 return (error);
3017 vp = (struct vnode *)fp->f_data;
3018 if ((error = vn_start_write(vp, &mp, V_WAIT | V_PCATCH)) != 0) {
3019 FILE_UNUSE(fp, p);
3020 return (error);
3021 }
3022 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
3023 error = VOP_FSYNC(vp, fp->f_cred, FSYNC_WAIT, 0, 0, p);
3024 if (error == 0 && bioops.io_fsync != NULL &&
3025 vp->v_mount && (vp->v_mount->mnt_flag & MNT_SOFTDEP))
3026 (*bioops.io_fsync)(vp);
3027 VOP_UNLOCK(vp, 0);
3028 vn_finished_write(mp, 0);
3029 FILE_UNUSE(fp, p);
3030 return (error);
3031 }
3032
3033 /*
3034 * Sync a range of file data. API modeled after that found in AIX.
3035 *
3036 * FDATASYNC indicates that we need only save enough metadata to be able
3037 * to re-read the written data. Note we duplicate AIX's requirement that
3038 * the file be open for writing.
3039 */
3040 /* ARGSUSED */
3041 int
3042 sys_fsync_range(l, v, retval)
3043 struct lwp *l;
3044 void *v;
3045 register_t *retval;
3046 {
3047 struct sys_fsync_range_args /* {
3048 syscallarg(int) fd;
3049 syscallarg(int) flags;
3050 syscallarg(off_t) start;
3051 syscallarg(int) length;
3052 } */ *uap = v;
3053 struct proc *p = l->l_proc;
3054 struct vnode *vp;
3055 struct file *fp;
3056 int flags, nflags;
3057 off_t s, e, len;
3058 int error;
3059
3060 /* getvnode() will use the descriptor for us */
3061 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
3062 return (error);
3063
3064 if ((fp->f_flag & FWRITE) == 0) {
3065 FILE_UNUSE(fp, p);
3066 return (EBADF);
3067 }
3068
3069 flags = SCARG(uap, flags);
3070 if (((flags & (FDATASYNC | FFILESYNC)) == 0) ||
3071 ((~flags & (FDATASYNC | FFILESYNC)) == 0)) {
3072 return (EINVAL);
3073 }
3074 /* Now set up the flags for value(s) to pass to VOP_FSYNC() */
3075 if (flags & FDATASYNC)
3076 nflags = FSYNC_DATAONLY | FSYNC_WAIT;
3077 else
3078 nflags = FSYNC_WAIT;
3079
3080 len = SCARG(uap, length);
3081 /* If length == 0, we do the whole file, and s = l = 0 will do that */
3082 if (len) {
3083 s = SCARG(uap, start);
3084 e = s + len;
3085 if (e < s) {
3086 FILE_UNUSE(fp, p);
3087 return (EINVAL);
3088 }
3089 } else {
3090 e = 0;
3091 s = 0;
3092 }
3093
3094 vp = (struct vnode *)fp->f_data;
3095 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
3096 error = VOP_FSYNC(vp, fp->f_cred, nflags, s, e, p);
3097
3098 if (error == 0 && bioops.io_fsync != NULL &&
3099 vp->v_mount && (vp->v_mount->mnt_flag & MNT_SOFTDEP))
3100 (*bioops.io_fsync)(vp);
3101
3102 VOP_UNLOCK(vp, 0);
3103 FILE_UNUSE(fp, p);
3104 return (error);
3105 }
3106
3107 /*
3108 * Sync the data of an open file.
3109 */
3110 /* ARGSUSED */
3111 int
3112 sys_fdatasync(l, v, retval)
3113 struct lwp *l;
3114 void *v;
3115 register_t *retval;
3116 {
3117 struct sys_fdatasync_args /* {
3118 syscallarg(int) fd;
3119 } */ *uap = v;
3120 struct proc *p = l->l_proc;
3121 struct vnode *vp;
3122 struct file *fp;
3123 int error;
3124
3125 /* getvnode() will use the descriptor for us */
3126 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
3127 return (error);
3128 if ((fp->f_flag & FWRITE) == 0) {
3129 FILE_UNUSE(fp, p);
3130 return (EBADF);
3131 }
3132 vp = (struct vnode *)fp->f_data;
3133 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
3134 error = VOP_FSYNC(vp, fp->f_cred, FSYNC_WAIT|FSYNC_DATAONLY, 0, 0, p);
3135 VOP_UNLOCK(vp, 0);
3136 FILE_UNUSE(fp, p);
3137 return (error);
3138 }
3139
3140 /*
3141 * Rename files, (standard) BSD semantics frontend.
3142 */
3143 /* ARGSUSED */
3144 int
3145 sys_rename(l, v, retval)
3146 struct lwp *l;
3147 void *v;
3148 register_t *retval;
3149 {
3150 struct sys_rename_args /* {
3151 syscallarg(const char *) from;
3152 syscallarg(const char *) to;
3153 } */ *uap = v;
3154 struct proc *p = l->l_proc;
3155
3156 return (rename_files(SCARG(uap, from), SCARG(uap, to), p, 0));
3157 }
3158
3159 /*
3160 * Rename files, POSIX semantics frontend.
