AMD64 - Refactor uio_resid and size_t assumptions.
[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 $
91fdc001 40 * $DragonFly: src/sys/kern/vfs_syscalls.c,v 1.135 2008/11/11 00:55:49 pavalos Exp $
984263bc
MD
41 */
42
984263bc
MD
43#include <sys/param.h>
44#include <sys/systm.h>
45#include <sys/buf.h>
16df2c74 46#include <sys/conf.h>
984263bc
MD
47#include <sys/sysent.h>
48#include <sys/malloc.h>
49#include <sys/mount.h>
432b8263 50#include <sys/mountctl.h>
984263bc 51#include <sys/sysproto.h>
984263bc
MD
52#include <sys/filedesc.h>
53#include <sys/kernel.h>
54#include <sys/fcntl.h>
55#include <sys/file.h>
56#include <sys/linker.h>
57#include <sys/stat.h>
58#include <sys/unistd.h>
59#include <sys/vnode.h>
60#include <sys/proc.h>
895c1f85 61#include <sys/priv.h>
9658f145 62#include <sys/jail.h>
dadab5e9 63#include <sys/namei.h>
21739618 64#include <sys/nlookup.h>
984263bc
MD
65#include <sys/dirent.h>
66#include <sys/extattr.h>
8fa76237 67#include <sys/spinlock.h>
8f6f8622 68#include <sys/kern_syscall.h>
70aac194 69#include <sys/objcache.h>
984263bc 70#include <sys/sysctl.h>
408357d8
MD
71
72#include <sys/buf2.h>
28623bf9
MD
73#include <sys/file2.h>
74#include <sys/spinlock2.h>
75
984263bc
MD
76#include <vm/vm.h>
77#include <vm/vm_object.h>
984263bc
MD
78#include <vm/vm_page.h>
79
28623bf9
MD
80#include <machine/limits.h>
81#include <machine/stdarg.h>
82
83#include <vfs/union/union.h>
dadab5e9 84
28623bf9 85static void mount_warning(struct mount *mp, const char *ctl, ...);
d3c546e6 86static int mount_path(struct proc *p, struct mount *mp, char **rb, char **fb);
e24b948e 87static int checkvp_chdir (struct vnode *vn, struct thread *td);
77480a97 88static void checkdirs (struct nchandle *old_nch, struct nchandle *new_nch);
402ed7e1 89static int chroot_refuse_vdir_fds (struct filedesc *fdp);
75ffff0d 90static int chroot_visible_mnt(struct mount *mp, struct proc *p);
402ed7e1
RG
91static int getutimes (const struct timeval *, struct timespec *);
92static int setfown (struct vnode *, uid_t, gid_t);
93static int setfmode (struct vnode *, int);
94static int setfflags (struct vnode *, int);
3a907475
MD
95static int setutimes (struct vnode *, struct vattr *,
96 const struct timespec *, int);
984263bc
MD
97static int usermount = 0; /* if 1, non-root can mount fs. */
98
402ed7e1 99int (*union_dircheckp) (struct thread *, struct vnode **, struct file *);
984263bc
MD
100
101SYSCTL_INT(_vfs, OID_AUTO, usermount, CTLFLAG_RW, &usermount, 0, "");
102
103/*
104 * Virtual File System System Calls
105 */
106
107/*
108 * Mount a file system.
109 */
41c20dac
MD
110/*
111 * mount_args(char *type, char *path, int flags, caddr_t data)
112 */
984263bc
MD
113/* ARGSUSED */
114int
753fd850 115sys_mount(struct mount_args *uap)
984263bc 116{
dadab5e9
MD
117 struct thread *td = curthread;
118 struct proc *p = td->td_proc;
984263bc 119 struct vnode *vp;
28623bf9 120 struct nchandle nch;
984263bc
MD
121 struct mount *mp;
122 struct vfsconf *vfsp;
123 int error, flag = 0, flag2 = 0;
1d505369 124 int hasmount;
984263bc 125 struct vattr va;
21739618 126 struct nlookupdata nd;
984263bc 127 char fstypename[MFSNAMELEN];
acde96db 128 struct ucred *cred = p->p_ucred;
984263bc 129
cd0d3e17 130 KKASSERT(p);
57962f80 131 if (jailed(cred))
84c3a710 132 return (EPERM);
895c1f85 133 if (usermount == 0 && (error = priv_check(td, PRIV_ROOT)))
984263bc
MD
134 return (error);
135 /*
136 * Do not allow NFS export by non-root users.
137 */
ab2eb4eb 138 if (uap->flags & MNT_EXPORTED) {
895c1f85 139 error = priv_check(td, PRIV_ROOT);
984263bc
MD
140 if (error)
141 return (error);
142 }
143 /*
144 * Silently enforce MNT_NOSUID and MNT_NODEV for non-root users
145 */
895c1f85 146 if (priv_check(td, PRIV_ROOT))
ab2eb4eb 147 uap->flags |= MNT_NOSUID | MNT_NODEV;
21739618 148
984263bc 149 /*
28623bf9 150 * Lookup the requested path and extract the nch and vnode.
984263bc 151 */
21739618
MD
152 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
153 if (error == 0) {
154 if ((error = nlookup(&nd)) == 0) {
28623bf9 155 if (nd.nl_nch.ncp->nc_vp == NULL)
21739618
MD
156 error = ENOENT;
157 }
158 }
159 if (error) {
160 nlookup_done(&nd);
984263bc 161 return (error);
21739618
MD
162 }
163
164 /*
165 * Extract the locked+refd ncp and cleanup the nd structure
166 */
28623bf9
MD
167 nch = nd.nl_nch;
168 cache_zero(&nd.nl_nch);
21739618
MD
169 nlookup_done(&nd);
170
28623bf9 171 if ((nch.ncp->nc_flag & NCF_ISMOUNTPT) && cache_findmount(&nch))
1d505369
MD
172 hasmount = 1;
173 else
174 hasmount = 0;
175
176
21739618 177 /*
28623bf9 178 * now we have the locked ref'd nch and unreferenced vnode.
21739618 179 */
28623bf9 180 vp = nch.ncp->nc_vp;
87de5057 181 if ((error = vget(vp, LK_EXCLUSIVE)) != 0) {
28623bf9 182 cache_put(&nch);
21739618
MD
183 return (error);
184 }
28623bf9 185 cache_unlock(&nch);
21739618
MD
186
187 /*
28623bf9 188 * Now we have an unlocked ref'd nch and a locked ref'd vp
21739618 189 */
ab2eb4eb 190 if (uap->flags & MNT_UPDATE) {
67863d04 191 if ((vp->v_flag & (VROOT|VPFSROOT)) == 0) {
28623bf9 192 cache_drop(&nch);
984263bc
MD
193 vput(vp);
194 return (EINVAL);
195 }
196 mp = vp->v_mount;
197 flag = mp->mnt_flag;
198 flag2 = mp->mnt_kern_flag;
199 /*
200 * We only allow the filesystem to be reloaded if it
201 * is currently mounted read-only.
202 */
ab2eb4eb 203 if ((uap->flags & MNT_RELOAD) &&
984263bc 204 ((mp->mnt_flag & MNT_RDONLY) == 0)) {
28623bf9 205 cache_drop(&nch);
984263bc
MD
206 vput(vp);
207 return (EOPNOTSUPP); /* Needs translation */
208 }
209 /*
210 * Only root, or the user that did the original mount is
211 * permitted to update it.
212 */
acde96db 213 if (mp->mnt_stat.f_owner != cred->cr_uid &&
895c1f85 214 (error = priv_check(td, PRIV_ROOT))) {
28623bf9 215 cache_drop(&nch);
984263bc
MD
216 vput(vp);
217 return (error);
218 }
f9642f56 219 if (vfs_busy(mp, LK_NOWAIT)) {
28623bf9 220 cache_drop(&nch);
984263bc
MD
221 vput(vp);
222 return (EBUSY);
223 }
1d505369 224 if ((vp->v_flag & VMOUNT) != 0 || hasmount) {
28623bf9 225 cache_drop(&nch);
f9642f56 226 vfs_unbusy(mp);
984263bc
MD
227 vput(vp);
228 return (EBUSY);
229 }
230 vp->v_flag |= VMOUNT;
984263bc 231 mp->mnt_flag |=
ab2eb4eb 232 uap->flags & (MNT_RELOAD | MNT_FORCE | MNT_UPDATE);
a11aaa81 233 vn_unlock(vp);
984263bc
MD
234 goto update;
235 }
236 /*
237 * If the user is not root, ensure that they own the directory
238 * onto which we are attempting to mount.
239 */
87de5057 240 if ((error = VOP_GETATTR(vp, &va)) ||
895c1f85 241 (va.va_uid != cred->cr_uid && (error = priv_check(td, PRIV_ROOT)))) {
28623bf9 242 cache_drop(&nch);
984263bc
MD
243 vput(vp);
244 return (error);
245 }
87de5057 246 if ((error = vinvalbuf(vp, V_SAVE, 0, 0)) != 0) {
28623bf9 247 cache_drop(&nch);
984263bc
MD
248 vput(vp);
249 return (error);
250 }
251 if (vp->v_type != VDIR) {
28623bf9 252 cache_drop(&nch);
984263bc
MD
253 vput(vp);
254 return (ENOTDIR);
255 }
5dc91765
MD
256 if (vp->v_mount->mnt_kern_flag & MNTK_NOSTKMNT) {
257 cache_drop(&nch);
258 vput(vp);
259 return (EPERM);
260 }
ab2eb4eb 261 if ((error = copyinstr(uap->type, fstypename, MFSNAMELEN, NULL)) != 0) {
28623bf9 262 cache_drop(&nch);
984263bc
MD
263 vput(vp);
264 return (error);
265 }
2613053d 266 vfsp = vfsconf_find_by_name(fstypename);
984263bc
MD
267 if (vfsp == NULL) {
268 linker_file_t lf;
269
270 /* Only load modules for root (very important!) */
895c1f85 271 if ((error = priv_check(td, PRIV_ROOT)) != 0) {
28623bf9 272 cache_drop(&nch);
984263bc
MD
273 vput(vp);
274 return error;
275 }
b8fb634a 276 error = linker_load_file(fstypename, &lf, 0);
984263bc 277 if (error || lf == NULL) {
28623bf9 278 cache_drop(&nch);
984263bc
MD
279 vput(vp);
280 if (lf == NULL)
281 error = ENODEV;
282 return error;
283 }
284 lf->userrefs++;
285 /* lookup again, see if the VFS was loaded */
2613053d 286 vfsp = vfsconf_find_by_name(fstypename);
984263bc
MD
287 if (vfsp == NULL) {
288 lf->userrefs--;
289 linker_file_unload(lf);
28623bf9 290 cache_drop(&nch);
984263bc
MD
291 vput(vp);
292 return (ENODEV);
293 }
294 }
1d505369 295 if ((vp->v_flag & VMOUNT) != 0 || hasmount) {
28623bf9 296 cache_drop(&nch);
984263bc
MD
297 vput(vp);
298 return (EBUSY);
299 }
300 vp->v_flag |= VMOUNT;
984263bc
MD
301
302 /*
303 * Allocate and initialize the filesystem.
304 */
efda3bd0 305 mp = kmalloc(sizeof(struct mount), M_MOUNT, M_ZERO|M_WAITOK);
984263bc
MD
306 TAILQ_INIT(&mp->mnt_nvnodelist);
307 TAILQ_INIT(&mp->mnt_reservedvnlist);
2281065e 308 TAILQ_INIT(&mp->mnt_jlist);
984263bc 309 mp->mnt_nvnodelistsize = 0;
ab6f251b 310 lockinit(&mp->mnt_lock, "vfslock", 0, 0);
f9642f56 311 vfs_busy(mp, LK_NOWAIT);
984263bc
MD
312 mp->mnt_op = vfsp->vfc_vfsops;
313 mp->mnt_vfc = vfsp;
314 vfsp->vfc_refcount++;
315 mp->mnt_stat.f_type = vfsp->vfc_typenum;
316 mp->mnt_flag |= vfsp->vfc_flags & MNT_VISFLAGMASK;
317 strncpy(mp->mnt_stat.f_fstypename, vfsp->vfc_name, MFSNAMELEN);
acde96db 318 mp->mnt_stat.f_owner = cred->cr_uid;
984263bc 319 mp->mnt_iosize_max = DFLTPHYS;
a11aaa81 320 vn_unlock(vp);
984263bc
MD
321update:
322 /*
323 * Set the mount level flags.
324 */
ab2eb4eb 325 if (uap->flags & MNT_RDONLY)
984263bc
MD
326 mp->mnt_flag |= MNT_RDONLY;
327 else if (mp->mnt_flag & MNT_RDONLY)
328 mp->mnt_kern_flag |= MNTK_WANTRDWR;
329 mp->mnt_flag &=~ (MNT_NOSUID | MNT_NOEXEC | MNT_NODEV |
330 MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | MNT_NOATIME |
331 MNT_NOSYMFOLLOW | MNT_IGNORE |
332 MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR);
ab2eb4eb 333 mp->mnt_flag |= uap->flags & (MNT_NOSUID | MNT_NOEXEC |
984263bc
MD
334 MNT_NODEV | MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | MNT_FORCE |
335 MNT_NOSYMFOLLOW | MNT_IGNORE |
336 MNT_NOATIME | MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR);
337 /*
338 * Mount the filesystem.
339 * XXX The final recipients of VFS_MOUNT just overwrite the ndp they
fad57d0e 340 * get.
984263bc 341 */
acde96db 342 error = VFS_MOUNT(mp, uap->path, uap->data, cred);
984263bc
MD
343 if (mp->mnt_flag & MNT_UPDATE) {
344 if (mp->mnt_kern_flag & MNTK_WANTRDWR)
345 mp->mnt_flag &= ~MNT_RDONLY;
346 mp->mnt_flag &=~ (MNT_UPDATE | MNT_RELOAD | MNT_FORCE);
347 mp->mnt_kern_flag &=~ MNTK_WANTRDWR;
348 if (error) {
349 mp->mnt_flag = flag;
350 mp->mnt_kern_flag = flag2;
351 }
f9642f56 352 vfs_unbusy(mp);
984263bc 353 vp->v_flag &= ~VMOUNT;
984263bc 354 vrele(vp);
28623bf9 355 cache_drop(&nch);
984263bc
MD
356 return (error);
357 }
ca466bae 358 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
984263bc 359 /*
21739618 360 * Put the new filesystem on the mount list after root. The mount
28623bf9
MD
361 * point gets its own mnt_ncmountpt (unless the VFS already set one
362 * up) which represents the root of the mount. The lookup code
363 * detects the mount point going forward and checks the root of
364 * the mount going backwards.
8e005a45
MD
365 *
366 * It is not necessary to invalidate or purge the vnode underneath
367 * because elements under the mount will be given their own glue
368 * namecache record.
984263bc 369 */
984263bc 370 if (!error) {
28623bf9
MD
371 if (mp->mnt_ncmountpt.ncp == NULL) {
372 /*
373 * allocate, then unlock, but leave the ref intact
374 */
375 cache_allocroot(&mp->mnt_ncmountpt, mp, NULL);
376 cache_unlock(&mp->mnt_ncmountpt);
377 }
378 mp->mnt_ncmounton = nch; /* inherits ref */
379 nch.ncp->nc_flag |= NCF_ISMOUNTPT;
380
381 /* XXX get the root of the fs and cache_setvp(mnt_ncmountpt...) */
984263bc 382 vp->v_flag &= ~VMOUNT;
861905fb 383 mountlist_insert(mp, MNTINS_LAST);
a11aaa81 384 vn_unlock(vp);
c0721f5f 385 checkdirs(&mp->mnt_ncmounton, &mp->mnt_ncmountpt);
41a01a4d 386 error = vfs_allocate_syncvnode(mp);
f9642f56 387 vfs_unbusy(mp);
1d505369
MD
388 error = VFS_START(mp, 0);
389 vrele(vp);
984263bc 390 } else {
66a1ddf5
MD
391 vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_coherency_ops);
392 vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_journal_ops);
393 vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_norm_ops);
394 vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_spec_ops);
395 vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_fifo_ops);
984263bc 396 vp->v_flag &= ~VMOUNT;
984263bc 397 mp->mnt_vfc->vfc_refcount--;
f9642f56 398 vfs_unbusy(mp);
efda3bd0 399 kfree(mp, M_MOUNT);
28623bf9 400 cache_drop(&nch);
984263bc
MD
401 vput(vp);
402 }
403 return (error);
404}
405
406/*
407 * Scan all active processes to see if any of them have a current
408 * or root directory onto which the new filesystem has just been
409 * mounted. If so, replace them with the new mount point.
