Introduce cratom(), remove crcopy().
[dragonfly.git] / sys / kern / vfs_syscalls.c
CommitLineData
984263bc
MD
1/*
2 * Copyright (c) 1989, 1993
3 * The Regents of the University of California. All rights reserved.
4 * (c) UNIX System Laboratories, Inc.
5 * All or some portions of this file are derived from material licensed
6 * to the University of California by American Telephone and Telegraph
7 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
8 * the permission of UNIX System Laboratories, Inc.
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 * 3. All advertising materials mentioning features or use of this software
19 * must display the following acknowledgement:
20 * This product includes software developed by the University of
21 * California, Berkeley and its contributors.
22 * 4. Neither the name of the University nor the names of its contributors
23 * may be used to endorse or promote products derived from this software
24 * without specific prior written permission.
25 *
26 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
27 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
28 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
29 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
30 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
31 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
32 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
33 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
34 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
35 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36 * SUCH DAMAGE.
37 *
38 * @(#)vfs_syscalls.c 8.13 (Berkeley) 4/15/94
39 * $FreeBSD: src/sys/kern/vfs_syscalls.c,v 1.151.2.18 2003/04/04 20:35:58 tegge Exp $
ab1c1714 40 * $DragonFly: src/sys/kern/vfs_syscalls.c,v 1.5 2003/06/25 05:22:32 dillon Exp $
984263bc
MD
41 */
42
43/* For 4.3 integer FS ID compatibility */
44#include "opt_compat.h"
45
46#include <sys/param.h>
47#include <sys/systm.h>
48#include <sys/buf.h>
49#include <sys/sysent.h>
50#include <sys/malloc.h>
51#include <sys/mount.h>
52#include <sys/sysproto.h>
984263bc
MD
53#include <sys/filedesc.h>
54#include <sys/kernel.h>
55#include <sys/fcntl.h>
56#include <sys/file.h>
57#include <sys/linker.h>
58#include <sys/stat.h>
59#include <sys/unistd.h>
60#include <sys/vnode.h>
61#include <sys/proc.h>
dadab5e9 62#include <sys/namei.h>
984263bc
MD
63#include <sys/dirent.h>
64#include <sys/extattr.h>
65
66#include <machine/limits.h>
67#include <miscfs/union/union.h>
68#include <sys/sysctl.h>
69#include <vm/vm.h>
70#include <vm/vm_object.h>
71#include <vm/vm_zone.h>
72#include <vm/vm_page.h>
73
dadab5e9
MD
74#include <sys/file2.h>
75
76static int change_dir __P((struct nameidata *ndp, struct thread *td));
984263bc
MD
77static void checkdirs __P((struct vnode *olddp));
78static int chroot_refuse_vdir_fds __P((struct filedesc *fdp));
79static int getutimes __P((const struct timeval *, struct timespec *));
41c20dac
MD
80static int setfown __P((struct vnode *, uid_t, gid_t));
81static int setfmode __P((struct vnode *, int));
82static int setfflags __P((struct vnode *, int));
83static int setutimes __P((struct vnode *, const struct timespec *, int));
984263bc
MD
84static int usermount = 0; /* if 1, non-root can mount fs. */
85
dadab5e9 86int (*union_dircheckp) __P((struct thread *, struct vnode **, struct file *));
984263bc
MD
87
88SYSCTL_INT(_vfs, OID_AUTO, usermount, CTLFLAG_RW, &usermount, 0, "");
89
90/*
91 * Virtual File System System Calls
92 */
93
94/*
95 * Mount a file system.
96 */
97#ifndef _SYS_SYSPROTO_H_
98struct mount_args {
99 char *type;
100 char *path;
101 int flags;
102 caddr_t data;
103};
104#endif
41c20dac
MD
105/*
106 * mount_args(char *type, char *path, int flags, caddr_t data)
107 */
984263bc
MD
108/* ARGSUSED */
109int
41c20dac 110mount(struct mount_args *uap)
984263bc 111{
dadab5e9
MD
112 struct thread *td = curthread;
113 struct proc *p = td->td_proc;
984263bc
MD
114 struct vnode *vp;
115 struct mount *mp;
116 struct vfsconf *vfsp;
117 int error, flag = 0, flag2 = 0;
118 struct vattr va;
119#ifdef COMPAT_43
120 u_long fstypenum;
121#endif
122 struct nameidata nd;
123 char fstypename[MFSNAMELEN];
124
dadab5e9 125 if (usermount == 0 && (error = suser(td)))
984263bc
MD
126 return (error);
127 /*
128 * Do not allow NFS export by non-root users.
129 */
130 if (SCARG(uap, flags) & MNT_EXPORTED) {
dadab5e9 131 error = suser(td);
984263bc
MD
132 if (error)
133 return (error);
134 }
135 /*
136 * Silently enforce MNT_NOSUID and MNT_NODEV for non-root users
137 */
dadab5e9 138 if (suser(td))
984263bc
MD
139 SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
140 /*
141 * Get vnode to be covered
142 */
143 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
dadab5e9 144 SCARG(uap, path), td);
984263bc
MD
145 if ((error = namei(&nd)) != 0)
146 return (error);
147 NDFREE(&nd, NDF_ONLY_PNBUF);
148 vp = nd.ni_vp;
149 if (SCARG(uap, flags) & MNT_UPDATE) {
150 if ((vp->v_flag & VROOT) == 0) {
151 vput(vp);
152 return (EINVAL);
153 }
154 mp = vp->v_mount;
155 flag = mp->mnt_flag;
156 flag2 = mp->mnt_kern_flag;
157 /*
158 * We only allow the filesystem to be reloaded if it
159 * is currently mounted read-only.
160 */
161 if ((SCARG(uap, flags) & MNT_RELOAD) &&
162 ((mp->mnt_flag & MNT_RDONLY) == 0)) {
163 vput(vp);
164 return (EOPNOTSUPP); /* Needs translation */
165 }
166 /*
167 * Only root, or the user that did the original mount is
168 * permitted to update it.
169 */
170 if (mp->mnt_stat.f_owner != p->p_ucred->cr_uid &&
dadab5e9 171 (error = suser(td))) {
984263bc
MD
172 vput(vp);
173 return (error);
174 }
dadab5e9 175 if (vfs_busy(mp, LK_NOWAIT, 0, td)) {
984263bc
MD
176 vput(vp);
177 return (EBUSY);
178 }
179 simple_lock(&vp->v_interlock);
180 if ((vp->v_flag & VMOUNT) != 0 ||
181 vp->v_mountedhere != NULL) {
182 simple_unlock(&vp->v_interlock);
dadab5e9 183 vfs_unbusy(mp, td);
984263bc
MD
184 vput(vp);
185 return (EBUSY);
186 }
187 vp->v_flag |= VMOUNT;
188 simple_unlock(&vp->v_interlock);
189 mp->mnt_flag |=
190 SCARG(uap, flags) & (MNT_RELOAD | MNT_FORCE | MNT_UPDATE);
dadab5e9 191 VOP_UNLOCK(vp, 0, td);
984263bc
MD
192 goto update;
193 }
194 /*
195 * If the user is not root, ensure that they own the directory
196 * onto which we are attempting to mount.
197 */
dadab5e9 198 if ((error = VOP_GETATTR(vp, &va, p->p_ucred, td)) ||
984263bc 199 (va.va_uid != p->p_ucred->cr_uid &&
dadab5e9 200 (error = suser(td)))) {
984263bc
MD
201 vput(vp);
202 return (error);
203 }
dadab5e9 204 if ((error = vinvalbuf(vp, V_SAVE, p->p_ucred, td, 0, 0)) != 0) {
984263bc
MD
205 vput(vp);
206 return (error);
207 }
208 if (vp->v_type != VDIR) {
209 vput(vp);
210 return (ENOTDIR);
211 }
212#ifdef COMPAT_43
213 /*
214 * Historically filesystem types were identified by number. If we
215 * get an integer for the filesystem type instead of a string, we
216 * check to see if it matches one of the historic filesystem types.
217 */
218 fstypenum = (uintptr_t)SCARG(uap, type);
219 if (fstypenum < maxvfsconf) {
220 for (vfsp = vfsconf; vfsp; vfsp = vfsp->vfc_next)
221 if (vfsp->vfc_typenum == fstypenum)
222 break;
223 if (vfsp == NULL) {
224 vput(vp);
225 return (ENODEV);
226 }
227 strncpy(fstypename, vfsp->vfc_name, MFSNAMELEN);
228 } else
229#endif /* COMPAT_43 */
230 if ((error = copyinstr(SCARG(uap, type), fstypename, MFSNAMELEN, NULL)) != 0) {
231 vput(vp);
232 return (error);
233 }
234 for (vfsp = vfsconf; vfsp; vfsp = vfsp->vfc_next)
235 if (!strcmp(vfsp->vfc_name, fstypename))
236 break;
237 if (vfsp == NULL) {
238 linker_file_t lf;
239
240 /* Only load modules for root (very important!) */
dadab5e9 241 if ((error = suser(td)) != 0) {
984263bc
MD
242 vput(vp);
243 return error;
244 }
245 error = linker_load_file(fstypename, &lf);
246 if (error || lf == NULL) {
247 vput(vp);
248 if (lf == NULL)
249 error = ENODEV;
250 return error;
251 }
252 lf->userrefs++;
253 /* lookup again, see if the VFS was loaded */
254 for (vfsp = vfsconf; vfsp; vfsp = vfsp->vfc_next)
255 if (!strcmp(vfsp->vfc_name, fstypename))
256 break;
257 if (vfsp == NULL) {
258 lf->userrefs--;
259 linker_file_unload(lf);
260 vput(vp);
261 return (ENODEV);
262 }
263 }
264 simple_lock(&vp->v_interlock);
265 if ((vp->v_flag & VMOUNT) != 0 ||
266 vp->v_mountedhere != NULL) {
267 simple_unlock(&vp->v_interlock);
268 vput(vp);
269 return (EBUSY);
270 }
271 vp->v_flag |= VMOUNT;
272 simple_unlock(&vp->v_interlock);
273
274 /*
275 * Allocate and initialize the filesystem.
276 */
277 mp = malloc(sizeof(struct mount), M_MOUNT, M_WAITOK);
278 bzero((char *)mp, (u_long)sizeof(struct mount));
279 TAILQ_INIT(&mp->mnt_nvnodelist);
280 TAILQ_INIT(&mp->mnt_reservedvnlist);
281 mp->mnt_nvnodelistsize = 0;
282 lockinit(&mp->mnt_lock, PVFS, "vfslock", 0, LK_NOPAUSE);
dadab5e9 283 (void)vfs_busy(mp, LK_NOWAIT, 0, td);
984263bc
MD
284 mp->mnt_op = vfsp->vfc_vfsops;
285 mp->mnt_vfc = vfsp;
286 vfsp->vfc_refcount++;
287 mp->mnt_stat.f_type = vfsp->vfc_typenum;
288 mp->mnt_flag |= vfsp->vfc_flags & MNT_VISFLAGMASK;
289 strncpy(mp->mnt_stat.f_fstypename, vfsp->vfc_name, MFSNAMELEN);
290 mp->mnt_vnodecovered = vp;
291 mp->mnt_stat.f_owner = p->p_ucred->cr_uid;
292 mp->mnt_iosize_max = DFLTPHYS;
dadab5e9 293 VOP_UNLOCK(vp, 0, td);
984263bc
MD
294update:
295 /*
296 * Set the mount level flags.
297 */
298 if (SCARG(uap, flags) & MNT_RDONLY)
299 mp->mnt_flag |= MNT_RDONLY;
300 else if (mp->mnt_flag & MNT_RDONLY)
301 mp->mnt_kern_flag |= MNTK_WANTRDWR;
302 mp->mnt_flag &=~ (MNT_NOSUID | MNT_NOEXEC | MNT_NODEV |
303 MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | MNT_NOATIME |
304 MNT_NOSYMFOLLOW | MNT_IGNORE |
305 MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR);
306 mp->mnt_flag |= SCARG(uap, flags) & (MNT_NOSUID | MNT_NOEXEC |
307 MNT_NODEV | MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | MNT_FORCE |
308 MNT_NOSYMFOLLOW | MNT_IGNORE |
309 MNT_NOATIME | MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR);
310 /*
311 * Mount the filesystem.
312 * XXX The final recipients of VFS_MOUNT just overwrite the ndp they
313 * get. No freeing of cn_pnbuf.
314 */
dadab5e9 315 error = VFS_MOUNT(mp, SCARG(uap, path), SCARG(uap, data), &nd, td);
984263bc
MD
316 if (mp->mnt_flag & MNT_UPDATE) {
317 if (mp->mnt_kern_flag & MNTK_WANTRDWR)
318 mp->mnt_flag &= ~MNT_RDONLY;
319 mp->mnt_flag &=~ (MNT_UPDATE | MNT_RELOAD | MNT_FORCE);
320 mp->mnt_kern_flag &=~ MNTK_WANTRDWR;
321 if (error) {
322 mp->mnt_flag = flag;
323 mp->mnt_kern_flag = flag2;
324 }
325 if ((mp->mnt_flag & MNT_RDONLY) == 0) {
326 if (mp->mnt_syncer == NULL)
327 error = vfs_allocate_syncvnode(mp);
328 } else {
329 if (mp->mnt_syncer != NULL)
330 vrele(mp->mnt_syncer);
331 mp->mnt_syncer = NULL;
332 }
dadab5e9 333 vfs_unbusy(mp, td);
984263bc
MD
334 simple_lock(&vp->v_interlock);
335 vp->v_flag &= ~VMOUNT;
336 simple_unlock(&vp->v_interlock);
337 vrele(vp);
338 return (error);
339 }
dadab5e9 340 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
984263bc
MD
341 /*
342 * Put the new filesystem on the mount list after root.
343 */
344 cache_purge(vp);
345 if (!error) {
346 simple_lock(&vp->v_interlock);
347 vp->v_flag &= ~VMOUNT;
348 vp->v_mountedhere = mp;
349 simple_unlock(&vp->v_interlock);
350 simple_lock(&mountlist_slock);
351 TAILQ_INSERT_TAIL(&mountlist, mp, mnt_list);
352 simple_unlock(&mountlist_slock);
353 checkdirs(vp);
dadab5e9 354 VOP_UNLOCK(vp, 0, td);
984263bc
MD
355 if ((mp->mnt_flag & MNT_RDONLY) == 0)
356 error = vfs_allocate_syncvnode(mp);
dadab5e9
MD
357 vfs_unbusy(mp, td);
358 if ((error = VFS_START(mp, 0, td)) != 0)
984263bc
MD
359 vrele(vp);
360 } else {
361 simple_lock(&vp->v_interlock);
362 vp->v_flag &= ~VMOUNT;
363 simple_unlock(&vp->v_interlock);
364 mp->mnt_vfc->vfc_refcount--;
dadab5e9 365 vfs_unbusy(mp, td);
984263bc
MD
366 free((caddr_t)mp, M_MOUNT);
367 vput(vp);
368 }
369 return (error);
370}
371
372/*
373 * Scan all active processes to see if any of them have a current
374 * or root directory onto which the new filesystem has just been
375 * mounted. If so, replace them with the new mount point.
376 */
377static void
dadab5e9 378checkdirs(struct vnode *olddp)
984263bc
MD
379{
380 struct filedesc *fdp;
381 struct vnode *newdp;
382 struct proc *p;
383
384 if (olddp->v_usecount == 1)
385 return;
386 if (VFS_ROOT(olddp->v_mountedhere, &newdp))
387 panic("mount: lost mount");
388 LIST_FOREACH(p, &allproc, p_list) {
389 fdp = p->p_fd;
390 if (fdp->fd_cdir == olddp) {
391 vrele(fdp->fd_cdir);
392 VREF(newdp);
393 fdp->fd_cdir = newdp;
394 }
395 if (fdp->fd_rdir == olddp) {
396 vrele(fdp->fd_rdir);
397 VREF(newdp);
398 fdp->fd_rdir = newdp;
399 }
400 }
401 if (rootvnode == olddp) {
402 vrele(rootvnode);
403 VREF(newdp);
404 rootvnode = newdp;
405 }
406 vput(newdp);
407}
408
409/*
410 * Unmount a file system.