3161 */
3162 /* ARGSUSED */
3163 int
3164 sys___posix_rename(l, v, retval)
3165 struct lwp *l;
3166 void *v;
3167 register_t *retval;
3168 {
3169 struct sys___posix_rename_args /* {
3170 syscallarg(const char *) from;
3171 syscallarg(const char *) to;
3172 } */ *uap = v;
3173 struct proc *p = l->l_proc;
3174
3175 return (rename_files(SCARG(uap, from), SCARG(uap, to), p, 1));
3176 }
3177
3178 /*
3179 * Rename files. Source and destination must either both be directories,
3180 * or both not be directories. If target is a directory, it must be empty.
3181 * If `from' and `to' refer to the same object, the value of the `retain'
3182 * argument is used to determine whether `from' will be
3183 *
3184 * (retain == 0) deleted unless `from' and `to' refer to the same
3185 * object in the file system's name space (BSD).
3186 * (retain == 1) always retained (POSIX).
3187 */
3188 static int
3189 rename_files(from, to, p, retain)
3190 const char *from, *to;
3191 struct proc *p;
3192 int retain;
3193 {
3194 struct mount *mp = NULL;
3195 struct vnode *tvp, *fvp, *tdvp;
3196 struct nameidata fromnd, tond;
3197 int error;
3198
3199 NDINIT(&fromnd, DELETE, WANTPARENT | SAVESTART, UIO_USERSPACE,
3200 from, p);
3201 if ((error = namei(&fromnd)) != 0)
3202 return (error);
3203 fvp = fromnd.ni_vp;
3204 error = vn_start_write(fvp, &mp, V_WAIT | V_PCATCH);
3205 if (error != 0) {
3206 VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
3207 vrele(fromnd.ni_dvp);
3208 vrele(fvp);
3209 if (fromnd.ni_startdir)
3210 vrele(fromnd.ni_startdir);
3211 PNBUF_PUT(fromnd.ni_cnd.cn_pnbuf);
3212 return (error);
3213 }
3214 NDINIT(&tond, RENAME, LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART |
3215 (fvp->v_type == VDIR ? CREATEDIR : 0), UIO_USERSPACE, to, p);
3216 if ((error = namei(&tond)) != 0) {
3217 VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
3218 vrele(fromnd.ni_dvp);
3219 vrele(fvp);
3220 goto out1;
3221 }
3222 tdvp = tond.ni_dvp;
3223 tvp = tond.ni_vp;
3224
3225 if (tvp != NULL) {
3226 if (fvp->v_type == VDIR && tvp->v_type != VDIR) {
3227 error = ENOTDIR;
3228 goto out;
3229 } else if (fvp->v_type != VDIR && tvp->v_type == VDIR) {
3230 error = EISDIR;
3231 goto out;
3232 }
3233 }
3234
3235 if (fvp == tdvp)
3236 error = EINVAL;
3237
3238 /*
3239 * Source and destination refer to the same object.
3240 */
3241 if (fvp == tvp) {
3242 if (retain)
3243 error = -1;
3244 else if (fromnd.ni_dvp == tdvp &&
3245 fromnd.ni_cnd.cn_namelen == tond.ni_cnd.cn_namelen &&
3246 !memcmp(fromnd.ni_cnd.cn_nameptr,
3247 tond.ni_cnd.cn_nameptr,
3248 fromnd.ni_cnd.cn_namelen))
3249 error = -1;
3250 }
3251
3252 out:
3253 if (!error) {
3254 VOP_LEASE(tdvp, p, p->p_ucred, LEASE_WRITE);
3255 if (fromnd.ni_dvp != tdvp)
3256 VOP_LEASE(fromnd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
3257 if (tvp) {
3258 VOP_LEASE(tvp, p, p->p_ucred, LEASE_WRITE);
3259 }
3260 error = VOP_RENAME(fromnd.ni_dvp, fromnd.ni_vp, &fromnd.ni_cnd,
3261 tond.ni_dvp, tond.ni_vp, &tond.ni_cnd);
3262 } else {
3263 VOP_ABORTOP(tond.ni_dvp, &tond.ni_cnd);
3264 if (tdvp == tvp)
3265 vrele(tdvp);
3266 else
3267 vput(tdvp);
3268 if (tvp)
3269 vput(tvp);
3270 VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
3271 vrele(fromnd.ni_dvp);
3272 vrele(fvp);
3273 }
3274 vrele(tond.ni_startdir);
3275 PNBUF_PUT(tond.ni_cnd.cn_pnbuf);
3276 out1:
3277 vn_finished_write(mp, 0);
3278 if (fromnd.ni_startdir)
3279 vrele(fromnd.ni_startdir);
3280 PNBUF_PUT(fromnd.ni_cnd.cn_pnbuf);
3281 return (error == -1 ? 0 : error);
3282 }
3283
3284 /*
3285 * Make a directory file.