21739618
MD
410 *
411 * The passed ncp is ref'd and locked (from the mount code) and
412 * must be associated with the vnode representing the root of the
413 * mount point.
984263bc 414 */
8fa76237 415struct checkdirs_info {
77480a97
MD
416 struct nchandle old_nch;
417 struct nchandle new_nch;
418 struct vnode *old_vp;
419 struct vnode *new_vp;
8fa76237
MD
420};
421
422static int checkdirs_callback(struct proc *p, void *data);
423
984263bc 424static void
77480a97 425checkdirs(struct nchandle *old_nch, struct nchandle *new_nch)
984263bc 426{
8fa76237 427 struct checkdirs_info info;
77480a97 428 struct vnode *olddp;
984263bc 429 struct vnode *newdp;
690a3127 430 struct mount *mp;
984263bc 431
77480a97
MD
432 /*
433 * If the old mount point's vnode has a usecount of 1, it is not
434 * being held as a descriptor anywhere.
435 */
436 olddp = old_nch->ncp->nc_vp;
3c37c940 437 if (olddp == NULL || olddp->v_sysref.refcnt == 1)
984263bc 438 return;
77480a97
MD
439
440 /*
441 * Force the root vnode of the new mount point to be resolved
442 * so we can update any matching processes.
443 */
444 mp = new_nch->mount;
690a3127 445 if (VFS_ROOT(mp, &newdp))
984263bc 446 panic("mount: lost mount");
77480a97
MD
447 cache_setunresolved(new_nch);
448 cache_setvp(new_nch, newdp);
21739618 449
77480a97
MD
450 /*
451 * Special handling of the root node
452 */
21739618
MD
453 if (rootvnode == olddp) {
454 vref(newdp);
77480a97 455 vfs_cache_setroot(newdp, cache_hold(new_nch));
21739618
MD
456 }
457
77480a97
MD
458 /*
459 * Pass newdp separately so the callback does not have to access
460 * it via new_nch->ncp->nc_vp.
461 */
462 info.old_nch = *old_nch;
463 info.new_nch = *new_nch;
464 info.new_vp = newdp;
8fa76237
MD
465 allproc_scan(checkdirs_callback, &info);
466 vput(newdp);
467}
468
469/*
470 * NOTE: callback is not MP safe because the scanned process's filedesc
471 * structure can be ripped out from under us, amoung other things.
472 */
473static int
474checkdirs_callback(struct proc *p, void *data)
475{
476 struct checkdirs_info *info = data;
477 struct filedesc *fdp;
28623bf9
MD
478 struct nchandle ncdrop1;
479 struct nchandle ncdrop2;
8fa76237
MD
480 struct vnode *vprele1;
481 struct vnode *vprele2;
482
483 if ((fdp = p->p_fd) != NULL) {
28623bf9
MD
484 cache_zero(&ncdrop1);
485 cache_zero(&ncdrop2);
8fa76237
MD
486 vprele1 = NULL;
487 vprele2 = NULL;
488
489 /*
490 * MPUNSAFE - XXX fdp can be pulled out from under a
491 * foreign process.
492 *
493 * A shared filedesc is ok, we don't have to copy it
494 * because we are making this change globally.
495 */
496 spin_lock_wr(&fdp->fd_spin);
77480a97
MD
497 if (fdp->fd_ncdir.mount == info->old_nch.mount &&
498 fdp->fd_ncdir.ncp == info->old_nch.ncp) {
8fa76237 499 vprele1 = fdp->fd_cdir;
77480a97
MD
500 vref(info->new_vp);
501 fdp->fd_cdir = info->new_vp;
8fa76237 502 ncdrop1 = fdp->fd_ncdir;
77480a97 503 cache_copy(&info->new_nch, &fdp->fd_ncdir);
984263bc 504 }
77480a97
MD
505 if (fdp->fd_nrdir.mount == info->old_nch.mount &&
506 fdp->fd_nrdir.ncp == info->old_nch.ncp) {
8fa76237 507 vprele2 = fdp->fd_rdir;
77480a97
MD
508 vref(info->new_vp);
509 fdp->fd_rdir = info->new_vp;
8fa76237 510 ncdrop2 = fdp->fd_nrdir;
77480a97 511 cache_copy(&info->new_nch, &fdp->fd_nrdir);
984263bc 512 }
8fa76237 513 spin_unlock_wr(&fdp->fd_spin);
28623bf9
MD
514 if (ncdrop1.ncp)
515 cache_drop(&ncdrop1);
516 if (ncdrop2.ncp)
517 cache_drop(&ncdrop2);
8fa76237
MD
518 if (vprele1)
519 vrele(vprele1);
520 if (vprele2)
521 vrele(vprele2);
984263bc 522 }
8fa76237 523 return(0);
984263bc
MD
524}
525
526/*
527 * Unmount a file system.
528 *
529 * Note: unmount takes a path to the vnode mounted on as argument,
530 * not special file (as before).
531 */
41c20dac
MD
532/*
533 * umount_args(char *path, int flags)
534 */
984263bc
MD
535/* ARGSUSED */
536int
753fd850 537sys_unmount(struct unmount_args *uap)
984263bc 538{
dadab5e9
MD
539 struct thread *td = curthread;
540 struct proc *p = td->td_proc;
d2293868 541 struct mount *mp = NULL;
984263bc 542 int error;
fad57d0e 543 struct nlookupdata nd;
984263bc 544
dadab5e9 545 KKASSERT(p);
84c3a710
HP
546 if (p->p_ucred->cr_prison != NULL)
547 return (EPERM);
895c1f85 548 if (usermount == 0 && (error = priv_check(td, PRIV_ROOT)))
84c3a710
HP
549 return (error);
550
ab2eb4eb 551 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
fad57d0e
MD
552 if (error == 0)
553 error = nlookup(&nd);
fad57d0e 554 if (error)
d2293868 555 goto out;
fad57d0e 556
28623bf9 557 mp = nd.nl_nch.mount;
984263bc
MD
558
559 /*
560 * Only root, or the user that did the original mount is
561 * permitted to unmount this filesystem.
562 */
563 if ((mp->mnt_stat.f_owner != p->p_ucred->cr_uid) &&
895c1f85 564 (error = priv_check(td, PRIV_ROOT)))
d2293868 565 goto out;
984263bc
MD
566
567 /*
568 * Don't allow unmounting the root file system.
569 */
570 if (mp->mnt_flag & MNT_ROOTFS) {
d2293868
MD
571 error = EINVAL;
572 goto out;
984263bc
MD
573 }
574
575 /*
576 * Must be the root of the filesystem
577 */
28623bf9 578 if (nd.nl_nch.ncp != mp->mnt_ncmountpt.ncp) {
d2293868
MD
579 error = EINVAL;
580 goto out;
984263bc 581 }
d2293868
MD
582
583out:
584 nlookup_done(&nd);
585 if (error)
586 return (error);
acde96db 587 return (dounmount(mp, uap->flags));
984263bc
MD
588}
589
590/*
591 * Do the actual file system unmount.
592 */
861905fb
MD
593static int
594dounmount_interlock(struct mount *mp)
595{
596 if (mp->mnt_kern_flag & MNTK_UNMOUNT)
597 return (EBUSY);
598 mp->mnt_kern_flag |= MNTK_UNMOUNT;
599 return(0);
600}
601
984263bc 602int
acde96db 603dounmount(struct mount *mp, int flags)
984263bc 604{
dcf26f79 605 struct namecache *ncp;
28623bf9 606 struct nchandle nch;
2ec4b00d 607 struct vnode *vp;
984263bc
MD
608 int error;
609 int async_flag;
2df9419c 610 int lflags;
28623bf9 611 int freeok = 1;
984263bc 612
861905fb
MD
613 /*
614 * Exclusive access for unmounting purposes
615 */
616 if ((error = mountlist_interlock(dounmount_interlock, mp)) != 0)
617 return (error);
618
619 /*
620 * Allow filesystems to detect that a forced unmount is in progress.
621 */
984263bc
MD
622 if (flags & MNT_FORCE)
623 mp->mnt_kern_flag |= MNTK_UNMOUNTF;
2df9419c 624 lflags = LK_EXCLUSIVE | ((flags & MNT_FORCE) ? 0 : LK_NOWAIT);
df4f70a6 625 error = lockmgr(&mp->mnt_lock, lflags);
984263bc
MD
626 if (error) {
627 mp->mnt_kern_flag &= ~(MNTK_UNMOUNT | MNTK_UNMOUNTF);
628 if (mp->mnt_kern_flag & MNTK_MWAIT)
e38462a2 629 wakeup(mp);
984263bc
MD
630 return (error);
631 }
632
633 if (mp->mnt_flag & MNT_EXPUBLIC)
634 vfs_setpublicfs(NULL, NULL, NULL);
635
636 vfs_msync(mp, MNT_WAIT);
637 async_flag = mp->mnt_flag & MNT_ASYNC;
638 mp->mnt_flag &=~ MNT_ASYNC;
dcf26f79
MD
639
640 /*
28623bf9
MD
641 * If this filesystem isn't aliasing other filesystems,
642 * try to invalidate any remaining namecache entries and
643 * check the count afterwords.
dcf26f79 644 */
28623bf9
MD
645 if ((mp->mnt_kern_flag & MNTK_NCALIASED) == 0) {
646 cache_lock(&mp->mnt_ncmountpt);
647 cache_inval(&mp->mnt_ncmountpt, CINV_DESTROY|CINV_CHILDREN);
648 cache_unlock(&mp->mnt_ncmountpt);
649
650 if ((ncp = mp->mnt_ncmountpt.ncp) != NULL &&
651 (ncp->nc_refs != 1 || TAILQ_FIRST(&ncp->nc_list))) {
dcf26f79
MD
652
653 if ((flags & MNT_FORCE) == 0) {
654 error = EBUSY;
28623bf9
MD
655 mount_warning(mp, "Cannot unmount: "
656 "%d namecache "
657 "references still "
658 "present",
659 ncp->nc_refs - 1);
dcf26f79 660 } else {
28623bf9
MD
661 mount_warning(mp, "Forced unmount: "
662 "%d namecache "
663 "references still "
664 "present",
665 ncp->nc_refs - 1);
666 freeok = 0;
dcf26f79
MD
667 }
668 }
669 }
670
28623bf9
MD
671 /*
672 * nchandle records ref the mount structure. Expect a count of 1
673 * (our mount->mnt_ncmountpt).
674 */
675 if (mp->mnt_refs != 1) {
676 if ((flags & MNT_FORCE) == 0) {
677 mount_warning(mp, "Cannot unmount: "
678 "%d process references still "
679 "present", mp->mnt_refs);
680 error = EBUSY;
681 } else {
682 mount_warning(mp, "Forced unmount: "
683 "%d process references still "
684 "present", mp->mnt_refs);
685 freeok = 0;
686 }
687 }
688
2ec4b00d
MD
689 /*
690 * Decomission our special mnt_syncer vnode. This also stops
691 * the vnlru code. If we are unable to unmount we recommission
692 * the vnode.
693 */
dcf26f79 694 if (error == 0) {
2ec4b00d
MD
695 if ((vp = mp->mnt_syncer) != NULL) {
696 mp->mnt_syncer = NULL;
697 vrele(vp);
698 }
dcf26f79
MD
699 if (((mp->mnt_flag & MNT_RDONLY) ||
700 (error = VFS_SYNC(mp, MNT_WAIT)) == 0) ||
701 (flags & MNT_FORCE)) {
702 error = VFS_UNMOUNT(mp, flags);
703 }
704 }
984263bc 705 if (error) {
41a01a4d
MD
706 if (mp->mnt_syncer == NULL)
707 vfs_allocate_syncvnode(mp);
984263bc
MD
708 mp->mnt_kern_flag &= ~(MNTK_UNMOUNT | MNTK_UNMOUNTF);
709 mp->mnt_flag |= async_flag;
df4f70a6 710 lockmgr(&mp->mnt_lock, LK_RELEASE);
984263bc 711 if (mp->mnt_kern_flag & MNTK_MWAIT)
e38462a2 712 wakeup(mp);
984263bc
MD
713 return (error);
714 }
432b8263
MD
715 /*
716 * Clean up any journals still associated with the mount after
717 * filesystem activity has ceased.
718 */
719 journal_remove_all_journals(mp,
720 ((flags & MNT_FORCE) ? MC_JOURNAL_STOP_IMM : 0));
721
861905fb 722 mountlist_remove(mp);
0961aa92
MD
723
724 /*
725 * Remove any installed vnode ops here so the individual VFSs don't
726 * have to.
727 */
66a1ddf5
MD
728 vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_coherency_ops);
729 vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_journal_ops);
730 vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_norm_ops);
731 vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_spec_ops);
732 vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_fifo_ops);
0961aa92 733
28623bf9
MD
734 if (mp->mnt_ncmountpt.ncp != NULL) {
735 nch = mp->mnt_ncmountpt;
736 cache_zero(&mp->mnt_ncmountpt);
737 cache_clrmountpt(&nch);
738 cache_drop(&nch);
739 }
740 if (mp->mnt_ncmounton.ncp != NULL) {
741 nch = mp->mnt_ncmounton;
742 cache_zero(&mp->mnt_ncmounton);
743 cache_clrmountpt(&nch);
744 cache_drop(&nch);
984263bc 745 }
1d505369 746
984263bc
MD
747 mp->mnt_vfc->vfc_refcount--;
748 if (!TAILQ_EMPTY(&mp->mnt_nvnodelist))
749 panic("unmount: dangling vnode");
df4f70a6 750 lockmgr(&mp->mnt_lock, LK_RELEASE);
984263bc 751 if (mp->mnt_kern_flag & MNTK_MWAIT)
e38462a2 752 wakeup(mp);
28623bf9
MD
753 if (freeok)
754 kfree(mp, M_MOUNT);
984263bc
MD
755 return (0);
756}
757
28623bf9
MD
758static
759void
760mount_warning(struct mount *mp, const char *ctl, ...)
761{
762 char *ptr;
763 char *buf;
764 __va_list va;
765
766 __va_start(va, ctl);
767 if (cache_fullpath(NULL, &mp->mnt_ncmounton, &ptr, &buf) == 0) {
6ea70f76 768 kprintf("unmount(%s): ", ptr);
379210cb 769 kvprintf(ctl, va);
6ea70f76 770 kprintf("\n");
28623bf9
MD
771 kfree(buf, M_TEMP);
772 } else {
56bcacad
MD
773 kprintf("unmount(%p", mp);
774 if (mp->mnt_ncmounton.ncp && mp->mnt_ncmounton.ncp->nc_name)
775 kprintf(",%s", mp->mnt_ncmounton.ncp->nc_name);
776 kprintf("): ");
379210cb 777 kvprintf(ctl, va);
6ea70f76 778 kprintf("\n");
28623bf9
MD
779 }
780 __va_end(va);
781}
782
984263bc 783/*
d3c546e6
MD
784 * Shim cache_fullpath() to handle the case where a process is chrooted into
785 * a subdirectory of a mount. In this case if the root mount matches the
786 * process root directory's mount we have to specify the process's root
787 * directory instead of the mount point, because the mount point might
788 * be above the root directory.
789 */
790static
791int
792mount_path(struct proc *p, struct mount *mp, char **rb, char **fb)
793{
794 struct nchandle *nch;
d3c546e6
MD
795
796 if (p && p->p_fd->fd_nrdir.mount == mp)
797 nch = &p->p_fd->fd_nrdir;
798 else
799 nch = &mp->mnt_ncmountpt;
800 return(cache_fullpath(p, nch, rb, fb));
801}
802
803/*
984263bc
MD
804 * Sync each mounted filesystem.