411 *
412 * Note: unmount takes a path to the vnode mounted on as argument,
413 * not special file (as before).
414 */
415#ifndef _SYS_SYSPROTO_H_
416struct unmount_args {
417 char *path;
418 int flags;
419};
420#endif
41c20dac
MD
421/*
422 * umount_args(char *path, int flags)
423 */
984263bc
MD
424/* ARGSUSED */
425int
41c20dac 426unmount(struct unmount_args *uap)
984263bc 427{
dadab5e9
MD
428 struct thread *td = curthread;
429 struct proc *p = td->td_proc;
41c20dac 430 struct vnode *vp;
984263bc
MD
431 struct mount *mp;
432 int error;
433 struct nameidata nd;
434
dadab5e9 435 KKASSERT(p);
984263bc 436 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
dadab5e9 437 SCARG(uap, path), td);
984263bc
MD
438 if ((error = namei(&nd)) != 0)
439 return (error);
440 vp = nd.ni_vp;
441 NDFREE(&nd, NDF_ONLY_PNBUF);
442 mp = vp->v_mount;
443
444 /*
445 * Only root, or the user that did the original mount is
446 * permitted to unmount this filesystem.
447 */
448 if ((mp->mnt_stat.f_owner != p->p_ucred->cr_uid) &&
dadab5e9 449 (error = suser(td))) {
984263bc
MD
450 vput(vp);
451 return (error);
452 }
453
454 /*
455 * Don't allow unmounting the root file system.
456 */
457 if (mp->mnt_flag & MNT_ROOTFS) {
458 vput(vp);
459 return (EINVAL);
460 }
461
462 /*
463 * Must be the root of the filesystem
464 */
465 if ((vp->v_flag & VROOT) == 0) {
466 vput(vp);
467 return (EINVAL);
468 }
469 vput(vp);
dadab5e9 470 return (dounmount(mp, SCARG(uap, flags), td));
984263bc
MD
471}
472
473/*
474 * Do the actual file system unmount.
475 */
476int
dadab5e9 477dounmount(struct mount *mp, int flags, struct thread *td)
984263bc
MD
478{
479 struct vnode *coveredvp;
480 int error;
481 int async_flag;
dadab5e9
MD
482 struct proc *p = td->td_proc;
483
484 KKASSERT(p);
984263bc
MD
485
486 simple_lock(&mountlist_slock);
487 if (mp->mnt_kern_flag & MNTK_UNMOUNT) {
488 simple_unlock(&mountlist_slock);
489 return (EBUSY);
490 }
491 mp->mnt_kern_flag |= MNTK_UNMOUNT;
492 /* Allow filesystems to detect that a forced unmount is in progress. */
493 if (flags & MNT_FORCE)
494 mp->mnt_kern_flag |= MNTK_UNMOUNTF;
495 error = lockmgr(&mp->mnt_lock, LK_DRAIN | LK_INTERLOCK |
dadab5e9 496 ((flags & MNT_FORCE) ? 0 : LK_NOWAIT), &mountlist_slock, td);
984263bc
MD
497 if (error) {
498 mp->mnt_kern_flag &= ~(MNTK_UNMOUNT | MNTK_UNMOUNTF);
499 if (mp->mnt_kern_flag & MNTK_MWAIT)
500 wakeup((caddr_t)mp);
501 return (error);
502 }
503
504 if (mp->mnt_flag & MNT_EXPUBLIC)
505 vfs_setpublicfs(NULL, NULL, NULL);
506
507 vfs_msync(mp, MNT_WAIT);
508 async_flag = mp->mnt_flag & MNT_ASYNC;
509 mp->mnt_flag &=~ MNT_ASYNC;
510 cache_purgevfs(mp); /* remove cache entries for this file sys */
511 if (mp->mnt_syncer != NULL)
512 vrele(mp->mnt_syncer);
513 if (((mp->mnt_flag & MNT_RDONLY) ||
dadab5e9 514 (error = VFS_SYNC(mp, MNT_WAIT, p->p_ucred, td)) == 0) ||
984263bc 515 (flags & MNT_FORCE))
dadab5e9 516 error = VFS_UNMOUNT(mp, flags, td);
984263bc
MD
517 simple_lock(&mountlist_slock);
518 if (error) {
519 if ((mp->mnt_flag & MNT_RDONLY) == 0 && mp->mnt_syncer == NULL)
520 (void) vfs_allocate_syncvnode(mp);
521 mp->mnt_kern_flag &= ~(MNTK_UNMOUNT | MNTK_UNMOUNTF);
522 mp->mnt_flag |= async_flag;
523 lockmgr(&mp->mnt_lock, LK_RELEASE | LK_INTERLOCK | LK_REENABLE,
dadab5e9 524 &mountlist_slock, td);
984263bc
MD
525 if (mp->mnt_kern_flag & MNTK_MWAIT)
526 wakeup((caddr_t)mp);
527 return (error);
528 }
529 TAILQ_REMOVE(&mountlist, mp, mnt_list);
530 if ((coveredvp = mp->mnt_vnodecovered) != NULLVP) {
531 coveredvp->v_mountedhere = (struct mount *)0;
532 vrele(coveredvp);
533 }
534 mp->mnt_vfc->vfc_refcount--;
535 if (!TAILQ_EMPTY(&mp->mnt_nvnodelist))
536 panic("unmount: dangling vnode");
dadab5e9 537 lockmgr(&mp->mnt_lock, LK_RELEASE | LK_INTERLOCK, &mountlist_slock, td);
984263bc
MD
538 if (mp->mnt_kern_flag & MNTK_MWAIT)
539 wakeup((caddr_t)mp);
540 free((caddr_t)mp, M_MOUNT);
541 return (0);
542}
543
544/*
545 * Sync each mounted filesystem.
546 */
547#ifndef _SYS_SYSPROTO_H_
548struct sync_args {
549 int dummy;
550};
551#endif
552
553#ifdef DEBUG
554static int syncprt = 0;
555SYSCTL_INT(_debug, OID_AUTO, syncprt, CTLFLAG_RW, &syncprt, 0, "");
556#endif
557
558/* ARGSUSED */
559int
41c20dac 560sync(struct sync_args *uap)
984263bc 561{
dadab5e9
MD
562 struct thread *td = curthread;
563 struct proc *p = td->td_proc;
41c20dac 564 struct mount *mp, *nmp;
984263bc
MD
565 int asyncflag;
566
567 simple_lock(&mountlist_slock);
568 for (mp = TAILQ_FIRST(&mountlist); mp != NULL; mp = nmp) {
dadab5e9 569 if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock, td)) {
984263bc
MD
570 nmp = TAILQ_NEXT(mp, mnt_list);
571 continue;
572 }
573 if ((mp->mnt_flag & MNT_RDONLY) == 0) {
574 asyncflag = mp->mnt_flag & MNT_ASYNC;
575 mp->mnt_flag &= ~MNT_ASYNC;
576 vfs_msync(mp, MNT_NOWAIT);
577 VFS_SYNC(mp, MNT_NOWAIT,
dadab5e9 578 ((p != NULL) ? p->p_ucred : NOCRED), td);
984263bc
MD
579 mp->mnt_flag |= asyncflag;
580 }
581 simple_lock(&mountlist_slock);
582 nmp = TAILQ_NEXT(mp, mnt_list);
dadab5e9 583 vfs_unbusy(mp, td);
984263bc
MD
584 }
585 simple_unlock(&mountlist_slock);
586#if 0
587/*
588 * XXX don't call vfs_bufstats() yet because that routine
589 * was not imported in the Lite2 merge.
590 */
591#ifdef DIAGNOSTIC
592 if (syncprt)
593 vfs_bufstats();
594#endif /* DIAGNOSTIC */
595#endif
596 return (0);
597}
598
599/* XXX PRISON: could be per prison flag */
600static int prison_quotas;
601#if 0
602SYSCTL_INT(_kern_prison, OID_AUTO, quotas, CTLFLAG_RW, &prison_quotas, 0, "");
603#endif
604
605/*
41c20dac
MD
606 * quotactl_args(char *path, int fcmd, int uid, caddr_t arg)
607 *
984263bc
MD
608 * Change filesystem quotas.
609 */
984263bc
MD
610/* ARGSUSED */
611int
41c20dac 612quotactl(struct quotactl_args *uap)
984263bc 613{
dadab5e9
MD
614 struct thread *td = curthread;
615 struct proc *p = td->td_proc;
41c20dac 616 struct mount *mp;
984263bc
MD
617 int error;
618 struct nameidata nd;
619
dadab5e9 620 KKASSERT(p);
41c20dac 621 if (p->p_ucred->cr_prison && !prison_quotas)
984263bc 622 return (EPERM);
dadab5e9 623 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), td);
984263bc
MD
624 if ((error = namei(&nd)) != 0)
625 return (error);
626 mp = nd.ni_vp->v_mount;
627 NDFREE(&nd, NDF_ONLY_PNBUF);
628 vrele(nd.ni_vp);
629 return (VFS_QUOTACTL(mp, SCARG(uap, cmd), SCARG(uap, uid),
dadab5e9 630 SCARG(uap, arg), td));
984263bc
MD
631}
632
633/*
41c20dac
MD
634 * statfs_args(char *path, struct statfs *buf)
635 *
984263bc
MD
636 * Get filesystem statistics.
637 */
984263bc
MD
638/* ARGSUSED */
639int
41c20dac 640statfs(struct statfs_args *uap)
984263bc 641{
dadab5e9 642 struct thread *td = curthread;
41c20dac
MD
643 struct mount *mp;
644 struct statfs *sp;
984263bc
MD
645 int error;
646 struct nameidata nd;
647 struct statfs sb;
648
dadab5e9 649 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), td);
984263bc
MD
650 if ((error = namei(&nd)) != 0)
651 return (error);
652 mp = nd.ni_vp->v_mount;
653 sp = &mp->mnt_stat;
654 NDFREE(&nd, NDF_ONLY_PNBUF);
655 vrele(nd.ni_vp);
dadab5e9 656 error = VFS_STATFS(mp, sp, td);
984263bc
MD
657 if (error)
658 return (error);
659 sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
dadab5e9 660 if (suser(td)) {
984263bc
MD
661 bcopy((caddr_t)sp, (caddr_t)&sb, sizeof(sb));
662 sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0;
663 sp = &sb;
664 }
665 return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
666}
667
668/*
41c20dac
MD
669 * fstatfs_args(int fd, struct statfs *buf)
670 *
984263bc
MD
671 * Get filesystem statistics.
672 */
984263bc
MD
673/* ARGSUSED */
674int
41c20dac 675fstatfs(struct fstatfs_args *uap)
984263bc 676{
dadab5e9
MD
677 struct thread *td = curthread;
678 struct proc *p = td->td_proc;
984263bc
MD
679 struct file *fp;
680 struct mount *mp;
681 register struct statfs *sp;
682 int error;
683 struct statfs sb;
684
dadab5e9 685 KKASSERT(p);
984263bc
MD
686 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
687 return (error);
688 mp = ((struct vnode *)fp->f_data)->v_mount;
689 if (mp == NULL)
690 return (EBADF);
691 sp = &mp->mnt_stat;
dadab5e9 692 error = VFS_STATFS(mp, sp, td);
984263bc
MD
693 if (error)
694 return (error);
695 sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
dadab5e9 696 if (suser(td)) {
984263bc
MD
697 bcopy((caddr_t)sp, (caddr_t)&sb, sizeof(sb));
698 sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0;
699 sp = &sb;
700 }
701 return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
702}
703
704/*
41c20dac
MD
705 * getfsstat_args(struct statfs *buf, long bufsize, int flags)
706 *
984263bc
MD
707 * Get statistics on all filesystems.
708 */
41c20dac 709/* ARGSUSED */
984263bc 710int
41c20dac 711getfsstat(struct getfsstat_args *uap)
984263bc 712{
dadab5e9
MD
713 struct thread *td = curthread;
714 struct proc *p = td->td_proc;
41c20dac
MD
715 struct mount *mp, *nmp;
716 struct statfs *sp;
984263bc
MD
717 caddr_t sfsp;
718 long count, maxcount, error;
719
720 maxcount = SCARG(uap, bufsize) / sizeof(struct statfs);
721 sfsp = (caddr_t)SCARG(uap, buf);
722 count = 0;
723 simple_lock(&mountlist_slock);
724 for (mp = TAILQ_FIRST(&mountlist); mp != NULL; mp = nmp) {
dadab5e9 725 if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock, td)) {
984263bc
MD
726 nmp = TAILQ_NEXT(mp, mnt_list);
727 continue;
728 }
729 if (sfsp && count < maxcount) {
730 sp = &mp->mnt_stat;
731 /*
732 * If MNT_NOWAIT or MNT_LAZY is specified, do not
733 * refresh the fsstat cache. MNT_NOWAIT or MNT_LAZY
734 * overrides MNT_WAIT.
735 */
736 if (((SCARG(uap, flags) & (MNT_LAZY|MNT_NOWAIT)) == 0 ||
737 (SCARG(uap, flags) & MNT_WAIT)) &&
dadab5e9 738 (error = VFS_STATFS(mp, sp, td))) {
984263bc
MD
739 simple_lock(&mountlist_slock);
740 nmp = TAILQ_NEXT(mp, mnt_list);
dadab5e9 741 vfs_unbusy(mp, td);
984263bc
MD
742 continue;
743 }
744 sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
745 error = copyout((caddr_t)sp, sfsp, sizeof(*sp));
746 if (error) {
dadab5e9 747 vfs_unbusy(mp, td);
984263bc
MD
748 return (error);
749 }
750 sfsp += sizeof(*sp);
751 }
752 count++;
753 simple_lock(&mountlist_slock);
754 nmp = TAILQ_NEXT(mp, mnt_list);
dadab5e9 755 vfs_unbusy(mp, td);
984263bc
MD
756 }
757 simple_unlock(&mountlist_slock);
758 if (sfsp && count > maxcount)
759 p->p_retval[0] = maxcount;
760 else
761 p->p_retval[0] = count;
762 return (0);
763}
764
765/*
41c20dac
MD
766 * fchdir_args(int fd)
767 *
984263bc
MD
768 * Change current working directory to a given file descriptor.
769 */
984263bc
MD
770/* ARGSUSED */
771int
41c20dac 772fchdir(struct fchdir_args *uap)
984263bc 773{
dadab5e9
MD
774 struct thread *td = curthread;
775 struct proc *p = td->td_proc;
41c20dac 776 struct filedesc *fdp = p->p_fd;
984263bc
MD
777 struct vnode *vp, *tdp;
778 struct mount *mp;
779 struct file *fp;
780 int error;
781
782 if ((error = getvnode(fdp, SCARG(uap, fd), &fp)) != 0)
783 return (error);
784 vp = (struct vnode *)fp->f_data;
785 VREF(vp);
dadab5e9 786 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
984263bc
MD
787 if (vp->v_type != VDIR)
788 error = ENOTDIR;
789 else
dadab5e9 790 error = VOP_ACCESS(vp, VEXEC, p->p_ucred, td);
984263bc 791 while (!error && (mp = vp->v_mountedhere) != NULL) {
dadab5e9 792 if (vfs_busy(mp, 0, 0, td))
984263bc
MD
793 continue;
794 error = VFS_ROOT(mp, &tdp);
dadab5e9 795 vfs_unbusy(mp, td);
984263bc
MD
796 if (error)
797 break;
798 vput(vp);
799 vp = tdp;
800 }
801 if (error) {
802 vput(vp);
803 return (error);
804 }
dadab5e9 805 VOP_UNLOCK(vp, 0, td);
984263bc
MD
806 vrele(fdp->fd_cdir);
807 fdp->fd_cdir = vp;
808 return (0);
809}
810
811/*
41c20dac
MD
812 * chdir_args(char *path)
813 *
984263bc
MD
814 * Change current working directory (``.'').
