vfs_syscalls.c revision 1.518.4.4
1 /* $NetBSD: vfs_syscalls.c,v 1.518.4.4 2020/04/21 18:42:42 martin Exp $ */ 2 3 /*- 4 * Copyright (c) 2008, 2009, 2019, 2020 The NetBSD Foundation, Inc. 5 * All rights reserved. 6 * 7 * This code is derived from software contributed to The NetBSD Foundation 8 * by Andrew Doran. 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 * 19 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 20 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 21 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 22 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 23 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 24 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 25 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 26 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 27 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 28 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 29 * POSSIBILITY OF SUCH DAMAGE. 30 */ 31 32 /* 33 * Copyright (c) 1989, 1993 34 * The Regents of the University of California. All rights reserved. 35 * (c) UNIX System Laboratories, Inc. 36 * All or some portions of this file are derived from material licensed 37 * to the University of California by American Telephone and Telegraph 38 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 39 * the permission of UNIX System Laboratories, Inc. 40 * 41 * Redistribution and use in source and binary forms, with or without 42 * modification, are permitted provided that the following conditions 43 * are met: 44 * 1. Redistributions of source code must retain the above copyright 45 * notice, this list of conditions and the following disclaimer. 46 * 2. Redistributions in binary form must reproduce the above copyright 47 * notice, this list of conditions and the following disclaimer in the 48 * documentation and/or other materials provided with the distribution. 49 * 3. Neither the name of the University nor the names of its contributors 50 * may be used to endorse or promote products derived from this software 51 * without specific prior written permission. 52 * 53 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 54 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 55 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 56 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 57 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 58 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 59 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 61 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 62 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 63 * SUCH DAMAGE. 64 * 65 * @(#)vfs_syscalls.c 8.42 (Berkeley) 7/31/95 66 */ 67 68 /* 69 * Virtual File System System Calls 70 */ 71 72 #include <sys/cdefs.h> 73 __KERNEL_RCSID(0, "$NetBSD: vfs_syscalls.c,v 1.518.4.4 2020/04/21 18:42:42 martin Exp $"); 74 75 #ifdef _KERNEL_OPT 76 #include "opt_fileassoc.h" 77 #include "veriexec.h" 78 #endif 79 80 #include <sys/param.h> 81 #include <sys/systm.h> 82 #include <sys/namei.h> 83 #include <sys/filedesc.h> 84 #include <sys/kernel.h> 85 #include <sys/file.h> 86 #include <sys/fcntl.h> 87 #include <sys/stat.h> 88 #include <sys/vnode.h> 89 #include <sys/mount.h> 90 #include <sys/fstrans.h> 91 #include <sys/proc.h> 92 #include <sys/uio.h> 93 #include <sys/kmem.h> 94 #include <sys/dirent.h> 95 #include <sys/sysctl.h> 96 #include <sys/syscallargs.h> 97 #include <sys/vfs_syscalls.h> 98 #include <sys/quota.h> 99 #include <sys/quotactl.h> 100 #include <sys/ktrace.h> 101 #ifdef FILEASSOC 102 #include <sys/fileassoc.h> 103 #endif /* FILEASSOC */ 104 #include <sys/extattr.h> 105 #include <sys/verified_exec.h> 106 #include <sys/kauth.h> 107 #include <sys/atomic.h> 108 #include <sys/module.h> 109 #include <sys/buf.h> 110 #include <sys/event.h> 111 #include <sys/compat_stub.h> 112 113 #include <miscfs/genfs/genfs.h> 114 #include <miscfs/specfs/specdev.h> 115 116 #include <nfs/rpcv2.h> 117 #include <nfs/nfsproto.h> 118 #include <nfs/nfs.h> 119 #include <nfs/nfs_var.h> 120 121 /* XXX this shouldn't be here */ 122 #ifndef OFF_T_MAX 123 #define OFF_T_MAX __type_max(off_t) 124 #endif 125 126 static int change_flags(struct vnode *, u_long, struct lwp *); 127 static int change_mode(struct vnode *, int, struct lwp *); 128 static int change_owner(struct vnode *, uid_t, gid_t, struct lwp *, int); 129 static int do_sys_openat(lwp_t *, int, const char *, int, int, int *); 130 static int do_sys_mkdirat(struct lwp *l, int, const char *, mode_t, 131 enum uio_seg); 132 static int do_sys_mkfifoat(struct lwp *, int, const char *, mode_t); 133 static int do_sys_symlinkat(struct lwp *, const char *, int, const char *, 134 enum uio_seg); 135 static int do_sys_renameat(struct lwp *l, int, const char *, int, const char *, 136 enum uio_seg, int); 137 static int do_sys_readlinkat(struct lwp *, int, const char *, char *, 138 size_t, register_t *); 139 static int do_sys_unlinkat(struct lwp *, int, const char *, int, enum uio_seg); 140 141 static int fd_nameiat(struct lwp *, int, struct nameidata *); 142 static int fd_nameiat_simple_user(struct lwp *, int, const char *, 143 namei_simple_flags_t, struct vnode **); 144 145 /* 146 * This table is used to maintain compatibility with 4.3BSD 147 * and NetBSD 0.9 mount syscalls - and possibly other systems. 148 * Note, the order is important! 149 * 150 * Do not modify this table. It should only contain filesystems 151 * supported by NetBSD 0.9 and 4.3BSD. 152 */ 153 const char * const mountcompatnames[] = { 154 NULL, /* 0 = MOUNT_NONE */ 155 MOUNT_FFS, /* 1 = MOUNT_UFS */ 156 MOUNT_NFS, /* 2 */ 157 MOUNT_MFS, /* 3 */ 158 MOUNT_MSDOS, /* 4 */ 159 MOUNT_CD9660, /* 5 = MOUNT_ISOFS */ 160 MOUNT_FDESC, /* 6 */ 161 MOUNT_KERNFS, /* 7 */ 162 NULL, /* 8 = MOUNT_DEVFS */ 163 MOUNT_AFS, /* 9 */ 164 }; 165 166 const u_int nmountcompatnames = __arraycount(mountcompatnames); 167 168 static int 169 fd_nameiat(struct lwp *l, int fdat, struct nameidata *ndp) 170 { 171 file_t *dfp; 172 int error; 173 174 if (fdat != AT_FDCWD) { 175 if ((error = fd_getvnode(fdat, &dfp)) != 0) 176 goto out; 177 178 NDAT(ndp, dfp->f_vnode); 179 } 180 181 error = namei(ndp); 182 183 if (fdat != AT_FDCWD) 184 fd_putfile(fdat); 185 out: 186 return error; 187 } 188 189 static int 190 fd_nameiat_simple_user(struct lwp *l, int fdat, const char *path, 191 namei_simple_flags_t sflags, struct vnode **vp_ret) 192 { 193 file_t *dfp; 194 struct vnode *dvp; 195 int error; 196 197 if (fdat != AT_FDCWD) { 198 if ((error = fd_getvnode(fdat, &dfp)) != 0) 199 goto out; 200 201 dvp = dfp->f_vnode; 202 } else { 203 dvp = NULL; 204 } 205 206 error = nameiat_simple_user(dvp, path, sflags, vp_ret); 207 208 if (fdat != AT_FDCWD) 209 fd_putfile(fdat); 210 out: 211 return error; 212 } 213 214 static int 215 open_setfp(struct lwp *l, file_t *fp, struct vnode *vp, int indx, int flags) 216 { 217 int error; 218 219 fp->f_flag = flags & FMASK; 220 fp->f_type = DTYPE_VNODE; 221 fp->f_ops = &vnops; 222 fp->f_vnode = vp; 223 224 if (flags & (O_EXLOCK | O_SHLOCK)) { 225 struct flock lf; 226 int type; 227 228 lf.l_whence = SEEK_SET; 229 lf.l_start = 0; 230 lf.l_len = 0; 231 if (flags & O_EXLOCK) 232 lf.l_type = F_WRLCK; 233 else 234 lf.l_type = F_RDLCK; 235 type = F_FLOCK; 236 if ((flags & FNONBLOCK) == 0) 237 type |= F_WAIT; 238 VOP_UNLOCK(vp); 239 error = VOP_ADVLOCK(vp, fp, F_SETLK, &lf, type); 240 if (error) { 241 (void) vn_close(vp, fp->f_flag, fp->f_cred); 242 fd_abort(l->l_proc, fp, indx); 243 return error; 244 } 245 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 246 atomic_or_uint(&fp->f_flag, FHASLOCK); 247 } 248 if (flags & O_CLOEXEC) 249 fd_set_exclose(l, indx, true); 250 return 0; 251 } 252 253 static int 254 mount_update(struct lwp *l, struct vnode *vp, const char *path, int flags, 255 void *data, size_t *data_len) 256 { 257 struct mount *mp; 258 int error = 0, saved_flags; 259 260 mp = vp->v_mount; 261 saved_flags = mp->mnt_flag; 262 263 /* We can operate only on VV_ROOT nodes. */ 264 if ((vp->v_vflag & VV_ROOT) == 0) { 265 error = EINVAL; 266 goto out; 267 } 268 269 /* 270 * We only allow the filesystem to be reloaded if it 271 * is currently mounted read-only. Additionally, we 272 * prevent read-write to read-only downgrades. 273 */ 274 if ((flags & (MNT_RELOAD | MNT_RDONLY)) != 0 && 275 (mp->mnt_flag & MNT_RDONLY) == 0 && 276 (mp->mnt_iflag & IMNT_CAN_RWTORO) == 0) { 277 error = EOPNOTSUPP; /* Needs translation */ 278 goto out; 279 } 280 281 /* 282 * Enabling MNT_UNION requires a covered mountpoint and 283 * must not happen on the root mount. 284 */ 285 if ((flags & MNT_UNION) != 0 && mp->mnt_vnodecovered == NULLVP) { 286 error = EOPNOTSUPP; 287 goto out; 288 } 289 290 error = kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_MOUNT, 291 KAUTH_REQ_SYSTEM_MOUNT_UPDATE, mp, KAUTH_ARG(flags), data); 292 if (error) 293 goto out; 294 295 error = vfs_suspend(mp, 0); 296 if (error) 297 goto out; 298 299 mutex_enter(mp->mnt_updating); 300 301 mp->mnt_flag &= ~MNT_OP_FLAGS; 302 mp->mnt_flag |= flags & MNT_OP_FLAGS; 303 304 /* 305 * Set the mount level flags. 306 */ 307 if ((flags & MNT_RDONLY) != (mp->mnt_flag & MNT_RDONLY)) { 308 if ((flags & MNT_RDONLY)) 309 mp->mnt_iflag |= IMNT_WANTRDONLY; 310 else 311 mp->mnt_iflag |= IMNT_WANTRDWR; 312 } 313 mp->mnt_flag &= ~MNT_BASIC_FLAGS; 314 mp->mnt_flag |= flags & MNT_BASIC_FLAGS; 315 if ((mp->mnt_iflag & IMNT_WANTRDONLY)) 316 mp->mnt_flag &= ~MNT_RDONLY; 317 318 error = VFS_MOUNT(mp, path, data, data_len); 319 320 if (error && data != NULL) { 321 int error2; 322 323 /* 324 * Update failed; let's try and see if it was an 325 * export request. For compat with 3.0 and earlier. 326 */ 327 error2 = vfs_hooks_reexport(mp, path, data); 328 329 /* 330 * Only update error code if the export request was 331 * understood but some problem occurred while 332 * processing it. 333 */ 334 if (error2 != EJUSTRETURN) 335 error = error2; 336 } 337 338 if (error == 0 && (mp->mnt_iflag & IMNT_WANTRDONLY)) 339 mp->mnt_flag |= MNT_RDONLY; 340 if (error) 341 mp->mnt_flag = saved_flags; 342 mp->mnt_flag &= ~MNT_OP_FLAGS; 343 mp->mnt_iflag &= ~(IMNT_WANTRDONLY | IMNT_WANTRDWR); 344 if ((mp->mnt_flag & (MNT_RDONLY | MNT_ASYNC)) == 0) { 345 if ((mp->mnt_iflag & IMNT_ONWORKLIST) == 0) 346 vfs_syncer_add_to_worklist(mp); 347 } else { 348 if ((mp->mnt_iflag & IMNT_ONWORKLIST) != 0) 349 vfs_syncer_remove_from_worklist(mp); 350 } 351 mutex_exit(mp->mnt_updating); 352 vfs_resume(mp); 353 354 if ((error == 0) && !