805 */
984263bc
MD
806
807#ifdef DEBUG
808static int syncprt = 0;
809SYSCTL_INT(_debug, OID_AUTO, syncprt, CTLFLAG_RW, &syncprt, 0, "");
6bdbb368 810#endif /* DEBUG */
984263bc 811
861905fb
MD
812static int sync_callback(struct mount *mp, void *data);
813
984263bc
MD
814/* ARGSUSED */
815int
753fd850 816sys_sync(struct sync_args *uap)
984263bc 817{
861905fb 818 mountlist_scan(sync_callback, NULL, MNTSCAN_FORWARD);
6bdbb368 819#ifdef DEBUG
861905fb
MD
820 /*
821 * print out buffer pool stat information on each sync() call.
822 */
984263bc
MD
823 if (syncprt)
824 vfs_bufstats();
6bdbb368 825#endif /* DEBUG */
984263bc
MD
826 return (0);
827}
828
861905fb
MD
829static
830int
831sync_callback(struct mount *mp, void *data __unused)
832{
833 int asyncflag;
834
835 if ((mp->mnt_flag & MNT_RDONLY) == 0) {
836 asyncflag = mp->mnt_flag & MNT_ASYNC;
837 mp->mnt_flag &= ~MNT_ASYNC;
838 vfs_msync(mp, MNT_NOWAIT);
87de5057 839 VFS_SYNC(mp, MNT_NOWAIT);
861905fb
MD
840 mp->mnt_flag |= asyncflag;
841 }
842 return(0);
843}
844
984263bc
MD
845/* XXX PRISON: could be per prison flag */
846static int prison_quotas;
847#if 0
848SYSCTL_INT(_kern_prison, OID_AUTO, quotas, CTLFLAG_RW, &prison_quotas, 0, "");
849#endif
850
851/*
41c20dac
MD
852 * quotactl_args(char *path, int fcmd, int uid, caddr_t arg)
853 *
984263bc
MD
854 * Change filesystem quotas.
855 */
984263bc
MD
856/* ARGSUSED */
857int
753fd850 858sys_quotactl(struct quotactl_args *uap)
984263bc 859{
fad57d0e
MD
860 struct nlookupdata nd;
861 struct thread *td;
862 struct proc *p;
41c20dac 863 struct mount *mp;
984263bc 864 int error;
984263bc 865
fad57d0e
MD
866 td = curthread;
867 p = td->td_proc;
41c20dac 868 if (p->p_ucred->cr_prison && !prison_quotas)
984263bc 869 return (EPERM);
fad57d0e 870
ab2eb4eb 871 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
fad57d0e
MD
872 if (error == 0)
873 error = nlookup(&nd);
874 if (error == 0) {
28623bf9 875 mp = nd.nl_nch.mount;
ab2eb4eb 876 error = VFS_QUOTACTL(mp, uap->cmd, uap->uid,
acde96db 877 uap->arg, nd.nl_cred);
fad57d0e
MD
878 }
879 nlookup_done(&nd);
880 return (error);
984263bc
MD
881}
882
949ecb9b 883/*
2281065e 884 * mountctl(char *path, int op, int fd, const void *ctl, int ctllen,
949ecb9b
MD
885 * void *buf, int buflen)
886 *
887 * This function operates on a mount point and executes the specified
888 * operation using the specified control data, and possibly returns data.
889 *
890 * The actual number of bytes stored in the result buffer is returned, 0
891 * if none, otherwise an error is returned.
892 */
893/* ARGSUSED */
894int
753fd850 895sys_mountctl(struct mountctl_args *uap)
949ecb9b
MD
896{
897 struct thread *td = curthread;
898 struct proc *p = td->td_proc;
2281065e 899 struct file *fp;
949ecb9b
MD
900 void *ctl = NULL;
901 void *buf = NULL;
902 char *path = NULL;
903 int error;
904
905 /*
906 * Sanity and permissions checks. We must be root.
907 */
908 KKASSERT(p);
909 if (p->p_ucred->cr_prison != NULL)
910 return (EPERM);
dad088a5
MD
911 if ((uap->op != MOUNTCTL_MOUNTFLAGS) &&
912 (error = priv_check(td, PRIV_ROOT)) != 0)
949ecb9b
MD
913 return (error);
914
915 /*
916 * Argument length checks
917 */
39b13188 918 if (uap->ctllen < 0 || uap->ctllen > 1024)
949ecb9b 919 return (EINVAL);
39b13188 920 if (uap->buflen < 0 || uap->buflen > 16 * 1024)
949ecb9b
MD
921 return (EINVAL);
922 if (uap->path == NULL)
923 return (EINVAL);
924
925 /*
926 * Allocate the necessary buffers and copyin data
927 */
70aac194 928 path = objcache_get(namei_oc, M_WAITOK);
949ecb9b
MD
929 error = copyinstr(uap->path, path, MAXPATHLEN, NULL);
930 if (error)
931 goto done;
932
933 if (uap->ctllen) {
efda3bd0 934 ctl = kmalloc(uap->ctllen + 1, M_TEMP, M_WAITOK|M_ZERO);
949ecb9b
MD
935 error = copyin(uap->ctl, ctl, uap->ctllen);
936 if (error)
937 goto done;
938 }
939 if (uap->buflen)
efda3bd0 940 buf = kmalloc(uap->buflen + 1, M_TEMP, M_WAITOK|M_ZERO);
949ecb9b
MD
941
942 /*
2281065e
MD
943 * Validate the descriptor
944 */
ecda6326
MD
945 if (uap->fd >= 0) {
946 fp = holdfp(p->p_fd, uap->fd, -1);
947 if (fp == NULL) {
948 error = EBADF;
949 goto done;
950 }
951 } else {
952 fp = NULL;
2281065e 953 }
2281065e
MD
954
955 /*
949ecb9b
MD
956 * Execute the internal kernel function and clean up.
957 */
2281065e
MD
958 error = kern_mountctl(path, uap->op, fp, ctl, uap->ctllen, buf, uap->buflen, &uap->sysmsg_result);
959 if (fp)
9f87144f 960 fdrop(fp);
949ecb9b
MD
961 if (error == 0 && uap->sysmsg_result > 0)
962 error = copyout(buf, uap->buf, uap->sysmsg_result);
963done:
964 if (path)
70aac194 965 objcache_put(namei_oc, path);
949ecb9b 966 if (ctl)
efda3bd0 967 kfree(ctl, M_TEMP);
949ecb9b 968 if (buf)
efda3bd0 969 kfree(buf, M_TEMP);
949ecb9b
MD
970 return (error);
971}
972
973/*
974 * Execute a mount control operation by resolving the path to a mount point
975 * and calling vop_mountctl().
67863d04
MD
976 *
977 * Use the mount point from the nch instead of the vnode so nullfs mounts
978 * can properly spike the VOP.
949ecb9b
MD
979 */
980int
2281065e
MD
981kern_mountctl(const char *path, int op, struct file *fp,
982 const void *ctl, int ctllen,
949ecb9b
MD
983 void *buf, int buflen, int *res)
984{
949ecb9b
MD
985 struct vnode *vp;
986 struct mount *mp;
987 struct nlookupdata nd;
988 int error;
989
990 *res = 0;
991 vp = NULL;
992 error = nlookup_init(&nd, path, UIO_SYSSPACE, NLC_FOLLOW);
993 if (error == 0)
994 error = nlookup(&nd);
995 if (error == 0)
28623bf9 996 error = cache_vget(&nd.nl_nch, nd.nl_cred, LK_EXCLUSIVE, &vp);
67863d04 997 mp = nd.nl_nch.mount;
949ecb9b
MD
998 nlookup_done(&nd);
999 if (error)
1000 return (error);
1001
949ecb9b
MD
1002 /*
1003 * Must be the root of the filesystem
1004 */
67863d04 1005 if ((vp->v_flag & (VROOT|VPFSROOT)) == 0) {
949ecb9b
MD
1006 vput(vp);
1007 return (EINVAL);
1008 }
2281065e 1009 error = vop_mountctl(mp->mnt_vn_use_ops, op, fp, ctl, ctllen,
949ecb9b
MD
1010 buf, buflen, res);
1011 vput(vp);
1012 return (error);
1013}
1014
984263bc 1015int
fad57d0e 1016kern_statfs(struct nlookupdata *nd, struct statfs *buf)
984263bc 1017{
dadab5e9 1018 struct thread *td = curthread;
75ffff0d 1019 struct proc *p = td->td_proc;
41c20dac
MD
1020 struct mount *mp;
1021 struct statfs *sp;
75ffff0d 1022 char *fullpath, *freepath;
984263bc 1023 int error;
984263bc 1024
fad57d0e 1025 if ((error = nlookup(nd)) != 0)
984263bc 1026 return (error);
28623bf9 1027 mp = nd->nl_nch.mount;
984263bc 1028 sp = &mp->mnt_stat;
acde96db 1029 if ((error = VFS_STATFS(mp, sp, nd->nl_cred)) != 0)
984263bc 1030 return (error);
75ffff0d 1031
d3c546e6 1032 error = mount_path(p, mp, &fullpath, &freepath);
75ffff0d
JS
1033 if (error)
1034 return(error);
1035 bzero(sp->f_mntonname, sizeof(sp->f_mntonname));
1036 strlcpy(sp->f_mntonname, fullpath, sizeof(sp->f_mntonname));
efda3bd0 1037 kfree(freepath, M_TEMP);
75ffff0d 1038
984263bc 1039 sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
9697c509
DRJ
1040 bcopy(sp, buf, sizeof(*buf));
1041 /* Only root should have access to the fsid's. */
895c1f85 1042 if (priv_check(td, PRIV_ROOT))
9697c509
DRJ
1043 buf->f_fsid.val[0] = buf->f_fsid.val[1] = 0;
1044 return (0);
984263bc
MD
1045}
1046
1047/*
9697c509 1048 * statfs_args(char *path, struct statfs *buf)
41c20dac 1049 *
984263bc
MD
1050 * Get filesystem statistics.
1051 */
984263bc 1052int
753fd850 1053sys_statfs(struct statfs_args *uap)
9697c509 1054{
fad57d0e 1055 struct nlookupdata nd;
9697c509
DRJ
1056 struct statfs buf;
1057 int error;
1058
fad57d0e
MD
1059 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
1060 if (error == 0)
1061 error = kern_statfs(&nd, &buf);
1062 nlookup_done(&nd);
9697c509
DRJ
1063 if (error == 0)
1064 error = copyout(&buf, uap->buf, sizeof(*uap->buf));
1065 return (error);
1066}
1067
1068int
1069kern_fstatfs(int fd, struct statfs *buf)
984263bc 1070{
dadab5e9
MD
1071 struct thread *td = curthread;
1072 struct proc *p = td->td_proc;
984263bc
MD
1073 struct file *fp;
1074 struct mount *mp;
1fd87d54 1075 struct statfs *sp;
75ffff0d 1076 char *fullpath, *freepath;
984263bc 1077 int error;
984263bc 1078
dadab5e9 1079 KKASSERT(p);
5b287bba 1080 if ((error = holdvnode(p->p_fd, fd, &fp)) != 0)
984263bc
MD
1081 return (error);
1082 mp = ((struct vnode *)fp->f_data)->v_mount;
5b287bba
MD
1083 if (mp == NULL) {
1084 error = EBADF;
1085 goto done;
1086 }
1087 if (fp->f_cred == NULL) {
1088 error = EINVAL;
1089 goto done;
1090 }
984263bc 1091 sp = &mp->mnt_stat;
5b287bba
MD
1092 if ((error = VFS_STATFS(mp, sp, fp->f_cred)) != 0)
1093 goto done;
75ffff0d 1094
d3c546e6 1095 if ((error = mount_path(p, mp, &fullpath, &freepath)) != 0)
5b287bba 1096 goto done;
75ffff0d
JS
1097 bzero(sp->f_mntonname, sizeof(sp->f_mntonname));
1098 strlcpy(sp->f_mntonname, fullpath, sizeof(sp->f_mntonname));
efda3bd0 1099 kfree(freepath, M_TEMP);
75ffff0d 1100
984263bc 1101 sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
9697c509 1102 bcopy(sp, buf, sizeof(*buf));
75ffff0d 1103
9697c509 1104 /* Only root should have access to the fsid's. */
895c1f85 1105 if (priv_check(td, PRIV_ROOT))
9697c509 1106 buf->f_fsid.val[0] = buf->f_fsid.val[1] = 0;
5b287bba
MD
1107 error = 0;
1108done:
1109 fdrop(fp);
1110 return (error);
9697c509
DRJ
1111}
1112
1113/*
1114 * fstatfs_args(int fd, struct statfs *buf)
1115 *
1116 * Get filesystem statistics.
1117 */
1118int
753fd850 1119sys_fstatfs(struct fstatfs_args *uap)
9697c509
DRJ
1120{
1121 struct statfs buf;
1122 int error;
1123
1124 error = kern_fstatfs(uap->fd, &buf);
1125
1126 if (error == 0)
1127 error = copyout(&buf, uap->buf, sizeof(*uap->buf));
1128 return (error);
984263bc
MD
1129}
1130
d9fad06e
MD
1131int
1132kern_statvfs(struct nlookupdata *nd, struct statvfs *buf)
1133{
1134 struct mount *mp;
1135 struct statvfs *sp;
1136 int error;
1137
1138 if ((error = nlookup(nd)) != 0)
1139 return (error);
1140 mp = nd->nl_nch.mount;
1141 sp = &mp->mnt_vstat;
1142 if ((error = VFS_STATVFS(mp, sp, nd->nl_cred)) != 0)
1143 return (error);
1144
1145 sp->f_flag = 0;
1146 if (mp->mnt_flag & MNT_RDONLY)
1147 sp->f_flag |= ST_RDONLY;
1148 if (mp->mnt_flag & MNT_NOSUID)
1149 sp->f_flag |= ST_NOSUID;
1150 bcopy(sp, buf, sizeof(*buf));
1151 return (0);
1152}
1153
1154/*
1155 * statfs_args(char *path, struct statfs *buf)
1156 *
1157 * Get filesystem statistics.
1158 */
1159int
1160sys_statvfs(struct statvfs_args *uap)
1161{
1162 struct nlookupdata nd;
1163 struct statvfs buf;
1164 int error;
1165
1166 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
1167 if (error == 0)
1168 error = kern_statvfs(&nd, &buf);
1169 nlookup_done(&nd);
1170 if (error == 0)
1171 error = copyout(&buf, uap->buf, sizeof(*uap->buf));
1172 return (error);
1173}
1174
1175int
1176kern_fstatvfs(int fd, struct statvfs *buf)
1177{
1178 struct thread *td = curthread;
1179 struct proc *p = td->td_proc;
1180 struct file *fp;
1181 struct mount *mp;
1182 struct statvfs *sp;
1183 int error;
1184
1185 KKASSERT(p);
1186 if ((error = holdvnode(p->p_fd, fd, &fp)) != 0)
1187 return (error);
1188 mp = ((struct vnode *)fp->f_data)->v_mount;
1189 if (mp == NULL) {
1190 error = EBADF;
1191 goto done;
1192 }
1193 if (fp->f_cred == NULL) {
1194 error = EINVAL;
1195 goto done;
1196 }
1197 sp = &mp->mnt_vstat;
1198 if ((error = VFS_STATVFS(mp, sp, fp->f_cred)) != 0)
1199 goto done;
1200
1201 sp->f_flag = 0;
1202 if (mp->mnt_flag & MNT_RDONLY)
1203 sp->f_flag |= ST_RDONLY;
1204 if (mp->mnt_flag & MNT_NOSUID)
1205 sp->f_flag |= ST_NOSUID;
1206
1207 bcopy(sp, buf, sizeof(*buf));
1208 error = 0;
1209done:
1210 fdrop(fp);
1211 return (error);
1212}
1213
1214/*
1215 * fstatfs_args(int fd, struct statfs *buf)
1216 *
1217 * Get filesystem statistics.