815 */
984263bc
MD
816/* ARGSUSED */
817int
41c20dac 818chdir(struct chdir_args *uap)
984263bc 819{
dadab5e9
MD
820 struct thread *td = curthread;
821 struct proc *p = td->td_proc;
41c20dac 822 struct filedesc *fdp = p->p_fd;
984263bc
MD
823 int error;
824 struct nameidata nd;
825
826 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
dadab5e9
MD
827 SCARG(uap, path), td);
828 if ((error = change_dir(&nd, td)) != 0)
984263bc
MD
829 return (error);
830 NDFREE(&nd, NDF_ONLY_PNBUF);
831 vrele(fdp->fd_cdir);
832 fdp->fd_cdir = nd.ni_vp;
833 return (0);
834}
835
836/*
837 * Helper function for raised chroot(2) security function: Refuse if
838 * any filedescriptors are open directories.
839 */
840static int
841chroot_refuse_vdir_fds(fdp)
842 struct filedesc *fdp;
843{
844 struct vnode *vp;
845 struct file *fp;
846 int error;
847 int fd;
848
849 for (fd = 0; fd < fdp->fd_nfiles ; fd++) {
850 error = getvnode(fdp, fd, &fp);
851 if (error)
852 continue;
853 vp = (struct vnode *)fp->f_data;
854 if (vp->v_type != VDIR)
855 continue;
856 return(EPERM);
857 }
858 return (0);
859}
860
861/*
862 * This sysctl determines if we will allow a process to chroot(2) if it
863 * has a directory open:
864 * 0: disallowed for all processes.
865 * 1: allowed for processes that were not already chroot(2)'ed.
866 * 2: allowed for all processes.
867 */
868
869static int chroot_allow_open_directories = 1;
870
871SYSCTL_INT(_kern, OID_AUTO, chroot_allow_open_directories, CTLFLAG_RW,
872 &chroot_allow_open_directories, 0, "");
873
874/*
41c20dac
MD
875 * chroot_args(char *path)
876 *
984263bc
MD
877 * Change notion of root (``/'') directory.
878 */
984263bc
MD
879/* ARGSUSED */
880int
41c20dac 881chroot(struct chroot_args *uap)
984263bc 882{
dadab5e9
MD
883 struct thread *td = curthread;
884 struct proc *p = td->td_proc;
41c20dac 885 struct filedesc *fdp = p->p_fd;
984263bc
MD
886 int error;
887 struct nameidata nd;
888
dadab5e9
MD
889 KKASSERT(p);
890 error = suser_cred(p->p_ucred, PRISON_ROOT);
984263bc
MD
891 if (error)
892 return (error);
893 if (chroot_allow_open_directories == 0 ||
894 (chroot_allow_open_directories == 1 && fdp->fd_rdir != rootvnode))
895 error = chroot_refuse_vdir_fds(fdp);
896 if (error)
897 return (error);
898 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
dadab5e9
MD
899 SCARG(uap, path), td);
900 if ((error = change_dir(&nd, td)) != 0)
984263bc
MD
901 return (error);
902 NDFREE(&nd, NDF_ONLY_PNBUF);
903 vrele(fdp->fd_rdir);
904 fdp->fd_rdir = nd.ni_vp;
905 if (!fdp->fd_jdir) {
906 fdp->fd_jdir = nd.ni_vp;
907 VREF(fdp->fd_jdir);
908 }
909 return (0);
910}
911
912/*
913 * Common routine for chroot and chdir.
914 */
915static int
dadab5e9 916change_dir(struct nameidata *ndp, struct thread *td)
984263bc
MD
917{
918 struct vnode *vp;
919 int error;
920
921 error = namei(ndp);
922 if (error)
923 return (error);
924 vp = ndp->ni_vp;
925 if (vp->v_type != VDIR)
926 error = ENOTDIR;
927 else
dadab5e9 928 error = VOP_ACCESS(vp, VEXEC, ndp->ni_cnd.cn_cred, td);
984263bc
MD
929 if (error)
930 vput(vp);
931 else
dadab5e9 932 VOP_UNLOCK(vp, 0, td);
984263bc
MD
933 return (error);
934}
935
936/*
41c20dac
MD
937 * open_args(char *path, int flags, int mode)
938 *
984263bc
MD
939 * Check permissions, allocate an open file structure,
940 * and call the device open routine if any.
941 */
984263bc 942int
41c20dac 943open(struct open_args *uap)
984263bc 944{
dadab5e9
MD
945 struct thread *td = curthread;
946 struct proc *p = td->td_proc;
41c20dac
MD
947 struct filedesc *fdp = p->p_fd;
948 struct file *fp;
949 struct vnode *vp;
984263bc
MD
950 int cmode, flags, oflags;
951 struct file *nfp;
952 int type, indx, error;
953 struct flock lf;
954 struct nameidata nd;
955
956 oflags = SCARG(uap, flags);
957 if ((oflags & O_ACCMODE) == O_ACCMODE)
958 return (EINVAL);
959 flags = FFLAGS(oflags);
960 error = falloc(p, &nfp, &indx);
961 if (error)
962 return (error);
963 fp = nfp;
964 cmode = ((SCARG(uap, mode) &~ fdp->fd_cmask) & ALLPERMS) &~ S_ISTXT;
dadab5e9 965 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), td);
984263bc
MD
966 p->p_dupfd = -indx - 1; /* XXX check for fdopen */
967 /*
968 * Bump the ref count to prevent another process from closing
969 * the descriptor while we are blocked in vn_open()
970 */
971 fhold(fp);
972 error = vn_open(&nd, flags, cmode);
973 if (error) {
974 /*
975 * release our own reference
976 */
dadab5e9 977 fdrop(fp, td);
984263bc
MD
978
979 /*
980 * handle special fdopen() case. bleh. dupfdopen() is
981 * responsible for dropping the old contents of ofiles[indx]
982 * if it succeeds.
983 */
984 if ((error == ENODEV || error == ENXIO) &&
985 p->p_dupfd >= 0 && /* XXX from fdopen */
986 (error =
41c20dac 987 dupfdopen(fdp, indx, p->p_dupfd, flags, error)) == 0) {
984263bc
MD
988 p->p_retval[0] = indx;
989 return (0);
990 }
991 /*
992 * Clean up the descriptor, but only if another thread hadn't
993 * replaced or closed it.
994 */
995 if (fdp->fd_ofiles[indx] == fp) {
996 fdp->fd_ofiles[indx] = NULL;
dadab5e9 997 fdrop(fp, td);
984263bc
MD
998 }
999
1000 if (error == ERESTART)
1001 error = EINTR;
1002 return (error);
1003 }
1004 p->p_dupfd = 0;
1005 NDFREE(&nd, NDF_ONLY_PNBUF);
1006 vp = nd.ni_vp;
1007
1008 /*
1009 * There should be 2 references on the file, one from the descriptor
1010 * table, and one for us.
1011 *
1012 * Handle the case where someone closed the file (via its file
1013 * descriptor) while we were blocked. The end result should look
1014 * like opening the file succeeded but it was immediately closed.
1015 */
1016 if (fp->f_count == 1) {
1017 KASSERT(fdp->fd_ofiles[indx] != fp,
1018 ("Open file descriptor lost all refs"));
dadab5e9
MD
1019 VOP_UNLOCK(vp, 0, td);
1020 vn_close(vp, flags & FMASK, fp->f_cred, td);
1021 fdrop(fp, td);
984263bc
MD
1022 p->p_retval[0] = indx;
1023 return 0;
1024 }
1025
1026 fp->f_data = (caddr_t)vp;
1027 fp->f_flag = flags & FMASK;
1028 fp->f_ops = &vnops;
1029 fp->f_type = (vp->v_type == VFIFO ? DTYPE_FIFO : DTYPE_VNODE);
1030 if (flags & (O_EXLOCK | O_SHLOCK)) {
1031 lf.l_whence = SEEK_SET;
1032 lf.l_start = 0;
1033 lf.l_len = 0;
1034 if (flags & O_EXLOCK)
1035 lf.l_type = F_WRLCK;
1036 else
1037 lf.l_type = F_RDLCK;
1038 type = F_FLOCK;
1039 if ((flags & FNONBLOCK) == 0)
1040 type |= F_WAIT;
dadab5e9 1041 VOP_UNLOCK(vp, 0, td);
984263bc
MD
1042 if ((error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type)) != 0) {
1043 /*
1044 * lock request failed. Normally close the descriptor
1045 * but handle the case where someone might have dup()d
1046 * it when we weren't looking. One reference is
1047 * owned by the descriptor array, the other by us.
1048 */
1049 if (fdp->fd_ofiles[indx] == fp) {
1050 fdp->fd_ofiles[indx] = NULL;
dadab5e9 1051 fdrop(fp, td);
984263bc 1052 }
dadab5e9 1053 fdrop(fp, td);
984263bc
MD
1054 return (error);
1055 }
dadab5e9 1056 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
984263bc
MD
1057 fp->f_flag |= FHASLOCK;
1058 }
1059 /* assert that vn_open created a backing object if one is needed */
1060 KASSERT(!vn_canvmio(vp) || VOP_GETVOBJECT(vp, NULL) == 0,
1061 ("open: vmio vnode has no backing object after vn_open"));
dadab5e9 1062 VOP_UNLOCK(vp, 0, td);
984263bc
MD
1063
1064 /*
1065 * release our private reference, leaving the one associated with the
1066 * descriptor table intact.
1067 */
dadab5e9 1068 fdrop(fp, td);
984263bc
MD
1069 p->p_retval[0] = indx;
1070 return (0);
1071}
1072
1073#ifdef COMPAT_43
1074/*
41c20dac
MD
1075 * ocreat(char *path, int mode)
1076 *
984263bc
MD
1077 * Create a file.
1078 */
984263bc 1079int
41c20dac 1080ocreat(struct ocreat_args *uap)
984263bc
MD
1081{
1082 struct open_args /* {
1083 syscallarg(char *) path;
1084 syscallarg(int) flags;
1085 syscallarg(int) mode;
1086 } */ nuap;
1087
1088 SCARG(&nuap, path) = SCARG(uap, path);
1089 SCARG(&nuap, mode) = SCARG(uap, mode);
1090 SCARG(&nuap, flags) = O_WRONLY | O_CREAT | O_TRUNC;
41c20dac 1091 return (open(&nuap));
984263bc
MD
1092}
1093#endif /* COMPAT_43 */
1094
1095/*
41c20dac
MD
1096 * mknod_args(char *path, int mode, int dev)
1097 *
984263bc
MD
1098 * Create a special file.
1099 */
984263bc
MD
1100/* ARGSUSED */
1101int
41c20dac 1102mknod(struct mknod_args *uap)
984263bc 1103{
dadab5e9
MD
1104 struct thread *td = curthread;
1105 struct proc *p = td->td_proc;
41c20dac 1106 struct vnode *vp;
984263bc
MD
1107 struct vattr vattr;
1108 int error;
1109 int whiteout = 0;
1110 struct nameidata nd;
1111
dadab5e9
MD
1112 KKASSERT(p);
1113
984263bc
MD
1114 switch (SCARG(uap, mode) & S_IFMT) {
1115 case S_IFCHR:
1116 case S_IFBLK:
dadab5e9 1117 error = suser(td);
984263bc
MD
1118 break;
1119 default:
dadab5e9 1120 error = suser_cred(p->p_ucred, PRISON_ROOT);
984263bc
MD
1121 break;
1122 }
1123 if (error)
1124 return (error);
1125 bwillwrite();
dadab5e9 1126 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), td);
984263bc
MD
1127 if ((error = namei(&nd)) != 0)
1128 return (error);
1129 vp = nd.ni_vp;
1130 if (vp != NULL)
1131 error = EEXIST;
1132 else {
1133 VATTR_NULL(&vattr);
1134 vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
1135 vattr.va_rdev = SCARG(uap, dev);
1136 whiteout = 0;
1137
1138 switch (SCARG(uap, mode) & S_IFMT) {
1139 case S_IFMT: /* used by badsect to flag bad sectors */
1140 vattr.va_type = VBAD;
1141 break;
1142 case S_IFCHR:
1143 vattr.va_type = VCHR;
1144 break;
1145 case S_IFBLK:
1146 vattr.va_type = VBLK;
1147 break;
1148 case S_IFWHT:
1149 whiteout = 1;
1150 break;
1151 default:
1152 error = EINVAL;
1153 break;
1154 }
1155 }
1156 if (!error) {
dadab5e9 1157 VOP_LEASE(nd.ni_dvp, td, p->p_ucred, LEASE_WRITE);
984263bc
MD
1158 if (whiteout)
1159 error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, CREATE);
1160 else {
1161 error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp,
1162 &nd.ni_cnd, &vattr);
1163 if (error == 0)
1164 vput(nd.ni_vp);
1165 }
1166 NDFREE(&nd, NDF_ONLY_PNBUF);
1167 vput(nd.ni_dvp);
1168 } else {
1169 NDFREE(&nd, NDF_ONLY_PNBUF);
1170 if (nd.ni_dvp == vp)
1171 vrele(nd.ni_dvp);
1172 else
1173 vput(nd.ni_dvp);
1174 if (vp)
1175 vrele(vp);
1176 }
1177 ASSERT_VOP_UNLOCKED(nd.ni_dvp, "mknod");
1178 ASSERT_VOP_UNLOCKED(nd.ni_vp, "mknod");
1179 return (error);
1180}
1181
1182/*
41c20dac
MD
1183 * mkfifo_args(char *path, int mode)
1184 *
984263bc
MD
1185 * Create a named pipe.
1186 */
984263bc
MD
1187/* ARGSUSED */
1188int
41c20dac 1189mkfifo(struct mkfifo_args *uap)
984263bc 1190{
dadab5e9
MD
1191 struct thread *td = curthread;
1192 struct proc *p = td->td_proc;
984263bc
MD
1193 struct vattr vattr;
1194 int error;
1195 struct nameidata nd;
1196
1197 bwillwrite();
dadab5e9 1198 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), td);
984263bc
MD
1199 if ((error = namei(&nd)) != 0)
1200 return (error);
1201 if (nd.ni_vp != NULL) {
1202 NDFREE(&nd, NDF_ONLY_PNBUF);
1203 if (nd.ni_dvp == nd.ni_vp)
1204 vrele(nd.ni_dvp);
1205 else
1206 vput(nd.ni_dvp);
1207 vrele(nd.ni_vp);
1208 return (EEXIST);
1209 }
1210 VATTR_NULL(&vattr);
1211 vattr.va_type = VFIFO;
1212 vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
dadab5e9 1213 VOP_LEASE(nd.ni_dvp, td, p->p_ucred, LEASE_WRITE);
984263bc
MD
1214 error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
1215 if (error == 0)
1216 vput(nd.ni_vp);
1217 NDFREE(&nd, NDF_ONLY_PNBUF);
1218 vput(nd.ni_dvp);
1219 return (error);
1220}
1221
1222/*
41c20dac
MD
1223 * link_args(char *path, char *link)
1224 *
984263bc
MD
1225 * Make a hard file link.