(saved_flags & MNT_EXTATTR) && 355 (flags & MNT_EXTATTR)) { 356 if (VFS_EXTATTRCTL(mp, EXTATTR_CMD_START, 357 NULL, 0, NULL) != 0) { 358 printf("%s: failed to start extattr, error = %d", 359 mp->mnt_stat.f_mntonname, error); 360 mp->mnt_flag &= ~MNT_EXTATTR; 361 } 362 } 363 364 if ((error == 0) && (saved_flags & MNT_EXTATTR) && 365 !(flags & MNT_EXTATTR)) { 366 if (VFS_EXTATTRCTL(mp, EXTATTR_CMD_STOP, 367 NULL, 0, NULL) != 0) { 368 printf("%s: failed to stop extattr, error = %d", 369 mp->mnt_stat.f_mntonname, error); 370 mp->mnt_flag |= MNT_RDONLY; 371 } 372 } 373 out: 374 return (error); 375 } 376 377 static int 378 mount_get_vfsops(const char *fstype, enum uio_seg type_seg, 379 struct vfsops **vfsops) 380 { 381 char fstypename[sizeof(((struct statvfs *)NULL)->f_fstypename)]; 382 int error; 383 384 if (type_seg == UIO_USERSPACE) { 385 /* Copy file-system type from userspace. */ 386 error = copyinstr(fstype, fstypename, sizeof(fstypename), NULL); 387 } else { 388 error = copystr(fstype, fstypename, sizeof(fstypename), NULL); 389 KASSERT(error == 0); 390 } 391 392 if (error) { 393 /* 394 * Historically, filesystem types were identified by numbers. 395 * If we get an integer for the filesystem type instead of a 396 * string, we check to see if it matches one of the historic 397 * filesystem types. 398 */ 399 u_long fsindex = (u_long)fstype; 400 if (fsindex >= nmountcompatnames || 401 mountcompatnames[fsindex] == NULL) 402 return ENODEV; 403 strlcpy(fstypename, mountcompatnames[fsindex], 404 sizeof(fstypename)); 405 } 406 407 /* Accept `ufs' as an alias for `ffs', for compatibility. */ 408 if (strcmp(fstypename, "ufs") == 0) 409 fstypename[0] = 'f'; 410 411 if ((*vfsops = vfs_getopsbyname(fstypename)) != NULL) 412 return 0; 413 414 /* If we can autoload a vfs module, try again */ 415 (void)module_autoload(fstypename, MODULE_CLASS_VFS); 416 417 if ((*vfsops = vfs_getopsbyname(fstypename)) != NULL) 418 return 0; 419 420 return ENODEV; 421 } 422 423 static int 424 mount_getargs(struct lwp *l, struct vnode *vp, const char *path, int flags, 425 void *data, size_t *data_len) 426 { 427 struct mount *mp; 428 int error; 429 430 /* If MNT_GETARGS is specified, it should be the only flag. */ 431 if (flags & ~MNT_GETARGS) 432 return EINVAL; 433 434 mp = vp->v_mount; 435 436 /* XXX: probably some notion of "can see" here if we want isolation. */ 437 error = kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_MOUNT, 438 KAUTH_REQ_SYSTEM_MOUNT_GET, mp, data, NULL); 439 if (error) 440 return error; 441 442 if ((vp->v_vflag & VV_ROOT) == 0) 443 return EINVAL; 444 445 if (vfs_busy(mp)) 446 return EPERM; 447 448 mutex_enter(mp->mnt_updating); 449 mp->mnt_flag &= ~MNT_OP_FLAGS; 450 mp->mnt_flag |= MNT_GETARGS; 451 error = VFS_MOUNT(mp, path, data, data_len); 452 mp->mnt_flag &= ~MNT_OP_FLAGS; 453 mutex_exit(mp->mnt_updating); 454 455 vfs_unbusy(mp); 456 return (error); 457 } 458 459 int 460 sys___mount50(struct lwp *l, const struct sys___mount50_args *uap, register_t *retval) 461 { 462 /* { 463 syscallarg(const char *) type; 464 syscallarg(const char *) path; 465 syscallarg(int) flags; 466 syscallarg(void *) data; 467 syscallarg(size_t) data_len; 468 } */ 469 470 return do_sys_mount(l, SCARG(uap, type), UIO_USERSPACE, SCARG(uap, path), 471 SCARG(uap, flags), SCARG(uap, data), UIO_USERSPACE, 472 SCARG(uap, data_len), retval); 473 } 474 475 int 476 do_sys_mount(struct lwp *l, const char *type, enum uio_seg type_seg, 477 const char *path, int flags, void *data, enum uio_seg data_seg, 478 size_t data_len, register_t *retval) 479 { 480 struct vfsops *vfsops = NULL; /* XXX gcc4.8 */ 481 struct vnode *vp; 482 void *data_buf = data; 483 bool vfsopsrele = false; 484 size_t alloc_sz = 0; 485 int error; 486 487 /* 488 * Get vnode to be covered 489 */ 490 error = namei_simple_user(path, NSM_FOLLOW_TRYEMULROOT, &vp); 491 if (error != 0) { 492 vp = NULL; 493 goto done; 494 } 495 496 if (flags & (MNT_GETARGS | MNT_UPDATE)) { 497 vfsops = vp->v_mount->mnt_op; 498 } else { 499 /* 'type' is userspace */ 500 error = mount_get_vfsops(type, type_seg, &vfsops); 501 if (error != 0) 502 goto done; 503 vfsopsrele = true; 504 } 505 506 /* 507 * We allow data to be NULL, even for userspace. Some fs's don't need 508 * it. The others will handle NULL. 509 */ 510 if (data != NULL && data_seg == UIO_USERSPACE) { 511 if (data_len == 0) { 512 /* No length supplied, use default for filesystem */ 513 data_len = vfsops->vfs_min_mount_data; 514 515 /* 516 * Hopefully a longer buffer won't make copyin() fail. 517 * For compatibility with 3.0 and earlier. 518 */ 519 if (flags & MNT_UPDATE 520 && data_len < sizeof (struct mnt_export_args30)) 521 data_len = sizeof (struct mnt_export_args30); 522 } 523 if ((data_len == 0) || (data_len > VFS_MAX_MOUNT_DATA)) { 524 error = EINVAL; 525 goto done; 526 } 527 alloc_sz = data_len; 528 data_buf = kmem_alloc(alloc_sz, KM_SLEEP); 529 530 /* NFS needs the buffer even for mnt_getargs .... */ 531 error = copyin(data, data_buf, data_len); 532 if (error != 0) 533 goto done; 534 } 535 536 if (flags & MNT_GETARGS) { 537 if (data_len == 0) { 538 error = EINVAL; 539 goto done; 540 } 541 error = mount_getargs(l, vp, path, flags, data_buf, &data_len); 542 if (error != 0) 543 goto done; 544 if (data_seg == UIO_USERSPACE) 545 error = copyout(data_buf, data, data_len); 546 *retval = data_len; 547 } else if (flags & MNT_UPDATE) { 548 error = mount_update(l, vp, path, flags, data_buf, &data_len); 549 } else { 550 /* Locking is handled internally in mount_domount(). */ 551 KASSERT(vfsopsrele == true); 552 error = mount_domount(l, &vp, vfsops, path, flags, data_buf, 553 &data_len); 554 vfsopsrele = false; 555 } 556 if (!error) 557 KNOTE(&fs_klist, VQ_MOUNT); 558 559 done: 560 if (vfsopsrele) 561 vfs_delref(vfsops); 562 if (vp != NULL) { 563 vrele(vp); 564 } 565 if (data_buf != data) 566 kmem_free(data_buf, alloc_sz); 567 return (error); 568 } 569 570 /* 571 * Unmount a file system. 572 * 573 * Note: unmount takes a path to the vnode mounted on as argument, 574 * not special file (as before). 575 */ 576 /* ARGSUSED */ 577 int 578 sys_unmount(struct lwp *l, const struct sys_unmount_args *uap, register_t *retval) 579 { 580 /* { 581 syscallarg(const char *) path; 582 syscallarg(int) flags; 583 } */ 584 struct vnode *vp; 585 struct mount *mp; 586 int error; 587 struct pathbuf *pb; 588 struct nameidata nd; 589 590 error = pathbuf_copyin(SCARG(uap, path), &pb); 591 if (error) { 592 return error; 593 } 594 595 NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | TRYEMULROOT, pb); 596 if ((error = namei(&nd)) != 0) { 597 pathbuf_destroy(pb); 598 return error; 599 } 600 vp = nd.ni_vp; 601 pathbuf_destroy(pb); 602 603 mp = vp->v_mount; 604 vfs_ref(mp); 605 VOP_UNLOCK(vp); 606 607 error = kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_MOUNT, 608 KAUTH_REQ_SYSTEM_MOUNT_UNMOUNT, mp, NULL, NULL); 609 if (error) { 610 vrele(vp); 611 vfs_rele(mp); 612 return (error); 613 } 614 615 /* 616 * Don't allow unmounting the root file system. 617 */ 618 if (mp->mnt_flag & MNT_ROOTFS) { 619 vrele(vp); 620 vfs_rele(mp); 621 return (EINVAL); 622 } 623 624 /* 625 * Must be the root of the filesystem 626 */ 627 if ((vp->v_vflag & VV_ROOT) == 0) { 628 vrele(vp); 629 vfs_rele(mp); 630 return (EINVAL); 631 } 632 633 vrele(vp); 634 error = dounmount(mp, SCARG(uap, flags), l); 635 vfs_rele(mp); 636 if (!error) 637 KNOTE(&fs_klist, VQ_UNMOUNT); 638 return error; 639 } 640 641 /* 642 * Sync each mounted filesystem. 643 */ 644 #ifdef DEBUG 645 int syncprt = 0; 646 struct ctldebug debug0 = { "syncprt", &syncprt }; 647 #endif 648 649 void 650 do_sys_sync(struct lwp *l) 651 { 652 mount_iterator_t *iter; 653 struct mount *mp; 654 int asyncflag; 655 656 mountlist_iterator_init(&iter); 657 while ((mp = mountlist_iterator_next(iter)) != NULL) { 658 mutex_enter(mp->mnt_updating); 659 if ((mp->mnt_flag & MNT_RDONLY) == 0) { 660 asyncflag = mp->mnt_flag & MNT_ASYNC; 661 mp->mnt_flag &= ~MNT_ASYNC; 662 VFS_SYNC(mp, MNT_NOWAIT, l->l_cred); 663 if (asyncflag) 664 mp->mnt_flag |= MNT_ASYNC; 665 } 666 mutex_exit(mp->mnt_updating); 667 } 668 mountlist_iterator_destroy(iter); 669 #ifdef DEBUG 670 if (syncprt) 671 vfs_bufstats(); 672 #endif /* DEBUG */ 673 } 674 675 static bool 676 sync_vnode_filter(void *cookie, vnode_t *vp) 677 { 678 679 if (vp->v_numoutput > 0) { 680 ++*(int *)cookie; 681 } 682 return false; 683 } 684 685 int 686 vfs_syncwait(void) 687 { 688 int nbusy, nbusy_prev, iter; 689 struct vnode_iterator *vniter; 690 mount_iterator_t *mpiter; 691 struct mount *mp; 692 693 for (nbusy_prev = 0, iter = 0; iter < 20;) { 694 nbusy = 0; 695 mountlist_iterator_init(&mpiter); 696 while ((mp = mountlist_iterator_next(mpiter)) != NULL) { 697 vnode_t *vp __diagused; 698 vfs_vnode_iterator_init(mp, &vniter); 699 vp = vfs_vnode_iterator_next(vniter, 700 sync_vnode_filter, &nbusy); 701 KASSERT(vp == NULL); 702 vfs_vnode_iterator_destroy(vniter); 703 } 704 mountlist_iterator_destroy(mpiter); 705 706 if (nbusy == 0) 707 break; 708 if (nbusy_prev == 0) 709 nbusy_prev = nbusy; 710 printf("%d ", nbusy); 711 kpause("syncwait", false, MAX(1, hz / 25 * iter), NULL); 712 if (nbusy >= nbusy_prev) /* we didn't flush anything */ 713 iter++; 714 else 715 nbusy_prev = nbusy; 716 } 717 718 if (nbusy) { 719 #if defined(DEBUG) || defined(DEBUG_HALT_BUSY) 720 printf("giving up\nPrinting vnodes for busy buffers\n"); 721 mountlist_iterator_init(&mpiter); 722 while ((mp = mountlist_iterator_next(mpiter)) != NULL) { 723 vnode_t *vp; 724 vfs_vnode_iterator_init(mp, &vniter); 725 vp = vfs_vnode_iterator_next(vniter, 726 NULL, NULL); 727 mutex_enter(vp->v_interlock); 728 if (vp->v_numoutput > 0) 729 vprint(NULL, vp); 730 mutex_exit(vp->v_interlock); 731 vrele(vp); 732 vfs_vnode_iterator_destroy(vniter); 733 } 734 mountlist_iterator_destroy(mpiter); 735 #endif 736 } 737 738 return nbusy; 739 } 740 741 /* ARGSUSED */ 742 int 743 sys_sync(struct lwp *l, const void *v, register_t *retval) 744 { 745 do_sys_sync(l); 746 return (0); 747 } 748 749 750 /* 751 * Access or change filesystem quotas. 