1218 */
1219int
1220sys_fstatvfs(struct fstatvfs_args *uap)
1221{
1222 struct statvfs buf;
1223 int error;
1224
1225 error = kern_fstatvfs(uap->fd, &buf);
1226
1227 if (error == 0)
1228 error = copyout(&buf, uap->buf, sizeof(*uap->buf));
1229 return (error);
1230}
1231
984263bc 1232/*
41c20dac
MD
1233 * getfsstat_args(struct statfs *buf, long bufsize, int flags)
1234 *
984263bc
MD
1235 * Get statistics on all filesystems.
1236 */
861905fb
MD
1237
1238struct getfsstat_info {
1239 struct statfs *sfsp;
1240 long count;
1241 long maxcount;
1242 int error;
1243 int flags;
861905fb
MD
1244 struct proc *p;
1245};
1246
1247static int getfsstat_callback(struct mount *, void *);
1248
41c20dac 1249/* ARGSUSED */
984263bc 1250int
753fd850 1251sys_getfsstat(struct getfsstat_args *uap)
984263bc 1252{
dadab5e9 1253 struct thread *td = curthread;
75ffff0d 1254 struct proc *p = td->td_proc;
861905fb 1255 struct getfsstat_info info;
75ffff0d 1256
861905fb 1257 bzero(&info, sizeof(info));
861905fb
MD
1258
1259 info.maxcount = uap->bufsize / sizeof(struct statfs);
1260 info.sfsp = uap->buf;
1261 info.count = 0;
1262 info.flags = uap->flags;
861905fb
MD
1263 info.p = p;
1264
1265 mountlist_scan(getfsstat_callback, &info, MNTSCAN_FORWARD);
1266 if (info.sfsp && info.count > info.maxcount)
1267 uap->sysmsg_result = info.maxcount;
75ffff0d 1268 else
861905fb
MD
1269 uap->sysmsg_result = info.count;
1270 return (info.error);
1271}
1272
1273static int
1274getfsstat_callback(struct mount *mp, void *data)
1275{
1276 struct getfsstat_info *info = data;
1277 struct statfs *sp;
1278 char *freepath;
1279 char *fullpath;
1280 int error;
1281
1282 if (info->sfsp && info->count < info->maxcount) {
28623bf9 1283 if (info->p && !chroot_visible_mnt(mp, info->p))
861905fb
MD
1284 return(0);
1285 sp = &mp->mnt_stat;
1286
1287 /*
1288 * If MNT_NOWAIT or MNT_LAZY is specified, do not
1289 * refresh the fsstat cache. MNT_NOWAIT or MNT_LAZY
1290 * overrides MNT_WAIT.
1291 */
1292 if (((info->flags & (MNT_LAZY|MNT_NOWAIT)) == 0 ||
1293 (info->flags & MNT_WAIT)) &&
acde96db 1294 (error = VFS_STATFS(mp, sp, info->p->p_ucred))) {
861905fb
MD
1295 return(0);
1296 }
1297 sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
1298
d3c546e6 1299 error = mount_path(info->p, mp, &fullpath, &freepath);
861905fb
MD
1300 if (error) {
1301 info->error = error;
1302 return(-1);
984263bc 1303 }
861905fb
MD
1304 bzero(sp->f_mntonname, sizeof(sp->f_mntonname));
1305 strlcpy(sp->f_mntonname, fullpath, sizeof(sp->f_mntonname));
efda3bd0 1306 kfree(freepath, M_TEMP);
861905fb
MD
1307
1308 error = copyout(sp, info->sfsp, sizeof(*sp));
1309 if (error) {
1310 info->error = error;
1311 return (-1);
984263bc 1312 }
861905fb 1313 ++info->sfsp;
984263bc 1314 }
861905fb
MD
1315 info->count++;
1316 return(0);
984263bc
MD
1317}
1318
1319/*
00fe9d48
MD
1320 * getvfsstat_args(struct statfs *buf, struct statvfs *vbuf,
1321 long bufsize, int flags)
1322 *
1323 * Get statistics on all filesystems.
1324 */
1325
1326struct getvfsstat_info {
1327 struct statfs *sfsp;
1328 struct statvfs *vsfsp;
1329 long count;
1330 long maxcount;
1331 int error;
1332 int flags;
1333 struct proc *p;
1334};
1335
1336static int getvfsstat_callback(struct mount *, void *);
1337
1338/* ARGSUSED */
1339int
1340sys_getvfsstat(struct getvfsstat_args *uap)
1341{
1342 struct thread *td = curthread;
1343 struct proc *p = td->td_proc;
1344 struct getvfsstat_info info;
1345
1346 bzero(&info, sizeof(info));
1347
1348 info.maxcount = uap->vbufsize / sizeof(struct statvfs);
1349 info.sfsp = uap->buf;
1350 info.vsfsp = uap->vbuf;
1351 info.count = 0;
1352 info.flags = uap->flags;
1353 info.p = p;
1354
1355 mountlist_scan(getvfsstat_callback, &info, MNTSCAN_FORWARD);
1356 if (info.vsfsp && info.count > info.maxcount)
1357 uap->sysmsg_result = info.maxcount;
1358 else
1359 uap->sysmsg_result = info.count;
1360 return (info.error);
1361}
1362
1363static int
1364getvfsstat_callback(struct mount *mp, void *data)
1365{
1366 struct getvfsstat_info *info = data;
1367 struct statfs *sp;
1368 struct statvfs *vsp;
1369 char *freepath;
1370 char *fullpath;
1371 int error;
1372
1373 if (info->vsfsp && info->count < info->maxcount) {
1374 if (info->p && !chroot_visible_mnt(mp, info->p))
1375 return(0);
1376 sp = &mp->mnt_stat;
1377 vsp = &mp->mnt_vstat;
1378
1379 /*
1380 * If MNT_NOWAIT or MNT_LAZY is specified, do not
1381 * refresh the fsstat cache. MNT_NOWAIT or MNT_LAZY
1382 * overrides MNT_WAIT.
1383 */
1384 if (((info->flags & (MNT_LAZY|MNT_NOWAIT)) == 0 ||
1385 (info->flags & MNT_WAIT)) &&
1386 (error = VFS_STATFS(mp, sp, info->p->p_ucred))) {
1387 return(0);
1388 }
1389 sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
1390
1391 if (((info->flags & (MNT_LAZY|MNT_NOWAIT)) == 0 ||
1392 (info->flags & MNT_WAIT)) &&
1393 (error = VFS_STATVFS(mp, vsp, info->p->p_ucred))) {
1394 return(0);
1395 }
1396 vsp->f_flag = 0;
1397 if (mp->mnt_flag & MNT_RDONLY)
1398 vsp->f_flag |= ST_RDONLY;
1399 if (mp->mnt_flag & MNT_NOSUID)
1400 vsp->f_flag |= ST_NOSUID;
1401
1402 error = mount_path(info->p, mp, &fullpath, &freepath);
1403 if (error) {
1404 info->error = error;
1405 return(-1);
1406 }
1407 bzero(sp->f_mntonname, sizeof(sp->f_mntonname));
1408 strlcpy(sp->f_mntonname, fullpath, sizeof(sp->f_mntonname));
1409 kfree(freepath, M_TEMP);
1410
1411 error = copyout(sp, info->sfsp, sizeof(*sp));
1412 if (error == 0)
1413 error = copyout(vsp, info->vsfsp, sizeof(*vsp));
1414 if (error) {
1415 info->error = error;
1416 return (-1);
1417 }
1418 ++info->sfsp;
1419 ++info->vsfsp;
1420 }
1421 info->count++;
1422 return(0);
1423}
1424
1425
1426/*
41c20dac
MD
1427 * fchdir_args(int fd)
1428 *
984263bc
MD
1429 * Change current working directory to a given file descriptor.
1430 */
984263bc
MD
1431/* ARGSUSED */
1432int
753fd850 1433sys_fchdir(struct fchdir_args *uap)
984263bc 1434{
dadab5e9
MD
1435 struct thread *td = curthread;
1436 struct proc *p = td->td_proc;
41c20dac 1437 struct filedesc *fdp = p->p_fd;
21739618 1438 struct vnode *vp, *ovp;
984263bc
MD
1439 struct mount *mp;
1440 struct file *fp;
28623bf9 1441 struct nchandle nch, onch, tnch;
984263bc
MD
1442 int error;
1443
5b287bba 1444 if ((error = holdvnode(fdp, uap->fd, &fp)) != 0)
984263bc
MD
1445 return (error);
1446 vp = (struct vnode *)fp->f_data;
597aea93 1447 vref(vp);
ca466bae 1448 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
28623bf9 1449 if (vp->v_type != VDIR || fp->f_nchandle.ncp == NULL)
984263bc
MD
1450 error = ENOTDIR;
1451 else
87de5057 1452 error = VOP_ACCESS(vp, VEXEC, p->p_ucred);
984263bc
MD
1453 if (error) {
1454 vput(vp);
5b287bba 1455 fdrop(fp);
984263bc
MD
1456 return (error);
1457 }
28623bf9 1458 cache_copy(&fp->f_nchandle, &nch);
1d505369
MD
1459
1460 /*
1461 * If the ncp has become a mount point, traverse through
1462 * the mount point.
1463 */
1464
28623bf9
MD
1465 while (!error && (nch.ncp->nc_flag & NCF_ISMOUNTPT) &&
1466 (mp = cache_findmount(&nch)) != NULL
1d505369 1467 ) {
28623bf9 1468 error = nlookup_mp(mp, &tnch);
21739618 1469 if (error == 0) {
28623bf9 1470 cache_unlock(&tnch); /* leave ref intact */
21739618 1471 vput(vp);
28623bf9 1472 vp = tnch.ncp->nc_vp;
87de5057 1473 error = vget(vp, LK_SHARED);
21739618 1474 KKASSERT(error == 0);
28623bf9
MD
1475 cache_drop(&nch);
1476 nch = tnch;
21739618
MD
1477 }
1478 }
1479 if (error == 0) {
1480 ovp = fdp->fd_cdir;
28623bf9 1481 onch = fdp->fd_ncdir;
a11aaa81 1482 vn_unlock(vp); /* leave ref intact */
21739618 1483 fdp->fd_cdir = vp;
28623bf9
MD
1484 fdp->fd_ncdir = nch;
1485 cache_drop(&onch);
21739618
MD
1486 vrele(ovp);
1487 } else {
28623bf9 1488 cache_drop(&nch);
21739618
MD
1489 vput(vp);
1490 }
5b287bba 1491 fdrop(fp);
21739618 1492 return (error);
984263bc
MD
1493}
1494
9697c509 1495int
21739618 1496kern_chdir(struct nlookupdata *nd)
9697c509
DRJ
1497{
1498 struct thread *td = curthread;
1499 struct proc *p = td->td_proc;
1500 struct filedesc *fdp = p->p_fd;
21739618 1501 struct vnode *vp, *ovp;
28623bf9 1502 struct nchandle onch;
9697c509
DRJ
1503 int error;
1504
21739618 1505 if ((error = nlookup(nd)) != 0)
9697c509 1506 return (error);
28623bf9 1507 if ((vp = nd->nl_nch.ncp->nc_vp) == NULL)
21739618 1508 return (ENOENT);
87de5057 1509 if ((error = vget(vp, LK_SHARED)) != 0)
21739618
MD
1510 return (error);
1511
1512 error = checkvp_chdir(vp, td);
a11aaa81 1513 vn_unlock(vp);
21739618
MD
1514 if (error == 0) {
1515 ovp = fdp->fd_cdir;
28623bf9
MD
1516 onch = fdp->fd_ncdir;
1517 cache_unlock(&nd->nl_nch); /* leave reference intact */
1518 fdp->fd_ncdir = nd->nl_nch;
21739618 1519 fdp->fd_cdir = vp;
28623bf9 1520 cache_drop(&onch);
21739618 1521 vrele(ovp);
28623bf9 1522 cache_zero(&nd->nl_nch);
21739618
MD
1523 } else {
1524 vrele(vp);
e24b948e 1525 }
e24b948e 1526 return (error);
9697c509
DRJ
1527}
1528
984263bc 1529/*
41c20dac
MD
1530 * chdir_args(char *path)
1531 *
984263bc
MD
1532 * Change current working directory (``.'').
1533 */
984263bc 1534int
753fd850 1535sys_chdir(struct chdir_args *uap)
984263bc 1536{
21739618 1537 struct nlookupdata nd;
9697c509 1538 int error;
984263bc 1539
21739618 1540 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
fad57d0e 1541 if (error == 0)
21739618 1542 error = kern_chdir(&nd);
fad57d0e 1543 nlookup_done(&nd);
9697c509 1544 return (error);
984263bc
MD
1545}
1546
1547/*
1548 * Helper function for raised chroot(2) security function: Refuse if
1549 * any filedescriptors are open directories.
1550 */
1551static int
c972a82f 1552chroot_refuse_vdir_fds(struct filedesc *fdp)
984263bc
MD
1553{
1554 struct vnode *vp;
1555 struct file *fp;
1556 int error;
1557 int fd;
1558
1559 for (fd = 0; fd < fdp->fd_nfiles ; fd++) {
5b287bba 1560 if ((error = holdvnode(fdp, fd, &fp)) != 0)
984263bc
MD
1561 continue;
1562 vp = (struct vnode *)fp->f_data;
5b287bba
MD
1563 if (vp->v_type != VDIR) {
1564 fdrop(fp);
984263bc 1565 continue;
5b287bba
MD
1566 }
1567 fdrop(fp);
984263bc
MD
1568 return(EPERM);
1569 }
1570 return (0);
1571}
1572
1573/*
1574 * This sysctl determines if we will allow a process to chroot(2) if it
1575 * has a directory open:
1576 * 0: disallowed for all processes.
1577 * 1: allowed for processes that were not already chroot(2)'ed.
1578 * 2: allowed for all processes.
1579 */
1580
1581static int chroot_allow_open_directories = 1;
1582
1583SYSCTL_INT(_kern, OID_AUTO, chroot_allow_open_directories, CTLFLAG_RW,
1584 &chroot_allow_open_directories, 0, "");
1585
1586/*
21739618
MD
1587 * chroot to the specified namecache entry. We obtain the vp from the
1588 * namecache data. The passed ncp must be locked and referenced and will
1589 * remain locked and referenced on return.
e24b948e 1590 */
53dd6631 1591int
28623bf9 1592kern_chroot(struct nchandle *nch)
e24b948e
MD
1593{
1594 struct thread *td = curthread;
1595 struct proc *p = td->td_proc;
1596 struct filedesc *fdp = p->p_fd;
21739618 1597 struct vnode *vp;
e24b948e
MD
1598 int error;
1599
1600 /*
ca3cd02d 1601 * Only privileged user can chroot
e24b948e 1602 */
3a591c90
MN
1603 error = priv_check_cred(p->p_ucred, PRIV_VFS_CHROOT, 0);
1604 if (error)
e24b948e
MD
1605 return (error);
1606
1607 /*
1608 * Disallow open directory descriptors (fchdir() breakouts).
1609 */
1610 if (chroot_allow_open_directories == 0 ||
1611 (chroot_allow_open_directories == 1 && fdp->fd_rdir != rootvnode)) {
1612 if ((error = chroot_refuse_vdir_fds(fdp)) != 0)
1613 return (error);
1614 }
28623bf9 1615 if ((vp = nch->ncp->nc_vp) == NULL)
21739618
MD
1616 return (ENOENT);
1617
87de5057 1618 if ((error = vget(vp, LK_SHARED)) != 0)
21739618 1619 return (error);
e24b948e
MD
1620
1621 /*
1622 * Check the validity of vp as a directory to change to and
1623 * associate it with rdir/jdir.
1624 */
21739618 1625 error = checkvp_chdir(vp, td);
a11aaa81 1626 vn_unlock(vp); /* leave reference intact */
21739618 1627 if (error == 0) {
e24b948e 1628 vrele(fdp->fd_rdir);
21739618 1629 fdp->fd_rdir = vp; /* reference inherited by fd_rdir */
28623bf9
MD
1630 cache_drop(&fdp->fd_nrdir);
1631 cache_copy(nch, &fdp->fd_nrdir);
e24b948e
MD
1632 if (fdp->fd_jdir == NULL) {
1633 fdp->fd_jdir = vp;
597aea93 1634 vref(fdp->fd_jdir);
28623bf9 1635 cache_copy(nch, &fdp->fd_njdir);
e24b948e 1636 }
21739618
MD
1637 } else {
1638 vrele(vp);
e24b948e
MD
1639 }
1640 return (error);
1641}
1642
1643/*
41c20dac
MD
1644 * chroot_args(char *path)
1645 *
984263bc
MD
1646 * Change notion of root (``/'') directory.