1226 */
984263bc
MD
1227/* ARGSUSED */
1228int
41c20dac 1229link(struct link_args *uap)
984263bc 1230{
dadab5e9
MD
1231 struct thread *td = curthread;
1232 struct proc *p = td->td_proc;
41c20dac 1233 struct vnode *vp;
984263bc
MD
1234 struct nameidata nd;
1235 int error;
1236
1237 bwillwrite();
dadab5e9 1238 NDINIT(&nd, LOOKUP, FOLLOW|NOOBJ, UIO_USERSPACE, SCARG(uap, path), td);
984263bc
MD
1239 if ((error = namei(&nd)) != 0)
1240 return (error);
1241 NDFREE(&nd, NDF_ONLY_PNBUF);
1242 vp = nd.ni_vp;
1243 if (vp->v_type == VDIR)
1244 error = EPERM; /* POSIX */
1245 else {
dadab5e9 1246 NDINIT(&nd, CREATE, LOCKPARENT|NOOBJ, UIO_USERSPACE, SCARG(uap, link), td);
984263bc
MD
1247 error = namei(&nd);
1248 if (!error) {
1249 if (nd.ni_vp != NULL) {
1250 if (nd.ni_vp)
1251 vrele(nd.ni_vp);
1252 error = EEXIST;
1253 } else {
dadab5e9 1254 VOP_LEASE(nd.ni_dvp, td, p->p_ucred,
984263bc 1255 LEASE_WRITE);
dadab5e9 1256 VOP_LEASE(vp, td, p->p_ucred, LEASE_WRITE);
984263bc
MD
1257 error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd);
1258 }
1259 NDFREE(&nd, NDF_ONLY_PNBUF);
1260 if (nd.ni_dvp == nd.ni_vp)
1261 vrele(nd.ni_dvp);
1262 else
1263 vput(nd.ni_dvp);
1264 }
1265 }
1266 vrele(vp);
1267 ASSERT_VOP_UNLOCKED(nd.ni_dvp, "link");
1268 ASSERT_VOP_UNLOCKED(nd.ni_vp, "link");
1269 return (error);
1270}
1271
1272/*
41c20dac
MD
1273 * symlink(char *path, char *link)
1274 *
984263bc
MD
1275 * Make a symbolic link.
1276 */
984263bc
MD
1277/* ARGSUSED */
1278int
41c20dac 1279symlink(struct symlink_args *uap)
984263bc 1280{
dadab5e9
MD
1281 struct thread *td = curthread;
1282 struct proc *p = td->td_proc;
984263bc
MD
1283 struct vattr vattr;
1284 char *path;
1285 int error;
1286 struct nameidata nd;
1287
1288 path = zalloc(namei_zone);
1289 if ((error = copyinstr(SCARG(uap, path), path, MAXPATHLEN, NULL)) != 0)
1290 goto out;
1291 bwillwrite();
dadab5e9 1292 NDINIT(&nd, CREATE, LOCKPARENT|NOOBJ, UIO_USERSPACE, SCARG(uap, link), td);
984263bc
MD
1293 if ((error = namei(&nd)) != 0)
1294 goto out;
1295 if (nd.ni_vp) {
1296 NDFREE(&nd, NDF_ONLY_PNBUF);
1297 if (nd.ni_dvp == nd.ni_vp)
1298 vrele(nd.ni_dvp);
1299 else
1300 vput(nd.ni_dvp);
1301 vrele(nd.ni_vp);
1302 error = EEXIST;
1303 goto out;
1304 }
1305 VATTR_NULL(&vattr);
1306 vattr.va_mode = ACCESSPERMS &~ p->p_fd->fd_cmask;
dadab5e9 1307 VOP_LEASE(nd.ni_dvp, td, p->p_ucred, LEASE_WRITE);
984263bc
MD
1308 error = VOP_SYMLINK(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr, path);
1309 NDFREE(&nd, NDF_ONLY_PNBUF);
1310 if (error == 0)
1311 vput(nd.ni_vp);
1312 vput(nd.ni_dvp);
1313 ASSERT_VOP_UNLOCKED(nd.ni_dvp, "symlink");
1314 ASSERT_VOP_UNLOCKED(nd.ni_vp, "symlink");
1315out:
1316 zfree(namei_zone, path);
1317 return (error);
1318}
1319
1320/*
41c20dac
MD
1321 * undelete_args(char *path)
1322 *
984263bc
MD
1323 * Delete a whiteout from the filesystem.
1324 */
1325/* ARGSUSED */
1326int
41c20dac 1327undelete(struct undelete_args *uap)
984263bc 1328{
dadab5e9
MD
1329 struct thread *td = curthread;
1330 struct proc *p = td->td_proc;
984263bc
MD
1331 int error;
1332 struct nameidata nd;
1333
1334 bwillwrite();
1335 NDINIT(&nd, DELETE, LOCKPARENT|DOWHITEOUT, UIO_USERSPACE,
dadab5e9 1336 SCARG(uap, path), td);
984263bc
MD
1337 error = namei(&nd);
1338 if (error)
1339 return (error);
1340
1341 if (nd.ni_vp != NULLVP || !(nd.ni_cnd.cn_flags & ISWHITEOUT)) {
1342 NDFREE(&nd, NDF_ONLY_PNBUF);
1343 if (nd.ni_dvp == nd.ni_vp)
1344 vrele(nd.ni_dvp);
1345 else
1346 vput(nd.ni_dvp);
1347 if (nd.ni_vp)
1348 vrele(nd.ni_vp);
1349 return (EEXIST);
1350 }
1351
dadab5e9 1352 VOP_LEASE(nd.ni_dvp, td, p->p_ucred, LEASE_WRITE);
984263bc
MD
1353 error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, DELETE);
1354 NDFREE(&nd, NDF_ONLY_PNBUF);
1355 vput(nd.ni_dvp);
1356 ASSERT_VOP_UNLOCKED(nd.ni_dvp, "undelete");
1357 ASSERT_VOP_UNLOCKED(nd.ni_vp, "undelete");
1358 return (error);
1359}
1360
1361/*
41c20dac
MD
1362 * unlink_args(char *path)
1363 *
984263bc
MD
1364 * Delete a name from the filesystem.
1365 */
984263bc 1366int
41c20dac 1367unlink(struct unlink_args *uap)
984263bc 1368{
dadab5e9
MD
1369 struct thread *td = curthread;
1370 struct proc *p = td->td_proc;
41c20dac 1371 struct vnode *vp;
984263bc
MD
1372 int error;
1373 struct nameidata nd;
1374
1375 bwillwrite();
dadab5e9 1376 NDINIT(&nd, DELETE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), td);
984263bc
MD
1377 if ((error = namei(&nd)) != 0)
1378 return (error);
1379 vp = nd.ni_vp;
dadab5e9
MD
1380 VOP_LEASE(vp, td, p->p_ucred, LEASE_WRITE);
1381 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
984263bc
MD
1382
1383 if (vp->v_type == VDIR)
1384 error = EPERM; /* POSIX */
1385 else {
1386 /*
1387 * The root of a mounted filesystem cannot be deleted.
1388 *
1389 * XXX: can this only be a VDIR case?
1390 */
1391 if (vp->v_flag & VROOT)
1392 error = EBUSY;
1393 }
1394
1395 if (!error) {
dadab5e9 1396 VOP_LEASE(nd.ni_dvp, td, p->p_ucred, LEASE_WRITE);
984263bc
MD
1397 error = VOP_REMOVE(nd.ni_dvp, vp, &nd.ni_cnd);
1398 }
1399 NDFREE(&nd, NDF_ONLY_PNBUF);
1400 if (nd.ni_dvp == vp)
1401 vrele(nd.ni_dvp);
1402 else
1403 vput(nd.ni_dvp);
1404 if (vp != NULLVP)
1405 vput(vp);
1406 ASSERT_VOP_UNLOCKED(nd.ni_dvp, "unlink");
1407 ASSERT_VOP_UNLOCKED(nd.ni_vp, "unlink");
1408 return (error);
1409}
1410
1411/*
41c20dac
MD
1412 * lseek_args(int fd, int pad, off_t offset, int whence)
1413 *
984263bc
MD
1414 * Reposition read/write file offset.
1415 */
984263bc 1416int
41c20dac 1417lseek(struct lseek_args *uap)
984263bc 1418{
dadab5e9
MD
1419 struct thread *td = curthread;
1420 struct proc *p = td->td_proc;
984263bc 1421 struct ucred *cred = p->p_ucred;
41c20dac
MD
1422 struct filedesc *fdp = p->p_fd;
1423 struct file *fp;
984263bc
MD
1424 struct vattr vattr;
1425 int error;
1426
1427 if ((u_int)SCARG(uap, fd) >= fdp->fd_nfiles ||
1428 (fp = fdp->fd_ofiles[SCARG(uap, fd)]) == NULL)
1429 return (EBADF);
1430 if (fp->f_type != DTYPE_VNODE)
1431 return (ESPIPE);
1432 switch (SCARG(uap, whence)) {
1433 case L_INCR:
1434 fp->f_offset += SCARG(uap, offset);
1435 break;
1436 case L_XTND:
dadab5e9 1437 error=VOP_GETATTR((struct vnode *)fp->f_data, &vattr, cred, td);
984263bc
MD
1438 if (error)
1439 return (error);
1440 fp->f_offset = SCARG(uap, offset) + vattr.va_size;
1441 break;
1442 case L_SET:
1443 fp->f_offset = SCARG(uap, offset);
1444 break;
1445 default:
1446 return (EINVAL);
1447 }
1448 *(off_t *)(p->p_retval) = fp->f_offset;
1449 return (0);
1450}
1451
1452#if defined(COMPAT_43) || defined(COMPAT_SUNOS)
1453/*
1454 * Reposition read/write file offset.
41c20dac
MD
1455 *
1456 * olseek_args(int fd, long offset, int whence)
984263bc 1457 */
984263bc 1458int
41c20dac 1459olseek(struct olseek_args *uap)
984263bc
MD
1460{
1461 struct lseek_args /* {
1462 syscallarg(int) fd;
1463 syscallarg(int) pad;
1464 syscallarg(off_t) offset;
1465 syscallarg(int) whence;
1466 } */ nuap;
1467 int error;
1468
1469 SCARG(&nuap, fd) = SCARG(uap, fd);
1470 SCARG(&nuap, offset) = SCARG(uap, offset);
1471 SCARG(&nuap, whence) = SCARG(uap, whence);
41c20dac 1472 error = lseek(&nuap);
984263bc
MD
1473 return (error);
1474}
1475#endif /* COMPAT_43 */
1476
1477/*
41c20dac
MD
1478 * access_args(char *path, int flags)
1479 *
984263bc
MD
1480 * Check access permissions.
1481 */
984263bc 1482int
41c20dac 1483access(struct access_args *uap)
984263bc 1484{
dadab5e9
MD
1485 struct thread *td = curthread;
1486 struct proc *p = td->td_proc;
984263bc 1487 struct ucred *cred, *tmpcred;
41c20dac 1488 struct vnode *vp;
984263bc
MD
1489 int error, flags;
1490 struct nameidata nd;
1491
1492 cred = p->p_ucred;
1493 /*
1494 * Create and modify a temporary credential instead of one that
1495 * is potentially shared. This could also mess up socket
1496 * buffer accounting which can run in an interrupt context.
1497 */
1498 tmpcred = crdup(cred);
41c20dac
MD
1499 tmpcred->cr_uid = p->p_ucred->cr_ruid;
1500 tmpcred->cr_groups[0] = p->p_ucred->cr_rgid;
984263bc
MD
1501 p->p_ucred = tmpcred;
1502 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
dadab5e9 1503 SCARG(uap, path), td);
984263bc
MD
1504 if ((error = namei(&nd)) != 0)
1505 goto out1;
1506 vp = nd.ni_vp;
1507
1508 /* Flags == 0 means only check for existence. */
1509 if (SCARG(uap, flags)) {
1510 flags = 0;
1511 if (SCARG(uap, flags) & R_OK)
1512 flags |= VREAD;
1513 if (SCARG(uap, flags) & W_OK)
1514 flags |= VWRITE;
1515 if (SCARG(uap, flags) & X_OK)
1516 flags |= VEXEC;
1517 if ((flags & VWRITE) == 0 || (error = vn_writechk(vp)) == 0)
dadab5e9 1518 error = VOP_ACCESS(vp, flags, tmpcred, td);
984263bc
MD
1519 }
1520 NDFREE(&nd, NDF_ONLY_PNBUF);
1521 vput(vp);
1522out1:
1523 p->p_ucred = cred;
1524 crfree(tmpcred);
1525 return (error);
1526}
1527
1528#if defined(COMPAT_43) || defined(COMPAT_SUNOS)
1529/*
41c20dac
MD
1530 * ostat_args(char *path, struct ostat *ub)
1531 *
984263bc
MD
1532 * Get file status; this version follows links.
1533 */
984263bc
MD
1534/* ARGSUSED */
1535int
41c20dac 1536ostat(struct ostat_args *uap)
984263bc 1537{
dadab5e9 1538 struct thread *td = curthread;
984263bc
MD
1539 struct stat sb;
1540 struct ostat osb;
1541 int error;
1542 struct nameidata nd;
1543
1544 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
dadab5e9 1545 SCARG(uap, path), td);
984263bc
MD
1546 if ((error = namei(&nd)) != 0)
1547 return (error);
1548 NDFREE(&nd, NDF_ONLY_PNBUF);
dadab5e9 1549 error = vn_stat(nd.ni_vp, &sb, td);
984263bc
MD
1550 vput(nd.ni_vp);
1551 if (error)
1552 return (error);
1553 cvtstat(&sb, &osb);
1554 error = copyout((caddr_t)&osb, (caddr_t)SCARG(uap, ub), sizeof (osb));
1555 return (error);
1556}
1557
1558/*
41c20dac
MD
1559 * olstat_args(char *path, struct ostat *ub)
1560 *
984263bc
MD
1561 * Get file status; this version does not follow links.
1562 */
984263bc
MD
1563/* ARGSUSED */
1564int
41c20dac 1565olstat(struct olstat_args *uap)
984263bc 1566{
dadab5e9 1567 struct thread *td = curthread;
984263bc
MD
1568 struct vnode *vp;
1569 struct stat sb;
1570 struct ostat osb;
1571 int error;
1572 struct nameidata nd;
1573
1574 NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
dadab5e9 1575 SCARG(uap, path), td);
984263bc
MD
1576 if ((error = namei(&nd)) != 0)
1577 return (error);
1578 vp = nd.ni_vp;
dadab5e9 1579 error = vn_stat(vp, &sb, td);
984263bc
MD
1580 NDFREE(&nd, NDF_ONLY_PNBUF);
1581 vput(vp);
1582 if (error)
1583 return (error);
1584 cvtstat(&sb, &osb);
1585 error = copyout((caddr_t)&osb, (caddr_t)SCARG(uap, ub), sizeof (osb));
1586 return (error);
1587}
1588
1589/*
1590 * Convert from an old to a new stat structure.
1591 */
1592void
1593cvtstat(st, ost)
1594 struct stat *st;
1595 struct ostat *ost;
1596{
984263bc
MD
1597 ost->st_dev = st->st_dev;
1598 ost->st_ino = st->st_ino;
1599 ost->st_mode = st->st_mode;
1600 ost->st_nlink = st->st_nlink;
1601 ost->st_uid = st->st_uid;
1602 ost->st_gid = st->st_gid;
1603 ost->st_rdev = st->st_rdev;
1604 if (st->st_size < (quad_t)1 << 32)
1605 ost->st_size = st->st_size;
1606 else
1607 ost->st_size = -2;
1608 ost->st_atime = st->st_atime;
1609 ost->st_mtime = st->st_mtime;
1610 ost->st_ctime = st->st_ctime;
1611 ost->st_blksize = st->st_blksize;
1612 ost->st_blocks = st->st_blocks;
1613 ost->st_flags = st->st_flags;
1614 ost->st_gen = st->st_gen;
1615}
1616#endif /* COMPAT_43 || COMPAT_SUNOS */
1617
1618/*
41c20dac
MD
1619 * stat_args(char *path, struct stat *ub)
1620 *
984263bc
MD
1621 * Get file status; this version follows links.
1622 */
984263bc
MD
1623/* ARGSUSED */
1624int
41c20dac 1625stat(struct stat_args *uap)
984263bc 1626{
dadab5e9 1627 struct thread *td = curthread;
984263bc
MD
1628 struct stat sb;
1629 int error;
1630 struct nameidata nd;
1631
1632 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
dadab5e9 1633 SCARG(uap, path), td);
984263bc
MD
1634 if ((error = namei(&nd)) != 0)
1635 return (error);
dadab5e9 1636 error = vn_stat(nd.ni_vp, &sb, td);
984263bc
MD
1637 NDFREE(&nd, NDF_ONLY_PNBUF);
1638 vput(nd.ni_vp);
1639 if (error)
1640 return (error);
1641 error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
1642 return (error);
1643}
1644
1645/*
41c20dac
MD
1646 * lstat_args(char *path, struct stat *ub)
1647 *
984263bc
MD
1648 * Get file status; this version does not follow links.