752 * 753 * (this is really 14 different calls bundled into one) 754 */ 755 756 static int 757 do_sys_quotactl_stat(struct mount *mp, struct quotastat *info_u) 758 { 759 struct quotastat info_k; 760 int error; 761 762 /* ensure any padding bytes are cleared */ 763 memset(&info_k, 0, sizeof(info_k)); 764 765 error = vfs_quotactl_stat(mp, &info_k); 766 if (error) { 767 return error; 768 } 769 770 return copyout(&info_k, info_u, sizeof(info_k)); 771 } 772 773 static int 774 do_sys_quotactl_idtypestat(struct mount *mp, int idtype, 775 struct quotaidtypestat *info_u) 776 { 777 struct quotaidtypestat info_k; 778 int error; 779 780 /* ensure any padding bytes are cleared */ 781 memset(&info_k, 0, sizeof(info_k)); 782 783 error = vfs_quotactl_idtypestat(mp, idtype, &info_k); 784 if (error) { 785 return error; 786 } 787 788 return copyout(&info_k, info_u, sizeof(info_k)); 789 } 790 791 static int 792 do_sys_quotactl_objtypestat(struct mount *mp, int objtype, 793 struct quotaobjtypestat *info_u) 794 { 795 struct quotaobjtypestat info_k; 796 int error; 797 798 /* ensure any padding bytes are cleared */ 799 memset(&info_k, 0, sizeof(info_k)); 800 801 error = vfs_quotactl_objtypestat(mp, objtype, &info_k); 802 if (error) { 803 return error; 804 } 805 806 return copyout(&info_k, info_u, sizeof(info_k)); 807 } 808 809 static int 810 do_sys_quotactl_get(struct mount *mp, const struct quotakey *key_u, 811 struct quotaval *val_u) 812 { 813 struct quotakey key_k; 814 struct quotaval val_k; 815 int error; 816 817 /* ensure any padding bytes are cleared */ 818 memset(&val_k, 0, sizeof(val_k)); 819 820 error = copyin(key_u, &key_k, sizeof(key_k)); 821 if (error) { 822 return error; 823 } 824 825 error = vfs_quotactl_get(mp, &key_k, &val_k); 826 if (error) { 827 return error; 828 } 829 830 return copyout(&val_k, val_u, sizeof(val_k)); 831 } 832 833 static int 834 do_sys_quotactl_put(struct mount *mp, const struct quotakey *key_u, 835 const struct quotaval *val_u) 836 { 837 struct quotakey key_k; 838 struct quotaval val_k; 839 int error; 840 841 error = copyin(key_u, &key_k, sizeof(key_k)); 842 if (error) { 843 return error; 844 } 845 846 error = copyin(val_u, &val_k, sizeof(val_k)); 847 if (error) { 848 return error; 849 } 850 851 return vfs_quotactl_put(mp, &key_k, &val_k); 852 } 853 854 static int 855 do_sys_quotactl_del(struct mount *mp, const struct quotakey *key_u) 856 { 857 struct quotakey key_k; 858 int error; 859 860 error = copyin(key_u, &key_k, sizeof(key_k)); 861 if (error) { 862 return error; 863 } 864 865 return vfs_quotactl_del(mp, &key_k); 866 } 867 868 static int 869 do_sys_quotactl_cursoropen(struct mount *mp, struct quotakcursor *cursor_u) 870 { 871 struct quotakcursor cursor_k; 872 int error; 873 874 /* ensure any padding bytes are cleared */ 875 memset(&cursor_k, 0, sizeof(cursor_k)); 876 877 error = vfs_quotactl_cursoropen(mp, &cursor_k); 878 if (error) { 879 return error; 880 } 881 882 return copyout(&cursor_k, cursor_u, sizeof(cursor_k)); 883 } 884 885 static int 886 do_sys_quotactl_cursorclose(struct mount *mp, struct quotakcursor *cursor_u) 887 { 888 struct quotakcursor cursor_k; 889 int error; 890 891 error = copyin(cursor_u, &cursor_k, sizeof(cursor_k)); 892 if (error) { 893 return error; 894 } 895 896 return vfs_quotactl_cursorclose(mp, &cursor_k); 897 } 898 899 static int 900 do_sys_quotactl_cursorskipidtype(struct mount *mp, 901 struct quotakcursor *cursor_u, int idtype) 902 { 903 struct quotakcursor cursor_k; 904 int error; 905 906 error = copyin(cursor_u, &cursor_k, sizeof(cursor_k)); 907 if (error) { 908 return error; 909 } 910 911 error = vfs_quotactl_cursorskipidtype(mp, &cursor_k, idtype); 912 if (error) { 913 return error; 914 } 915 916 return copyout(&cursor_k, cursor_u, sizeof(cursor_k)); 917 } 918 919 static int 920 do_sys_quotactl_cursorget(struct mount *mp, struct quotakcursor *cursor_u, 921 struct quotakey *keys_u, struct quotaval *vals_u, unsigned maxnum, 922 unsigned *ret_u) 923 { 924 #define CGET_STACK_MAX 8 925 struct quotakcursor cursor_k; 926 struct quotakey stackkeys[CGET_STACK_MAX]; 927 struct quotaval stackvals[CGET_STACK_MAX]; 928 struct quotakey *keys_k; 929 struct quotaval *vals_k; 930 unsigned ret_k; 931 int error; 932 933 if (maxnum > 128) { 934 maxnum = 128; 935 } 936 937 error = copyin(cursor_u, &cursor_k, sizeof(cursor_k)); 938 if (error) { 939 return error; 940 } 941 942 if (maxnum <= CGET_STACK_MAX) { 943 keys_k = stackkeys; 944 vals_k = stackvals; 945 /* ensure any padding bytes are cleared */ 946 memset(keys_k, 0, maxnum * sizeof(keys_k[0])); 947 memset(vals_k, 0, maxnum * sizeof(vals_k[0])); 948 } else { 949 keys_k = kmem_zalloc(maxnum * sizeof(keys_k[0]), KM_SLEEP); 950 vals_k = kmem_zalloc(maxnum * sizeof(vals_k[0]), KM_SLEEP); 951 } 952 953 error = vfs_quotactl_cursorget(mp, &cursor_k, keys_k, vals_k, maxnum, 954 &ret_k); 955 if (error) { 956 goto fail; 957 } 958 959 error = copyout(keys_k, keys_u, ret_k * sizeof(keys_k[0])); 960 if (error) { 961 goto fail; 962 } 963 964 error = copyout(vals_k, vals_u, ret_k * sizeof(vals_k[0])); 965 if (error) { 966 goto fail; 967 } 968 969 error = copyout(&ret_k, ret_u, sizeof(ret_k)); 970 if (error) { 971 goto fail; 972 } 973 974 /* do last to maximize the chance of being able to recover a failure */ 975 error = copyout(&cursor_k, cursor_u, sizeof(cursor_k)); 976 977 fail: 978 if (keys_k != stackkeys) { 979 kmem_free(keys_k, maxnum * sizeof(keys_k[0])); 980 } 981 if (vals_k != stackvals) { 982 kmem_free(vals_k, maxnum * sizeof(vals_k[0])); 983 } 984 return error; 985 } 986 987 static int 988 do_sys_quotactl_cursoratend(struct mount *mp, struct quotakcursor *cursor_u, 989 int *ret_u) 990 { 991 struct quotakcursor cursor_k; 992 int ret_k; 993 int error; 994 995 error = copyin(cursor_u, &cursor_k, sizeof(cursor_k)); 996 if (error) { 997 return error; 998 } 999 1000 error = vfs_quotactl_cursoratend(mp, &cursor_k, &ret_k); 1001 if (error) { 1002 return error; 1003 } 1004 1005 error = copyout(&ret_k, ret_u, sizeof(ret_k)); 1006 if (error) { 1007 return error; 1008 } 1009 1010 return copyout(&cursor_k, cursor_u, sizeof(cursor_k)); 1011 } 1012 1013 static int 1014 do_sys_quotactl_cursorrewind(struct mount *mp, struct quotakcursor *cursor_u) 1015 { 1016 struct quotakcursor cursor_k; 1017 int error; 1018 1019 error = copyin(cursor_u, &cursor_k, sizeof(cursor_k)); 1020 if (error) { 1021 return error; 1022 } 1023 1024 error = vfs_quotactl_cursorrewind(mp, &cursor_k); 1025 if (error) { 1026 return error; 1027 } 1028 1029 return copyout(&cursor_k, cursor_u, sizeof(cursor_k)); 1030 } 1031 1032 static int 1033 do_sys_quotactl_quotaon(struct mount *mp, int idtype, const char *path_u) 1034 { 1035 char *path_k; 1036 int error; 1037 1038 /* XXX this should probably be a struct pathbuf */ 1039 path_k = PNBUF_GET(); 1040 error = copyin(path_u, path_k, PATH_MAX); 1041 if (error) { 1042 PNBUF_PUT(path_k); 1043 return error; 1044 } 1045 1046 error = vfs_quotactl_quotaon(mp, idtype, path_k); 1047 1048 PNBUF_PUT(path_k); 1049 return error; 1050 } 1051 1052 static int 1053 do_sys_quotactl_quotaoff(struct mount *mp, int idtype) 1054 { 1055 return vfs_quotactl_quotaoff(mp, idtype); 1056 } 1057 1058 int 1059 do_sys_quotactl(const char *path_u, const struct quotactl_args *args) 1060 { 1061 struct mount *mp; 1062 struct vnode *vp; 1063 int error; 1064 1065 error = namei_simple_user(path_u, NSM_FOLLOW_TRYEMULROOT, &vp); 1066 if (error != 0) 1067 return (error); 1068 mp = vp->v_mount; 1069 1070 switch (args->qc_op) { 1071 case QUOTACTL_STAT: 1072 error = do_sys_quotactl_stat(mp, args->u.stat.qc_info); 1073 break; 1074 case QUOTACTL_IDTYPESTAT: 1075 error = do_sys_quotactl_idtypestat(mp, 1076 args->u.idtypestat.qc_idtype, 1077 args->u.idtypestat.qc_info); 1078 break; 1079 case QUOTACTL_OBJTYPESTAT: 1080 error = do_sys_quotactl_objtypestat(mp, 1081 args->u.objtypestat.qc_objtype, 1082 args->u.objtypestat.qc_info); 1083 break; 1084 case QUOTACTL_GET: 1085 error = do_sys_quotactl_get(mp, 1086 args->u.get.qc_key, 1087 args->u.get.qc_val); 1088 break; 1089 case QUOTACTL_PUT: 1090 error = do_sys_quotactl_put(mp, 1091 args->u.put.qc_key, 1092 args->u.put.qc_val); 1093 break; 1094 case QUOTACTL_DEL: 1095 error = do_sys_quotactl_del(mp, args->u.del.qc_key); 1096 break; 1097 case QUOTACTL_CURSOROPEN: 1098 error = do_sys_quotactl_cursoropen(mp, 1099 args->u.cursoropen.qc_cursor); 1100 break; 1101 case QUOTACTL_CURSORCLOSE: 1102 error = do_sys_quotactl_cursorclose(mp, 1103 args->u.cursorclose.qc_cursor); 1104 break; 1105 case QUOTACTL_CURSORSKIPIDTYPE: 1106 error = do_sys_quotactl_cursorskipidtype(mp, 1107 args->u.cursorskipidtype.qc_cursor, 1108 args->u.cursorskipidtype.qc_idtype); 1109 break; 1110 case QUOTACTL_CURSORGET: 1111 error = do_sys_quotactl_cursorget(mp, 1112 args->u.cursorget.qc_cursor, 1113 args->u.cursorget.qc_keys, 1114 args->u.cursorget.qc_vals, 1115 args->u.cursorget.qc_maxnum, 1116 args->u.cursorget.qc_ret); 1117 break; 1118 case QUOTACTL_CURSORATEND: 1119 error = do_sys_quotactl_cursoratend(mp, 1120 args->u.cursoratend.qc_cursor, 1121 args->u.cursoratend.qc_ret); 1122 break; 1123 case QUOTACTL_CURSORREWIND: 1124 error = do_sys_quotactl_cursorrewind(mp, 1125 args->u.cursorrewind.qc_cursor); 1126 break; 1127 case QUOTACTL_QUOTAON: 1128 error = do_sys_quotactl_quotaon(mp, 1129 args->u.quotaon.qc_idtype, 1130 args->u.quotaon.qc_quotafile); 1131 break; 1132 case QUOTACTL_QUOTAOFF: 1133 error = do_sys_quotactl_quotaoff(mp, 1134 args->u.quotaoff.qc_idtype); 1135 break; 1136 default: 1137 error = EINVAL; 1138 break; 1139 } 1140 1141 vrele(vp); 1142 return error; 1143 } 1144 1145 /* ARGSUSED */ 1146 int 1147 sys___quotactl(struct lwp *l, const struct sys___quotactl_args *uap, 1148 register_t *retval) 1149 { 1150 /* { 1151 syscallarg(const char *) path; 1152 syscallarg(struct quotactl_args *) args; 1153 } */ 1154 struct quotactl_args args; 1155 int error; 1156 1157 error = copyin(SCARG(uap, args), &args, sizeof(args)); 1158 if (error) { 1159 return error; 1160 } 1161 1162 return do_sys_quotactl(SCARG(uap, path), &args); 1163 } 1164 1165 int 1166 dostatvfs(struct mount *mp, struct statvfs *sp, struct lwp *l, int flags, 1167 int root) 1168 { 1169 struct vnode *rvp; 1170 int error = 0; 1171 1172 /* 1173 * If MNT_NOWAIT or MNT_LAZY is specified, do not 1174 * refresh the fsstat cache. MNT_WAIT or MNT_LAZY 1175 * overrides MNT_NOWAIT. 