3286 */
3287 /* ARGSUSED */
3288 int
3289 sys_mkdir(l, v, retval)
3290 struct lwp *l;
3291 void *v;
3292 register_t *retval;
3293 {
3294 struct sys_mkdir_args /* {
3295 syscallarg(const char *) path;
3296 syscallarg(int) mode;
3297 } */ *uap = v;
3298 struct proc *p = l->l_proc;
3299 struct mount *mp;
3300 struct vnode *vp;
3301 struct vattr vattr;
3302 int error;
3303 struct nameidata nd;
3304
3305 restart:
3306 NDINIT(&nd, CREATE, LOCKPARENT | CREATEDIR, UIO_USERSPACE,
3307 SCARG(uap, path), p);
3308 if ((error = namei(&nd)) != 0)
3309 return (error);
3310 vp = nd.ni_vp;
3311 if (vp != NULL) {
3312 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
3313 if (nd.ni_dvp == vp)
3314 vrele(nd.ni_dvp);
3315 else
3316 vput(nd.ni_dvp);
3317 vrele(vp);
3318 return (EEXIST);
3319 }
3320 if (vn_start_write(nd.ni_dvp, &mp, V_NOWAIT) != 0) {
3321 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
3322 if (nd.ni_dvp == vp)
3323 vrele(nd.ni_dvp);
3324 else
3325 vput(nd.ni_dvp);
3326 if ((error = vn_start_write(NULL, &mp,
3327 V_WAIT | V_SLEEPONLY | V_PCATCH)) != 0)
3328 return (error);
3329 goto restart;
3330 }
3331 VATTR_NULL(&vattr);
3332 vattr.va_type = VDIR;
3333 vattr.va_mode =
3334 (SCARG(uap, mode) & ACCESSPERMS) &~ p->p_cwdi->cwdi_cmask;
3335 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
3336 error = VOP_MKDIR(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
3337 if (!error)
3338 vput(nd.ni_vp);
3339 vn_finished_write(mp, 0);
3340 return (error);
3341 }
3342
3343 /*
3344 * Remove a directory file.
3345 */
3346 /* ARGSUSED */
3347 int
3348 sys_rmdir(l, v, retval)
3349 struct lwp *l;
3350 void *v;
3351 register_t *retval;
3352 {
3353 struct sys_rmdir_args /* {
3354 syscallarg(const char *) path;
3355 } */ *uap = v;
3356 struct proc *p = l->l_proc;
3357 struct mount *mp;
3358 struct vnode *vp;
3359 int error;
3360 struct nameidata nd;
3361
3362 restart:
3363 NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
3364 SCARG(uap, path), p);
3365 if ((error = namei(&nd)) != 0)
3366 return (error);
3367 vp = nd.ni_vp;
3368 if (vp->v_type != VDIR) {
3369 error = ENOTDIR;
3370 goto out;
3371 }
3372 /*
3373 * No rmdir "." please.
3374 */
3375 if (nd.ni_dvp == vp) {
3376 error = EINVAL;
3377 goto out;
3378 }
3379 /*
3380 * The root of a mounted filesystem cannot be deleted.
3381 */
3382 if (vp->v_flag & VROOT) {
3383 error = EBUSY;
3384 goto out;
3385 }
3386 if (vn_start_write(nd.ni_dvp, &mp, V_NOWAIT) != 0) {
3387 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
3388 if (nd.ni_dvp == vp)
3389 vrele(nd.ni_dvp);
3390 else
3391 vput(nd.ni_dvp);
3392 vput(vp);
3393 if ((error = vn_start_write(NULL, &mp,
3394 V_WAIT | V_SLEEPONLY | V_PCATCH)) != 0)
3395 return (error);
3396 goto restart;
3397 }
3398 VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
3399 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
3400 error = VOP_RMDIR(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
3401 vn_finished_write(mp, 0);
3402 return (error);
3403
3404 out:
3405 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
3406 if (nd.ni_dvp == vp)
3407 vrele(nd.ni_dvp);
3408 else
3409 vput(nd.ni_dvp);
3410 vput(vp);
3411 return (error);
3412 }
3413
3414 /*
3415 * Read a block of directory entries in a file system independent format.