1647 */
984263bc
MD
1648/* ARGSUSED */
1649int
753fd850 1650sys_chroot(struct chroot_args *uap)
984263bc 1651{
dadab5e9 1652 struct thread *td = curthread;
21739618 1653 struct nlookupdata nd;
e24b948e 1654 int error;
984263bc 1655
e24b948e 1656 KKASSERT(td->td_proc);
21739618 1657 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
53dd6631
JS
1658 if (error) {
1659 nlookup_done(&nd);
1660 return(error);
1661 }
3a907475 1662 nd.nl_flags |= NLC_EXEC;
53dd6631
JS
1663 error = nlookup(&nd);
1664 if (error == 0)
28623bf9 1665 error = kern_chroot(&nd.nl_nch);
fad57d0e 1666 nlookup_done(&nd);
53dd6631 1667 return(error);
984263bc
MD
1668}
1669
1670/*
e24b948e
MD
1671 * Common routine for chroot and chdir. Given a locked, referenced vnode,
1672 * determine whether it is legal to chdir to the vnode. The vnode's state
1673 * is not changed by this call.
984263bc 1674 */
e24b948e
MD
1675int
1676checkvp_chdir(struct vnode *vp, struct thread *td)
984263bc 1677{
984263bc
MD
1678 int error;
1679
984263bc
MD
1680 if (vp->v_type != VDIR)
1681 error = ENOTDIR;
1682 else
87de5057 1683 error = VOP_ACCESS(vp, VEXEC, td->td_proc->p_ucred);
984263bc
MD
1684 return (error);
1685}
1686
984263bc 1687int
fad57d0e 1688kern_open(struct nlookupdata *nd, int oflags, int mode, int *res)
984263bc 1689{
dadab5e9
MD
1690 struct thread *td = curthread;
1691 struct proc *p = td->td_proc;
fbfe4e7d 1692 struct lwp *lp = td->td_lwp;
41c20dac 1693 struct filedesc *fdp = p->p_fd;
9697c509 1694 int cmode, flags;
984263bc 1695 struct file *nfp;
fad57d0e
MD
1696 struct file *fp;
1697 struct vnode *vp;
984263bc
MD
1698 int type, indx, error;
1699 struct flock lf;
984263bc 1700
984263bc
MD
1701 if ((oflags & O_ACCMODE) == O_ACCMODE)
1702 return (EINVAL);
1703 flags = FFLAGS(oflags);
fad57d0e 1704 error = falloc(p, &nfp, NULL);
984263bc
MD
1705 if (error)
1706 return (error);
1707 fp = nfp;
3a907475 1708 cmode = ((mode &~ fdp->fd_cmask) & ALLPERMS) & ~S_ISTXT;
fad57d0e 1709
984263bc 1710 /*
fad57d0e
MD
1711 * XXX p_dupfd is a real mess. It allows a device to return a
1712 * file descriptor to be duplicated rather then doing the open
1713 * itself.
984263bc 1714 */
fbfe4e7d 1715 lp->lwp_dupfd = -1;
21739618
MD
1716
1717 /*
fad57d0e
MD
1718 * Call vn_open() to do the lookup and assign the vnode to the
1719 * file pointer. vn_open() does not change the ref count on fp
1720 * and the vnode, on success, will be inherited by the file pointer
1721 * and unlocked.
21739618 1722 */
fad57d0e
MD
1723 nd->nl_flags |= NLC_LOCKVP;
1724 error = vn_open(nd, fp, flags, cmode);
1725 nlookup_done(nd);
984263bc
MD
1726 if (error) {
1727 /*
984263bc
MD
1728 * handle special fdopen() case. bleh. dupfdopen() is
1729 * responsible for dropping the old contents of ofiles[indx]
1730 * if it succeeds.
fad57d0e 1731 *
259b8ea0
MD
1732 * Note that fsetfd() will add a ref to fp which represents
1733 * the fd_files[] assignment. We must still drop our
1734 * reference.
984263bc 1735 */
fbfe4e7d 1736 if ((error == ENODEV || error == ENXIO) && lp->lwp_dupfd >= 0) {
259b8ea0
MD
1737 if (fdalloc(p, 0, &indx) == 0) {
1738 error = dupfdopen(p, indx, lp->lwp_dupfd, flags, error);
fad57d0e
MD
1739 if (error == 0) {
1740 *res = indx;
9f87144f 1741 fdrop(fp); /* our ref */
fad57d0e
MD
1742 return (0);
1743 }
259b8ea0 1744 fsetfd(p, NULL, indx);
fad57d0e 1745 }
984263bc 1746 }
9f87144f 1747 fdrop(fp); /* our ref */
984263bc
MD
1748 if (error == ERESTART)
1749 error = EINTR;
1750 return (error);
1751 }
fad57d0e
MD
1752
1753 /*
1754 * ref the vnode for ourselves so it can't be ripped out from under
1755 * is. XXX need an ND flag to request that the vnode be returned
1756 * anyway.
259b8ea0
MD
1757 *
1758 * Reserve a file descriptor but do not assign it until the open
1759 * succeeds.
fad57d0e
MD
1760 */
1761 vp = (struct vnode *)fp->f_data;
1762 vref(vp);
259b8ea0 1763 if ((error = fdalloc(p, 0, &indx)) != 0) {
9f87144f 1764 fdrop(fp);
fad57d0e
MD
1765 vrele(vp);
1766 return (error);
1767 }
1768
1769 /*
1770 * If no error occurs the vp will have been assigned to the file
1771 * pointer.
1772 */
fbfe4e7d 1773 lp->lwp_dupfd = 0;
984263bc 1774
984263bc
MD
1775 if (flags & (O_EXLOCK | O_SHLOCK)) {
1776 lf.l_whence = SEEK_SET;
1777 lf.l_start = 0;
1778 lf.l_len = 0;
1779 if (flags & O_EXLOCK)
1780 lf.l_type = F_WRLCK;
1781 else
1782 lf.l_type = F_RDLCK;
71c18fe3
MD
1783 if (flags & FNONBLOCK)
1784 type = 0;
1785 else
1786 type = F_WAIT;
fad57d0e 1787
984263bc
MD
1788 if ((error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type)) != 0) {
1789 /*
259b8ea0
MD
1790 * lock request failed. Clean up the reserved
1791 * descriptor.
984263bc 1792 */
fad57d0e 1793 vrele(vp);
259b8ea0 1794 fsetfd(p, NULL, indx);
9f87144f 1795 fdrop(fp);
984263bc
MD
1796 return (error);
1797 }
984263bc
MD
1798 fp->f_flag |= FHASLOCK;
1799 }
7540ab49
MD
1800#if 0
1801 /*
1802 * Assert that all regular file vnodes were created with a object.
1803 */
1804 KASSERT(vp->v_type != VREG || vp->v_object != NULL,
1805 ("open: regular file has no backing object after vn_open"));
1806#endif
984263bc 1807
fad57d0e 1808 vrele(vp);
21739618
MD
1809
1810 /*
984263bc
MD
1811 * release our private reference, leaving the one associated with the
1812 * descriptor table intact.
1813 */
259b8ea0 1814 fsetfd(p, fp, indx);
9f87144f 1815 fdrop(fp);
9697c509 1816 *res = indx;
984263bc
MD
1817 return (0);
1818}
1819
984263bc 1820/*
9697c509 1821 * open_args(char *path, int flags, int mode)
41c20dac 1822 *
9697c509
DRJ
1823 * Check permissions, allocate an open file structure,
1824 * and call the device open routine if any.
984263bc 1825 */
984263bc 1826int
753fd850 1827sys_open(struct open_args *uap)
984263bc 1828{
fad57d0e 1829 struct nlookupdata nd;
9697c509
DRJ
1830 int error;
1831
85dbab7f 1832 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
fad57d0e
MD
1833 if (error == 0) {
1834 error = kern_open(&nd, uap->flags,
1835 uap->mode, &uap->sysmsg_result);
1836 }
1837 nlookup_done(&nd);
9697c509 1838 return (error);
984263bc 1839}
984263bc 1840
83a11774
NT
1841/*
1842 * openat_args(int fd, char *path, int flags, int mode)
1843 */
1844int
1845sys_openat(struct openat_args *uap)
1846{
83a11774
NT
1847 struct nlookupdata nd;
1848 int error;
1db695af 1849 struct file *fp;
83a11774 1850
1db695af
NT
1851 error = nlookup_init_at(&nd, &fp, uap->fd, uap->path, UIO_USERSPACE, 0);
1852 if (error == 0) {
1853 error = kern_open(&nd, uap->flags, uap->mode,
1854 &uap->sysmsg_result);
83a11774 1855 }
1db695af 1856 nlookup_done_at(&nd, fp);
83a11774
NT
1857 return (error);
1858}
1859
984263bc 1860int
0e9b9130 1861kern_mknod(struct nlookupdata *nd, int mode, int rmajor, int rminor)
984263bc 1862{
dadab5e9
MD
1863 struct thread *td = curthread;
1864 struct proc *p = td->td_proc;
41c20dac 1865 struct vnode *vp;
984263bc
MD
1866 struct vattr vattr;
1867 int error;
1868 int whiteout = 0;
984263bc 1869
dadab5e9
MD
1870 KKASSERT(p);
1871
fad57d0e
MD
1872 VATTR_NULL(&vattr);
1873 vattr.va_mode = (mode & ALLPERMS) &~ p->p_fd->fd_cmask;
0e9b9130
MD
1874 vattr.va_rmajor = rmajor;
1875 vattr.va_rminor = rminor;
fad57d0e
MD
1876
1877 switch (mode & S_IFMT) {
1878 case S_IFMT: /* used by badsect to flag bad sectors */
3a591c90 1879 error = priv_check_cred(p->p_ucred, PRIV_VFS_MKNOD_BAD, 0);
fad57d0e
MD
1880 vattr.va_type = VBAD;
1881 break;
1882 case S_IFCHR:
44a36778 1883 error = priv_check(td, PRIV_VFS_MKNOD_DEV);
fad57d0e
MD
1884 vattr.va_type = VCHR;
1885 break;
1886 case S_IFBLK:
44a36778 1887 error = priv_check(td, PRIV_VFS_MKNOD_DEV);
fad57d0e
MD
1888 vattr.va_type = VBLK;
1889 break;
1890 case S_IFWHT:
3a591c90 1891 error = priv_check_cred(p->p_ucred, PRIV_VFS_MKNOD_WHT, 0);
fad57d0e
MD
1892 whiteout = 1;
1893 break;
44a36778 1894 case S_IFDIR: /* special directories support for HAMMER */
3a591c90 1895 error = priv_check_cred(p->p_ucred, PRIV_VFS_MKNOD_DIR, 0);
bed9e372
MD
1896 vattr.va_type = VDIR;
1897 break;
fad57d0e
MD
1898 default:
1899 error = EINVAL;
1900 break;
984263bc 1901 }
44a36778
MN
1902
1903 if (error)
1904 return (error);
1905
1906 bwillinode(1);
1907 nd->nl_flags |= NLC_CREATE | NLC_REFDVP;
1908 if ((error = nlookup(nd)) != 0)
1909 return (error);
1910 if (nd->nl_nch.ncp->nc_vp)
1911 return (EEXIST);
1912 if ((error = ncp_writechk(&nd->nl_nch)) != 0)
1913 return (error);
1914
1915 if (whiteout) {
1916 error = VOP_NWHITEOUT(&nd->nl_nch, nd->nl_dvp,
1917 nd->nl_cred, NAMEI_CREATE);
1918 } else {
1919 vp = NULL;
1920 error = VOP_NMKNOD(&nd->nl_nch, nd->nl_dvp,
1921 &vp, nd->nl_cred, &vattr);
1922 if (error == 0)
1923 vput(vp);
984263bc 1924 }
9697c509
DRJ
1925 return (error);
1926}
1927
1928/*
1929 * mknod_args(char *path, int mode, int dev)
1930 *
1931 * Create a special file.
1932 */
1933int
753fd850 1934sys_mknod(struct mknod_args *uap)
9697c509 1935{
fad57d0e 1936 struct nlookupdata nd;
9697c509
DRJ
1937 int error;
1938
fad57d0e 1939 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
0e9b9130
MD
1940 if (error == 0) {
1941 error = kern_mknod(&nd, uap->mode,
1942 umajor(uap->dev), uminor(uap->dev));
1943 }
fad57d0e 1944 nlookup_done(&nd);
984263bc
MD
1945 return (error);
1946}
1947
984263bc 1948int
fad57d0e 1949kern_mkfifo(struct nlookupdata *nd, int mode)
984263bc 1950{
dadab5e9
MD
1951 struct thread *td = curthread;
1952 struct proc *p = td->td_proc;
984263bc 1953 struct vattr vattr;
fad57d0e 1954 struct vnode *vp;
984263bc 1955 int error;
984263bc 1956
c4df9635 1957 bwillinode(1);
fad57d0e 1958
5312fa43 1959 nd->nl_flags |= NLC_CREATE | NLC_REFDVP;
fad57d0e 1960 if ((error = nlookup(nd)) != 0)
984263bc 1961 return (error);
28623bf9 1962 if (nd->nl_nch.ncp->nc_vp)
984263bc 1963 return (EEXIST);
28623bf9 1964 if ((error = ncp_writechk(&nd->nl_nch)) != 0)
72310cfb 1965 return (error);
fad57d0e 1966
984263bc
MD
1967 VATTR_NULL(&vattr);
1968 vattr.va_type = VFIFO;
136178b3 1969 vattr.va_mode = (mode & ALLPERMS) &~ p->p_fd->fd_cmask;
fad57d0e 1970 vp = NULL;
5312fa43 1971 error = VOP_NMKNOD(&nd->nl_nch, nd->nl_dvp, &vp, nd->nl_cred, &vattr);
984263bc 1972 if (error == 0)
fad57d0e 1973 vput(vp);
136178b3
DRJ
1974 return (error);
1975}
1976
1977/*
1978 * mkfifo_args(char *path, int mode)
1979 *
1980 * Create a named pipe.
1981 */
1982int
753fd850 1983sys_mkfifo(struct mkfifo_args *uap)
136178b3 1984{
fad57d0e 1985 struct nlookupdata nd;
136178b3
DRJ
1986 int error;
1987
fad57d0e
MD
1988 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
1989 if (error == 0)
1990 error = kern_mkfifo(&nd, uap->mode);
1991 nlookup_done(&nd);
984263bc
MD
1992 return (error);
1993}
1994
8c40f81e 1995static int hardlink_check_uid = 0;
43a0f7ae 1996SYSCTL_INT(_security, OID_AUTO, hardlink_check_uid, CTLFLAG_RW,
8c40f81e
MD
1997 &hardlink_check_uid, 0,
1998 "Unprivileged processes cannot create hard links to files owned by other "
1999 "users");
2000static int hardlink_check_gid = 0;
43a0f7ae 2001SYSCTL_INT(_security, OID_AUTO, hardlink_check_gid, CTLFLAG_RW,
8c40f81e
MD
2002 &hardlink_check_gid, 0,
2003 "Unprivileged processes cannot create hard links to files owned by other "
2004 "groups");
2005
2006static int
2007can_hardlink(struct vnode *vp, struct thread *td, struct ucred *cred)
2008{
2009 struct vattr va;
2010 int error;
2011
2012 /*
2013 * Shortcut if disabled
2014 */
2015 if (hardlink_check_uid == 0 && hardlink_check_gid == 0)
2016 return (0);
2017
2018 /*
f6df0641 2019 * Privileged user can always hardlink
8c40f81e 2020 */
f6df0641 2021 if (priv_check_cred(cred, PRIV_VFS_LINK, 0) == 0)
8c40f81e
MD
2022 return (0);
2023
2024 /*
2025 * Otherwise only if the originating file is owned by the
2026 * same user or group. Note that any group is allowed if
2027 * the file is owned by the caller.