1176 */ 1177 KASSERT(l == curlwp); 1178 rvp = cwdrdir(); 1179 if (flags == MNT_NOWAIT || flags == MNT_LAZY || 1180 (flags != MNT_WAIT && flags != 0)) { 1181 memcpy(sp, &mp->mnt_stat, sizeof(*sp)); 1182 } else { 1183 /* Get the filesystem stats now */ 1184 memset(sp, 0, sizeof(*sp)); 1185 if ((error = VFS_STATVFS(mp, sp)) != 0) { 1186 if (rvp) 1187 vrele(rvp); 1188 return error; 1189 } 1190 if (rvp == NULL) 1191 (void)memcpy(&mp->mnt_stat, sp, sizeof(mp->mnt_stat)); 1192 } 1193 1194 if (rvp != NULL) { 1195 size_t len; 1196 char *bp; 1197 char c; 1198 char *path = PNBUF_GET(); 1199 1200 bp = path + MAXPATHLEN; 1201 *--bp = '\0'; 1202 error = getcwd_common(rvp, rootvnode, &bp, path, 1203 MAXPATHLEN / 2, 0, l); 1204 if (error) { 1205 PNBUF_PUT(path); 1206 vrele(rvp); 1207 return error; 1208 } 1209 len = strlen(bp); 1210 if (len != 1) { 1211 /* 1212 * for mount points that are below our root, we can see 1213 * them, so we fix up the pathname and return them. The 1214 * rest we cannot see, so we don't allow viewing the 1215 * data. 1216 */ 1217 if (strncmp(bp, sp->f_mntonname, len) == 0 && 1218 ((c = sp->f_mntonname[len]) == '/' || c == '\0')) { 1219 (void)strlcpy(sp->f_mntonname, 1220 c == '\0' ? "/" : &sp->f_mntonname[len], 1221 sizeof(sp->f_mntonname)); 1222 } else { 1223 if (root) 1224 (void)strlcpy(sp->f_mntonname, "/", 1225 sizeof(sp->f_mntonname)); 1226 else 1227 error = EPERM; 1228 } 1229 } 1230 PNBUF_PUT(path); 1231 vrele(rvp); 1232 } 1233 sp->f_flag = mp->mnt_flag & MNT_VISFLAGMASK; 1234 return error; 1235 } 1236 1237 /* 1238 * Get filesystem statistics by path. 1239 */ 1240 int 1241 do_sys_pstatvfs(struct lwp *l, const char *path, int flags, struct statvfs *sb) 1242 { 1243 struct mount *mp; 1244 int error; 1245 struct vnode *vp; 1246 1247 error = namei_simple_user(path, NSM_FOLLOW_TRYEMULROOT, &vp); 1248 if (error != 0) 1249 return error; 1250 mp = vp->v_mount; 1251 error = dostatvfs(mp, sb, l, flags, 1); 1252 vrele(vp); 1253 return error; 1254 } 1255 1256 /* ARGSUSED */ 1257 int 1258 sys___statvfs190(struct lwp *l, const struct sys___statvfs190_args *uap, register_t *retval) 1259 { 1260 /* { 1261 syscallarg(const char *) path; 1262 syscallarg(struct statvfs *) buf; 1263 syscallarg(int) flags; 1264 } */ 1265 struct statvfs *sb; 1266 int error; 1267 1268 sb = STATVFSBUF_GET(); 1269 error = do_sys_pstatvfs(l, SCARG(uap, path), SCARG(uap, flags), sb); 1270 if (error == 0) 1271 error = copyout(sb, SCARG(uap, buf), sizeof(*sb)); 1272 STATVFSBUF_PUT(sb); 1273 return error; 1274 } 1275 1276 /* 1277 * Get filesystem statistics by fd. 1278 */ 1279 int 1280 do_sys_fstatvfs(struct lwp *l, int fd, int flags, struct statvfs *sb) 1281 { 1282 file_t *fp; 1283 struct mount *mp; 1284 int error; 1285 1286 /* fd_getvnode() will use the descriptor for us */ 1287 if ((error = fd_getvnode(fd, &fp)) != 0) 1288 return (error); 1289 mp = fp->f_vnode->v_mount; 1290 error = dostatvfs(mp, sb, curlwp, flags, 1); 1291 fd_putfile(fd); 1292 return error; 1293 } 1294 1295 /* ARGSUSED */ 1296 int 1297 sys___fstatvfs190(struct lwp *l, const struct sys___fstatvfs190_args *uap, register_t *retval) 1298 { 1299 /* { 1300 syscallarg(int) fd; 1301 syscallarg(struct statvfs *) buf; 1302 syscallarg(int) flags; 1303 } */ 1304 struct statvfs *sb; 1305 int error; 1306 1307 sb = STATVFSBUF_GET(); 1308 error = do_sys_fstatvfs(l, SCARG(uap, fd), SCARG(uap, flags), sb); 1309 if (error == 0) 1310 error = copyout(sb, SCARG(uap, buf), sizeof(*sb)); 1311 STATVFSBUF_PUT(sb); 1312 return error; 1313 } 1314 1315 1316 /* 1317 * Get statistics on all filesystems. 1318 */ 1319 int 1320 do_sys_getvfsstat(struct lwp *l, void *sfsp, size_t bufsize, int flags, 1321 int (*copyfn)(const void *, void *, size_t), size_t entry_sz, 1322 register_t *retval) 1323 { 1324 int root = 0; 1325 mount_iterator_t *iter; 1326 struct proc *p = l->l_proc; 1327 struct mount *mp; 1328 struct statvfs *sb; 1329 size_t count, maxcount; 1330 int error = 0; 1331 1332 sb = STATVFSBUF_GET(); 1333 maxcount = bufsize / entry_sz; 1334 count = 0; 1335 mountlist_iterator_init(&iter); 1336 while ((mp = mountlist_iterator_next(iter)) != NULL) { 1337 if (sfsp && count < maxcount) { 1338 error = dostatvfs(mp, sb, l, flags, 0); 1339 if (error) { 1340 error = 0; 1341 continue; 1342 } 1343 error = copyfn(sb, sfsp, entry_sz); 1344 if (error) 1345 goto out; 1346 sfsp = (char *)sfsp + entry_sz; 1347 root |= strcmp(sb->f_mntonname, "/") == 0; 1348 } 1349 count++; 1350 } 1351 1352 if (root == 0 && p->p_cwdi->cwdi_rdir) { 1353 /* 1354 * fake a root entry 1355 */ 1356 error = dostatvfs(p->p_cwdi->cwdi_rdir->v_mount, 1357 sb, l, flags, 1); 1358 if (error != 0) 1359 goto out; 1360 if (sfsp) { 1361 error = copyfn(sb, sfsp, entry_sz); 1362 if (error != 0) 1363 goto out; 1364 } 1365 count++; 1366 } 1367 if (sfsp && count > maxcount) 1368 *retval = maxcount; 1369 else 1370 *retval = count; 1371 out: 1372 mountlist_iterator_destroy(iter); 1373 STATVFSBUF_PUT(sb); 1374 return error; 1375 } 1376 1377 int 1378 sys___getvfsstat90(struct lwp *l, const struct sys___getvfsstat90_args *uap, 1379 register_t *retval) 1380 { 1381 /* { 1382 syscallarg(struct statvfs *) buf; 1383 syscallarg(size_t) bufsize; 1384 syscallarg(int) flags; 1385 } */ 1386 1387 return do_sys_getvfsstat(l, SCARG(uap, buf), SCARG(uap, bufsize), 1388 SCARG(uap, flags), copyout, sizeof (struct statvfs), retval); 1389 } 1390 1391 /* 1392 * Change current working directory to a given file descriptor. 1393 */ 1394 /* ARGSUSED */ 1395 int 1396 sys_fchdir(struct lwp *l, const struct sys_fchdir_args *uap, register_t *retval) 1397 { 1398 /* { 1399 syscallarg(int) fd; 1400 } */ 1401 struct cwdinfo *cwdi; 1402 struct vnode *vp, *tdp; 1403 struct mount *mp; 1404 file_t *fp; 1405 int error, fd; 1406 1407 /* fd_getvnode() will use the descriptor for us */ 1408 fd = SCARG(uap, fd); 1409 if ((error = fd_getvnode(fd, &fp)) != 0) 1410 return (error); 1411 vp = fp->f_vnode; 1412 1413 vref(vp); 1414 vn_lock(vp, LK_SHARED | LK_RETRY); 1415 if (vp->v_type != VDIR) 1416 error = ENOTDIR; 1417 else 1418 error = VOP_ACCESS(vp, VEXEC, l->l_cred); 1419 if (error) { 1420 vput(vp); 1421 goto out; 1422 } 1423 while ((mp = vp->v_mountedhere) != NULL) { 1424 error = vfs_busy(mp); 1425 vput(vp); 1426 if (error != 0) 1427 goto out; 1428 error = VFS_ROOT(mp, LK_SHARED, &tdp); 1429 vfs_unbusy(mp); 1430 if (error) 1431 goto out; 1432 vp = tdp; 1433 } 1434 VOP_UNLOCK(vp); 1435 1436 /* 1437 * Disallow changing to a directory not under the process's 1438 * current root directory (if there is one). 1439 */ 1440 cwdi = cwdenter(RW_WRITER); 1441 if (cwdi->cwdi_rdir && !vn_isunder(vp, NULL, l)) { 1442 vrele(vp); 1443 error = EPERM; /* operation not permitted */ 1444 } else { 1445 vrele(cwdi->cwdi_cdir); 1446 cwdi->cwdi_cdir = vp; 1447 } 1448 cwdexit(cwdi); 1449 1450 out: 1451 fd_putfile(fd); 1452 return (error); 1453 } 1454 1455 /* 1456 * Change this process's notion of the root directory to a given file 1457 * descriptor. 1458 */ 1459 int 1460 sys_fchroot(struct lwp *l, const struct sys_fchroot_args *uap, register_t *retval) 1461 { 1462 struct vnode *vp; 1463 file_t *fp; 1464 int error, fd = SCARG(uap, fd); 1465 1466 if ((error = kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_CHROOT, 1467 KAUTH_REQ_SYSTEM_CHROOT_FCHROOT, NULL, NULL, NULL)) != 0) 1468 return error; 1469 /* fd_getvnode() will use the descriptor for us */ 1470 if ((error = fd_getvnode(fd, &fp)) != 0) 1471 return error; 1472 vp = fp->f_vnode; 1473 vn_lock(vp, LK_SHARED | LK_RETRY); 1474 if (vp->v_type != VDIR) 1475 error = ENOTDIR; 1476 else 1477 error = VOP_ACCESS(vp, VEXEC, l->l_cred); 1478 VOP_UNLOCK(vp); 1479 if (error) 1480 goto out; 1481 vref(vp); 1482 change_root(vp); 1483 1484 out: 1485 fd_putfile(fd); 1486 return (error); 1487 } 1488 1489 /* 1490 * Change current working directory (``.''). 1491 */ 1492 /* ARGSUSED */ 1493 int 1494 sys_chdir(struct lwp *l, const struct sys_chdir_args *uap, register_t *retval) 1495 { 1496 /* { 1497 syscallarg(const char *) path; 1498 } */ 1499 struct cwdinfo *cwdi; 1500 int error; 1501 struct vnode *vp, *ovp; 1502 1503 error = chdir_lookup(SCARG(uap, path), UIO_USERSPACE, &vp, l); 1504 if (error != 0) 1505 return (error); 1506 1507 cwdi = cwdenter(RW_WRITER); 1508 ovp = cwdi->cwdi_cdir; 1509 cwdi->cwdi_cdir = vp; 1510 cwdexit(cwdi); 1511 vrele(ovp); 1512 return (0); 1513 } 1514 1515 /* 1516 * Change notion of root (``/'') directory. 1517 */ 1518 /* ARGSUSED */ 1519 int 1520 sys_chroot(struct lwp *l, const struct sys_chroot_args *uap, register_t *retval) 1521 { 1522 /* { 1523 syscallarg(const char *) path; 1524 } */ 1525 int error; 1526 struct vnode *vp; 1527 1528 if ((error = kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_CHROOT, 1529 KAUTH_REQ_SYSTEM_CHROOT_CHROOT, NULL, NULL, NULL)) != 0) 1530 return (error); 1531 1532 error = chdir_lookup(SCARG(uap, path), UIO_USERSPACE, &vp, l); 1533 if (error == 0) 1534 change_root(vp); 1535 return error; 1536 } 1537 1538 /* 1539 * Common routine for chroot and fchroot. 1540 * NB: callers need to properly authorize the change root operation. 1541 */ 1542 void 1543 change_root(struct vnode *vp) 1544 { 1545 struct cwdinfo *cwdi; 1546 kauth_cred_t ncred; 1547 struct lwp *l = curlwp; 1548 struct proc *p = l->l_proc; 1549 1550 ncred = kauth_cred_alloc(); 1551 1552 cwdi = cwdenter(RW_WRITER); 1553 if (cwdi->cwdi_rdir != NULL) 1554 vrele(cwdi->cwdi_rdir); 1555 cwdi->cwdi_rdir = vp; 1556 1557 /* 1558 * Prevent escaping from chroot by putting the root under 1559 * the working directory. Silently chdir to / if we aren't 1560 * already there. 1561 */ 1562 if (!vn_isunder(cwdi->cwdi_cdir, vp, l)) { 1563 /* 1564 * XXX would be more failsafe to change directory to a 1565 * deadfs node here instead 1566 */ 1567 vrele(cwdi->cwdi_cdir); 1568 vref(vp); 1569 cwdi->cwdi_cdir = vp; 1570 } 1571 cwdexit(cwdi); 1572 1573 /* Get a write lock on the process credential. */ 1574 proc_crmod_enter(); 1575 1576 kauth_cred_clone(p->p_cred, ncred); 1577 kauth_proc_chroot(ncred, p->p_cwdi); 1578 1579 /* Broadcast our credentials to the process and other LWPs. */ 1580 proc_crmod_leave(ncred, p->p_cred, true); 1581 } 1582 1583 /* 1584 * Common routine for chroot and chdir. 