3416 */
3417 int
3418 sys_getdents(l, v, retval)
3419 struct lwp *l;
3420 void *v;
3421 register_t *retval;
3422 {
3423 struct sys_getdents_args /* {
3424 syscallarg(int) fd;
3425 syscallarg(char *) buf;
3426 syscallarg(size_t) count;
3427 } */ *uap = v;
3428 struct proc *p = l->l_proc;
3429 struct file *fp;
3430 int error, done;
3431
3432 /* getvnode() will use the descriptor for us */
3433 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
3434 return (error);
3435 if ((fp->f_flag & FREAD) == 0) {
3436 error = EBADF;
3437 goto out;
3438 }
3439 error = vn_readdir(fp, SCARG(uap, buf), UIO_USERSPACE,
3440 SCARG(uap, count), &done, p, 0, 0);
3441 #ifdef KTRACE
3442 if (!error && KTRPOINT(p, KTR_GENIO)) {
3443 struct iovec iov;
3444 iov.iov_base = SCARG(uap, buf);
3445 iov.iov_len = done;
3446 ktrgenio(p, SCARG(uap, fd), UIO_READ, &iov, done, 0);
3447 }
3448 #endif
3449 *retval = done;
3450 out:
3451 FILE_UNUSE(fp, p);
3452 return (error);
3453 }
3454
3455 /*
3456 * Set the mode mask for creation of filesystem nodes.
3457 */
3458 int
3459 sys_umask(l, v, retval)
3460 struct lwp *l;
3461 void *v;
3462 register_t *retval;
3463 {
3464 struct sys_umask_args /* {
3465 syscallarg(mode_t) newmask;
3466 } */ *uap = v;
3467 struct proc *p = l->l_proc;
3468 struct cwdinfo *cwdi;
3469
3470 cwdi = p->p_cwdi;
3471 *retval = cwdi->cwdi_cmask;
3472 cwdi->cwdi_cmask = SCARG(uap, newmask) & ALLPERMS;
3473 return (0);
3474 }
3475
3476 /*
3477 * Void all references to file by ripping underlying filesystem
3478 * away from vnode.
3479 */
3480 /* ARGSUSED */
3481 int
3482 sys_revoke(l, v, retval)
3483 struct lwp *l;
3484 void *v;
3485 register_t *retval;
3486 {
3487 struct sys_revoke_args /* {
3488 syscallarg(const char *) path;
3489 } */ *uap = v;
3490 struct proc *p = l->l_proc;
3491 struct mount *mp;
3492 struct vnode *vp;
3493 struct vattr vattr;
3494 int error;
3495 struct nameidata nd;
3496
3497 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
3498 if ((error = namei(&nd)) != 0)
3499 return (error);
3500 vp = nd.ni_vp;
3501 if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)) != 0)
3502 goto out;
3503 if (p->p_ucred->cr_uid != vattr.va_uid &&
3504 (error = suser(p->p_ucred, &p->p_acflag)) != 0)
3505 goto out;
3506 if ((error = vn_start_write(vp, &mp, V_WAIT | V_PCATCH)) != 0)
3507 goto out;
3508 if (vp->v_usecount > 1 || (vp->v_flag & (VALIASED | VLAYER)))
3509 VOP_REVOKE(vp, REVOKEALL);
3510 vn_finished_write(mp, 0);
3511 out:
3512 vrele(vp);
3513 return (error);
3514 }
3515
3516 /*
3517 * Convert a user file descriptor to a kernel file entry.
3518 */
3519 int
3520 getvnode(fdp, fd, fpp)
3521 struct filedesc *fdp;
3522 int fd;
3523 struct file **fpp;
3524 {
3525 struct vnode *vp;
3526 struct file *fp;
3527
3528 if ((fp = fd_getfile(fdp, fd)) == NULL)
3529 return (EBADF);
3530
3531 FILE_USE(fp);
3532
3533 if (fp->f_type != DTYPE_VNODE) {
3534 FILE_UNUSE(fp, NULL);
3535 return (EINVAL);
3536 }
3537
3538 vp = (struct vnode *)fp->f_data;
3539 if (vp->v_type == VBAD) {
3540 FILE_UNUSE(fp, NULL);
3541 return (EBADF);
3542 }
3543
3544 *fpp = fp;
3545 return (0);
3546 }
3547
3548 /*
3549 * Push extended attribute configuration information into the VFS.
3550 *
3551 * NOTE: Not all file systems that support extended attributes will
3552 * require the use of this system call.