2028 */
87de5057 2029 error = VOP_GETATTR(vp, &va);
8c40f81e
MD
2030 if (error != 0)
2031 return (error);
2032
2033 if (hardlink_check_uid) {
2034 if (cred->cr_uid != va.va_uid)
2035 return (EPERM);
2036 }
2037
2038 if (hardlink_check_gid) {
2039 if (cred->cr_uid != va.va_uid && !groupmember(va.va_gid, cred))
2040 return (EPERM);
2041 }
2042
2043 return (0);
2044}
2045
984263bc 2046int
fad57d0e 2047kern_link(struct nlookupdata *nd, struct nlookupdata *linknd)
984263bc 2048{
dadab5e9 2049 struct thread *td = curthread;
41c20dac 2050 struct vnode *vp;
984263bc
MD
2051 int error;
2052
fad57d0e
MD
2053 /*
2054 * Lookup the source and obtained a locked vnode.
2055 *
3a907475
MD
2056 * You may only hardlink a file which you have write permission
2057 * on or which you own.
2058 *
fad57d0e
MD
2059 * XXX relookup on vget failure / race ?
2060 */
c4df9635 2061 bwillinode(1);
3a907475 2062 nd->nl_flags |= NLC_WRITE | NLC_OWN | NLC_HLINK;
fad57d0e 2063 if ((error = nlookup(nd)) != 0)
984263bc 2064 return (error);
28623bf9 2065 vp = nd->nl_nch.ncp->nc_vp;
fad57d0e 2066 KKASSERT(vp != NULL);
984263bc 2067 if (vp->v_type == VDIR)
fad57d0e 2068 return (EPERM); /* POSIX */
28623bf9 2069 if ((error = ncp_writechk(&nd->nl_nch)) != 0)
72310cfb 2070 return (error);
87de5057 2071 if ((error = vget(vp, LK_EXCLUSIVE)) != 0)
fad57d0e
MD
2072 return (error);
2073
2074 /*
2075 * Unlock the source so we can lookup the target without deadlocking
2076 * (XXX vp is locked already, possible other deadlock?). The target
2077 * must not exist.
2078 */
2079 KKASSERT(nd->nl_flags & NLC_NCPISLOCKED);
2080 nd->nl_flags &= ~NLC_NCPISLOCKED;
28623bf9 2081 cache_unlock(&nd->nl_nch);
fad57d0e 2082
5312fa43 2083 linknd->nl_flags |= NLC_CREATE | NLC_REFDVP;
fad57d0e
MD
2084 if ((error = nlookup(linknd)) != 0) {
2085 vput(vp);
2086 return (error);
984263bc 2087 }
28623bf9 2088 if (linknd->nl_nch.ncp->nc_vp) {
fad57d0e
MD
2089 vput(vp);
2090 return (EEXIST);
2091 }
2092
2093 /*
2094 * Finally run the new API VOP.
2095 */
8c40f81e 2096 error = can_hardlink(vp, td, td->td_proc->p_ucred);
5312fa43
MD
2097 if (error == 0) {
2098 error = VOP_NLINK(&linknd->nl_nch, linknd->nl_dvp,
2099 vp, linknd->nl_cred);
2100 }
fad57d0e 2101 vput(vp);
984263bc
MD
2102 return (error);
2103}
2104
2105/*
9697c509 2106 * link_args(char *path, char *link)
41c20dac 2107 *
9697c509 2108 * Make a hard file link.
984263bc 2109 */
984263bc 2110int
753fd850 2111sys_link(struct link_args *uap)
9697c509 2112{
fad57d0e 2113 struct nlookupdata nd, linknd;
9697c509
DRJ
2114 int error;
2115
fad57d0e
MD
2116 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
2117 if (error == 0) {
2118 error = nlookup_init(&linknd, uap->link, UIO_USERSPACE, 0);
2119 if (error == 0)
2120 error = kern_link(&nd, &linknd);
2121 nlookup_done(&linknd);
2122 }
2123 nlookup_done(&nd);
9697c509
DRJ
2124 return (error);
2125}
2126
2127int
fad57d0e 2128kern_symlink(struct nlookupdata *nd, char *path, int mode)
984263bc
MD
2129{
2130 struct vattr vattr;
fad57d0e 2131 struct vnode *vp;
dff430ab 2132 struct vnode *dvp;
984263bc 2133 int error;
984263bc 2134
c4df9635 2135 bwillinode(1);
5312fa43 2136 nd->nl_flags |= NLC_CREATE | NLC_REFDVP;
fad57d0e 2137 if ((error = nlookup(nd)) != 0)
9697c509 2138 return (error);
28623bf9 2139 if (nd->nl_nch.ncp->nc_vp)
9697c509 2140 return (EEXIST);
28623bf9 2141 if ((error = ncp_writechk(&nd->nl_nch)) != 0)
72310cfb 2142 return (error);
5312fa43 2143 dvp = nd->nl_dvp;
984263bc 2144 VATTR_NULL(&vattr);
fad57d0e 2145 vattr.va_mode = mode;
dff430ab 2146 error = VOP_NSYMLINK(&nd->nl_nch, dvp, &vp, nd->nl_cred, &vattr, path);
984263bc 2147 if (error == 0)
fad57d0e 2148 vput(vp);
9697c509
DRJ
2149 return (error);
2150}
2151
2152/*
2153 * symlink(char *path, char *link)
2154 *
2155 * Make a symbolic link.
2156 */
2157int
753fd850 2158sys_symlink(struct symlink_args *uap)
9697c509
DRJ
2159{
2160 struct thread *td = curthread;
fad57d0e 2161 struct nlookupdata nd;
9697c509
DRJ
2162 char *path;
2163 int error;
fad57d0e 2164 int mode;
9697c509 2165
70aac194 2166 path = objcache_get(namei_oc, M_WAITOK);
9697c509 2167 error = copyinstr(uap->path, path, MAXPATHLEN, NULL);
e24b948e 2168 if (error == 0) {
fad57d0e
MD
2169 error = nlookup_init(&nd, uap->link, UIO_USERSPACE, 0);
2170 if (error == 0) {
2171 mode = ACCESSPERMS & ~td->td_proc->p_fd->fd_cmask;
2172 error = kern_symlink(&nd, path, mode);
2173 }
2174 nlookup_done(&nd);
e24b948e 2175 }
70aac194 2176 objcache_put(namei_oc, path);
984263bc
MD
2177 return (error);
2178}
2179
2180/*
41c20dac
MD
2181 * undelete_args(char *path)
2182 *
984263bc
MD
2183 * Delete a whiteout from the filesystem.
2184 */
2185/* ARGSUSED */
2186int
753fd850 2187sys_undelete(struct undelete_args *uap)
984263bc 2188{
fad57d0e 2189 struct nlookupdata nd;
984263bc 2190 int error;
984263bc 2191
ab2eb4eb 2192 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
c4df9635 2193 bwillinode(1);
5312fa43 2194 nd.nl_flags |= NLC_DELETE | NLC_REFDVP;
fad57d0e
MD
2195 if (error == 0)
2196 error = nlookup(&nd);
b8997912 2197 if (error == 0)
28623bf9 2198 error = ncp_writechk(&nd.nl_nch);
dff430ab 2199 if (error == 0) {
5312fa43
MD
2200 error = VOP_NWHITEOUT(&nd.nl_nch, nd.nl_dvp, nd.nl_cred,
2201 NAMEI_DELETE);
dff430ab 2202 }
fad57d0e 2203 nlookup_done(&nd);
984263bc
MD
2204 return (error);
2205}
2206
984263bc 2207int
fad57d0e 2208kern_unlink(struct nlookupdata *nd)
984263bc 2209{
984263bc 2210 int error;
984263bc 2211
c4df9635 2212 bwillinode(1);
5312fa43 2213 nd->nl_flags |= NLC_DELETE | NLC_REFDVP;
fad57d0e 2214 if ((error = nlookup(nd)) != 0)
984263bc 2215 return (error);
28623bf9 2216 if ((error = ncp_writechk(&nd->nl_nch)) != 0)
72310cfb 2217 return (error);
5312fa43 2218 error = VOP_NREMOVE(&nd->nl_nch, nd->nl_dvp, nd->nl_cred);
984263bc
MD
2219 return (error);
2220}
2221
2222/*
9697c509 2223 * unlink_args(char *path)
41c20dac 2224 *
9697c509 2225 * Delete a name from the filesystem.
984263bc 2226 */
984263bc 2227int
753fd850 2228sys_unlink(struct unlink_args *uap)
9697c509 2229{
fad57d0e 2230 struct nlookupdata nd;
9697c509
DRJ
2231 int error;
2232
fad57d0e
MD
2233 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
2234 if (error == 0)
2235 error = kern_unlink(&nd);
2236 nlookup_done(&nd);
9697c509
DRJ
2237 return (error);
2238}
2239
2240int
5a3fe67d 2241kern_lseek(int fd, off_t offset, int whence, off_t *res)
984263bc 2242{
dadab5e9
MD
2243 struct thread *td = curthread;
2244 struct proc *p = td->td_proc;
41c20dac 2245 struct file *fp;
116edfbf 2246 struct vnode *vp;
984263bc 2247 struct vattr vattr;
116edfbf 2248 off_t new_offset;
984263bc
MD
2249 int error;
2250
228b401d
MD
2251 fp = holdfp(p->p_fd, fd, -1);
2252 if (fp == NULL)
984263bc 2253 return (EBADF);
228b401d
MD
2254 if (fp->f_type != DTYPE_VNODE) {
2255 error = ESPIPE;
2256 goto done;
2257 }
116edfbf 2258 vp = (struct vnode *)fp->f_data;
228b401d 2259
9697c509 2260 switch (whence) {
984263bc 2261 case L_INCR:
116edfbf 2262 new_offset = fp->f_offset + offset;
228b401d 2263 error = 0;
984263bc
MD
2264 break;
2265 case L_XTND:
116edfbf
MD
2266 error = VOP_GETATTR(vp, &vattr);
2267 new_offset = offset + vattr.va_size;
984263bc
MD
2268 break;
2269 case L_SET:
116edfbf 2270 new_offset = offset;
228b401d 2271 error = 0;
984263bc
MD
2272 break;
2273 default:
116edfbf 2274 new_offset = 0;
228b401d
MD
2275 error = EINVAL;
2276 break;
984263bc 2277 }
116edfbf
MD
2278
2279 /*
2280 * Validate the seek position. Negative offsets are not allowed
58675cb0
MD
2281 * for regular files or directories.
2282 *
2283 * Normally we would also not want to allow negative offsets for
2284 * character and block-special devices. However kvm addresses
2285 * on 64 bit architectures might appear to be negative and must
2286 * be allowed.
116edfbf
MD
2287 */
2288 if (error == 0) {
2289 if (new_offset < 0 &&
58675cb0 2290 (vp->v_type == VREG || vp->v_type == VDIR)) {
116edfbf 2291 error = EINVAL;
b06b3b87
MD
2292 } else {
2293 fp->f_offset = new_offset;
116edfbf 2294 }
116edfbf 2295 }
9697c509 2296 *res = fp->f_offset;
228b401d
MD
2297done:
2298 fdrop(fp);
2299 return (error);
984263bc
MD
2300}
2301
984263bc 2302/*
9697c509 2303 * lseek_args(int fd, int pad, off_t offset, int whence)
41c20dac 2304 *
9697c509 2305 * Reposition read/write file offset.
984263bc 2306 */
984263bc 2307int
753fd850 2308sys_lseek(struct lseek_args *uap)
984263bc 2309{
984263bc
MD
2310 int error;
2311
9697c509 2312 error = kern_lseek(uap->fd, uap->offset, uap->whence,
116edfbf 2313 &uap->sysmsg_offset);
9697c509 2314
984263bc
MD
2315 return (error);
2316}
984263bc 2317
984263bc 2318int
fad57d0e 2319kern_access(struct nlookupdata *nd, int aflags)
984263bc 2320{
41c20dac 2321 struct vnode *vp;
984263bc 2322 int error, flags;
984263bc 2323
fad57d0e
MD
2324 if ((error = nlookup(nd)) != 0)
2325 return (error);
806dcf9a 2326retry:
28623bf9 2327 error = cache_vget(&nd->nl_nch, nd->nl_cred, LK_EXCLUSIVE, &vp);
9697c509 2328 if (error)
fad57d0e 2329 return (error);
984263bc
MD
2330
2331 /* Flags == 0 means only check for existence. */
9697c509 2332 if (aflags) {
984263bc 2333 flags = 0;
9697c509 2334 if (aflags & R_OK)
984263bc 2335 flags |= VREAD;
9697c509 2336 if (aflags & W_OK)
984263bc 2337 flags |= VWRITE;
9697c509 2338 if (aflags & X_OK)
984263bc 2339 flags |= VEXEC;
468bb1f9 2340 if ((flags & VWRITE) == 0 ||
28623bf9 2341 (error = vn_writechk(vp, &nd->nl_nch)) == 0)
87de5057 2342 error = VOP_ACCESS(vp, flags, nd->nl_cred);
806dcf9a
MD
2343
2344 /*
2345 * If the file handle is stale we have to re-resolve the
2346 * entry. This is a hack at the moment.
2347 */
2348 if (error == ESTALE) {
5e1a76ea 2349 vput(vp);
28623bf9
MD
2350 cache_setunresolved(&nd->nl_nch);
2351 error = cache_resolve(&nd->nl_nch, nd->nl_cred);
806dcf9a 2352 if (error == 0) {
806dcf9a
MD
2353 vp = NULL;
2354 goto retry;
2355 }
5e1a76ea 2356 return(error);
806dcf9a 2357 }
984263bc 2358 }
984263bc 2359 vput(vp);
984263bc
MD
2360 return (error);
2361}
2362
984263bc 2363/*
9697c509 2364 * access_args(char *path, int flags)
41c20dac 2365 *
9697c509 2366 * Check access permissions.
984263bc 2367 */
984263bc 2368int
753fd850 2369sys_access(struct access_args *uap)
984263bc 2370{
fad57d0e 2371 struct nlookupdata nd;
9697c509 2372 int error;
984263bc 2373
fad57d0e
MD
2374 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
2375 if (error == 0)
2376 error = kern_access(&nd, uap->flags);
2377 nlookup_done(&nd);
984263bc
MD
2378 return (error);
2379}
2380
984263bc 2381int
21739618 2382kern_stat(struct nlookupdata *nd, struct stat *st)
984263bc 2383{
984263bc 2384 int error;
21739618
MD
2385 struct vnode *vp;
2386 thread_t td;
984263bc 2387
21739618 2388 if ((error = nlookup(nd)) != 0)
984263bc 2389 return (error);
806dcf9a 2390again:
28623bf9 2391 if ((vp = nd->nl_nch.ncp->nc_vp) == NULL)
21739618
MD
2392 return (ENOENT);
2393
2394 td = curthread;
87de5057 2395 if ((error = vget(vp, LK_SHARED)) != 0)
21739618 2396 return (error);
87de5057 2397 error = vn_stat(vp, st, nd->nl_cred);
806dcf9a
MD
2398
2399 /*
2400 * If the file handle is stale we have to re-resolve the entry. This
2401 * is a hack at the moment.
2402 */
2403 if (error == ESTALE) {
5dc91765 2404 vput(vp);
28623bf9
MD
2405 cache_setunresolved(&nd->nl_nch);
2406 error = cache_resolve(&nd->nl_nch, nd->nl_cred);
5dc91765 2407 if (error == 0)
806dcf9a 2408 goto again;
5dc91765
MD
2409 } else {
2410 vput(vp);
806dcf9a 2411 }
984263bc
MD
2412 return (error);
2413}
2414
2415/*
41c20dac
MD
2416 * stat_args(char *path, struct stat *ub)
2417 *
984263bc
MD
2418 * Get file status; this version follows links.