1585 * XXX "where" should be enum uio_seg 1586 */ 1587 int 1588 chdir_lookup(const char *path, int where, struct vnode **vpp, struct lwp *l) 1589 { 1590 struct pathbuf *pb; 1591 struct nameidata nd; 1592 int error; 1593 1594 error = pathbuf_maybe_copyin(path, where, &pb); 1595 if (error) { 1596 return error; 1597 } 1598 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | LOCKSHARED | TRYEMULROOT, pb); 1599 if ((error = namei(&nd)) != 0) { 1600 pathbuf_destroy(pb); 1601 return error; 1602 } 1603 *vpp = nd.ni_vp; 1604 pathbuf_destroy(pb); 1605 1606 if ((*vpp)->v_type != VDIR) 1607 error = ENOTDIR; 1608 else 1609 error = VOP_ACCESS(*vpp, VEXEC, l->l_cred); 1610 1611 if (error) 1612 vput(*vpp); 1613 else 1614 VOP_UNLOCK(*vpp); 1615 return (error); 1616 } 1617 1618 /* 1619 * Internals of sys_open - path has already been converted into a pathbuf 1620 * (so we can easily reuse this function from other parts of the kernel, 1621 * like posix_spawn post-processing). 1622 */ 1623 int 1624 do_open(lwp_t *l, struct vnode *dvp, struct pathbuf *pb, int open_flags, 1625 int open_mode, int *fd) 1626 { 1627 struct proc *p = l->l_proc; 1628 struct cwdinfo *cwdi = p->p_cwdi; 1629 file_t *fp; 1630 struct vnode *vp; 1631 int flags, cmode; 1632 int indx, error; 1633 struct nameidata nd; 1634 1635 if (open_flags & O_SEARCH) { 1636 open_flags &= ~(int)O_SEARCH; 1637 } 1638 1639 /* 1640 * Only one of the O_EXEC, O_RDONLY, O_WRONLY and O_RDWR flags 1641 * may be specified. 1642 */ 1643 if ((open_flags & O_EXEC) && (open_flags & O_ACCMODE)) 1644 return EINVAL; 1645 1646 flags = FFLAGS(open_flags); 1647 if ((flags & (FREAD | FWRITE)) == 0) 1648 return EINVAL; 1649 1650 if ((error = fd_allocfile(&fp, &indx)) != 0) { 1651 return error; 1652 } 1653 1654 /* We're going to read cwdi->cwdi_cmask unlocked here. */ 1655 cmode = ((open_mode &~ cwdi->cwdi_cmask) & ALLPERMS) &~ S_ISTXT; 1656 NDINIT(&nd, LOOKUP, FOLLOW | TRYEMULROOT, pb); 1657 if (dvp != NULL) 1658 NDAT(&nd, dvp); 1659 1660 l->l_dupfd = -indx - 1; /* XXX check for fdopen */ 1661 if ((error = vn_open(&nd, flags, cmode)) != 0) { 1662 fd_abort(p, fp, indx); 1663 if ((error == EDUPFD || error == EMOVEFD) && 1664 l->l_dupfd >= 0 && /* XXX from fdopen */ 1665 (error = 1666 fd_dupopen(l->l_dupfd, &indx, flags, error)) == 0) { 1667 *fd = indx; 1668 return 0; 1669 } 1670 if (error == ERESTART) 1671 error = EINTR; 1672 return error; 1673 } 1674 1675 l->l_dupfd = 0; 1676 vp = nd.ni_vp; 1677 1678 if ((error = open_setfp(l, fp, vp, indx, flags))) 1679 return error; 1680 1681 VOP_UNLOCK(vp); 1682 *fd = indx; 1683 fd_affix(p, fp, indx); 1684 return 0; 1685 } 1686 1687 int 1688 fd_open(const char *path, int open_flags, int open_mode, int *fd) 1689 { 1690 struct pathbuf *pb; 1691 int error, oflags; 1692 1693 oflags = FFLAGS(open_flags); 1694 if ((oflags & (FREAD | FWRITE)) == 0) 1695 return EINVAL; 1696 1697 pb = pathbuf_create(path); 1698 if (pb == NULL) 1699 return ENOMEM; 1700 1701 error = do_open(curlwp, NULL, pb, open_flags, open_mode, fd); 1702 pathbuf_destroy(pb); 1703 1704 return error; 1705 } 1706 1707 static int 1708 do_sys_openat(lwp_t *l, int fdat, const char *path, int flags, 1709 int mode, int *fd) 1710 { 1711 file_t *dfp = NULL; 1712 struct vnode *dvp = NULL; 1713 struct pathbuf *pb; 1714 const char *pathstring = NULL; 1715 int error; 1716 1717 if (path == NULL) { 1718 MODULE_HOOK_CALL(vfs_openat_10_hook, (&pb), enosys(), error); 1719 if (error == ENOSYS) 1720 goto no_compat; 1721 if (error) 1722 return error; 1723 } else { 1724 no_compat: 1725 error = pathbuf_copyin(path, &pb); 1726 if (error) 1727 return error; 1728 } 1729 1730 pathstring = pathbuf_stringcopy_get(pb); 1731 1732 /* 1733 * fdat is ignored if: 1734 * 1) if fdat is AT_FDCWD, which means use current directory as base. 1735 * 2) if path is absolute, then fdat is useless. 1736 */ 1737 if (fdat != AT_FDCWD && pathstring[0] != '/') { 1738 /* fd_getvnode() will use the descriptor for us */ 1739 if ((error = fd_getvnode(fdat, &dfp)) != 0) 1740 goto out; 1741 1742 dvp = dfp->f_vnode; 1743 } 1744 1745 error = do_open(l, dvp, pb, flags, mode, fd); 1746 1747 if (dfp != NULL) 1748 fd_putfile(fdat); 1749 out: 1750 pathbuf_stringcopy_put(pb, pathstring); 1751 pathbuf_destroy(pb); 1752 return error; 1753 } 1754 1755 int 1756 sys_open(struct lwp *l, const struct sys_open_args *uap, register_t *retval) 1757 { 1758 /* { 1759 syscallarg(const char *) path; 1760 syscallarg(int) flags; 1761 syscallarg(int) mode; 1762 } */ 1763 int error; 1764 int fd; 1765 1766 error = do_sys_openat(l, AT_FDCWD, SCARG(uap, path), 1767 SCARG(uap, flags), SCARG(uap, mode), &fd); 1768 1769 if (error == 0) 1770 *retval = fd; 1771 1772 return error; 1773 } 1774 1775 int 1776 sys_openat(struct lwp *l, const struct sys_openat_args *uap, register_t *retval) 1777 { 1778 /* { 1779 syscallarg(int) fd; 1780 syscallarg(const char *) path; 1781 syscallarg(int) oflags; 1782 syscallarg(int) mode; 1783 } */ 1784 int error; 1785 int fd; 1786 1787 error = do_sys_openat(l, SCARG(uap, fd), SCARG(uap, path), 1788 SCARG(uap, oflags), SCARG(uap, mode), &fd); 1789 1790 if (error == 0) 1791 *retval = fd; 1792 1793 return error; 1794 } 1795 1796 static void 1797 vfs__fhfree(fhandle_t *fhp) 1798 { 1799 size_t fhsize; 1800 1801 fhsize = FHANDLE_SIZE(fhp); 1802 kmem_free(fhp, fhsize); 1803 } 1804 1805 /* 1806 * vfs_composefh: compose a filehandle. 1807 */ 1808 1809 int 1810 vfs_composefh(struct vnode *vp, fhandle_t *fhp, size_t *fh_size) 1811 { 1812 struct mount *mp; 1813 struct fid *fidp; 1814 int error; 1815 size_t needfhsize; 1816 size_t fidsize; 1817 1818 mp = vp->v_mount; 1819 fidp = NULL; 1820 if (*fh_size < FHANDLE_SIZE_MIN) { 1821 fidsize = 0; 1822 } else { 1823 fidsize = *fh_size - offsetof(fhandle_t, fh_fid); 1824 if (fhp != NULL) { 1825 memset(fhp, 0, *fh_size); 1826 fhp->fh_fsid = mp->mnt_stat.f_fsidx; 1827 fidp = &fhp->fh_fid; 1828 } 1829 } 1830 error = VFS_VPTOFH(vp, fidp, &fidsize); 1831 needfhsize = FHANDLE_SIZE_FROM_FILEID_SIZE(fidsize); 1832 if (error == 0 && *fh_size < needfhsize) { 1833 error = E2BIG; 1834 } 1835 *fh_size = needfhsize; 1836 return error; 1837 } 1838 1839 int 1840 vfs_composefh_alloc(struct vnode *vp, fhandle_t **fhpp) 1841 { 1842 struct mount *mp; 1843 fhandle_t *fhp; 1844 size_t fhsize; 1845 size_t fidsize; 1846 int error; 1847 1848 mp = vp->v_mount; 1849 fidsize = 0; 1850 error = VFS_VPTOFH(vp, NULL, &fidsize); 1851 KASSERT(error != 0); 1852 if (error != E2BIG) { 1853 goto out; 1854 } 1855 fhsize = FHANDLE_SIZE_FROM_FILEID_SIZE(fidsize); 1856 fhp = kmem_zalloc(fhsize, KM_SLEEP); 1857 fhp->fh_fsid = mp->mnt_stat.f_fsidx; 1858 error = VFS_VPTOFH(vp, &fhp->fh_fid, &fidsize); 1859 if (error == 0) { 1860 KASSERT((FHANDLE_SIZE(fhp) == fhsize && 1861 FHANDLE_FILEID(fhp)->fid_len == fidsize)); 1862 *fhpp = fhp; 1863 } else { 1864 kmem_free(fhp, fhsize); 1865 } 1866 out: 1867 return error; 1868 } 1869 1870 void 1871 vfs_composefh_free(fhandle_t *fhp) 1872 { 1873 1874 vfs__fhfree(fhp); 1875 } 1876 1877 /* 1878 * vfs_fhtovp: lookup a vnode by a filehandle. 1879 */ 1880 1881 int 1882 vfs_fhtovp(fhandle_t *fhp, struct vnode **vpp) 1883 { 1884 struct mount *mp; 1885 int error; 1886 1887 *vpp = NULL; 1888 mp = vfs_getvfs(FHANDLE_FSID(fhp)); 1889 if (mp == NULL) { 1890 error = ESTALE; 1891 goto out; 1892 } 1893 if (mp->mnt_op->vfs_fhtovp == NULL) { 1894 error = EOPNOTSUPP; 1895 goto out; 1896 } 1897 error = VFS_FHTOVP(mp, FHANDLE_FILEID(fhp), LK_EXCLUSIVE, vpp); 1898 out: 1899 return error; 1900 } 1901 1902 /* 1903 * vfs_copyinfh_alloc: allocate and copyin a filehandle, given 1904 * the needed size. 1905 */ 1906 1907 int 1908 vfs_copyinfh_alloc(const void *ufhp, size_t fhsize, fhandle_t **fhpp) 1909 { 1910 fhandle_t *fhp; 1911 int error; 1912 1913 if (fhsize > FHANDLE_SIZE_MAX) { 1914 return EINVAL; 1915 } 1916 if (fhsize < FHANDLE_SIZE_MIN) { 1917 return EINVAL; 1918 } 1919 again: 1920 fhp = kmem_alloc(fhsize, KM_SLEEP); 1921 error = copyin(ufhp, fhp, fhsize); 1922 if (error == 0) { 1923 /* XXX this check shouldn't be here */ 1924 if (FHANDLE_SIZE(fhp) == fhsize) { 1925 *fhpp = fhp; 1926 return 0; 1927 } else if (fhsize == NFSX_V2FH && FHANDLE_SIZE(fhp) < fhsize) { 1928 /* 1929 * a kludge for nfsv2 padded handles. 1930 */ 1931 size_t sz; 1932 1933 sz = FHANDLE_SIZE(fhp); 1934 kmem_free(fhp, fhsize); 1935 fhsize = sz; 1936 goto again; 1937 } else { 1938 /* 1939 * userland told us wrong size. 1940 */ 1941 error = EINVAL; 1942 } 1943 } 1944 kmem_free(fhp, fhsize); 1945 return error; 1946 } 1947 1948 void 1949 vfs_copyinfh_free(fhandle_t *fhp) 1950 { 1951 1952 vfs__fhfree(fhp); 1953 } 1954 1955 /* 1956 * Get file handle system call 1957 */ 1958 int 1959 sys___getfh30(struct lwp *l, const struct sys___getfh30_args *uap, register_t *retval) 1960 { 1961 /* { 1962 syscallarg(char *) fname; 1963 syscallarg(fhandle_t *) fhp; 1964 syscallarg(size_t *) fh_size; 1965 } */ 1966 struct vnode *vp; 1967 fhandle_t *fh; 1968 int error; 1969 struct pathbuf *pb; 1970 struct nameidata nd; 1971 size_t sz; 1972 size_t usz; 1973 1974 /* 1975 * Must be super user 1976 */ 1977 error = kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_FILEHANDLE, 1978 0, NULL, NULL, NULL); 1979 if (error) 1980 return (error); 1981 1982 error = pathbuf_copyin(SCARG(uap, fname), &pb); 1983 if (error) { 1984 return error; 1985 } 1986 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | TRYEMULROOT, pb); 1987 error = namei(&nd); 1988 if (error) { 1989 pathbuf_destroy(pb); 1990 return error; 1991 } 1992 vp = nd.ni_vp; 1993 pathbuf_destroy(pb); 1994 1995 error = vfs_composefh_alloc(vp, &fh); 1996 vput(vp); 1997 if (error != 0) { 1998 return error; 1999 } 2000 error = copyin(SCARG(uap, fh_size), &usz, sizeof(size_t)); 2001 if (error != 0) { 2002 goto out; 2003 } 2004 sz = FHANDLE_SIZE(fh); 2005 error = copyout(&sz, SCARG(uap, fh_size), sizeof(size_t)); 2006 if (error != 0) { 2007 goto out; 2008 } 2009 if (usz >= sz) { 2010 error = copyout(fh, SCARG(uap, fhp), sz); 2011 } else { 2012 error = E2BIG; 2013 } 2014 out: 2015 vfs_composefh_free(fh); 2016 return (error); 2017 } 2018 2019 /* 2020 * Open a file given a file handle. 