3553 */
3554 int
3555 sys_extattrctl(struct lwp *l, void *v, register_t *retval)
3556 {
3557 struct sys_extattrctl_args /* {
3558 syscallarg(const char *) path;
3559 syscallarg(int) cmd;
3560 syscallarg(const char *) filename;
3561 syscallarg(int) attrnamespace;
3562 syscallarg(const char *) attrname;
3563 } */ *uap = v;
3564 struct proc *p = l->l_proc;
3565 struct vnode *vp;
3566 struct nameidata nd;
3567 struct mount *mp;
3568 char attrname[EXTATTR_MAXNAMELEN];
3569 int error;
3570
3571 if (SCARG(uap, attrname) != NULL) {
3572 error = copyinstr(SCARG(uap, attrname), attrname,
3573 sizeof(attrname), NULL);
3574 if (error)
3575 return (error);
3576 }
3577
3578 vp = NULL;
3579 if (SCARG(uap, filename) != NULL) {
3580 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
3581 SCARG(uap, filename), p);
3582 error = namei(&nd);
3583 if (error)
3584 return (error);
3585 vp = nd.ni_vp;
3586 }
3587
3588 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
3589 error = namei(&nd);
3590 if (error) {
3591 if (vp != NULL)
3592 vput(vp);
3593 return (error);
3594 }
3595
3596 error = vn_start_write(nd.ni_vp, &mp, V_WAIT | V_PCATCH);
3597 if (error) {
3598 if (vp != NULL)
3599 vput(vp);
3600 return (error);
3601 }
3602
3603 error = VFS_EXTATTRCTL(mp, SCARG(uap, cmd), vp,
3604 SCARG(uap, attrnamespace),
3605 SCARG(uap, attrname) != NULL ? attrname : NULL, p);
3606
3607 vn_finished_write(mp, 0);
3608
3609 if (vp != NULL)
3610 vrele(vp);
3611
3612 return (error);
3613 }
3614
3615 /*
3616 * Set a named extended attribute on a file or directory.
3617 */
3618 static int
3619 extattr_set_vp(struct vnode *vp, int attrnamespace, const char *attrname,
3620 const void *data, size_t nbytes, struct proc *p, register_t *retval)
3621 {
3622 struct mount *mp;
3623 struct uio auio;
3624 struct iovec aiov;
3625 ssize_t cnt;
3626 int error;
3627
3628 error = vn_start_write(vp, &mp, V_WAIT | V_PCATCH);
3629 if (error)
3630 return (error);
3631 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
3632 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
3633
3634 aiov.iov_base = (caddr_t) data; /* XXX kills const */
3635 aiov.iov_len = nbytes;
3636 auio.uio_iov = &aiov;
3637 auio.uio_iovcnt = 1;
3638 auio.uio_offset = 0;
3639 if (nbytes > INT_MAX) {
3640 error = EINVAL;
3641 goto done;
3642 }
3643 auio.uio_resid = nbytes;
3644 auio.uio_rw = UIO_WRITE;
3645 auio.uio_segflg = UIO_USERSPACE;
3646 auio.uio_procp = p;
3647 cnt = nbytes;
3648
3649 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, &auio,
3650 p->p_ucred, p);
3651 cnt -= auio.uio_resid;
3652 retval[0] = cnt;
3653
3654 done:
3655 VOP_UNLOCK(vp, 0);
3656 vn_finished_write(mp, 0);
3657 return (error);
3658 }
3659
3660 int
3661 sys_extattr_set_fd(struct lwp *l, void *v, register_t *retval)
3662 {
3663 struct sys_extattr_set_fd_args /* {
3664 syscallarg(int) fd;
3665 syscallarg(int) attrnamespace;
3666 syscallarg(const char *) attrname;
3667 syscallarg(const void *) data;
3668 syscallarg(size_t) nbytes;
3669 } */ *uap = v;
3670 struct proc *p = l->l_proc;
3671 struct file *fp;
3672 struct vnode *vp;
3673 char attrname[EXTATTR_MAXNAMELEN];
3674 int error;
3675
3676 error = copyinstr(SCARG(uap, attrname), attrname, sizeof(attrname),
3677 NULL);
3678 if (error)
3679 return (error);
3680
3681 error = getvnode(p->p_fd, SCARG(uap, fd), &fp);
3682 if (error)
3683 return (error);
3684 vp = (struct vnode *) fp->f_data;
3685
3686 error = extattr_set_vp(vp, SCARG(uap, attrnamespace), attrname,
3687 SCARG(uap, data), SCARG(uap, nbytes), p, retval);
3688
3689 FILE_UNUSE(fp, p);
3690 return (error);
3691 }
3692
3693 int
3694 sys_extattr_set_file(struct lwp *l, void *v, register_t *retval)
3695 {
3696 struct sys_extattr_set_file_args /* {
3697 syscallarg(const char *) path;
3698 syscallarg(int) attrnamespace;
3699 syscallarg(const char *) attrname;
3700 syscallarg(const void *) data;
3701 syscallarg(size_t) nbytes;
3702 } */ *uap = v;
3703 struct proc *p = l->l_proc;
3704 struct nameidata nd;
3705 char attrname[EXTATTR_MAXNAMELEN];
3706 int error;
3707
3708 error = copyinstr(SCARG(uap, attrname), attrname, sizeof(attrname),
3709 NULL);
3710 if (error)
3711 return (error);
3712
3713 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
3714 error = namei(&nd);
3715 if (error)
3716 return (error);
3717
3718 error = extattr_set_vp(nd.ni_vp, SCARG(uap, attrnamespace), attrname,
3719 SCARG(uap, data), SCARG(uap, nbytes), p, retval);
3720
3721 vrele(nd.ni_vp);
3722 return (error);
3723 }
3724
3725 int
3726 sys_extattr_set_link(struct lwp *l, void *v, register_t *retval)
3727 {
3728 struct sys_extattr_set_link_args /* {
3729 syscallarg(const char *) path;
3730 syscallarg(int) attrnamespace;
3731 syscallarg(const char *) attrname;
3732 syscallarg(const void *) data;
3733 syscallarg(size_t) nbytes;
3734 } */ *uap = v;
3735 struct proc *p = l->l_proc;
3736 struct nameidata nd;
3737 char attrname[EXTATTR_MAXNAMELEN];
3738 int error;
3739
3740 error = copyinstr(SCARG(uap, attrname), attrname, sizeof(attrname),
3741 NULL);
3742 if (error)
3743 return (error);
3744
3745 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
3746 error = namei(&nd);
3747 if (error)
3748 return (error);
3749
3750 error = extattr_set_vp(nd.ni_vp, SCARG(uap, attrnamespace), attrname,
3751 SCARG(uap, data), SCARG(uap, nbytes), p, retval);
3752
3753 vrele(nd.ni_vp);
3754 return (error);
3755 }
3756
3757 /*
3758 * Get a named extended attribute on a file or directory.