1649 */
984263bc
MD
1650/* ARGSUSED */
1651int
41c20dac 1652lstat(struct lstat_args *uap)
984263bc 1653{
dadab5e9 1654 struct thread *td = curthread;
984263bc
MD
1655 int error;
1656 struct vnode *vp;
1657 struct stat sb;
1658 struct nameidata nd;
1659
1660 NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
dadab5e9 1661 SCARG(uap, path), td);
984263bc
MD
1662 if ((error = namei(&nd)) != 0)
1663 return (error);
1664 vp = nd.ni_vp;
dadab5e9 1665 error = vn_stat(vp, &sb, td);
984263bc
MD
1666 NDFREE(&nd, NDF_ONLY_PNBUF);
1667 vput(vp);
1668 if (error)
1669 return (error);
1670 error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
1671 return (error);
1672}
1673
1674void
1675cvtnstat(sb, nsb)
1676 struct stat *sb;
1677 struct nstat *nsb;
1678{
1679 nsb->st_dev = sb->st_dev;
1680 nsb->st_ino = sb->st_ino;
1681 nsb->st_mode = sb->st_mode;
1682 nsb->st_nlink = sb->st_nlink;
1683 nsb->st_uid = sb->st_uid;
1684 nsb->st_gid = sb->st_gid;
1685 nsb->st_rdev = sb->st_rdev;
1686 nsb->st_atimespec = sb->st_atimespec;
1687 nsb->st_mtimespec = sb->st_mtimespec;
1688 nsb->st_ctimespec = sb->st_ctimespec;
1689 nsb->st_size = sb->st_size;
1690 nsb->st_blocks = sb->st_blocks;
1691 nsb->st_blksize = sb->st_blksize;
1692 nsb->st_flags = sb->st_flags;
1693 nsb->st_gen = sb->st_gen;
1694 nsb->st_qspare[0] = sb->st_qspare[0];
1695 nsb->st_qspare[1] = sb->st_qspare[1];
1696}
1697
41c20dac
MD
1698/*
1699 * nstat_args(char *path, struct nstat *ub)
1700 */
984263bc
MD
1701/* ARGSUSED */
1702int
41c20dac 1703nstat(struct nstat_args *uap)
984263bc 1704{
dadab5e9 1705 struct thread *td = curthread;
984263bc
MD
1706 struct stat sb;
1707 struct nstat nsb;
1708 int error;
1709 struct nameidata nd;
1710
1711 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
dadab5e9 1712 SCARG(uap, path), td);
984263bc
MD
1713 if ((error = namei(&nd)) != 0)
1714 return (error);
1715 NDFREE(&nd, NDF_ONLY_PNBUF);
dadab5e9 1716 error = vn_stat(nd.ni_vp, &sb, td);
984263bc
MD
1717 vput(nd.ni_vp);
1718 if (error)
1719 return (error);
1720 cvtnstat(&sb, &nsb);
1721 error = copyout((caddr_t)&nsb, (caddr_t)SCARG(uap, ub), sizeof (nsb));
1722 return (error);
1723}
1724
1725/*
41c20dac
MD
1726 * lstat_args(char *path, struct stat *ub)
1727 *
984263bc
MD
1728 * Get file status; this version does not follow links.
1729 */
984263bc
MD
1730/* ARGSUSED */
1731int
41c20dac 1732nlstat(struct nlstat_args *uap)
984263bc 1733{
dadab5e9 1734 struct thread *td = curthread;
984263bc
MD
1735 int error;
1736 struct vnode *vp;
1737 struct stat sb;
1738 struct nstat nsb;
1739 struct nameidata nd;
1740
1741 NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
dadab5e9 1742 SCARG(uap, path), td);
984263bc
MD
1743 if ((error = namei(&nd)) != 0)
1744 return (error);
1745 vp = nd.ni_vp;
1746 NDFREE(&nd, NDF_ONLY_PNBUF);
dadab5e9 1747 error = vn_stat(vp, &sb, td);
984263bc
MD
1748 vput(vp);
1749 if (error)
1750 return (error);
1751 cvtnstat(&sb, &nsb);
1752 error = copyout((caddr_t)&nsb, (caddr_t)SCARG(uap, ub), sizeof (nsb));
1753 return (error);
1754}
1755
1756/*
41c20dac
MD
1757 * pathconf_Args(char *path, int name)
1758 *
984263bc
MD
1759 * Get configurable pathname variables.
1760 */
984263bc
MD
1761/* ARGSUSED */
1762int
41c20dac 1763pathconf(struct pathconf_args *uap)
984263bc 1764{
dadab5e9
MD
1765 struct thread *td = curthread;
1766 struct proc *p = td->td_proc;
984263bc
MD
1767 int error;
1768 struct nameidata nd;
1769
1770 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
dadab5e9 1771 SCARG(uap, path), td);
984263bc
MD
1772 if ((error = namei(&nd)) != 0)
1773 return (error);
1774 NDFREE(&nd, NDF_ONLY_PNBUF);
1775 error = VOP_PATHCONF(nd.ni_vp, SCARG(uap, name), p->p_retval);
1776 vput(nd.ni_vp);
1777 return (error);
1778}
1779
1780/*
41c20dac
MD
1781 * readlink_args(char *path, char *buf, int count)
1782 *
984263bc
MD
1783 * Return target name of a symbolic link.
1784 */
984263bc
MD
1785/* ARGSUSED */
1786int
41c20dac 1787readlink(struct readlink_args *uap)
984263bc 1788{
dadab5e9
MD
1789 struct thread *td = curthread;
1790 struct proc *p = td->td_proc;
41c20dac 1791 struct vnode *vp;
984263bc
MD
1792 struct iovec aiov;
1793 struct uio auio;
1794 int error;
1795 struct nameidata nd;
1796
1797 NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | NOOBJ, UIO_USERSPACE,
dadab5e9 1798 SCARG(uap, path), td);
984263bc
MD
1799 if ((error = namei(&nd)) != 0)
1800 return (error);
1801 NDFREE(&nd, NDF_ONLY_PNBUF);
1802 vp = nd.ni_vp;
1803 if (vp->v_type != VLNK)
1804 error = EINVAL;
1805 else {
1806 aiov.iov_base = SCARG(uap, buf);
1807 aiov.iov_len = SCARG(uap, count);
1808 auio.uio_iov = &aiov;
1809 auio.uio_iovcnt = 1;
1810 auio.uio_offset = 0;
1811 auio.uio_rw = UIO_READ;
1812 auio.uio_segflg = UIO_USERSPACE;
dadab5e9 1813 auio.uio_td = td;
984263bc
MD
1814 auio.uio_resid = SCARG(uap, count);
1815 error = VOP_READLINK(vp, &auio, p->p_ucred);
1816 }
1817 vput(vp);
1818 p->p_retval[0] = SCARG(uap, count) - auio.uio_resid;
1819 return (error);
1820}
1821
1822static int
41c20dac 1823setfflags(struct vnode *vp, int flags)
984263bc 1824{
dadab5e9
MD
1825 struct thread *td = curthread;
1826 struct proc *p = td->td_proc;
984263bc
MD
1827 int error;
1828 struct vattr vattr;
1829
1830 /*
1831 * Prevent non-root users from setting flags on devices. When
1832 * a device is reused, users can retain ownership of the device
1833 * if they are allowed to set flags and programs assume that
1834 * chown can't fail when done as root.
1835 */
1836 if ((vp->v_type == VCHR || vp->v_type == VBLK) &&
dadab5e9 1837 ((error = suser_cred(p->p_ucred, PRISON_ROOT)) != 0))
984263bc
MD
1838 return (error);
1839
dadab5e9
MD
1840 VOP_LEASE(vp, td, p->p_ucred, LEASE_WRITE);
1841 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
984263bc
MD
1842 VATTR_NULL(&vattr);
1843 vattr.va_flags = flags;
dadab5e9
MD
1844 error = VOP_SETATTR(vp, &vattr, p->p_ucred, td);
1845 VOP_UNLOCK(vp, 0, td);
984263bc
MD
1846 return (error);
1847}
1848
1849/*
41c20dac
MD
1850 * chflags(char *path, int flags)
1851 *
984263bc
MD
1852 * Change flags of a file given a path name.
1853 */
984263bc
MD
1854/* ARGSUSED */
1855int
41c20dac 1856chflags(struct chflags_args *uap)
984263bc 1857{
dadab5e9 1858 struct thread *td = curthread;
984263bc
MD
1859 int error;
1860 struct nameidata nd;
1861
dadab5e9 1862 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), td);
984263bc
MD
1863 if ((error = namei(&nd)) != 0)
1864 return (error);
1865 NDFREE(&nd, NDF_ONLY_PNBUF);
41c20dac 1866 error = setfflags(nd.ni_vp, SCARG(uap, flags));
984263bc
MD
1867 vrele(nd.ni_vp);
1868 return error;
1869}
1870
1871/*
41c20dac
MD
1872 * fchflags_args(int fd, int flags)
1873 *
984263bc
MD
1874 * Change flags of a file given a file descriptor.
1875 */
984263bc
MD
1876/* ARGSUSED */
1877int
41c20dac 1878fchflags(struct fchflags_args *uap)
984263bc 1879{
dadab5e9
MD
1880 struct thread *td = curthread;
1881 struct proc *p = td->td_proc;
984263bc
MD
1882 struct file *fp;
1883 int error;
1884
1885 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
1886 return (error);
41c20dac 1887 return setfflags((struct vnode *) fp->f_data, SCARG(uap, flags));
984263bc
MD
1888}
1889
1890static int
41c20dac 1891setfmode(struct vnode *vp, int mode)
984263bc 1892{
dadab5e9
MD
1893 struct thread *td = curthread;
1894 struct proc *p = td->td_proc;
984263bc
MD
1895 int error;
1896 struct vattr vattr;
1897
dadab5e9
MD
1898 VOP_LEASE(vp, td, p->p_ucred, LEASE_WRITE);
1899 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
984263bc
MD
1900 VATTR_NULL(&vattr);
1901 vattr.va_mode = mode & ALLPERMS;
dadab5e9
MD
1902 error = VOP_SETATTR(vp, &vattr, p->p_ucred, td);
1903 VOP_UNLOCK(vp, 0, td);
984263bc
MD
1904 return error;
1905}
1906
1907/*
41c20dac
MD
1908 * chmod_args(char *path, int mode)
1909 *
984263bc
MD
1910 * Change mode of a file given path name.
1911 */
984263bc
MD
1912/* ARGSUSED */
1913int
41c20dac 1914chmod(struct chmod_args *uap)
984263bc 1915{
dadab5e9 1916 struct thread *td = curthread;
984263bc
MD
1917 int error;
1918 struct nameidata nd;
1919
dadab5e9 1920 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), td);
984263bc
MD
1921 if ((error = namei(&nd)) != 0)
1922 return (error);
1923 NDFREE(&nd, NDF_ONLY_PNBUF);
41c20dac 1924 error = setfmode(nd.ni_vp, SCARG(uap, mode));
984263bc
MD
1925 vrele(nd.ni_vp);
1926 return error;
1927}
1928
1929/*
41c20dac
MD
1930 * lchmod_args(char *path, int mode)
1931 *
984263bc
MD
1932 * Change mode of a file given path name (don't follow links.)
1933 */
984263bc
MD
1934/* ARGSUSED */
1935int
41c20dac 1936lchmod(struct lchmod_args *uap)
984263bc 1937{
dadab5e9 1938 struct thread *td = curthread;
984263bc
MD
1939 int error;
1940 struct nameidata nd;
1941
dadab5e9 1942 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), td);
984263bc
MD
1943 if ((error = namei(&nd)) != 0)
1944 return (error);
1945 NDFREE(&nd, NDF_ONLY_PNBUF);
41c20dac 1946 error = setfmode(nd.ni_vp, SCARG(uap, mode));
984263bc
MD
1947 vrele(nd.ni_vp);
1948 return error;
1949}
1950
1951/*
41c20dac
MD
1952 * fchmod_args(int fd, int mode)
1953 *
984263bc
MD
1954 * Change mode of a file given a file descriptor.
1955 */
984263bc
MD
1956/* ARGSUSED */
1957int
41c20dac 1958fchmod(struct fchmod_args *uap)
984263bc 1959{
dadab5e9
MD
1960 struct thread *td = curthread;
1961 struct proc *p = td->td_proc;
984263bc
MD
1962 struct file *fp;
1963 int error;
1964
1965 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
1966 return (error);
41c20dac 1967 return setfmode((struct vnode *)fp->f_data, SCARG(uap, mode));
984263bc
MD
1968}
1969
1970static int
41c20dac 1971setfown(struct vnode *vp, uid_t uid, gid_t gid)
984263bc 1972{
dadab5e9
MD
1973 struct thread *td = curthread;
1974 struct proc *p = td->td_proc;
984263bc
MD
1975 int error;
1976 struct vattr vattr;
1977
dadab5e9
MD
1978 VOP_LEASE(vp, td, p->p_ucred, LEASE_WRITE);
1979 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
984263bc
MD
1980 VATTR_NULL(&vattr);
1981 vattr.va_uid = uid;
1982 vattr.va_gid = gid;
dadab5e9
MD
1983 error = VOP_SETATTR(vp, &vattr, p->p_ucred, td);
1984 VOP_UNLOCK(vp, 0, td);
984263bc
MD
1985 return error;
1986}
1987
1988/*
41c20dac
MD
1989 * chown(char *path, int uid, int gid)
1990 *
984263bc
MD
1991 * Set ownership given a path name.
1992 */
984263bc
MD
1993/* ARGSUSED */
1994int
41c20dac 1995chown(struct chown_args *uap)
984263bc 1996{
dadab5e9 1997 struct thread *td = curthread;
984263bc
MD
1998 int error;
1999 struct nameidata nd;
2000
dadab5e9 2001 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), td);
984263bc
MD
2002 if ((error = namei(&nd)) != 0)
2003 return (error);
2004 NDFREE(&nd, NDF_ONLY_PNBUF);
41c20dac 2005 error = setfown(nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid));
984263bc
MD
2006 vrele(nd.ni_vp);
2007 return (error);
2008}
2009
2010/*
41c20dac
MD
2011 * lchown_args(char *path, int uid, int gid)
2012 *
984263bc
MD
2013 * Set ownership given a path name, do not cross symlinks.
2014 */
984263bc
MD
2015/* ARGSUSED */
2016int
41c20dac 2017lchown(struct lchown_args *uap)
984263bc 2018{
dadab5e9 2019 struct thread *td = curthread;
984263bc
MD
2020 int error;
2021 struct nameidata nd;
2022
dadab5e9 2023 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), td);
984263bc
MD
2024 if ((error = namei(&nd)) != 0)
2025 return (error);
2026 NDFREE(&nd, NDF_ONLY_PNBUF);
41c20dac 2027 error = setfown(nd.ni_vp, SCARG(uap, uid), SCARG(uap, gid));
984263bc
MD
2028 vrele(nd.ni_vp);
2029 return (error);
2030}
2031
2032/*
41c20dac
MD
2033 * fchown_args(int fd, int uid, int gid)
2034 *
984263bc
MD
2035 * Set ownership given a file descriptor.