2419 */
984263bc 2420int
753fd850 2421sys_stat(struct stat_args *uap)
984263bc 2422{
21739618 2423 struct nlookupdata nd;
9697c509
DRJ
2424 struct stat st;
2425 int error;
984263bc 2426
21739618
MD
2427 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
2428 if (error == 0) {
2429 error = kern_stat(&nd, &st);
2430 if (error == 0)
2431 error = copyout(&st, uap->ub, sizeof(*uap->ub));
21739618 2432 }
fad57d0e 2433 nlookup_done(&nd);
984263bc
MD
2434 return (error);
2435}
2436
2437/*
41c20dac
MD
2438 * lstat_args(char *path, struct stat *ub)
2439 *
984263bc
MD
2440 * Get file status; this version does not follow links.
2441 */
984263bc 2442int
753fd850 2443sys_lstat(struct lstat_args *uap)
984263bc 2444{
21739618 2445 struct nlookupdata nd;
9697c509
DRJ
2446 struct stat st;
2447 int error;
984263bc 2448
21739618
MD
2449 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
2450 if (error == 0) {
2451 error = kern_stat(&nd, &st);
2452 if (error == 0)
2453 error = copyout(&st, uap->ub, sizeof(*uap->ub));
21739618 2454 }
fad57d0e 2455 nlookup_done(&nd);
984263bc
MD
2456 return (error);
2457}
2458
984263bc 2459/*
e101a3f4
NT
2460 * fstatat_args(int fd, char *path, struct stat *sb, int flags)
2461 *
2462 * Get status of file pointed to by fd/path.
2463 */
2464int
2465sys_fstatat(struct fstatat_args *uap)
2466{
2467 struct nlookupdata nd;
2468 struct stat st;
2469 int error;
2470 int flags;
2471 struct file *fp;
2472
2473 if (uap->flags & ~_AT_SYMLINK_MASK)
2474 return (EINVAL);
2475
2476 flags = (uap->flags & AT_SYMLINK_NOFOLLOW) ? 0 : NLC_FOLLOW;
2477
2478 error = nlookup_init_at(&nd, &fp, uap->fd, uap->path,
2479 UIO_USERSPACE, flags);
2480 if (error == 0) {
2481 error = kern_stat(&nd, &st);
2482 if (error == 0)
2483 error = copyout(&st, uap->sb, sizeof(*uap->sb));
2484 }
2485 nlookup_done_at(&nd, fp);
2486 return (error);
2487}
2488
2489/*
41c20dac
MD
2490 * pathconf_Args(char *path, int name)
2491 *
984263bc
MD
2492 * Get configurable pathname variables.
2493 */
984263bc
MD
2494/* ARGSUSED */
2495int
753fd850 2496sys_pathconf(struct pathconf_args *uap)
984263bc 2497{
fad57d0e
MD
2498 struct nlookupdata nd;
2499 struct vnode *vp;
984263bc 2500 int error;
984263bc 2501
fad57d0e 2502 vp = NULL;
ab2eb4eb 2503 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
fad57d0e
MD
2504 if (error == 0)
2505 error = nlookup(&nd);
2506 if (error == 0)
28623bf9 2507 error = cache_vget(&nd.nl_nch, nd.nl_cred, LK_EXCLUSIVE, &vp);
fad57d0e
MD
2508 nlookup_done(&nd);
2509 if (error == 0) {
a385a915 2510 error = VOP_PATHCONF(vp, uap->name, &uap->sysmsg_reg);
fad57d0e
MD
2511 vput(vp);
2512 }
984263bc
MD
2513 return (error);
2514}
2515
bbab27f6
DRJ
2516/*
2517 * XXX: daver
2518 * kern_readlink isn't properly split yet. There is a copyin burried
2519 * in VOP_READLINK().
2520 */
984263bc 2521int
fad57d0e 2522kern_readlink(struct nlookupdata *nd, char *buf, int count, int *res)
984263bc 2523{
dadab5e9
MD
2524 struct thread *td = curthread;
2525 struct proc *p = td->td_proc;
41c20dac 2526 struct vnode *vp;
984263bc
MD
2527 struct iovec aiov;
2528 struct uio auio;
2529 int error;
984263bc 2530
fad57d0e
MD
2531 if ((error = nlookup(nd)) != 0)
2532 return (error);
28623bf9 2533 error = cache_vget(&nd->nl_nch, nd->nl_cred, LK_EXCLUSIVE, &vp);
9697c509 2534 if (error)
984263bc 2535 return (error);
fad57d0e 2536 if (vp->v_type != VLNK) {
984263bc 2537 error = EINVAL;
fad57d0e 2538 } else {
9697c509
DRJ
2539 aiov.iov_base = buf;
2540 aiov.iov_len = count;
984263bc
MD
2541 auio.uio_iov = &aiov;
2542 auio.uio_iovcnt = 1;
2543 auio.uio_offset = 0;
2544 auio.uio_rw = UIO_READ;
2545 auio.uio_segflg = UIO_USERSPACE;
dadab5e9 2546 auio.uio_td = td;
9697c509 2547 auio.uio_resid = count;
984263bc
MD
2548 error = VOP_READLINK(vp, &auio, p->p_ucred);
2549 }
2550 vput(vp);
9697c509
DRJ
2551 *res = count - auio.uio_resid;
2552 return (error);
2553}
2554
2555/*
2556 * readlink_args(char *path, char *buf, int count)
2557 *
2558 * Return target name of a symbolic link.
2559 */
2560int
753fd850 2561sys_readlink(struct readlink_args *uap)
9697c509 2562{
fad57d0e 2563 struct nlookupdata nd;
9697c509
DRJ
2564 int error;
2565
fad57d0e
MD
2566 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
2567 if (error == 0) {
2568 error = kern_readlink(&nd, uap->buf, uap->count,
2569 &uap->sysmsg_result);
2570 }
2571 nlookup_done(&nd);
984263bc
MD
2572 return (error);
2573}
2574
2575static int
41c20dac 2576setfflags(struct vnode *vp, int flags)
984263bc 2577{
dadab5e9
MD
2578 struct thread *td = curthread;
2579 struct proc *p = td->td_proc;
984263bc
MD
2580 int error;
2581 struct vattr vattr;
2582
2583 /*
2584 * Prevent non-root users from setting flags on devices. When
2585 * a device is reused, users can retain ownership of the device
2586 * if they are allowed to set flags and programs assume that
2587 * chown can't fail when done as root.
2588 */
2589 if ((vp->v_type == VCHR || vp->v_type == VBLK) &&
6dc79895 2590 ((error = priv_check_cred(p->p_ucred, PRIV_VFS_CHFLAGS_DEV, 0)) != 0))
984263bc
MD
2591 return (error);
2592
186fccd6
MD
2593 /*
2594 * note: vget is required for any operation that might mod the vnode
2595 * so VINACTIVE is properly cleared.
2596 */
87de5057 2597 if ((error = vget(vp, LK_EXCLUSIVE)) == 0) {
186fccd6
MD
2598 VATTR_NULL(&vattr);
2599 vattr.va_flags = flags;
87de5057 2600 error = VOP_SETATTR(vp, &vattr, p->p_ucred);
186fccd6
MD
2601 vput(vp);
2602 }
984263bc
MD
2603 return (error);
2604}
2605
2606/*
41c20dac
MD
2607 * chflags(char *path, int flags)
2608 *
984263bc
MD
2609 * Change flags of a file given a path name.
2610 */
984263bc
MD
2611/* ARGSUSED */
2612int
753fd850 2613sys_chflags(struct chflags_args *uap)
984263bc 2614{
fad57d0e
MD
2615 struct nlookupdata nd;
2616 struct vnode *vp;
984263bc 2617 int error;
984263bc 2618
fad57d0e 2619 vp = NULL;
ab2eb4eb 2620 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
fad57d0e
MD
2621 if (error == 0)
2622 error = nlookup(&nd);
2623 if (error == 0)
28623bf9 2624 error = ncp_writechk(&nd.nl_nch);
72310cfb 2625 if (error == 0)
28623bf9 2626 error = cache_vref(&nd.nl_nch, nd.nl_cred, &vp);
fad57d0e
MD
2627 nlookup_done(&nd);
2628 if (error == 0) {
ab2eb4eb 2629 error = setfflags(vp, uap->flags);
fad57d0e
MD
2630 vrele(vp);
2631 }
2632 return (error);
984263bc
MD
2633}
2634
2635/*
91fdc001
PA
2636 * lchflags(char *path, int flags)
2637 *
2638 * Change flags of a file given a path name, but don't follow symlinks.
2639 */
2640/* ARGSUSED */
2641int
2642sys_lchflags(struct lchflags_args *uap)
2643{
2644 struct nlookupdata nd;
2645 struct vnode *vp;
2646 int error;
2647
2648 vp = NULL;
2649 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
91fdc001
PA
2650 if (error == 0)
2651 error = nlookup(&nd);
2652 if (error == 0)
2653 error = ncp_writechk(&nd.nl_nch);
2654 if (error == 0)
2655 error = cache_vref(&nd.nl_nch, nd.nl_cred, &vp);
2656 nlookup_done(&nd);
2657 if (error == 0) {
2658 error = setfflags(vp, uap->flags);
2659 vrele(vp);
2660 }
2661 return (error);
2662}
2663
2664/*
41c20dac
MD
2665 * fchflags_args(int fd, int flags)
2666 *
984263bc
MD
2667 * Change flags of a file given a file descriptor.
2668 */
984263bc
MD
2669/* ARGSUSED */
2670int
753fd850 2671sys_fchflags(struct fchflags_args *uap)
984263bc 2672{
dadab5e9
MD
2673 struct thread *td = curthread;
2674 struct proc *p = td->td_proc;
984263bc
MD
2675 struct file *fp;
2676 int error;
2677
5b287bba 2678 if ((error = holdvnode(p->p_fd, uap->fd, &fp)) != 0)
984263bc 2679 return (error);
28623bf9
MD
2680 if (fp->f_nchandle.ncp)
2681 error = ncp_writechk(&fp->f_nchandle);
72310cfb
MD
2682 if (error == 0)
2683 error = setfflags((struct vnode *) fp->f_data, uap->flags);
5b287bba
MD
2684 fdrop(fp);
2685 return (error);
984263bc
MD
2686}
2687
2688static int
41c20dac 2689setfmode(struct vnode *vp, int mode)
984263bc 2690{
dadab5e9
MD
2691 struct thread *td = curthread;
2692 struct proc *p = td->td_proc;
984263bc
MD
2693 int error;
2694 struct vattr vattr;
2695
186fccd6
MD
2696 /*
2697 * note: vget is required for any operation that might mod the vnode
2698 * so VINACTIVE is properly cleared.
2699 */
87de5057 2700 if ((error = vget(vp, LK_EXCLUSIVE)) == 0) {
186fccd6
MD
2701 VATTR_NULL(&vattr);
2702 vattr.va_mode = mode & ALLPERMS;
87de5057 2703 error = VOP_SETATTR(vp, &vattr, p->p_ucred);
186fccd6
MD
2704 vput(vp);
2705 }
984263bc
MD
2706 return error;
2707}
2708
9697c509 2709int
fad57d0e 2710kern_chmod(struct nlookupdata *nd, int mode)
9697c509 2711{
fad57d0e 2712 struct vnode *vp;
9697c509
DRJ
2713 int error;
2714
fad57d0e 2715 if ((error = nlookup(nd)) != 0)
9697c509 2716 return (error);
28623bf9 2717 if ((error = cache_vref(&nd->nl_nch, nd->nl_cred, &vp)) != 0)
fad57d0e 2718 return (error);
28623bf9 2719 if ((error = ncp_writechk(&nd->nl_nch)) == 0)
72310cfb 2720 error = setfmode(vp, mode);
fad57d0e
MD
2721 vrele(vp);
2722 return (error);
9697c509
DRJ
2723}
2724
984263bc 2725/*
41c20dac
MD
2726 * chmod_args(char *path, int mode)
2727 *
984263bc
MD
2728 * Change mode of a file given path name.
2729 */
984263bc
MD
2730/* ARGSUSED */
2731int
753fd850 2732sys_chmod(struct chmod_args *uap)
984263bc 2733{
fad57d0e 2734 struct nlookupdata nd;
9697c509 2735 int error;
984263bc 2736
fad57d0e
MD
2737 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
2738 if (error == 0)
2739 error = kern_chmod(&nd, uap->mode);
2740 nlookup_done(&nd);
9697c509 2741 return (error);
984263bc
MD
2742}
2743
2744/*
41c20dac
MD
2745 * lchmod_args(char *path, int mode)
2746 *
984263bc
MD
2747 * Change mode of a file given path name (don't follow links.)
2748 */
984263bc
MD
2749/* ARGSUSED */
2750int
753fd850 2751sys_lchmod(struct lchmod_args *uap)
984263bc 2752{
fad57d0e 2753 struct nlookupdata nd;
984263bc 2754 int error;
984263bc 2755
fad57d0e
MD
2756 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
2757 if (error == 0)
2758 error = kern_chmod(&nd, uap->mode);
2759 nlookup_done(&nd);
2760 return (error);
984263bc
MD
2761}
2762
2763/*
41c20dac
MD
2764 * fchmod_args(int fd, int mode)
2765 *
984263bc
MD
2766 * Change mode of a file given a file descriptor.
2767 */
984263bc
MD
2768/* ARGSUSED */
2769int
753fd850 2770sys_fchmod(struct fchmod_args *uap)
984263bc 2771{
dadab5e9
MD
2772 struct thread *td = curthread;
2773 struct proc *p = td->td_proc;
984263bc
MD
2774 struct file *fp;
2775 int error;
2776
5b287bba 2777 if ((error = holdvnode(p->p_fd, uap->fd, &fp)) != 0)
984263bc 2778 return (error);
28623bf9
MD
2779 if (fp->f_nchandle.ncp)
2780 error = ncp_writechk(&fp->f_nchandle);
72310cfb
MD
2781 if (error == 0)
2782 error = setfmode((struct vnode *)fp->f_data, uap->mode);
5b287bba
MD
2783 fdrop(fp);
2784 return (error);
984263bc
MD
2785}
2786
8cde3c79
NT
2787/*
2788 * fchmodat_args(char *path, int mode)
2789 *
2790 * Change mode of a file pointed to by fd/path.
2791 */
2792int
2793sys_fchmodat(struct fchmodat_args *uap)
2794{
2795 struct nlookupdata nd;
2796 struct file *fp;
2797 int error;
2798 int flags;
2799
2800 if (uap->flags & ~_AT_SYMLINK_MASK)
2801 return (EINVAL);
2802 flags = (uap->flags & AT_SYMLINK_NOFOLLOW) ? 0 : NLC_FOLLOW;
2803
2804 error = nlookup_init_at(&nd, &fp, uap->fd, uap->path,
2805 UIO_USERSPACE, flags);
2806 if (error == 0)
2807 error = kern_chmod(&nd, uap->mode);
2808 nlookup_done_at(&nd, fp);
2809 return (error);
2810}
2811
984263bc 2812static int
41c20dac 2813setfown(struct vnode *vp, uid_t uid, gid_t gid)
984263bc 2814{
dadab5e9
MD
2815 struct thread *td = curthread;
2816 struct proc *p = td->td_proc;
984263bc
MD
2817 int error;
2818 struct vattr vattr;
2819
186fccd6
MD
2820 /*
2821 * note: vget is required for any operation that might mod the vnode
2822 * so VINACTIVE is properly cleared.
2823 */
87de5057 2824 if ((error = vget(vp, LK_EXCLUSIVE)) == 0) {
186fccd6
MD
2825 VATTR_NULL(&vattr);
2826 vattr.va_uid = uid;
2827 vattr.va_gid = gid;
87de5057 2828 error = VOP_SETATTR(vp, &vattr, p->p_ucred);
186fccd6
MD
2829 vput(vp);
2830 }
984263bc
MD
2831 return error;
2832}
2833
9697c509 2834int
fad57d0e 2835kern_chown(struct nlookupdata *nd, int uid, int gid)
9697c509 2836{
fad57d0e 2837 struct vnode *vp;
9697c509
DRJ
2838 int error;
2839
fad57d0e 2840 if ((error = nlookup(nd)) != 0)
9697c509 2841 return (error);
28623bf9 2842 if ((error = cache_vref(&nd->nl_nch, nd->nl_cred, &vp)) != 0)
fad57d0e 2843 return (error);
28623bf9 2844 if ((error = ncp_writechk(&nd->nl_nch)) == 0)
72310cfb 2845 error = setfown(vp, uid, gid);
fad57d0e 2846 vrele(vp);
9697c509
DRJ
2847 return (error);
2848}
2849
984263bc 2850/*
41c20dac
MD
2851 * chown(char *path, int uid, int gid)
2852 *
984263bc
MD
2853 * Set ownership given a path name.