2021 * 2022 * Check permissions, allocate an open file structure, 2023 * and call the device open routine if any. 2024 */ 2025 2026 int 2027 dofhopen(struct lwp *l, const void *ufhp, size_t fhsize, int oflags, 2028 register_t *retval) 2029 { 2030 file_t *fp; 2031 struct vnode *vp = NULL; 2032 kauth_cred_t cred = l->l_cred; 2033 file_t *nfp; 2034 int indx, error; 2035 struct vattr va; 2036 fhandle_t *fh; 2037 int flags; 2038 proc_t *p; 2039 2040 p = curproc; 2041 2042 /* 2043 * Must be super user 2044 */ 2045 if ((error = kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_FILEHANDLE, 2046 0, NULL, NULL, NULL))) 2047 return (error); 2048 2049 if (oflags & O_SEARCH) { 2050 oflags &= ~(int)O_SEARCH; 2051 } 2052 2053 flags = FFLAGS(oflags); 2054 if ((flags & (FREAD | FWRITE)) == 0) 2055 return (EINVAL); 2056 if ((flags & O_CREAT)) 2057 return (EINVAL); 2058 if ((error = fd_allocfile(&nfp, &indx)) != 0) 2059 return (error); 2060 fp = nfp; 2061 error = vfs_copyinfh_alloc(ufhp, fhsize, &fh); 2062 if (error != 0) { 2063 goto bad; 2064 } 2065 error = vfs_fhtovp(fh, &vp); 2066 vfs_copyinfh_free(fh); 2067 if (error != 0) { 2068 goto bad; 2069 } 2070 2071 /* Now do an effective vn_open */ 2072 2073 if (vp->v_type == VSOCK) { 2074 error = EOPNOTSUPP; 2075 goto bad; 2076 } 2077 error = vn_openchk(vp, cred, flags); 2078 if (error != 0) 2079 goto bad; 2080 if (flags & O_TRUNC) { 2081 VOP_UNLOCK(vp); /* XXX */ 2082 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); /* XXX */ 2083 vattr_null(&va); 2084 va.va_size = 0; 2085 error = VOP_SETATTR(vp, &va, cred); 2086 if (error) 2087 goto bad; 2088 } 2089 if ((error = VOP_OPEN(vp, flags, cred)) != 0) 2090 goto bad; 2091 if (flags & FWRITE) { 2092 mutex_enter(vp->v_interlock); 2093 vp->v_writecount++; 2094 mutex_exit(vp->v_interlock); 2095 } 2096 2097 /* done with modified vn_open, now finish what sys_open does. */ 2098 if ((error = open_setfp(l, fp, vp, indx, flags))) 2099 return error; 2100 2101 VOP_UNLOCK(vp); 2102 *retval = indx; 2103 fd_affix(p, fp, indx); 2104 return (0); 2105 2106 bad: 2107 fd_abort(p, fp, indx); 2108 if (vp != NULL) 2109 vput(vp); 2110 return (error); 2111 } 2112 2113 int 2114 sys___fhopen40(struct lwp *l, const struct sys___fhopen40_args *uap, register_t *retval) 2115 { 2116 /* { 2117 syscallarg(const void *) fhp; 2118 syscallarg(size_t) fh_size; 2119 syscallarg(int) flags; 2120 } */ 2121 2122 return dofhopen(l, SCARG(uap, fhp), SCARG(uap, fh_size), 2123 SCARG(uap, flags), retval); 2124 } 2125 2126 int 2127 do_fhstat(struct lwp *l, const void *ufhp, size_t fhsize, struct stat *sb) 2128 { 2129 int error; 2130 fhandle_t *fh; 2131 struct vnode *vp; 2132 2133 /* 2134 * Must be super user 2135 */ 2136 if ((error = kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_FILEHANDLE, 2137 0, NULL, NULL, NULL))) 2138 return (error); 2139 2140 error = vfs_copyinfh_alloc(ufhp, fhsize, &fh); 2141 if (error != 0) 2142 return error; 2143 2144 error = vfs_fhtovp(fh, &vp); 2145 vfs_copyinfh_free(fh); 2146 if (error != 0) 2147 return error; 2148 2149 error = vn_stat(vp, sb); 2150 vput(vp); 2151 return error; 2152 } 2153 2154 2155 /* ARGSUSED */ 2156 int 2157 sys___fhstat50(struct lwp *l, const struct sys___fhstat50_args *uap, register_t *retval) 2158 { 2159 /* { 2160 syscallarg(const void *) fhp; 2161 syscallarg(size_t) fh_size; 2162 syscallarg(struct stat *) sb; 2163 } */ 2164 struct stat sb; 2165 int error; 2166 2167 error = do_fhstat(l, SCARG(uap, fhp), SCARG(uap, fh_size), &sb); 2168 if (error) 2169 return error; 2170 return copyout(&sb, SCARG(uap, sb), sizeof(sb)); 2171 } 2172 2173 int 2174 do_fhstatvfs(struct lwp *l, const void *ufhp, size_t fhsize, struct statvfs *sb, 2175 int flags) 2176 { 2177 fhandle_t *fh; 2178 struct mount *mp; 2179 struct vnode *vp; 2180 int error; 2181 2182 /* 2183 * Must be super user 2184 */ 2185 if ((error = kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_FILEHANDLE, 2186 0, NULL, NULL, NULL))) 2187 return error; 2188 2189 error = vfs_copyinfh_alloc(ufhp, fhsize, &fh); 2190 if (error != 0) 2191 return error; 2192 2193 error = vfs_fhtovp(fh, &vp); 2194 vfs_copyinfh_free(fh); 2195 if (error != 0) 2196 return error; 2197 2198 mp = vp->v_mount; 2199 error = dostatvfs(mp, sb, l, flags, 1); 2200 vput(vp); 2201 return error; 2202 } 2203 2204 /* ARGSUSED */ 2205 int 2206 sys___fhstatvfs190(struct lwp *l, const struct sys___fhstatvfs190_args *uap, register_t *retval) 2207 { 2208 /* { 2209 syscallarg(const void *) fhp; 2210 syscallarg(size_t) fh_size; 2211 syscallarg(struct statvfs *) buf; 2212 syscallarg(int) flags; 2213 } */ 2214 struct statvfs *sb = STATVFSBUF_GET(); 2215 int error; 2216 2217 error = do_fhstatvfs(l, SCARG(uap, fhp), SCARG(uap, fh_size), sb, 2218 SCARG(uap, flags)); 2219 if (error == 0) 2220 error = copyout(sb, SCARG(uap, buf), sizeof(*sb)); 2221 STATVFSBUF_PUT(sb); 2222 return error; 2223 } 2224 2225 int 2226 do_posix_mknodat(struct lwp *l, int fdat, const char *pathname, mode_t mode, 2227 dev_t dev) 2228 { 2229 2230 /* 2231 * The POSIX mknod(2) call is an alias for mkfifo(2) for S_IFIFO 2232 * in mode and dev=0. 2233 * 2234 * In all the other cases it's implementation defined behavior. 2235 */ 2236 2237 if ((mode & S_IFIFO) && dev == 0) 2238 return do_sys_mkfifoat(l, fdat, pathname, mode); 2239 else 2240 return do_sys_mknodat(l, fdat, pathname, mode, dev, 2241 UIO_USERSPACE); 2242 } 2243 2244 /* 2245 * Create a special file. 2246 */ 2247 /* ARGSUSED */ 2248 int 2249 sys___mknod50(struct lwp *l, const struct sys___mknod50_args *uap, 2250 register_t *retval) 2251 { 2252 /* { 2253 syscallarg(const char *) path; 2254 syscallarg(mode_t) mode; 2255 syscallarg(dev_t) dev; 2256 } */ 2257 return do_posix_mknodat(l, AT_FDCWD, SCARG(uap, path), 2258 SCARG(uap, mode), SCARG(uap, dev)); 2259 } 2260 2261 int 2262 sys_mknodat(struct lwp *l, const struct sys_mknodat_args *uap, 2263 register_t *retval) 2264 { 2265 /* { 2266 syscallarg(int) fd; 2267 syscallarg(const char *) path; 2268 syscallarg(mode_t) mode; 2269 syscallarg(int) pad; 2270 syscallarg(dev_t) dev; 2271 } */ 2272 2273 return do_posix_mknodat(l, SCARG(uap, fd), SCARG(uap, path), 2274 SCARG(uap, mode), SCARG(uap, dev)); 2275 } 2276 2277 int 2278 do_sys_mknod(struct lwp *l, const char *pathname, mode_t mode, dev_t dev, 2279 enum uio_seg seg) 2280 { 2281 return do_sys_mknodat(l, AT_FDCWD, pathname, mode, dev, seg); 2282 } 2283 2284 int 2285 do_sys_mknodat(struct lwp *l, int fdat, const char *pathname, mode_t mode, 2286 dev_t dev, enum uio_seg seg) 2287 { 2288 struct proc *p = l->l_proc; 2289 struct vnode *vp; 2290 struct vattr vattr; 2291 int error, optype; 2292 struct pathbuf *pb; 2293 struct nameidata nd; 2294 const char *pathstring; 2295 2296 if ((error = kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_MKNOD, 2297 0, NULL, NULL, NULL)) != 0) 2298 return (error); 2299 2300 optype = VOP_MKNOD_DESCOFFSET; 2301 2302 error = pathbuf_maybe_copyin(pathname, seg, &pb); 2303 if (error) { 2304 return error; 2305 } 2306 pathstring = pathbuf_stringcopy_get(pb); 2307 if (pathstring == NULL) { 2308 pathbuf_destroy(pb); 2309 return ENOMEM; 2310 } 2311 2312 NDINIT(&nd, CREATE, LOCKPARENT | TRYEMULROOT, pb); 2313 2314 if ((error = fd_nameiat(l, fdat, &nd)) != 0) 2315 goto out; 2316 vp = nd.ni_vp; 2317 2318 if (vp != NULL) 2319 error = EEXIST; 2320 else { 2321 vattr_null(&vattr); 2322 /* We will read cwdi->cwdi_cmask unlocked. */ 2323 vattr.va_mode = (mode & ALLPERMS) &~ p->p_cwdi->cwdi_cmask; 2324 vattr.va_rdev = dev; 2325 2326 switch (mode & S_IFMT) { 2327 case S_IFMT: /* used by badsect to flag bad sectors */ 2328 vattr.va_type = VBAD; 2329 break; 2330 case S_IFCHR: 2331 vattr.va_type = VCHR; 2332 break; 2333 case S_IFBLK: 2334 vattr.va_type = VBLK; 2335 break; 2336 case S_IFWHT: 2337 optype = VOP_WHITEOUT_DESCOFFSET; 2338 break; 2339 case S_IFREG: 2340 #if NVERIEXEC > 0 2341 error = veriexec_openchk(l, nd.ni_vp, pathstring, 2342 O_CREAT); 2343 #endif /* NVERIEXEC > 0 */ 2344 vattr.va_type = VREG; 2345 vattr.va_rdev = VNOVAL; 2346 optype = VOP_CREATE_DESCOFFSET; 2347 break; 2348 default: 2349 error = EINVAL; 2350 break; 2351 } 2352 2353 if (error == 0 && optype == VOP_MKNOD_DESCOFFSET && 2354 vattr.va_rdev == VNOVAL) 2355 error = EINVAL; 2356 } 2357 2358 if (!error) { 2359 switch (optype) { 2360 case VOP_WHITEOUT_DESCOFFSET: 2361 error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, CREATE); 2362 if (error) 2363 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd); 2364 vput(nd.ni_dvp); 2365 break; 2366 2367 case VOP_MKNOD_DESCOFFSET: 2368 error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, 2369 &nd.ni_cnd, &vattr); 2370 if (error == 0) 2371 vrele(nd.ni_vp); 2372 vput(nd.ni_dvp); 2373 break; 2374 2375 case VOP_CREATE_DESCOFFSET: 2376 error = VOP_CREATE(nd.ni_dvp, &nd.ni_vp, 2377 &nd.ni_cnd, &vattr); 2378 if (error == 0) 2379 vrele(nd.ni_vp); 2380 vput(nd.ni_dvp); 2381 break; 2382 } 2383 } else { 2384 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd); 2385 if (nd.ni_dvp == vp) 2386 vrele(nd.ni_dvp); 2387 else 2388 vput(nd.ni_dvp); 2389 if (vp) 2390 vrele(vp); 2391 } 2392 out: 2393 pathbuf_stringcopy_put(pb, pathstring); 2394 pathbuf_destroy(pb); 2395 return (error); 2396 } 2397 2398 /* 2399 * Create a named pipe. 2400 */ 2401 /* ARGSUSED */ 2402 int 2403 sys_mkfifo(struct lwp *l, const struct sys_mkfifo_args *uap, register_t *retval) 2404 { 2405 /* { 2406 syscallarg(const char *) path; 2407 syscallarg(int) mode; 2408 } */ 2409 return do_sys_mkfifoat(l, AT_FDCWD, SCARG(uap, path), SCARG(uap, mode)); 2410 } 2411 2412 int 2413 sys_mkfifoat(struct lwp *l, const struct sys_mkfifoat_args *uap, 2414 register_t *retval) 2415 { 2416 /* { 2417 syscallarg(int) fd; 2418 syscallarg(const char *) path; 2419 syscallarg(int) mode; 2420 } */ 2421 2422 return do_sys_mkfifoat(l, SCARG(uap, fd), SCARG(uap, path), 2423 SCARG(uap, mode)); 2424 } 2425 2426 static int 2427 do_sys_mkfifoat(struct lwp *l, int fdat, const char *path, mode_t mode) 2428 { 2429 struct proc *p = l->l_proc; 2430 struct vattr vattr; 2431 int error; 2432 struct pathbuf *pb; 2433 struct nameidata nd; 2434 2435 error = pathbuf_copyin(path, &pb); 2436 if (error) { 2437 return error; 2438 } 2439 NDINIT(&nd, CREATE, LOCKPARENT | TRYEMULROOT, pb); 2440 2441 if ((error = fd_nameiat(l, fdat, &nd)) != 0) { 2442 pathbuf_destroy(pb); 2443 return error; 2444 } 2445 if (nd.ni_vp != NULL) { 2446 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd); 2447 if (nd.ni_dvp == nd.ni_vp) 2448 vrele(nd.ni_dvp); 2449 else 2450 vput(nd.ni_dvp); 2451 vrele(nd.ni_vp); 2452 pathbuf_destroy(pb); 2453 return (EEXIST); 2454 } 2455 vattr_null(&vattr); 2456 vattr.va_type = VFIFO; 2457 /* We will read cwdi->cwdi_cmask unlocked. */ 2458 vattr.va_mode = (mode & ALLPERMS) &~ p->p_cwdi->cwdi_cmask; 2459 error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr); 2460 if (error == 0) 2461 vrele(nd.ni_vp); 2462 vput(nd.ni_dvp); 2463 pathbuf_destroy(pb); 2464 return (error); 2465 } 2466 2467 /* 2468 * Make a hard file link. 2469 */ 2470 /* ARGSUSED */ 2471 int 2472 do_sys_linkat(struct