2036 */
984263bc
MD
2037/* ARGSUSED */
2038int
41c20dac 2039fchown(struct fchown_args *uap)
984263bc 2040{
dadab5e9
MD
2041 struct thread *td = curthread;
2042 struct proc *p = td->td_proc;
984263bc
MD
2043 struct file *fp;
2044 int error;
2045
2046 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2047 return (error);
41c20dac 2048 return setfown((struct vnode *)fp->f_data,
984263bc
MD
2049 SCARG(uap, uid), SCARG(uap, gid));
2050}
2051
2052static int
41c20dac 2053getutimes(const struct timeval *usrtvp, struct timespec *tsp)
984263bc
MD
2054{
2055 struct timeval tv[2];
2056 int error;
2057
2058 if (usrtvp == NULL) {
2059 microtime(&tv[0]);
2060 TIMEVAL_TO_TIMESPEC(&tv[0], &tsp[0]);
2061 tsp[1] = tsp[0];
2062 } else {
2063 if ((error = copyin(usrtvp, tv, sizeof (tv))) != 0)
2064 return (error);
2065 TIMEVAL_TO_TIMESPEC(&tv[0], &tsp[0]);
2066 TIMEVAL_TO_TIMESPEC(&tv[1], &tsp[1]);
2067 }
2068 return 0;
2069}
2070
2071static int
41c20dac 2072setutimes(struct vnode *vp, const struct timespec *ts, int nullflag)
984263bc 2073{
dadab5e9
MD
2074 struct thread *td = curthread;
2075 struct proc *p = td->td_proc;
984263bc
MD
2076 int error;
2077 struct vattr vattr;
2078
dadab5e9
MD
2079 VOP_LEASE(vp, td, p->p_ucred, LEASE_WRITE);
2080 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
984263bc
MD
2081 VATTR_NULL(&vattr);
2082 vattr.va_atime = ts[0];
2083 vattr.va_mtime = ts[1];
2084 if (nullflag)
2085 vattr.va_vaflags |= VA_UTIMES_NULL;
dadab5e9
MD
2086 error = VOP_SETATTR(vp, &vattr, p->p_ucred, td);
2087 VOP_UNLOCK(vp, 0, td);
984263bc
MD
2088 return error;
2089}
2090
2091/*
41c20dac
MD
2092 * utimes_args(char *path, struct timeval *tptr)
2093 *
984263bc
MD
2094 * Set the access and modification times of a file.
2095 */
984263bc
MD
2096/* ARGSUSED */
2097int
41c20dac 2098utimes(struct utimes_args *uap)
984263bc 2099{
dadab5e9 2100 struct thread *td = curthread;
984263bc
MD
2101 struct timespec ts[2];
2102 struct timeval *usrtvp;
2103 int error;
2104 struct nameidata nd;
2105
2106 usrtvp = SCARG(uap, tptr);
2107 if ((error = getutimes(usrtvp, ts)) != 0)
2108 return (error);
dadab5e9 2109 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), td);
984263bc
MD
2110 if ((error = namei(&nd)) != 0)
2111 return (error);
2112 NDFREE(&nd, NDF_ONLY_PNBUF);
41c20dac 2113 error = setutimes(nd.ni_vp, ts, usrtvp == NULL);
984263bc
MD
2114 vrele(nd.ni_vp);
2115 return (error);
2116}
2117
2118/*
41c20dac
MD
2119 * lutimes_args(char *path, struct timeval *tptr)
2120 *
984263bc
MD
2121 * Set the access and modification times of a file.
2122 */
984263bc
MD
2123/* ARGSUSED */
2124int
41c20dac 2125lutimes(struct lutimes_args *uap)
984263bc 2126{
dadab5e9 2127 struct thread *td = curthread;
984263bc
MD
2128 struct timespec ts[2];
2129 struct timeval *usrtvp;
2130 int error;
2131 struct nameidata nd;
2132
2133 usrtvp = SCARG(uap, tptr);
2134 if ((error = getutimes(usrtvp, ts)) != 0)
2135 return (error);
dadab5e9 2136 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), td);
984263bc
MD
2137 if ((error = namei(&nd)) != 0)
2138 return (error);
2139 NDFREE(&nd, NDF_ONLY_PNBUF);
41c20dac 2140 error = setutimes(nd.ni_vp, ts, usrtvp == NULL);
984263bc
MD
2141 vrele(nd.ni_vp);
2142 return (error);
2143}
2144
2145/*
41c20dac
MD
2146 * futimes_args(int fd, struct timeval *tptr)
2147 *
984263bc
MD
2148 * Set the access and modification times of a file.
2149 */
984263bc
MD
2150/* ARGSUSED */
2151int
41c20dac 2152futimes(struct futimes_args *uap)
984263bc 2153{
dadab5e9
MD
2154 struct thread *td = curthread;
2155 struct proc *p = td->td_proc;
984263bc
MD
2156 struct timespec ts[2];
2157 struct file *fp;
2158 struct timeval *usrtvp;
2159 int error;
2160
2161 usrtvp = SCARG(uap, tptr);
2162 if ((error = getutimes(usrtvp, ts)) != 0)
2163 return (error);
2164 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2165 return (error);
41c20dac 2166 return setutimes((struct vnode *)fp->f_data, ts, usrtvp == NULL);
984263bc
MD
2167}
2168
2169/*
41c20dac
MD
2170 * truncate(char *path, int pad, off_t length)
2171 *
984263bc
MD
2172 * Truncate a file given its path name.
2173 */
984263bc
MD
2174/* ARGSUSED */
2175int
41c20dac 2176truncate(struct truncate_args *uap)
984263bc 2177{
dadab5e9
MD
2178 struct thread *td = curthread;
2179 struct proc *p = td->td_proc;
41c20dac 2180 struct vnode *vp;
984263bc
MD
2181 struct vattr vattr;
2182 int error;
2183 struct nameidata nd;
2184
2185 if (uap->length < 0)
2186 return(EINVAL);
dadab5e9 2187 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), td);
984263bc
MD
2188 if ((error = namei(&nd)) != 0)
2189 return (error);
2190 vp = nd.ni_vp;
2191 NDFREE(&nd, NDF_ONLY_PNBUF);
dadab5e9
MD
2192 VOP_LEASE(vp, td, p->p_ucred, LEASE_WRITE);
2193 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
984263bc
MD
2194 if (vp->v_type == VDIR)
2195 error = EISDIR;
2196 else if ((error = vn_writechk(vp)) == 0 &&
dadab5e9 2197 (error = VOP_ACCESS(vp, VWRITE, p->p_ucred, td)) == 0) {
984263bc
MD
2198 VATTR_NULL(&vattr);
2199 vattr.va_size = SCARG(uap, length);
dadab5e9 2200 error = VOP_SETATTR(vp, &vattr, p->p_ucred, td);
984263bc
MD
2201 }
2202 vput(vp);
2203 return (error);
2204}
2205
2206/*
41c20dac
MD
2207 * ftruncate_args(int fd, int pad, off_t length)
2208 *
984263bc
MD
2209 * Truncate a file given a file descriptor.
2210 */
984263bc
MD
2211/* ARGSUSED */
2212int
41c20dac 2213ftruncate(struct ftruncate_args *uap)
984263bc 2214{
dadab5e9
MD
2215 struct thread *td = curthread;
2216 struct proc *p = td->td_proc;
984263bc
MD
2217 struct vattr vattr;
2218 struct vnode *vp;
2219 struct file *fp;
2220 int error;
2221
2222 if (uap->length < 0)
2223 return(EINVAL);
2224 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2225 return (error);
2226 if ((fp->f_flag & FWRITE) == 0)
2227 return (EINVAL);
2228 vp = (struct vnode *)fp->f_data;
dadab5e9
MD
2229 VOP_LEASE(vp, td, p->p_ucred, LEASE_WRITE);
2230 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
984263bc
MD
2231 if (vp->v_type == VDIR)
2232 error = EISDIR;
2233 else if ((error = vn_writechk(vp)) == 0) {
2234 VATTR_NULL(&vattr);
2235 vattr.va_size = SCARG(uap, length);
dadab5e9 2236 error = VOP_SETATTR(vp, &vattr, fp->f_cred, td);
984263bc 2237 }
dadab5e9 2238 VOP_UNLOCK(vp, 0, td);
984263bc
MD
2239 return (error);
2240}
2241
2242#if defined(COMPAT_43) || defined(COMPAT_SUNOS)
2243/*
41c20dac
MD
2244 * otruncate_args(char *path, long length)
2245 *
984263bc
MD
2246 * Truncate a file given its path name.
2247 */
984263bc
MD
2248/* ARGSUSED */
2249int
41c20dac 2250otruncate(struct otruncate_args *uap)
984263bc
MD
2251{
2252 struct truncate_args /* {
2253 syscallarg(char *) path;
2254 syscallarg(int) pad;
2255 syscallarg(off_t) length;
2256 } */ nuap;
2257
2258 SCARG(&nuap, path) = SCARG(uap, path);
2259 SCARG(&nuap, length) = SCARG(uap, length);
41c20dac 2260 return (truncate(&nuap));
984263bc
MD
2261}
2262
2263/*
41c20dac
MD
2264 * oftruncate_args(int fd, long length)
2265 *
984263bc
MD
2266 * Truncate a file given a file descriptor.
2267 */
984263bc
MD
2268/* ARGSUSED */
2269int
41c20dac 2270oftruncate(struct oftruncate_args *uap)
984263bc
MD
2271{
2272 struct ftruncate_args /* {
2273 syscallarg(int) fd;
2274 syscallarg(int) pad;
2275 syscallarg(off_t) length;
2276 } */ nuap;
2277
2278 SCARG(&nuap, fd) = SCARG(uap, fd);
2279 SCARG(&nuap, length) = SCARG(uap, length);
41c20dac 2280 return (ftruncate(&nuap));
984263bc
MD
2281}
2282#endif /* COMPAT_43 || COMPAT_SUNOS */
2283
2284/*
41c20dac
MD
2285 * fsync(int fd)
2286 *
984263bc
MD
2287 * Sync an open file.
2288 */
984263bc
MD
2289/* ARGSUSED */
2290int
41c20dac 2291fsync(struct fsync_args *uap)
984263bc 2292{
dadab5e9
MD
2293 struct thread *td = curthread;
2294 struct proc *p = td->td_proc;
41c20dac 2295 struct vnode *vp;
984263bc
MD
2296 struct file *fp;
2297 vm_object_t obj;
2298 int error;
2299
2300 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2301 return (error);
2302 vp = (struct vnode *)fp->f_data;
dadab5e9 2303 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
984263bc
MD
2304 if (VOP_GETVOBJECT(vp, &obj) == 0)
2305 vm_object_page_clean(obj, 0, 0, 0);
dadab5e9 2306 if ((error = VOP_FSYNC(vp, fp->f_cred, MNT_WAIT, td)) == 0 &&
984263bc
MD
2307 vp->v_mount && (vp->v_mount->mnt_flag & MNT_SOFTDEP) &&
2308 bioops.io_fsync)
2309 error = (*bioops.io_fsync)(vp);
dadab5e9 2310 VOP_UNLOCK(vp, 0, td);
984263bc
MD
2311 return (error);
2312}
2313
2314/*
41c20dac
MD
2315 * rename_args(char *from, char *to)
2316 *
984263bc
MD
2317 * Rename files. Source and destination must either both be directories,
2318 * or both not be directories. If target is a directory, it must be empty.
2319 */
984263bc
MD
2320/* ARGSUSED */
2321int
41c20dac 2322rename(struct rename_args *uap)
984263bc 2323{
dadab5e9
MD
2324 struct thread *td = curthread;
2325 struct proc *p = td->td_proc;
41c20dac 2326 struct vnode *tvp, *fvp, *tdvp;
984263bc
MD
2327 struct nameidata fromnd, tond;
2328 int error;
2329
2330 bwillwrite();
2331 NDINIT(&fromnd, DELETE, WANTPARENT | SAVESTART, UIO_USERSPACE,
dadab5e9 2332 SCARG(uap, from), td);
984263bc
MD
2333 if ((error = namei(&fromnd)) != 0)
2334 return (error);
2335 fvp = fromnd.ni_vp;
2336 NDINIT(&tond, RENAME, LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART | NOOBJ,
dadab5e9 2337 UIO_USERSPACE, SCARG(uap, to), td);
984263bc
MD
2338 if (fromnd.ni_vp->v_type == VDIR)
2339 tond.ni_cnd.cn_flags |= WILLBEDIR;
2340 if ((error = namei(&tond)) != 0) {
2341 /* Translate error code for rename("dir1", "dir2/."). */
2342 if (error == EISDIR && fvp->v_type == VDIR)
2343 error = EINVAL;
2344 NDFREE(&fromnd, NDF_ONLY_PNBUF);
2345 vrele(fromnd.ni_dvp);
2346 vrele(fvp);
2347 goto out1;
2348 }
2349 tdvp = tond.ni_dvp;
2350 tvp = tond.ni_vp;
2351 if (tvp != NULL) {
2352 if (fvp->v_type == VDIR && tvp->v_type != VDIR) {
2353 error = ENOTDIR;
2354 goto out;
2355 } else if (fvp->v_type != VDIR && tvp->v_type == VDIR) {
2356 error = EISDIR;
2357 goto out;
2358 }
2359 }
2360 if (fvp == tdvp)
2361 error = EINVAL;
2362 /*
2363 * If the source is the same as the destination (that is, if they
2364 * are links to the same vnode), then there is nothing to do.
2365 */
2366 if (fvp == tvp)
2367 error = -1;
2368out:
2369 if (!error) {
dadab5e9 2370 VOP_LEASE(tdvp, td, p->p_ucred, LEASE_WRITE);
984263bc 2371 if (fromnd.ni_dvp != tdvp) {
dadab5e9 2372 VOP_LEASE(fromnd.ni_dvp, td, p->p_ucred, LEASE_WRITE);
984263bc
MD
2373 }
2374 if (tvp) {
dadab5e9 2375 VOP_LEASE(tvp, td, p->p_ucred, LEASE_WRITE);
984263bc
MD
2376 }
2377 error = VOP_RENAME(fromnd.ni_dvp, fromnd.ni_vp, &fromnd.ni_cnd,
2378 tond.ni_dvp, tond.ni_vp, &tond.ni_cnd);
2379 NDFREE(&fromnd, NDF_ONLY_PNBUF);
2380 NDFREE(&tond, NDF_ONLY_PNBUF);
2381 } else {
2382 NDFREE(&fromnd, NDF_ONLY_PNBUF);
2383 NDFREE(&tond, NDF_ONLY_PNBUF);
2384 if (tdvp == tvp)
2385 vrele(tdvp);
2386 else
2387 vput(tdvp);
2388 if (tvp)
2389 vput(tvp);
2390 vrele(fromnd.ni_dvp);
2391 vrele(fvp);
2392 }
2393 vrele(tond.ni_startdir);
2394 ASSERT_VOP_UNLOCKED(fromnd.ni_dvp, "rename");
2395 ASSERT_VOP_UNLOCKED(fromnd.ni_vp, "rename");
2396 ASSERT_VOP_UNLOCKED(tond.ni_dvp, "rename");
2397 ASSERT_VOP_UNLOCKED(tond.ni_vp, "rename");
2398out1:
2399 if (fromnd.ni_startdir)
2400 vrele(fromnd.ni_startdir);
2401 if (error == -1)
2402 return (0);
2403 return (error);
2404}
2405
2406/*
41c20dac
MD
2407 * mkdir_args(char *path, int mode)
2408 *
984263bc
MD
2409 * Make a directory file.