2854 */
984263bc 2855int
753fd850 2856sys_chown(struct chown_args *uap)
984263bc 2857{
fad57d0e 2858 struct nlookupdata nd;
9697c509
DRJ
2859 int error;
2860
fad57d0e
MD
2861 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
2862 if (error == 0)
2863 error = kern_chown(&nd, uap->uid, uap->gid);
2864 nlookup_done(&nd);
984263bc
MD
2865 return (error);
2866}
2867
2868/*
41c20dac
MD
2869 * lchown_args(char *path, int uid, int gid)
2870 *
984263bc
MD
2871 * Set ownership given a path name, do not cross symlinks.
2872 */
984263bc 2873int
753fd850 2874sys_lchown(struct lchown_args *uap)
984263bc 2875{
fad57d0e 2876 struct nlookupdata nd;
984263bc 2877 int error;
9697c509 2878
fad57d0e
MD
2879 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
2880 if (error == 0)
2881 error = kern_chown(&nd, uap->uid, uap->gid);
2882 nlookup_done(&nd);
984263bc
MD
2883 return (error);
2884}
2885
2886/*
41c20dac
MD
2887 * fchown_args(int fd, int uid, int gid)
2888 *
984263bc
MD
2889 * Set ownership given a file descriptor.
2890 */
984263bc
MD
2891/* ARGSUSED */
2892int
753fd850 2893sys_fchown(struct fchown_args *uap)
984263bc 2894{
dadab5e9
MD
2895 struct thread *td = curthread;
2896 struct proc *p = td->td_proc;
984263bc
MD
2897 struct file *fp;
2898 int error;
2899
5b287bba 2900 if ((error = holdvnode(p->p_fd, uap->fd, &fp)) != 0)
984263bc 2901 return (error);
28623bf9
MD
2902 if (fp->f_nchandle.ncp)
2903 error = ncp_writechk(&fp->f_nchandle);
72310cfb
MD
2904 if (error == 0)
2905 error = setfown((struct vnode *)fp->f_data, uap->uid, uap->gid);
5b287bba
MD
2906 fdrop(fp);
2907 return (error);
984263bc
MD
2908}
2909
32013176
NT
2910/*
2911 * fchownat(int fd, char *path, int uid, int gid, int flags)
2912 *
2913 * Set ownership of file pointed to by fd/path.
2914 */
2915int
2916sys_fchownat(struct fchownat_args *uap)
2917{
2918 struct nlookupdata nd;
2919 struct file *fp;
2920 int error;
2921 int flags;
2922
2923 if (uap->flags & ~_AT_SYMLINK_MASK)
2924 return (EINVAL);
2925 flags = (uap->flags & AT_SYMLINK_NOFOLLOW) ? 0 : NLC_FOLLOW;
2926
2927 error = nlookup_init_at(&nd, &fp, uap->fd, uap->path,
2928 UIO_USERSPACE, flags);
2929 if (error == 0)
2930 error = kern_chown(&nd, uap->uid, uap->gid);
2931 nlookup_done_at(&nd, fp);
2932 return (error);
2933}
2934
2935
984263bc 2936static int
9697c509 2937getutimes(const struct timeval *tvp, struct timespec *tsp)
984263bc
MD
2938{
2939 struct timeval tv[2];
984263bc 2940
9697c509 2941 if (tvp == NULL) {
984263bc
MD
2942 microtime(&tv[0]);
2943 TIMEVAL_TO_TIMESPEC(&tv[0], &tsp[0]);
2944 tsp[1] = tsp[0];
2945 } else {
9697c509
DRJ
2946 TIMEVAL_TO_TIMESPEC(&tvp[0], &tsp[0]);
2947 TIMEVAL_TO_TIMESPEC(&tvp[1], &tsp[1]);
984263bc
MD
2948 }
2949 return 0;
2950}
2951
2952static int
3a907475
MD
2953setutimes(struct vnode *vp, struct vattr *vattr,
2954 const struct timespec *ts, int nullflag)
984263bc 2955{
dadab5e9
MD
2956 struct thread *td = curthread;
2957 struct proc *p = td->td_proc;
984263bc 2958 int error;
984263bc 2959
3a907475
MD
2960 VATTR_NULL(vattr);
2961 vattr->va_atime = ts[0];
2962 vattr->va_mtime = ts[1];
2963 if (nullflag)
2964 vattr->va_vaflags |= VA_UTIMES_NULL;
2965 error = VOP_SETATTR(vp, vattr, p->p_ucred);
2966
984263bc
MD
2967 return error;
2968}
2969
9697c509 2970int
fad57d0e 2971kern_utimes(struct nlookupdata *nd, struct timeval *tptr)
9697c509
DRJ
2972{
2973 struct timespec ts[2];
fad57d0e 2974 struct vnode *vp;
3a907475 2975 struct vattr vattr;
9697c509
DRJ
2976 int error;
2977
fad57d0e 2978 if ((error = getutimes(tptr, ts)) != 0)
9697c509 2979 return (error);
3a907475
MD
2980
2981 /*
2982 * NOTE: utimes() succeeds for the owner even if the file
2983 * is not user-writable.
2984 */
2985 nd->nl_flags |= NLC_OWN | NLC_WRITE;
2986
fad57d0e 2987 if ((error = nlookup(nd)) != 0)
9697c509 2988 return (error);
28623bf9 2989 if ((error = ncp_writechk(&nd->nl_nch)) != 0)
72310cfb 2990 return (error);
28623bf9 2991 if ((error = cache_vref(&nd->nl_nch, nd->nl_cred, &vp)) != 0)
fad57d0e 2992 return (error);
ee89633d
MD
2993
2994 /*
3a907475
MD
2995 * note: vget is required for any operation that might mod the vnode
2996 * so VINACTIVE is properly cleared.
ee89633d 2997 */
3a907475
MD
2998 if ((error = vn_writechk(vp, &nd->nl_nch)) == 0) {
2999 error = vget(vp, LK_EXCLUSIVE);
3000 if (error == 0) {
3001 error = setutimes(vp, &vattr, ts, (tptr == NULL));
3002 vput(vp);
3003 }
918101d3 3004 }
fad57d0e 3005 vrele(vp);
9697c509
DRJ
3006 return (error);
3007}
3008
984263bc 3009/*
41c20dac
MD
3010 * utimes_args(char *path, struct timeval *tptr)
3011 *
984263bc
MD
3012 * Set the access and modification times of a file.
3013 */
984263bc 3014int
753fd850 3015sys_utimes(struct utimes_args *uap)
984263bc 3016{
9697c509 3017 struct timeval tv[2];
fad57d0e 3018 struct nlookupdata nd;
9697c509
DRJ
3019 int error;
3020
3021 if (uap->tptr) {
3022 error = copyin(uap->tptr, tv, sizeof(tv));
3023 if (error)
3024 return (error);
3025 }
fad57d0e
MD
3026 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
3027 if (error == 0)
3028 error = kern_utimes(&nd, uap->tptr ? tv : NULL);
3029 nlookup_done(&nd);
984263bc
MD
3030 return (error);
3031}
3032
3033/*
41c20dac
MD
3034 * lutimes_args(char *path, struct timeval *tptr)
3035 *
984263bc
MD
3036 * Set the access and modification times of a file.
3037 */
984263bc 3038int
753fd850 3039sys_lutimes(struct lutimes_args *uap)
984263bc 3040{
9697c509 3041 struct timeval tv[2];
fad57d0e 3042 struct nlookupdata nd;
9697c509
DRJ
3043 int error;
3044
3045 if (uap->tptr) {
3046 error = copyin(uap->tptr, tv, sizeof(tv));
3047 if (error)
3048 return (error);
3049 }
fad57d0e
MD
3050 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
3051 if (error == 0)
3052 error = kern_utimes(&nd, uap->tptr ? tv : NULL);
3053 nlookup_done(&nd);
984263bc
MD
3054 return (error);
3055}
3056
3a907475
MD
3057/*
3058 * Set utimes on a file descriptor. The creds used to open the
3059 * file are used to determine whether the operation is allowed
3060 * or not.
3061 */
984263bc 3062int
9697c509 3063kern_futimes(int fd, struct timeval *tptr)
984263bc 3064{
dadab5e9
MD
3065 struct thread *td = curthread;
3066 struct proc *p = td->td_proc;
984263bc
MD
3067 struct timespec ts[2];
3068 struct file *fp;
3a907475
MD
3069 struct vnode *vp;
3070 struct vattr vattr;
984263bc
MD
3071 int error;
3072
9697c509
DRJ
3073 error = getutimes(tptr, ts);
3074 if (error)
984263bc 3075 return (error);
5b287bba 3076 if ((error = holdvnode(p->p_fd, fd, &fp)) != 0)
984263bc 3077 return (error);
28623bf9
MD
3078 if (fp->f_nchandle.ncp)
3079 error = ncp_writechk(&fp->f_nchandle);
3a907475
MD
3080 if (error == 0) {
3081 vp = fp->f_data;
3082 error = vget(vp, LK_EXCLUSIVE);
3083 if (error == 0) {
3084 error = VOP_GETATTR(vp, &vattr);
3085 if (error == 0) {
3086 error = naccess_va(&vattr, NLC_OWN | NLC_WRITE,
3087 fp->f_cred);
3088 }
3089 if (error == 0) {
3090 error = setutimes(vp, &vattr, ts,
3091 (tptr == NULL));
3092 }
3093 vput(vp);
3094 }
3095 }
5b287bba 3096 fdrop(fp);
9697c509 3097 return (error);
984263bc
MD
3098}
3099
3100/*
9697c509 3101 * futimes_args(int fd, struct timeval *tptr)
41c20dac 3102 *
9697c509 3103 * Set the access and modification times of a file.
984263bc 3104 */
984263bc 3105int
753fd850 3106sys_futimes(struct futimes_args *uap)
9697c509
DRJ
3107{
3108 struct timeval tv[2];
3109 int error;
3110
3111 if (uap->tptr) {
3112 error = copyin(uap->tptr, tv, sizeof(tv));
3113 if (error)
3114 return (error);
3115 }
3116
b651c485 3117 error = kern_futimes(uap->fd, uap->tptr ? tv : NULL);
9697c509
DRJ
3118
3119 return (error);
3120}
3121
3122int
fad57d0e 3123kern_truncate(struct nlookupdata *nd, off_t length)
984263bc 3124{
41c20dac 3125 struct vnode *vp;
984263bc
MD
3126 struct vattr vattr;
3127 int error;
984263bc 3128
9697c509 3129 if (length < 0)
984263bc 3130 return(EINVAL);
3a907475 3131 nd->nl_flags |= NLC_WRITE | NLC_TRUNCATE;
fad57d0e 3132 if ((error = nlookup(nd)) != 0)
984263bc 3133 return (error);
28623bf9 3134 if ((error = ncp_writechk(&nd->nl_nch)) != 0)
72310cfb 3135 return (error);
28623bf9 3136 if ((error = cache_vref(&nd->nl_nch, nd->nl_cred, &vp)) != 0)
fad57d0e 3137 return (error);
ca466bae 3138 if ((error = vn_lock(vp, LK_EXCLUSIVE | LK_RETRY)) != 0) {
fad57d0e
MD
3139 vrele(vp);
3140 return (error);
3141 }
3142 if (vp->v_type == VDIR) {
984263bc 3143 error = EISDIR;
3a907475 3144 } else if ((error = vn_writechk(vp, &nd->nl_nch)) == 0) {
984263bc 3145 VATTR_NULL(&vattr);
9697c509 3146 vattr.va_size = length;
87de5057 3147 error = VOP_SETATTR(vp, &vattr, nd->nl_cred);
984263bc
MD
3148 }
3149 vput(vp);
3150 return (error);
3151}
3152
9697c509
DRJ
3153/*
3154 * truncate(char *path, int pad, off_t length)
3155 *
3156 * Truncate a file given its path name.
3157 */
3158int
753fd850 3159sys_truncate(struct truncate_args *uap)
9697c509 3160{
fad57d0e 3161 struct nlookupdata nd;
9697c509
DRJ
3162 int error;
3163
fad57d0e
MD
3164 error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
3165 if (error == 0)
3166 error = kern_truncate(&nd, uap->length);
3167 nlookup_done(&nd);
9697c509
DRJ
3168 return error;
3169}
3170
984263bc 3171int
8f6f8622 3172kern_ftruncate(int fd, off_t length)
984263bc 3173{
dadab5e9
MD
3174 struct thread *td = curthread;
3175 struct proc *p = td->td_proc;
984263bc
MD
3176 struct vattr vattr;
3177 struct vnode *vp;
3178 struct file *fp;
3179 int error;
3180
8f6f8622 3181 if (length < 0)
984263bc 3182 return(EINVAL);
5b287bba 3183 if ((error = holdvnode(p->p_fd, fd, &fp)) != 0)
984263bc 3184 return (error);
28623bf9
MD
3185 if (fp->f_nchandle.ncp) {
3186 error = ncp_writechk(&fp->f_nchandle);
72310cfb
MD
3187 if (error)
3188 goto done;
3189 }
5b287bba
MD
3190 if ((fp->f_flag & FWRITE) == 0) {
3191 error = EINVAL;
3192 goto done;
3193 }
3a907475
MD
3194 if (fp->f_flag & FAPPENDONLY) { /* inode was set s/uapnd */
3195 error = EINVAL;
3196 goto done;
3197 }
984263bc 3198 vp = (struct vnode *)fp->f_data;
ca466bae 3199 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
5b287bba 3200 if (vp->v_type == VDIR) {
984263bc 3201 error = EISDIR;
468bb1f9 3202 } else if ((error = vn_writechk(vp, NULL)) == 0) {
984263bc 3203 VATTR_NULL(&vattr);
8f6f8622 3204 vattr.va_size = length;
87de5057 3205 error = VOP_SETATTR(vp, &vattr, fp->f_cred);
984263bc 3206 }
a11aaa81 3207 vn_unlock(vp);
5b287bba
MD
3208done:
3209 fdrop(fp);
984263bc
MD
3210 return (error);
3211}
3212
984263bc 3213/*
8f6f8622 3214 * ftruncate_args(int fd, int pad, off_t length)
41c20dac 3215 *
8f6f8622 3216 * Truncate a file given a file descriptor.
984263bc 3217 */
984263bc 3218int
753fd850 3219sys_ftruncate(struct ftruncate_args *uap)
984263bc 3220{
8f6f8622 3221 int error;
984263bc 3222
8f6f8622
DRJ
3223 error = kern_ftruncate(uap->fd, uap->length);
3224
3225 return (error);
984263bc
MD
3226}
3227
984263bc 3228/*
41c20dac
MD
3229 * fsync(int fd)
3230 *
984263bc
MD
3231 * Sync an open file.
3232 */
984263bc
MD
3233/* ARGSUSED */
3234int
753fd850 3235sys_fsync(struct fsync_args *uap)
984263bc 3236{
dadab5e9
MD
3237 struct thread *td = curthread;
3238 struct proc *p = td->td_proc;
41c20dac 3239 struct vnode *vp;
984263bc
MD
3240 struct file *fp;
3241 vm_object_t obj;
3242 int error;
3243
5b287bba 3244 if ((error = holdvnode(p->p_fd, uap->fd, &fp)) != 0)
984263bc
MD
3245 return (error);
3246 vp = (struct vnode *)fp->f_data;
ca466bae 3247 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
7540ab49 3248 if ((obj = vp->v_object) != NULL)
984263bc 3249 vm_object_page_clean(obj, 0, 0, 0);
408357d8
MD
3250 if ((error = VOP_FSYNC(vp, MNT_WAIT)) == 0 && vp->v_mount)
3251 error = buf_fsync(vp);
a11aaa81 3252 vn_unlock(vp);
5b287bba