3759 */
3760 static int
3761 extattr_get_vp(struct vnode *vp, int attrnamespace, const char *attrname,
3762 void *data, size_t nbytes, struct proc *p, register_t *retval)
3763 {
3764 struct uio auio, *auiop;
3765 struct iovec aiov;
3766 ssize_t cnt;
3767 size_t size, *sizep;
3768 int error;
3769
3770 VOP_LEASE(vp, p, p->p_ucred, LEASE_READ);
3771 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
3772
3773 /*
3774 * Slightly unusual semantics: if the user provides a NULL data
3775 * pointer, they don't want to receive the data, just the maximum
3776 * read length.
3777 */
3778 auiop = NULL;
3779 sizep = NULL;
3780 cnt = 0;
3781 if (data != NULL) {
3782 aiov.iov_base = data;
3783 aiov.iov_len = nbytes;
3784 auio.uio_iov = &aiov;
3785 auio.uio_offset = 0;
3786 if (nbytes > INT_MAX) {
3787 error = EINVAL;
3788 goto done;
3789 }
3790 auio.uio_resid = nbytes;
3791 auio.uio_rw = UIO_READ;
3792 auio.uio_segflg = UIO_USERSPACE;
3793 auio.uio_procp = p;
3794 auiop = &auio;
3795 cnt = nbytes;
3796 } else
3797 sizep = &size;
3798
3799 error = VOP_GETEXTATTR(vp, attrnamespace, attrname, auiop, sizep,
3800 p->p_ucred, p);
3801
3802 if (auiop != NULL) {
3803 cnt -= auio.uio_resid;
3804 retval[0] = cnt;
3805 } else
3806 retval[0] = size;
3807
3808 done:
3809 VOP_UNLOCK(vp, 0);
3810 return (error);
3811 }
3812
3813 int
3814 sys_extattr_get_fd(struct lwp *l, void *v, register_t *retval)
3815 {
3816 struct sys_extattr_get_fd_args /* {
3817 syscallarg(int) fd;
3818 syscallarg(int) attrnamespace;
3819 syscallarg(const char *) attrname;
3820 syscallarg(void *) data;
3821 syscallarg(size_t) nbytes;
3822 } */ *uap = v;
3823 struct proc *p = l->l_proc;
3824 struct file *fp;
3825 struct vnode *vp;
3826 char attrname[EXTATTR_MAXNAMELEN];
3827 int error;
3828
3829 error = copyinstr(SCARG(uap, attrname), attrname, sizeof(attrname),
3830 NULL);
3831 if (error)
3832 return (error);
3833
3834 error = getvnode(p->p_fd, SCARG(uap, fd), &fp);
3835 if (error)
3836 return (error);
3837 vp = (struct vnode *) fp->f_data;
3838
3839 error = extattr_get_vp(vp, SCARG(uap, attrnamespace), attrname,
3840 SCARG(uap, data), SCARG(uap, nbytes), p, retval);
3841
3842 FILE_UNUSE(fp, p);
3843 return (error);
3844 }
3845
3846 int
3847 sys_extattr_get_file(struct lwp *l, void *v, register_t *retval)
3848 {
3849 struct sys_extattr_get_file_args /* {
3850 syscallarg(const char *) path;
3851 syscallarg(int) attrnamespace;
3852 syscallarg(const char *) attrname;
3853 syscallarg(void *) data;
3854 syscallarg(size_t) nbytes;
3855 } */ *uap = v;
3856 struct proc *p = l->l_proc;
3857 struct nameidata nd;
3858 char attrname[EXTATTR_MAXNAMELEN];
3859 int error;
3860
3861 error = copyinstr(SCARG(uap, attrname), attrname, sizeof(attrname),
3862 NULL);
3863 if (error)
3864 return (error);
3865
3866 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
3867 error = namei(&nd);
3868 if (error)
3869 return (error);
3870
3871 error = extattr_get_vp(nd.ni_vp, SCARG(uap, attrnamespace), attrname,
3872 SCARG(uap, data), SCARG(uap, nbytes), p, retval);
3873
3874 vrele(nd.ni_vp);
3875 return (error);
3876 }
3877
3878 int
3879 sys_extattr_get_link(struct lwp *l, void *v, register_t *retval)
3880 {
3881 struct sys_extattr_get_link_args /* {
3882 syscallarg(const char *) path;
3883 syscallarg(int) attrnamespace;
3884 syscallarg(const char *) attrname;
3885 syscallarg(void *) data;
3886 syscallarg(size_t) nbytes;
3887 } */ *uap = v;
3888 struct proc *p = l->l_proc;
3889 struct nameidata nd;
3890 char attrname[EXTATTR_MAXNAMELEN];
3891 int error;
3892
3893 error = copyinstr(SCARG(uap, attrname), attrname, sizeof(attrname),