2410 */
984263bc
MD
2411/* ARGSUSED */
2412int
41c20dac 2413mkdir(struct mkdir_args *uap)
984263bc 2414{
dadab5e9
MD
2415 struct thread *td = curthread;
2416 struct proc *p = td->td_proc;
41c20dac 2417 struct vnode *vp;
984263bc
MD
2418 struct vattr vattr;
2419 int error;
2420 struct nameidata nd;
2421
2422 bwillwrite();
dadab5e9 2423 NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), td);
984263bc
MD
2424 nd.ni_cnd.cn_flags |= WILLBEDIR;
2425 if ((error = namei(&nd)) != 0)
2426 return (error);
2427 vp = nd.ni_vp;
2428 if (vp != NULL) {
2429 NDFREE(&nd, NDF_ONLY_PNBUF);
2430 if (nd.ni_dvp == vp)
2431 vrele(nd.ni_dvp);
2432 else
2433 vput(nd.ni_dvp);
2434 vrele(vp);
2435 return (EEXIST);
2436 }
2437 VATTR_NULL(&vattr);
2438 vattr.va_type = VDIR;
2439 vattr.va_mode = (SCARG(uap, mode) & ACCESSPERMS) &~ p->p_fd->fd_cmask;
dadab5e9 2440 VOP_LEASE(nd.ni_dvp, td, p->p_ucred, LEASE_WRITE);
984263bc
MD
2441 error = VOP_MKDIR(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
2442 NDFREE(&nd, NDF_ONLY_PNBUF);
2443 vput(nd.ni_dvp);
2444 if (!error)
2445 vput(nd.ni_vp);
2446 ASSERT_VOP_UNLOCKED(nd.ni_dvp, "mkdir");
2447 ASSERT_VOP_UNLOCKED(nd.ni_vp, "mkdir");
2448 return (error);
2449}
2450
2451/*
41c20dac
MD
2452 * rmdir_args(char *path)
2453 *
984263bc
MD
2454 * Remove a directory file.
2455 */
984263bc
MD
2456/* ARGSUSED */
2457int
41c20dac 2458rmdir(struct rmdir_args *uap)
984263bc 2459{
dadab5e9
MD
2460 struct thread *td = curthread;
2461 struct proc *p = td->td_proc;
41c20dac 2462 struct vnode *vp;
984263bc
MD
2463 int error;
2464 struct nameidata nd;
2465
2466 bwillwrite();
2467 NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
dadab5e9 2468 SCARG(uap, path), td);
984263bc
MD
2469 if ((error = namei(&nd)) != 0)
2470 return (error);
2471 vp = nd.ni_vp;
2472 if (vp->v_type != VDIR) {
2473 error = ENOTDIR;
2474 goto out;
2475 }
2476 /*
2477 * No rmdir "." please.
2478 */
2479 if (nd.ni_dvp == vp) {
2480 error = EINVAL;
2481 goto out;
2482 }
2483 /*
2484 * The root of a mounted filesystem cannot be deleted.
2485 */
2486 if (vp->v_flag & VROOT)
2487 error = EBUSY;
2488 else {
dadab5e9
MD
2489 VOP_LEASE(nd.ni_dvp, td, p->p_ucred, LEASE_WRITE);
2490 VOP_LEASE(vp, td, p->p_ucred, LEASE_WRITE);
984263bc
MD
2491 error = VOP_RMDIR(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
2492 }
2493out:
2494 NDFREE(&nd, NDF_ONLY_PNBUF);
2495 if (nd.ni_dvp == vp)
2496 vrele(nd.ni_dvp);
2497 else
2498 vput(nd.ni_dvp);
2499 if (vp != NULLVP)
2500 vput(vp);
2501 ASSERT_VOP_UNLOCKED(nd.ni_dvp, "rmdir");
2502 ASSERT_VOP_UNLOCKED(nd.ni_vp, "rmdir");
2503 return (error);
2504}
2505
2506#ifdef COMPAT_43
2507/*
41c20dac
MD
2508 * ogetdirentries_args(int fd, char *buf, u_int count, long *basep)
2509 *
984263bc
MD
2510 * Read a block of directory entries in a file system independent format.
2511 */
984263bc 2512int
41c20dac 2513ogetdirentries(struct ogetdirentries_args *uap)
984263bc 2514{
dadab5e9
MD
2515 struct thread *td = curthread;
2516 struct proc *p = td->td_proc;
984263bc
MD
2517 struct vnode *vp;
2518 struct file *fp;
2519 struct uio auio, kuio;
2520 struct iovec aiov, kiov;
2521 struct dirent *dp, *edp;
2522 caddr_t dirbuf;
2523 int error, eofflag, readcnt;
2524 long loff;
2525
2526 /* XXX arbitrary sanity limit on `count'. */
2527 if (SCARG(uap, count) > 64 * 1024)
2528 return (EINVAL);
2529 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2530 return (error);
2531 if ((fp->f_flag & FREAD) == 0)
2532 return (EBADF);
2533 vp = (struct vnode *)fp->f_data;
2534unionread:
2535 if (vp->v_type != VDIR)
2536 return (EINVAL);
2537 aiov.iov_base = SCARG(uap, buf);
2538 aiov.iov_len = SCARG(uap, count);
2539 auio.uio_iov = &aiov;
2540 auio.uio_iovcnt = 1;
2541 auio.uio_rw = UIO_READ;
2542 auio.uio_segflg = UIO_USERSPACE;
dadab5e9 2543 auio.uio_td = td;
984263bc 2544 auio.uio_resid = SCARG(uap, count);
dadab5e9 2545 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
984263bc
MD
2546 loff = auio.uio_offset = fp->f_offset;
2547# if (BYTE_ORDER != LITTLE_ENDIAN)
2548 if (vp->v_mount->mnt_maxsymlinklen <= 0) {
2549 error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag,
2550 NULL, NULL);
2551 fp->f_offset = auio.uio_offset;
2552 } else
2553# endif
2554 {
2555 kuio = auio;
2556 kuio.uio_iov = &kiov;
2557 kuio.uio_segflg = UIO_SYSSPACE;
2558 kiov.iov_len = SCARG(uap, count);
2559 MALLOC(dirbuf, caddr_t, SCARG(uap, count), M_TEMP, M_WAITOK);
2560 kiov.iov_base = dirbuf;
2561 error = VOP_READDIR(vp, &kuio, fp->f_cred, &eofflag,
2562 NULL, NULL);
2563 fp->f_offset = kuio.uio_offset;
2564 if (error == 0) {
2565 readcnt = SCARG(uap, count) - kuio.uio_resid;
2566 edp = (struct dirent *)&dirbuf[readcnt];
2567 for (dp = (struct dirent *)dirbuf; dp < edp; ) {
2568# if (BYTE_ORDER == LITTLE_ENDIAN)
2569 /*
2570 * The expected low byte of
2571 * dp->d_namlen is our dp->d_type.
2572 * The high MBZ byte of dp->d_namlen
2573 * is our dp->d_namlen.
2574 */
2575 dp->d_type = dp->d_namlen;
2576 dp->d_namlen = 0;
2577# else
2578 /*
2579 * The dp->d_type is the high byte
2580 * of the expected dp->d_namlen,
2581 * so must be zero'ed.
2582 */
2583 dp->d_type = 0;
2584# endif
2585 if (dp->d_reclen > 0) {
2586 dp = (struct dirent *)
2587 ((char *)dp + dp->d_reclen);
2588 } else {
2589 error = EIO;
2590 break;
2591 }
2592 }
2593 if (dp >= edp)
2594 error = uiomove(dirbuf, readcnt, &auio);
2595 }
2596 FREE(dirbuf, M_TEMP);
2597 }
dadab5e9 2598 VOP_UNLOCK(vp, 0, td);
984263bc
MD
2599 if (error)
2600 return (error);
2601 if (SCARG(uap, count) == auio.uio_resid) {
2602 if (union_dircheckp) {
dadab5e9 2603 error = union_dircheckp(td, &vp, fp);
984263bc
MD
2604 if (error == -1)
2605 goto unionread;
2606 if (error)
2607 return (error);
2608 }
2609 if ((vp->v_flag & VROOT) &&
2610 (vp->v_mount->mnt_flag & MNT_UNION)) {
2611 struct vnode *tvp = vp;
2612 vp = vp->v_mount->mnt_vnodecovered;
2613 VREF(vp);
2614 fp->f_data = (caddr_t) vp;
2615 fp->f_offset = 0;
2616 vrele(tvp);
2617 goto unionread;
2618 }
2619 }
2620 error = copyout((caddr_t)&loff, (caddr_t)SCARG(uap, basep),
2621 sizeof(long));
2622 p->p_retval[0] = SCARG(uap, count) - auio.uio_resid;
2623 return (error);
2624}
2625#endif /* COMPAT_43 */
2626
2627/*
41c20dac
MD
2628 * getdirentries_args(int fd, char *buf, u_int conut, long *basep)
2629 *
984263bc
MD
2630 * Read a block of directory entries in a file system independent format.
2631 */
984263bc 2632int
41c20dac 2633getdirentries(struct getdirentries_args *uap)
984263bc 2634{
dadab5e9
MD
2635 struct thread *td = curthread;
2636 struct proc *p = td->td_proc;
984263bc
MD
2637 struct vnode *vp;
2638 struct file *fp;
2639 struct uio auio;
2640 struct iovec aiov;
2641 long loff;
2642 int error, eofflag;
2643
2644 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
2645 return (error);
2646 if ((fp->f_flag & FREAD) == 0)
2647 return (EBADF);
2648 vp = (struct vnode *)fp->f_data;
2649unionread:
2650 if (vp->v_type != VDIR)
2651 return (EINVAL);
2652 aiov.iov_base = SCARG(uap, buf);
2653 aiov.iov_len = SCARG(uap, count);
2654 auio.uio_iov = &aiov;
2655 auio.uio_iovcnt = 1;
2656 auio.uio_rw = UIO_READ;
2657 auio.uio_segflg = UIO_USERSPACE;
dadab5e9 2658 auio.uio_td = td;
984263bc 2659 auio.uio_resid = SCARG(uap, count);
dadab5e9
MD
2660 /* vn_lock(vp, LK_SHARED | LK_RETRY, td); */
2661 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
984263bc
MD
2662 loff = auio.uio_offset = fp->f_offset;
2663 error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, NULL, NULL);
2664 fp->f_offset = auio.uio_offset;
dadab5e9 2665 VOP_UNLOCK(vp, 0, td);
984263bc
MD
2666 if (error)
2667 return (error);
2668 if (SCARG(uap, count) == auio.uio_resid) {
2669 if (union_dircheckp) {
dadab5e9 2670 error = union_dircheckp(td, &vp, fp);
984263bc
MD
2671 if (error == -1)
2672 goto unionread;
2673 if (error)
2674 return (error);
2675 }
2676 if ((vp->v_flag & VROOT) &&
2677 (vp->v_mount->mnt_flag & MNT_UNION)) {
2678 struct vnode *tvp = vp;
2679 vp = vp->v_mount->mnt_vnodecovered;
2680 VREF(vp);
2681 fp->f_data = (caddr_t) vp;
2682 fp->f_offset = 0;
2683 vrele(tvp);
2684 goto unionread;
2685 }
2686 }
2687 if (SCARG(uap, basep) != NULL) {
2688 error = copyout((caddr_t)&loff, (caddr_t)SCARG(uap, basep),
2689 sizeof(long));
2690 }
2691 p->p_retval[0] = SCARG(uap, count) - auio.uio_resid;
2692 return (error);
2693}
41c20dac
MD
2694
2695/*
2696 * getdents_args(int fd, char *buf, size_t count)
2697 */
984263bc 2698int
41c20dac 2699getdents(struct getdents_args *uap)
984263bc
MD
2700{
2701 struct getdirentries_args ap;
41c20dac 2702
984263bc
MD
2703 ap.fd = uap->fd;
2704 ap.buf = uap->buf;
2705 ap.count = uap->count;
2706 ap.basep = NULL;
41c20dac 2707 return getdirentries(&ap);
984263bc
MD
2708}
2709
2710/*
41c20dac
MD
2711 * umask(int newmask)
2712 *
984263bc
MD
2713 * Set the mode mask for creation of filesystem nodes.
2714 *
2715 * MP SAFE
2716 */
984263bc 2717int
41c20dac 2718umask(struct umask_args *uap)
984263bc 2719{
dadab5e9
MD
2720 struct thread *td = curthread;
2721 struct proc *p = td->td_proc;
41c20dac 2722 struct filedesc *fdp;
984263bc
MD
2723
2724 fdp = p->p_fd;
2725 p->p_retval[0] = fdp->fd_cmask;
2726 fdp->fd_cmask = SCARG(uap, newmask) & ALLPERMS;
2727 return (0);
2728}
2729
2730/*
41c20dac
MD
2731 * revoke(char *path)
2732 *
984263bc
MD
2733 * Void all references to file by ripping underlying filesystem
2734 * away from vnode.
2735 */
984263bc
MD
2736/* ARGSUSED */
2737int
41c20dac 2738revoke(struct revoke_args *uap)
984263bc 2739{
dadab5e9
MD
2740 struct thread *td = curthread;
2741 struct proc *p = td->td_proc;
41c20dac 2742 struct vnode *vp;
984263bc
MD
2743 struct vattr vattr;
2744 int error;
2745 struct nameidata nd;
2746
dadab5e9 2747 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), td);
984263bc
MD
2748 if ((error = namei(&nd)) != 0)
2749 return (error);
2750 vp = nd.ni_vp;
2751 NDFREE(&nd, NDF_ONLY_PNBUF);
2752 if (vp->v_type != VCHR && vp->v_type != VBLK) {
2753 error = EINVAL;
2754 goto out;
2755 }
dadab5e9 2756 if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, td)) != 0)
984263bc
MD
2757 goto out;
2758 if (p->p_ucred->cr_uid != vattr.va_uid &&
dadab5e9 2759 (error = suser_cred(p->p_ucred, PRISON_ROOT)))
984263bc
MD
2760 goto out;
2761 if (vcount(vp) > 1)
2762 VOP_REVOKE(vp, REVOKEALL);
2763out:
2764 vrele(vp);
2765 return (error);
2766}
2767
2768/*
2769 * Convert a user file descriptor to a kernel file entry.
2770 */
2771int
41c20dac 2772getvnode(struct filedesc *fdp, int fd, struct file **fpp)
984263bc
MD
2773{
2774 struct file *fp;
2775
2776 if ((u_int)fd >= fdp->fd_nfiles ||
2777 (fp = fdp->fd_ofiles[fd]) == NULL)
2778 return (EBADF);
2779 if (fp->f_type != DTYPE_VNODE && fp->f_type != DTYPE_FIFO)
2780 return (EINVAL);
2781 *fpp = fp;
2782 return (0);
2783}
2784/*
41c20dac
MD
2785 * getfh_args(char *fname, fhandle_t *fhp)
2786 *
984263bc
MD
2787 * Get (NFS) file handle
2788 */
984263bc 2789int
41c20dac 2790getfh(struct getfh_args *uap)
984263bc 2791{
dadab5e9 2792 struct thread *td = curthread;
984263bc
MD
2793 struct nameidata nd;
2794 fhandle_t fh;
41c20dac 2795 struct vnode *vp;
984263bc
MD
2796 int error;
2797
2798 /*
2799 * Must be super user
2800 */
dadab5e9 2801 error = suser(td);
984263bc
MD
2802 if (error)
2803 return (error);
dadab5e9 2804 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE, uap->fname, td);
984263bc
MD
2805 error = namei(&nd);
2806 if (error)
2807 return (error);
2808 NDFREE(&nd, NDF_ONLY_PNBUF);
2809 vp = nd.ni_vp;
2810 bzero(&fh, sizeof(fh));
2811 fh.fh_fsid = vp->v_mount->mnt_stat.f_fsid;
2812 error = VFS_VPTOFH(vp, &fh.fh_fid);
2813 vput(vp);
2814 if (error)
2815 return (error);
2816 error = copyout(&fh, uap->fhp, sizeof (fh));
2817 return (error);
2818}
2819
2820/*
41c20dac
MD
2821 * fhopen_args(const struct fhandle *u_fhp, int flags)
2822 *
984263bc
MD
2823 * syscall for the rpc.lockd to use to translate a NFS file handle into
2824 * an open descriptor.
2825 *
2826 * warning: do not remove the suser() call or this becomes one giant
2827 * security hole.