lwp *l, int fdpath, const char *path, int fdlink, 2473 const char *link, int follow, register_t *retval) 2474 { 2475 struct vnode *vp; 2476 struct pathbuf *linkpb; 2477 struct nameidata nd; 2478 namei_simple_flags_t ns_flags; 2479 int error; 2480 2481 if (follow & AT_SYMLINK_FOLLOW) 2482 ns_flags = NSM_FOLLOW_TRYEMULROOT; 2483 else 2484 ns_flags = NSM_NOFOLLOW_TRYEMULROOT; 2485 2486 error = fd_nameiat_simple_user(l, fdpath, path, ns_flags, &vp); 2487 if (error != 0) 2488 return (error); 2489 error = pathbuf_copyin(link, &linkpb); 2490 if (error) { 2491 goto out1; 2492 } 2493 NDINIT(&nd, CREATE, LOCKPARENT | TRYEMULROOT, linkpb); 2494 if ((error = fd_nameiat(l, fdlink, &nd)) != 0) 2495 goto out2; 2496 if (nd.ni_vp) { 2497 error = EEXIST; 2498 goto abortop; 2499 } 2500 /* Prevent hard links on directories. */ 2501 if (vp->v_type == VDIR) { 2502 error = EPERM; 2503 goto abortop; 2504 } 2505 /* Prevent cross-mount operation. */ 2506 if (nd.ni_dvp->v_mount != vp->v_mount) { 2507 error = EXDEV; 2508 goto abortop; 2509 } 2510 error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd); 2511 VOP_UNLOCK(nd.ni_dvp); 2512 vrele(nd.ni_dvp); 2513 out2: 2514 pathbuf_destroy(linkpb); 2515 out1: 2516 vrele(vp); 2517 return (error); 2518 abortop: 2519 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd); 2520 if (nd.ni_dvp == nd.ni_vp) 2521 vrele(nd.ni_dvp); 2522 else 2523 vput(nd.ni_dvp); 2524 if (nd.ni_vp != NULL) 2525 vrele(nd.ni_vp); 2526 goto out2; 2527 } 2528 2529 int 2530 sys_link(struct lwp *l, const struct sys_link_args *uap, register_t *retval) 2531 { 2532 /* { 2533 syscallarg(const char *) path; 2534 syscallarg(const char *) link; 2535 } */ 2536 const char *path = SCARG(uap, path); 2537 const char *link = SCARG(uap, link); 2538 2539 return do_sys_linkat(l, AT_FDCWD, path, AT_FDCWD, link, 2540 AT_SYMLINK_FOLLOW, retval); 2541 } 2542 2543 int 2544 sys_linkat(struct lwp *l, const struct sys_linkat_args *uap, 2545 register_t *retval) 2546 { 2547 /* { 2548 syscallarg(int) fd1; 2549 syscallarg(const char *) name1; 2550 syscallarg(int) fd2; 2551 syscallarg(const char *) name2; 2552 syscallarg(int) flags; 2553 } */ 2554 int fd1 = SCARG(uap, fd1); 2555 const char *name1 = SCARG(uap, name1); 2556 int fd2 = SCARG(uap, fd2); 2557 const char *name2 = SCARG(uap, name2); 2558 int follow; 2559 2560 follow = SCARG(uap, flags) & AT_SYMLINK_FOLLOW; 2561 2562 return do_sys_linkat(l, fd1, name1, fd2, name2, follow, retval); 2563 } 2564 2565 2566 int 2567 do_sys_symlink(const char *patharg, const char *link, enum uio_seg seg) 2568 { 2569 return do_sys_symlinkat(NULL, patharg, AT_FDCWD, link, seg); 2570 } 2571 2572 static int 2573 do_sys_symlinkat(struct lwp *l, const char *patharg, int fdat, 2574 const char *link, enum uio_seg seg) 2575 { 2576 struct proc *p = curproc; 2577 struct vattr vattr; 2578 char *path; 2579 int error; 2580 size_t len; 2581 struct pathbuf *linkpb; 2582 struct nameidata nd; 2583 2584 KASSERT(l != NULL || fdat == AT_FDCWD); 2585 2586 path = PNBUF_GET(); 2587 if (seg == UIO_USERSPACE) { 2588 if ((error = copyinstr(patharg, path, MAXPATHLEN, &len)) != 0) 2589 goto out1; 2590 if ((error = pathbuf_copyin(link, &linkpb)) != 0) 2591 goto out1; 2592 } else { 2593 len = strlen(patharg) + 1; 2594 KASSERT(len <= MAXPATHLEN); 2595 memcpy(path, patharg, len); 2596 linkpb = pathbuf_create(link); 2597 if (linkpb == NULL) { 2598 error = ENOMEM; 2599 goto out1; 2600 } 2601 } 2602 ktrkuser("symlink-target", path, len - 1); 2603 2604 NDINIT(&nd, CREATE, LOCKPARENT | TRYEMULROOT, linkpb); 2605 if ((error = fd_nameiat(l, fdat, &nd)) != 0) 2606 goto out2; 2607 if (nd.ni_vp) { 2608 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd); 2609 if (nd.ni_dvp == nd.ni_vp) 2610 vrele(nd.ni_dvp); 2611 else 2612 vput(nd.ni_dvp); 2613 vrele(nd.ni_vp); 2614 error = EEXIST; 2615 goto out2; 2616 } 2617 vattr_null(&vattr); 2618 vattr.va_type = VLNK; 2619 /* We will read cwdi->cwdi_cmask unlocked. */ 2620 vattr.va_mode = ACCESSPERMS &~ p->p_cwdi->cwdi_cmask; 2621 error = VOP_SYMLINK(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr, path); 2622 if (error == 0) 2623 vrele(nd.ni_vp); 2624 vput(nd.ni_dvp); 2625 out2: 2626 pathbuf_destroy(linkpb); 2627 out1: 2628 PNBUF_PUT(path); 2629 return (error); 2630 } 2631 2632 /* 2633 * Make a symbolic link. 2634 */ 2635 /* ARGSUSED */ 2636 int 2637 sys_symlink(struct lwp *l, const struct sys_symlink_args *uap, register_t *retval) 2638 { 2639 /* { 2640 syscallarg(const char *) path; 2641 syscallarg(const char *) link; 2642 } */ 2643 2644 return do_sys_symlinkat(l, SCARG(uap, path), AT_FDCWD, SCARG(uap, link), 2645 UIO_USERSPACE); 2646 } 2647 2648 int 2649 sys_symlinkat(struct lwp *l, const struct sys_symlinkat_args *uap, 2650 register_t *retval) 2651 { 2652 /* { 2653 syscallarg(const char *) path1; 2654 syscallarg(int) fd; 2655 syscallarg(const char *) path2; 2656 } */ 2657 2658 return do_sys_symlinkat(l, SCARG(uap, path1), SCARG(uap, fd), 2659 SCARG(uap, path2), UIO_USERSPACE); 2660 } 2661 2662 /* 2663 * Delete a whiteout from the filesystem. 2664 */ 2665 /* ARGSUSED */ 2666 int 2667 sys_undelete(struct lwp *l, const struct sys_undelete_args *uap, register_t *retval) 2668 { 2669 /* { 2670 syscallarg(const char *) path; 2671 } */ 2672 int error; 2673 struct pathbuf *pb; 2674 struct nameidata nd; 2675 2676 error = pathbuf_copyin(SCARG(uap, path), &pb); 2677 if (error) { 2678 return error; 2679 } 2680 2681 NDINIT(&nd, DELETE, LOCKPARENT | DOWHITEOUT | TRYEMULROOT, pb); 2682 error = namei(&nd); 2683 if (error) { 2684 pathbuf_destroy(pb); 2685 return (error); 2686 } 2687 2688 if (nd.ni_vp != NULLVP || !(nd.ni_cnd.cn_flags & ISWHITEOUT)) { 2689 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd); 2690 if (nd.ni_dvp == nd.ni_vp) 2691 vrele(nd.ni_dvp); 2692 else 2693 vput(nd.ni_dvp); 2694 if (nd.ni_vp) 2695 vrele(nd.ni_vp); 2696 pathbuf_destroy(pb); 2697 return (EEXIST); 2698 } 2699 if ((error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, DELETE)) != 0) 2700 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd); 2701 vput(nd.ni_dvp); 2702 pathbuf_destroy(pb); 2703 return (error); 2704 } 2705 2706 /* 2707 * Delete a name from the filesystem. 2708 */ 2709 /* ARGSUSED */ 2710 int 2711 sys_unlink(struct lwp *l, const struct sys_unlink_args *uap, register_t *retval) 2712 { 2713 /* { 2714 syscallarg(const char *) path; 2715 } */ 2716 2717 return do_sys_unlinkat(l, AT_FDCWD, SCARG(uap, path), 0, UIO_USERSPACE); 2718 } 2719 2720 int 2721 sys_unlinkat(struct lwp *l, const struct sys_unlinkat_args *uap, 2722 register_t *retval) 2723 { 2724 /* { 2725 syscallarg(int) fd; 2726 syscallarg(const char *) path; 2727 syscallarg(int) flag; 2728 } */ 2729 2730 return do_sys_unlinkat(l, SCARG(uap, fd), SCARG(uap, path), 2731 SCARG(uap, flag), UIO_USERSPACE); 2732 } 2733 2734 int 2735 do_sys_unlink(const char *arg, enum uio_seg seg) 2736 { 2737 return do_sys_unlinkat(NULL, AT_FDCWD, arg, 0, seg); 2738 } 2739 2740 static int 2741 do_sys_unlinkat(struct lwp *l, int fdat, const char *arg, int flags, 2742 enum uio_seg seg) 2743 { 2744 struct vnode *vp; 2745 int error; 2746 struct pathbuf *pb; 2747 struct nameidata nd; 2748 const char *pathstring; 2749 2750 KASSERT(l != NULL || fdat == AT_FDCWD); 2751 2752 error = pathbuf_maybe_copyin(arg, seg, &pb); 2753 if (error) { 2754 return error; 2755 } 2756 pathstring = pathbuf_stringcopy_get(pb); 2757 if (pathstring == NULL) { 2758 pathbuf_destroy(pb); 2759 return ENOMEM; 2760 } 2761 2762 NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF | TRYEMULROOT, pb); 2763 if ((error = fd_nameiat(l, fdat, &nd)) != 0) 2764 goto out; 2765 vp = nd.ni_vp; 2766 2767 /* 2768 * The root of a mounted filesystem cannot be deleted. 2769 */ 2770 if ((vp->v_vflag & VV_ROOT) != 0) { 2771 error = EBUSY; 2772 goto abort; 2773 } 2774 2775 if ((vp->v_type == VDIR) && (vp->v_mountedhere != NULL)) { 2776 error = EBUSY; 2777 goto abort; 2778 } 2779 2780 /* 2781 * No rmdir "." please. 2782 */ 2783 if (nd.ni_dvp == vp) { 2784 error = EINVAL; 2785 goto abort; 2786 } 2787 2788 /* 2789 * AT_REMOVEDIR is required to remove a directory 2790 */ 2791 if (vp->v_type == VDIR) { 2792 if (!(flags & AT_REMOVEDIR)) { 2793 error = EPERM; 2794 goto abort; 2795 } else { 2796 error = VOP_RMDIR(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd); 2797 vput(nd.ni_dvp); 2798 goto out; 2799 } 2800 } 2801 2802 /* 2803 * Starting here we only deal with non directories. 2804 */ 2805 if (flags & AT_REMOVEDIR) { 2806 error = ENOTDIR; 2807 goto abort; 2808 } 2809 2810 #if NVERIEXEC > 0 2811 /* Handle remove requests for veriexec entries. */ 2812 if ((error = veriexec_removechk(curlwp, nd.ni_vp, pathstring)) != 0) { 2813 goto abort; 2814 } 2815 #endif /* NVERIEXEC > 0 */ 2816 2817 #ifdef FILEASSOC 2818 (void)fileassoc_file_delete(vp); 2819 #endif /* FILEASSOC */ 2820 error = VOP_REMOVE(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd); 2821 vput(nd.ni_dvp); 2822 goto out; 2823 2824 abort: 2825 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd); 2826 if (nd.ni_dvp == vp) 2827 vrele(nd.ni_dvp); 2828 else 2829 vput(nd.ni_dvp); 2830 vput(vp); 2831 2832 out: 2833 pathbuf_stringcopy_put(pb, pathstring); 2834 pathbuf_destroy(pb); 2835 return (error); 2836 } 2837 2838 /* 2839 * Reposition read/write file offset. 2840 */ 2841 int 2842 sys_lseek(struct lwp *l, const struct sys_lseek_args *uap, register_t *retval) 2843 { 2844 /* { 2845 syscallarg(int) fd; 2846 syscallarg(int) pad; 2847 syscallarg(off_t) offset; 2848 syscallarg(int) whence; 2849 } */ 2850 kauth_cred_t cred = l->l_cred; 2851 file_t *fp; 2852 struct vnode *vp; 2853 struct vattr vattr; 2854 off_t newoff; 2855 int error, fd; 2856 2857 fd = SCARG(uap, fd); 2858 2859 if ((fp = fd_getfile(fd)) == NULL) 2860 return (EBADF); 2861 2862 vp = fp->f_vnode; 2863 if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) { 2864 error = ESPIPE; 2865 goto out; 2866 } 2867 2868 vn_lock(vp, LK_SHARED | LK_RETRY); 2869 2870 switch (SCARG(uap, whence)) { 2871 case SEEK_CUR: 2872 newoff = fp->f_offset + SCARG(uap, offset); 2873 break; 2874 case SEEK_END: 2875 error = VOP_GETATTR(vp, &vattr, cred); 2876 if (error) { 2877 VOP_UNLOCK(vp); 2878 goto out; 2879 } 2880 newoff = SCARG(uap, offset) + vattr.va_size; 2881 break; 2882 case SEEK_SET: 2883 newoff = SCARG(uap, offset); 2884 break; 2885 default: 2886 error = EINVAL; 2887 VOP_UNLOCK(vp); 2888 goto out; 2889 } 2890 VOP_UNLOCK(vp); 2891 if ((error = VOP_SEEK(vp, fp->f_offset, newoff, cred)) == 0) { 2892 *(off_t *)retval = fp->f_offset = newoff; 2893 } 2894 out: 2895 fd_putfile(fd); 2896 return (error); 2897 } 2898 2899 /* 2900 * Positional read system call. 2901 */ 2902 int 2903 sys_pread(struct lwp *l, const struct sys_pread_args *uap, register_t *retval) 2904 { 2905 /* { 2906 syscallarg(int) fd; 2907 syscallarg(void *) buf; 2908 syscallarg(size_t) nbyte; 2909 syscallarg(off_t) offset; 2910 } */ 2911 file_t *fp; 2912 struct vnode *vp; 2913 off_t offset; 2914 int error, fd = SCARG(uap, fd); 2915 2916 if ((fp = fd_getfile(fd)) == NULL) 2917 return (EBADF); 2918 2919 if ((fp->f_flag & FREAD) == 0) { 2920 fd_putfile(fd); 2921 return (EBADF); 2922 } 2923 2924 vp = fp->f_vnode; 2925 if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) { 2926 error = ESPIPE; 2927 goto out; 2928 } 2929 2930 offset = SCARG(uap, offset); 2931 2932 /* 2933 * XXX This works because no file systems actually 2934 * XXX take any action on the seek operation. 