3894 NULL);
3895 if (error)
3896 return (error);
3897
3898 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
3899 error = namei(&nd);
3900 if (error)
3901 return (error);
3902
3903 error = extattr_get_vp(nd.ni_vp, SCARG(uap, attrnamespace), attrname,
3904 SCARG(uap, data), SCARG(uap, nbytes), p, retval);
3905
3906 vrele(nd.ni_vp);
3907 return (error);
3908 }
3909
3910 /*
3911 * Delete a named extended attribute on a file or directory.
3912 */
3913 static int
3914 extattr_delete_vp(struct vnode *vp, int attrnamespace, const char *attrname,
3915 struct proc *p)
3916 {
3917 struct mount *mp;
3918 int error;
3919
3920 error = vn_start_write(vp, &mp, V_WAIT | V_PCATCH);
3921 if (error)
3922 return (error);
3923 VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
3924 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
3925
3926 error = VOP_DELETEEXTATTR(vp, attrnamespace, attrname, p->p_ucred, p);
3927 if (error == EOPNOTSUPP)
3928 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, NULL,
3929 p->p_ucred, p);
3930
3931 VOP_UNLOCK(vp, 0);
3932 vn_finished_write(mp, 0);
3933 return (error);
3934 }
3935
3936 int
3937 sys_extattr_delete_fd(struct lwp *l, void *v, register_t *retval)
3938 {
3939 struct sys_extattr_delete_fd_args /* {
3940 syscallarg(int) fd;
3941 syscallarg(int) attrnamespace;
3942 syscallarg(const char *) attrname;
3943 } */ *uap = v;
3944 struct proc *p = l->l_proc;
3945 struct file *fp;
3946 struct vnode *vp;
3947 char attrname[EXTATTR_MAXNAMELEN];
3948 int error;
3949
3950 error = copyinstr(SCARG(uap, attrname), attrname, sizeof(attrname),
3951 NULL);
3952 if (error)
3953 return (error);
3954
3955 error = getvnode(p->p_fd, SCARG(uap, fd), &fp);
3956 if (error)
3957 return (error);
3958 vp = (struct vnode *) fp->f_data;
3959
3960 error = extattr_delete_vp(vp, SCARG(uap, attrnamespace), attrname, p);
3961
3962 FILE_UNUSE(fp, p);
3963 return (error);
3964 }
3965
3966 int
3967 sys_extattr_delete_file(struct lwp *l, void *v, register_t *retval)
3968 {
3969 struct sys_extattr_delete_file_args /* {
3970 syscallarg(const char *) path;
3971 syscallarg(int) attrnamespace;
3972 syscallarg(const char *) attrname;
3973 } */ *uap = v;
3974 struct proc *p = l->l_proc;
3975 struct nameidata nd;
3976 char attrname[EXTATTR_MAXNAMELEN];
3977 int error;
3978
3979 error = copyinstr(SCARG(uap, attrname), attrname, sizeof(attrname),
3980 NULL);
3981 if (error)
3982 return (error);
3983
3984 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
3985 error = namei(&nd);
3986 if (error)
3987 return (error);
3988
3989 error = extattr_delete_vp(nd.ni_vp, SCARG(uap, attrnamespace), attrname,
3990 p);
3991
3992 vrele(nd.ni_vp);
3993 return (error);
3994 }
3995
3996 int
3997 sys_extattr_delete_link(struct lwp *l, void *v, register_t *retval)
3998 {
3999 struct sys_extattr_delete_link_args /* {
4000 syscallarg(const char *) path;
4001 syscallarg(int) attrnamespace;
4002 syscallarg(const char *) attrname;
4003 } */ *uap = v;
4004 struct proc *p = l->l_proc;
4005 struct nameidata nd;
4006 char attrname[EXTATTR_MAXNAMELEN];
4007 int error;
4008
4009 error = copyinstr(SCARG(uap, attrname), attrname, sizeof(attrname),
4010 NULL);
4011 if (error)
4012 return (error);
4013
4014 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
4015 error = namei(&nd);
4016 if (error)
4017 return (error);
4018
4019 error = extattr_delete_vp(nd.ni_vp, SCARG(uap, attrnamespace), attrname,
4020 p);
4021
4022 vrele(nd.ni_vp);
4023 return (error);
4024 }
4025
4026 /*
4027 * Retrieve a list of extended attributes on a file or directory.