2828 */
984263bc 2829int
41c20dac 2830fhopen(struct fhopen_args *uap)
984263bc 2831{
dadab5e9
MD
2832 struct thread *td = curthread;
2833 struct proc *p = td->td_proc;
984263bc
MD
2834 struct mount *mp;
2835 struct vnode *vp;
2836 struct fhandle fhp;
2837 struct vattr vat;
2838 struct vattr *vap = &vat;
2839 struct flock lf;
2840 struct file *fp;
41c20dac 2841 struct filedesc *fdp = p->p_fd;
984263bc
MD
2842 int fmode, mode, error, type;
2843 struct file *nfp;
2844 int indx;
2845
2846 /*
2847 * Must be super user
2848 */
dadab5e9 2849 error = suser(td);
984263bc
MD
2850 if (error)
2851 return (error);
2852
2853 fmode = FFLAGS(SCARG(uap, flags));
2854 /* why not allow a non-read/write open for our lockd? */
2855 if (((fmode & (FREAD | FWRITE)) == 0) || (fmode & O_CREAT))
2856 return (EINVAL);
2857 error = copyin(SCARG(uap,u_fhp), &fhp, sizeof(fhp));
2858 if (error)
2859 return(error);
2860 /* find the mount point */
2861 mp = vfs_getvfs(&fhp.fh_fsid);
2862 if (mp == NULL)
2863 return (ESTALE);
2864 /* now give me my vnode, it gets returned to me locked */
2865 error = VFS_FHTOVP(mp, &fhp.fh_fid, &vp);
2866 if (error)
2867 return (error);
2868 /*
2869 * from now on we have to make sure not
2870 * to forget about the vnode
2871 * any error that causes an abort must vput(vp)
2872 * just set error = err and 'goto bad;'.
2873 */
2874
2875 /*
2876 * from vn_open
2877 */
2878 if (vp->v_type == VLNK) {
2879 error = EMLINK;
2880 goto bad;
2881 }
2882 if (vp->v_type == VSOCK) {
2883 error = EOPNOTSUPP;
2884 goto bad;
2885 }
2886 mode = 0;
2887 if (fmode & (FWRITE | O_TRUNC)) {
2888 if (vp->v_type == VDIR) {
2889 error = EISDIR;
2890 goto bad;
2891 }
2892 error = vn_writechk(vp);
2893 if (error)
2894 goto bad;
2895 mode |= VWRITE;
2896 }
2897 if (fmode & FREAD)
2898 mode |= VREAD;
2899 if (mode) {
dadab5e9 2900 error = VOP_ACCESS(vp, mode, p->p_ucred, td);
984263bc
MD
2901 if (error)
2902 goto bad;
2903 }
2904 if (fmode & O_TRUNC) {
dadab5e9
MD
2905 VOP_UNLOCK(vp, 0, td); /* XXX */
2906 VOP_LEASE(vp, td, p->p_ucred, LEASE_WRITE);
2907 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td); /* XXX */
984263bc
MD
2908 VATTR_NULL(vap);
2909 vap->va_size = 0;
dadab5e9 2910 error = VOP_SETATTR(vp, vap, p->p_ucred, td);
984263bc
MD
2911 if (error)
2912 goto bad;
2913 }
dadab5e9 2914 error = VOP_OPEN(vp, fmode, p->p_ucred, td);
984263bc
MD
2915 if (error)
2916 goto bad;
2917 /*
2918 * Make sure that a VM object is created for VMIO support.
2919 */
2920 if (vn_canvmio(vp) == TRUE) {
dadab5e9 2921 if ((error = vfs_object_create(vp, td, p->p_ucred)) != 0)
984263bc
MD
2922 goto bad;
2923 }
2924 if (fmode & FWRITE)
2925 vp->v_writecount++;
2926
2927 /*
2928 * end of vn_open code
2929 */
2930
2931 if ((error = falloc(p, &nfp, &indx)) != 0) {
2932 if (fmode & FWRITE)
2933 vp->v_writecount--;
2934 goto bad;
2935 }
2936 fp = nfp;
2937
2938 /*
2939 * hold an extra reference to avoid having fp ripped out
2940 * from under us while we block in the lock op.
2941 */
2942 fhold(fp);
2943 nfp->f_data = (caddr_t)vp;
2944 nfp->f_flag = fmode & FMASK;
2945 nfp->f_ops = &vnops;
2946 nfp->f_type = DTYPE_VNODE;
2947 if (fmode & (O_EXLOCK | O_SHLOCK)) {
2948 lf.l_whence = SEEK_SET;
2949 lf.l_start = 0;
2950 lf.l_len = 0;
2951 if (fmode & O_EXLOCK)
2952 lf.l_type = F_WRLCK;
2953 else
2954 lf.l_type = F_RDLCK;
2955 type = F_FLOCK;
2956 if ((fmode & FNONBLOCK) == 0)
2957 type |= F_WAIT;
dadab5e9 2958 VOP_UNLOCK(vp, 0, td);
984263bc
MD
2959 if ((error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type)) != 0) {
2960 /*
2961 * lock request failed. Normally close the descriptor
2962 * but handle the case where someone might have dup()d
2963 * or close()d it when we weren't looking.
2964 */
2965 if (fdp->fd_ofiles[indx] == fp) {
2966 fdp->fd_ofiles[indx] = NULL;
dadab5e9 2967 fdrop(fp, td);
984263bc
MD
2968 }
2969
2970 /*
2971 * release our private reference.
2972 */
dadab5e9 2973 fdrop(fp, td);
984263bc
MD
2974 return (error);
2975 }
dadab5e9 2976 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
984263bc
MD
2977 fp->f_flag |= FHASLOCK;
2978 }
2979 if ((vp->v_type == VREG) && (VOP_GETVOBJECT(vp, NULL) != 0))
dadab5e9 2980 vfs_object_create(vp, td, p->p_ucred);
984263bc 2981
dadab5e9
MD
2982 VOP_UNLOCK(vp, 0, td);
2983 fdrop(fp, td);
984263bc
MD
2984 p->p_retval[0] = indx;
2985 return (0);
2986
2987bad:
2988 vput(vp);
2989 return (error);
2990}
2991
41c20dac
MD
2992/*
2993 * fhstat_args(struct fhandle *u_fhp, struct stat *sb)
2994 */
984263bc 2995int
41c20dac 2996fhstat(struct fhstat_args *uap)
984263bc 2997{
dadab5e9 2998 struct thread *td = curthread;
984263bc
MD
2999 struct stat sb;
3000 fhandle_t fh;
3001 struct mount *mp;
3002 struct vnode *vp;
3003 int error;
3004
3005 /*
3006 * Must be super user
3007 */
dadab5e9 3008 error = suser(td);
984263bc
MD
3009 if (error)
3010 return (error);
3011
3012 error = copyin(SCARG(uap, u_fhp), &fh, sizeof(fhandle_t));
3013 if (error)
3014 return (error);
3015
3016 if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL)
3017 return (ESTALE);
3018 if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)))
3019 return (error);
dadab5e9 3020 error = vn_stat(vp, &sb, td);
984263bc
MD
3021 vput(vp);
3022 if (error)
3023 return (error);
3024 error = copyout(&sb, SCARG(uap, sb), sizeof(sb));
3025 return (error);
3026}
3027
41c20dac
MD
3028/*
3029 * fhstatfs_args(struct fhandle *u_fhp, struct statfs *buf)
3030 */
984263bc 3031int
41c20dac 3032fhstatfs(struct fhstatfs_args *uap)
984263bc 3033{
dadab5e9 3034 struct thread *td = curthread;
984263bc
MD
3035 struct statfs *sp;
3036 struct mount *mp;
3037 struct vnode *vp;
3038 struct statfs sb;
3039 fhandle_t fh;
3040 int error;
3041
3042 /*
3043 * Must be super user
3044 */
dadab5e9 3045 if ((error = suser(td)))
984263bc
MD
3046 return (error);
3047
3048 if ((error = copyin(SCARG(uap, u_fhp), &fh, sizeof(fhandle_t))) != 0)
3049 return (error);
3050
3051 if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL)
3052 return (ESTALE);
3053 if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)))
3054 return (error);
3055 mp = vp->v_mount;
3056 sp = &mp->mnt_stat;
3057 vput(vp);
dadab5e9 3058 if ((error = VFS_STATFS(mp, sp, td)) != 0)
984263bc
MD
3059 return (error);
3060 sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
dadab5e9 3061 if (suser(td)) {
984263bc
MD
3062 bcopy((caddr_t)sp, (caddr_t)&sb, sizeof(sb));
3063 sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0;
3064 sp = &sb;
3065 }
3066 return (copyout(sp, SCARG(uap, buf), sizeof(*sp)));
3067}
3068
3069/*
3070 * Syscall to push extended attribute configuration information into the
3071 * VFS. Accepts a path, which it converts to a mountpoint, as well as
3072 * a command (int cmd), and attribute name and misc data. For now, the
3073 * attribute name is left in userspace for consumption by the VFS_op.
3074 * It will probably be changed to be copied into sysspace by the
3075 * syscall in the future, once issues with various consumers of the
3076 * attribute code have raised their hands.
3077 *
3078 * Currently this is used only by UFS Extended Attributes.
3079 */
3080int
41c20dac 3081extattrctl(struct extattrctl_args *uap)
984263bc 3082{
dadab5e9 3083 struct thread *td = curthread;
984263bc
MD
3084 struct nameidata nd;
3085 struct mount *mp;
3086 int error;
3087
dadab5e9 3088 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), td);
984263bc
MD
3089 if ((error = namei(&nd)) != 0)
3090 return (error);
3091 mp = nd.ni_vp->v_mount;
3092 NDFREE(&nd, 0);
3093 return (VFS_EXTATTRCTL(mp, SCARG(uap, cmd), SCARG(uap, attrname),
dadab5e9 3094 SCARG(uap, arg), td));
984263bc
MD
3095}
3096
3097/*
3098 * Syscall to set a named extended attribute on a file or directory.
3099 * Accepts attribute name, and a uio structure pointing to the data to set.
3100 * The uio is consumed in the style of writev(). The real work happens
3101 * in VOP_SETEXTATTR().
3102 */
3103int
41c20dac 3104extattr_set_file(struct extattr_set_file_args *uap)
984263bc 3105{
dadab5e9
MD
3106 struct thread *td = curthread;
3107 struct proc *p = td->td_proc;
984263bc
MD
3108 struct nameidata nd;
3109 struct uio auio;
3110 struct iovec *iov, *needfree = NULL, aiov[UIO_SMALLIOV];
3111 char attrname[EXTATTR_MAXNAMELEN];
3112 u_int iovlen, cnt;
3113 int error, i;
3114
3115 error = copyin(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN);
3116 if (error)
3117 return (error);
3118 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
dadab5e9 3119 SCARG(uap, path), td);
984263bc
MD
3120 if ((error = namei(&nd)) != 0)
3121 return(error);
3122 iovlen = uap->iovcnt * sizeof(struct iovec);
3123 if (uap->iovcnt > UIO_SMALLIOV) {
3124 if (uap->iovcnt > UIO_MAXIOV) {
3125 error = EINVAL;
3126 goto done;
3127 }
3128 MALLOC(iov, struct iovec *, iovlen, M_IOV, M_WAITOK);
3129 needfree = iov;
3130 } else
3131 iov = aiov;
3132 auio.uio_iov = iov;
3133 auio.uio_iovcnt = uap->iovcnt;
3134 auio.uio_rw = UIO_WRITE;
3135 auio.uio_segflg = UIO_USERSPACE;
dadab5e9 3136 auio.uio_td = td;
984263bc
MD
3137 auio.uio_offset = 0;
3138 if ((error = copyin((caddr_t)uap->iovp, (caddr_t)iov, iovlen)))
3139 goto done;
3140 auio.uio_resid = 0;
3141 for (i = 0; i < uap->iovcnt; i++) {
3142 if (iov->iov_len > INT_MAX - auio.uio_resid) {
3143 error = EINVAL;
3144 goto done;
3145 }
3146 auio.uio_resid += iov->iov_len;
3147 iov++;
3148 }
3149 cnt = auio.uio_resid;
dadab5e9 3150 error = VOP_SETEXTATTR(nd.ni_vp, attrname, &auio, p->p_ucred, td);
984263bc
MD
3151 cnt -= auio.uio_resid;
3152 p->p_retval[0] = cnt;
3153done:
3154 if (needfree)
3155 FREE(needfree, M_IOV);
3156 NDFREE(&nd, 0);
3157 return (error);
3158}
3159
3160/*
3161 * Syscall to get a named extended attribute on a file or directory.
3162 * Accepts attribute name, and a uio structure pointing to a buffer for the
3163 * data. The uio is consumed in the style of readv(). The real work
3164 * happens in VOP_GETEXTATTR();
3165 */
3166int
41c20dac 3167extattr_get_file(struct extattr_get_file_args *uap)
984263bc 3168{
dadab5e9
MD
3169 struct thread *td = curthread;
3170 struct proc *p = td->td_proc;
984263bc
MD
3171 struct nameidata nd;
3172 struct uio auio;
3173 struct iovec *iov, *needfree, aiov[UIO_SMALLIOV];
3174 char attrname[EXTATTR_MAXNAMELEN];
3175 u_int iovlen, cnt;
3176 int error, i;
3177
3178 error = copyin(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN);
3179 if (error)
3180 return (error);
3181 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
dadab5e9 3182 SCARG(uap, path), td);
984263bc
MD
3183 if ((error = namei(&nd)) != 0)
3184 return (error);
3185 iovlen = uap->iovcnt * sizeof (struct iovec);
3186 if (uap->iovcnt > UIO_SMALLIOV) {
3187 if (uap->iovcnt > UIO_MAXIOV) {
3188 NDFREE(&nd, 0);
3189 return (EINVAL);
3190 }
3191 MALLOC(iov, struct iovec *, iovlen, M_IOV, M_WAITOK);
3192 needfree = iov;
3193 } else {
3194 iov = aiov;
3195 needfree = NULL;
3196 }
3197 auio.uio_iov = iov;
3198 auio.uio_iovcnt = uap->iovcnt;
3199 auio.uio_rw = UIO_READ;
3200 auio.uio_segflg = UIO_USERSPACE;
dadab5e9 3201 auio.uio_td = td;
984263bc
MD
3202 auio.uio_offset = 0;
3203 if ((error = copyin((caddr_t)uap->iovp, (caddr_t)iov, iovlen)))
3204 goto done;
3205 auio.uio_resid = 0;
3206 for (i = 0; i < uap->iovcnt; i++) {
3207 if (iov->iov_len > INT_MAX - auio.uio_resid) {
3208 error = EINVAL;
3209 goto done;
3210 }
3211 auio.uio_resid += iov->iov_len;
3212 iov++;
3213 }
3214 cnt = auio.uio_resid;
dadab5e9 3215 error = VOP_GETEXTATTR(nd.ni_vp, attrname, &auio, p->p_ucred, td);
984263bc
MD
3216 cnt -= auio.uio_resid;
3217 p->p_retval[0] = cnt;
3218done:
3219 if (needfree)
3220 FREE(needfree, M_IOV);
3221 NDFREE(&nd, 0);
3222 return(error);
3223}
3224
3225/*
3226 * Syscall to delete a named extended attribute from a file or directory.
3227 * Accepts attribute name. The real work happens in VOP_SETEXTATTR().
3228 */
3229int
41c20dac 3230extattr_delete_file(struct extattr_delete_file_args *uap)
984263bc 3231{
dadab5e9
MD
3232 struct thread *td = curthread;
3233 struct proc *p = td->td_proc;
984263bc
MD
3234 struct nameidata nd;
3235 char attrname[EXTATTR_MAXNAMELEN];
3236 int error;
3237
3238 error = copyin(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN);
3239 if (error)
3240 return(error);
3241 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
dadab5e9 3242 SCARG(uap, path), td);
984263bc
MD
3243 if ((error = namei(&nd)) != 0)
3244 return(error);
dadab5e9 3245 error = VOP_SETEXTATTR(nd.ni_vp, attrname, NULL, p->p_ucred, td);
984263bc
MD
3246 NDFREE(&nd, 0);
3247 return(error);
3248}