4028 */
4029 static int
4030 extattr_list_vp(struct vnode *vp, int attrnamespace, void *data, size_t nbytes,
4031 struct proc *p, register_t *retval)
4032 {
4033 struct uio auio, *auiop;
4034 size_t size, *sizep;
4035 struct iovec aiov;
4036 ssize_t cnt;
4037 int error;
4038
4039 VOP_LEASE(vp, p, p->p_ucred, LEASE_READ);
4040 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
4041
4042 auiop = NULL;
4043 sizep = NULL;
4044 cnt = 0;
4045 if (data != NULL) {
4046 aiov.iov_base = data;
4047 aiov.iov_len = nbytes;
4048 auio.uio_iov = &aiov;
4049 auio.uio_offset = 0;
4050 if (nbytes > INT_MAX) {
4051 error = EINVAL;
4052 goto done;
4053 }
4054 auio.uio_resid = nbytes;
4055 auio.uio_rw = UIO_READ;
4056 auio.uio_segflg = UIO_USERSPACE;
4057 auio.uio_procp = p;
4058 auiop = &auio;
4059 cnt = nbytes;
4060 } else
4061 sizep = &size;
4062
4063 error = VOP_LISTEXTATTR(vp, attrnamespace, auiop, sizep,
4064 p->p_ucred, p);
4065
4066 if (auiop != NULL) {
4067 cnt -= auio.uio_resid;
4068 retval[0] = cnt;
4069 } else
4070 retval[0] = size;
4071
4072 done:
4073 VOP_UNLOCK(vp, 0);
4074 return (error);
4075 }
4076
4077 int
4078 sys_extattr_list_fd(struct lwp *l, void *v, register_t *retval)
4079 {
4080 struct sys_extattr_list_fd_args /* {
4081 syscallarg(int) fd;
4082 syscallarg(int) attrnamespace;
4083 syscallarg(void *) data;
4084 syscallarg(size_t) nbytes;
4085 } */ *uap = v;
4086 struct proc *p = l->l_proc;
4087 struct file *fp;
4088 struct vnode *vp;
4089 int error;
4090
4091 error = getvnode(p->p_fd, SCARG(uap, fd), &fp);
4092 if (error)
4093 return (error);
4094 vp = (struct vnode *) fp->f_data;
4095
4096 error = extattr_list_vp(vp, SCARG(uap, attrnamespace),
4097 SCARG(uap, data), SCARG(uap, nbytes), p, retval);
4098
4099 FILE_UNUSE(fp, p);
4100 return (error);
4101 }
4102
4103 int
4104 sys_extattr_list_file(struct lwp *l, void *v, register_t *retval)
4105 {
4106 struct sys_extattr_list_file_args /* {
4107 syscallarg(const char *) path;
4108 syscallarg(int) attrnamespace;
4109 syscallarg(void *) data;
4110 syscallarg(size_t) nbytes;
4111 } */ *uap = v;
4112 struct proc *p = l->l_proc;
4113 struct nameidata nd;
4114 int error;
4115
4116 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
4117 error = namei(&nd);
4118 if (error)
4119 return (error);
4120
4121 error = extattr_list_vp(nd.ni_vp, SCARG(uap, attrnamespace),
4122 SCARG(uap, data), SCARG(uap, nbytes), p, retval);
4123
4124 vrele(nd.ni_vp);
4125 return (error);
4126 }
4127
4128 int
4129 sys_extattr_list_link(struct lwp *l, void *v, register_t *retval)
4130 {
4131 struct sys_extattr_list_link_args /* {
4132 syscallarg(const char *) path;
4133 syscallarg(int) attrnamespace;
4134 syscallarg(void *) data;
4135 syscallarg(size_t) nbytes;
4136 } */ *uap = v;
4137 struct proc *p = l->l_proc;
4138 struct nameidata nd;
4139 int error;
4140
4141 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
4142 error = namei(&nd);
4143 if (error)
4144 return (error);
4145
4146 error = extattr_list_vp(nd.ni_vp, SCARG(uap, attrnamespace),
4147 SCARG(uap, data), SCARG(uap, nbytes), p, retval);
4148
4149 vrele(nd.ni_vp);
4150 return (error);
4151 }
4152