2935 */ 2936 if ((error = VOP_SEEK(vp, fp->f_offset, offset, fp->f_cred)) != 0) 2937 goto out; 2938 2939 /* dofileread() will unuse the descriptor for us */ 2940 return (dofileread(fd, fp, SCARG(uap, buf), SCARG(uap, nbyte), 2941 &offset, 0, retval)); 2942 2943 out: 2944 fd_putfile(fd); 2945 return (error); 2946 } 2947 2948 /* 2949 * Positional scatter read system call. 2950 */ 2951 int 2952 sys_preadv(struct lwp *l, const struct sys_preadv_args *uap, register_t *retval) 2953 { 2954 /* { 2955 syscallarg(int) fd; 2956 syscallarg(const struct iovec *) iovp; 2957 syscallarg(int) iovcnt; 2958 syscallarg(off_t) offset; 2959 } */ 2960 off_t offset = SCARG(uap, offset); 2961 2962 return do_filereadv(SCARG(uap, fd), SCARG(uap, iovp), 2963 SCARG(uap, iovcnt), &offset, 0, retval); 2964 } 2965 2966 /* 2967 * Positional write system call. 2968 */ 2969 int 2970 sys_pwrite(struct lwp *l, const struct sys_pwrite_args *uap, register_t *retval) 2971 { 2972 /* { 2973 syscallarg(int) fd; 2974 syscallarg(const void *) buf; 2975 syscallarg(size_t) nbyte; 2976 syscallarg(off_t) offset; 2977 } */ 2978 file_t *fp; 2979 struct vnode *vp; 2980 off_t offset; 2981 int error, fd = SCARG(uap, fd); 2982 2983 if ((fp = fd_getfile(fd)) == NULL) 2984 return (EBADF); 2985 2986 if ((fp->f_flag & FWRITE) == 0) { 2987 fd_putfile(fd); 2988 return (EBADF); 2989 } 2990 2991 vp = fp->f_vnode; 2992 if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) { 2993 error = ESPIPE; 2994 goto out; 2995 } 2996 2997 offset = SCARG(uap, offset); 2998 2999 /* 3000 * XXX This works because no file systems actually 3001 * XXX take any action on the seek operation. 3002 */ 3003 if ((error = VOP_SEEK(vp, fp->f_offset, offset, fp->f_cred)) != 0) 3004 goto out; 3005 3006 /* dofilewrite() will unuse the descriptor for us */ 3007 return (dofilewrite(fd, fp, SCARG(uap, buf), SCARG(uap, nbyte), 3008 &offset, 0, retval)); 3009 3010 out: 3011 fd_putfile(fd); 3012 return (error); 3013 } 3014 3015 /* 3016 * Positional gather write system call. 3017 */ 3018 int 3019 sys_pwritev(struct lwp *l, const struct sys_pwritev_args *uap, register_t *retval) 3020 { 3021 /* { 3022 syscallarg(int) fd; 3023 syscallarg(const struct iovec *) iovp; 3024 syscallarg(int) iovcnt; 3025 syscallarg(off_t) offset; 3026 } */ 3027 off_t offset = SCARG(uap, offset); 3028 3029 return do_filewritev(SCARG(uap, fd), SCARG(uap, iovp), 3030 SCARG(uap, iovcnt), &offset, 0, retval); 3031 } 3032 3033 /* 3034 * Check access permissions. 3035 */ 3036 int 3037 sys_access(struct lwp *l, const struct sys_access_args *uap, register_t *retval) 3038 { 3039 /* { 3040 syscallarg(const char *) path; 3041 syscallarg(int) flags; 3042 } */ 3043 3044 return do_sys_accessat(l, AT_FDCWD, SCARG(uap, path), 3045 SCARG(uap, flags), 0); 3046 } 3047 3048 int 3049 do_sys_accessat(struct lwp *l, int fdat, const char *path, 3050 int mode, int flags) 3051 { 3052 kauth_cred_t cred; 3053 struct vnode *vp; 3054 int error, nd_flag, vmode; 3055 struct pathbuf *pb; 3056 struct nameidata nd; 3057 3058 CTASSERT(F_OK == 0); 3059 if ((mode & ~(R_OK | W_OK | X_OK)) != 0) { 3060 /* nonsense mode */ 3061 return EINVAL; 3062 } 3063 3064 nd_flag = FOLLOW | LOCKLEAF | LOCKSHARED | TRYEMULROOT; 3065 if (flags & AT_SYMLINK_NOFOLLOW) 3066 nd_flag &= ~FOLLOW; 3067 3068 error = pathbuf_copyin(path, &pb); 3069 if (error) 3070 return error; 3071 3072 NDINIT(&nd, LOOKUP, nd_flag, pb); 3073 3074 /* Override default credentials */ 3075 cred = kauth_cred_dup(l->l_cred); 3076 if (!(flags & AT_EACCESS)) { 3077 kauth_cred_seteuid(cred, kauth_cred_getuid(l->l_cred)); 3078 kauth_cred_setegid(cred, kauth_cred_getgid(l->l_cred)); 3079 } 3080 nd.ni_cnd.cn_cred = cred; 3081 3082 if ((error = fd_nameiat(l, fdat, &nd)) != 0) { 3083 pathbuf_destroy(pb); 3084 goto out; 3085 } 3086 vp = nd.ni_vp; 3087 pathbuf_destroy(pb); 3088 3089 /* Flags == 0 means only check for existence. */ 3090 if (mode) { 3091 vmode = 0; 3092 if (mode & R_OK) 3093 vmode |= VREAD; 3094 if (mode & W_OK) 3095 vmode |= VWRITE; 3096 if (mode & X_OK) 3097 vmode |= VEXEC; 3098 3099 error = VOP_ACCESS(vp, vmode, cred); 3100 if (!error && (vmode & VWRITE)) 3101 error = vn_writechk(vp); 3102 } 3103 vput(vp); 3104 out: 3105 kauth_cred_free(cred); 3106 return (error); 3107 } 3108 3109 int 3110 sys_faccessat(struct lwp *l, const struct sys_faccessat_args *uap, 3111 register_t *retval) 3112 { 3113 /* { 3114 syscallarg(int) fd; 3115 syscallarg(const char *) path; 3116 syscallarg(int) amode; 3117 syscallarg(int) flag; 3118 } */ 3119 3120 return do_sys_accessat(l, SCARG(uap, fd), SCARG(uap, path), 3121 SCARG(uap, amode), SCARG(uap, flag)); 3122 } 3123 3124 /* 3125 * Common code for all sys_stat functions, including compat versions. 3126 */ 3127 int 3128 do_sys_stat(const char *userpath, unsigned int nd_flag, 3129 struct stat *sb) 3130 { 3131 return do_sys_statat(NULL, AT_FDCWD, userpath, nd_flag, sb); 3132 } 3133 3134 int 3135 do_sys_statat(struct lwp *l, int fdat, const char *userpath, 3136 unsigned int nd_flag, struct stat *sb) 3137 { 3138 int error; 3139 struct pathbuf *pb; 3140 struct nameidata nd; 3141 3142 KASSERT(l != NULL || fdat == AT_FDCWD); 3143 3144 error = pathbuf_copyin(userpath, &pb); 3145 if (error) { 3146 return error; 3147 } 3148 3149 NDINIT(&nd, LOOKUP, nd_flag | LOCKLEAF | TRYEMULROOT, pb); 3150 3151 error = fd_nameiat(l, fdat, &nd); 3152 if (error != 0) { 3153 pathbuf_destroy(pb); 3154 return error; 3155 } 3156 error = vn_stat(nd.ni_vp, sb); 3157 vput(nd.ni_vp); 3158 pathbuf_destroy(pb); 3159 return error; 3160 } 3161 3162 /* 3163 * Get file status; this version follows links. 3164 */ 3165 /* ARGSUSED */ 3166 int 3167 sys___stat50(struct lwp *l, const struct sys___stat50_args *uap, register_t *retval) 3168 { 3169 /* { 3170 syscallarg(const char *) path; 3171 syscallarg(struct stat *) ub; 3172 } */ 3173 struct stat sb; 3174 int error; 3175 3176 error = do_sys_statat(l, AT_FDCWD, SCARG(uap, path), FOLLOW, &sb); 3177 if (error) 3178 return error; 3179 return copyout(&sb, SCARG(uap, ub), sizeof(sb)); 3180 } 3181 3182 /* 3183 * Get file status; this version does not follow links. 3184 */ 3185 /* ARGSUSED */ 3186 int 3187 sys___lstat50(struct lwp *l, const struct sys___lstat50_args *uap, register_t *retval) 3188 { 3189 /* { 3190 syscallarg(const char *) path; 3191 syscallarg(struct stat *) ub; 3192 } */ 3193 struct stat sb; 3194 int error; 3195 3196 error = do_sys_statat(l, AT_FDCWD, SCARG(uap, path), NOFOLLOW, &sb); 3197 if (error) 3198 return error; 3199 return copyout(&sb, SCARG(uap, ub), sizeof(sb)); 3200 } 3201 3202 int 3203 sys_fstatat(struct lwp *l, const struct sys_fstatat_args *uap, 3204 register_t *retval) 3205 { 3206 /* { 3207 syscallarg(int) fd; 3208 syscallarg(const char *) path; 3209 syscallarg(struct stat *) buf; 3210 syscallarg(int) flag; 3211 } */ 3212 unsigned int nd_flag; 3213 struct stat sb; 3214 int error; 3215 3216 if (SCARG(uap, flag) & AT_SYMLINK_NOFOLLOW) 3217 nd_flag = NOFOLLOW; 3218 else 3219 nd_flag = FOLLOW; 3220 3221 error = do_sys_statat(l, SCARG(uap, fd), SCARG(uap, path), nd_flag, 3222 &sb); 3223 if (error) 3224 return error; 3225 return copyout(&sb, SCARG(uap, buf), sizeof(sb)); 3226 } 3227 3228 /* 3229 * Get configurable pathname variables. 3230 */ 3231 /* ARGSUSED */ 3232 int 3233 sys_pathconf(struct lwp *l, const struct sys_pathconf_args *uap, register_t *retval) 3234 { 3235 /* { 3236 syscallarg(const char *) path; 3237 syscallarg(int) name; 3238 } */ 3239 int error; 3240 struct pathbuf *pb; 3241 struct nameidata nd; 3242 3243 error = pathbuf_copyin(SCARG(uap, path), &pb); 3244 if (error) { 3245 return error; 3246 } 3247 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | TRYEMULROOT, pb); 3248 if ((error = namei(&nd)) != 0) { 3249 pathbuf_destroy(pb); 3250 return (error); 3251 } 3252 error = VOP_PATHCONF(nd.ni_vp, SCARG(uap, name), retval); 3253 vput(nd.ni_vp); 3254 pathbuf_destroy(pb); 3255 return (error); 3256 } 3257 3258 /* 3259 * Return target name of a symbolic link. 3260 */ 3261 /* ARGSUSED */ 3262 int 3263 sys_readlink(struct lwp *l, const struct sys_readlink_args *uap, 3264 register_t *retval) 3265 { 3266 /* { 3267 syscallarg(const char *) path; 3268 syscallarg(char *) buf; 3269 syscallarg(size_t) count; 3270 } */ 3271 return do_sys_readlinkat(l, AT_FDCWD, SCARG(uap, path), 3272 SCARG(uap, buf), SCARG(uap, count), retval); 3273 } 3274 3275 static int 3276 do_sys_readlinkat(struct lwp *l, int fdat, const char *path, char *buf, 3277 size_t count, register_t *retval) 3278 { 3279 struct vnode *vp; 3280 struct iovec aiov; 3281 struct uio auio; 3282 int error; 3283 struct pathbuf *pb; 3284 struct nameidata nd; 3285 3286 error = pathbuf_copyin(path, &pb); 3287 if (error) { 3288 return error; 3289 } 3290 NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | LOCKSHARED | TRYEMULROOT, pb); 3291 if ((error = fd_nameiat(l, fdat, &nd)) != 0) { 3292 pathbuf_destroy(pb); 3293 return error; 3294 } 3295 vp = nd.ni_vp; 3296 pathbuf_destroy(pb); 3297 if (vp->v_type != VLNK) 3298 error = EINVAL; 3299 else if (!(vp->v_mount->mnt_flag & MNT_SYMPERM) || 3300 (error = VOP_ACCESS(vp, VREAD, l->l_cred)) == 0) { 3301 aiov.iov_base = buf; 3302 aiov.iov_len = count; 3303 auio.uio_iov = &aiov; 3304 auio.uio_iovcnt = 1; 3305 auio.uio_offset = 0; 3306 auio.uio_rw = UIO_READ; 3307 KASSERT(l == curlwp); 3308 auio.uio_vmspace = l->l_proc->p_vmspace; 3309 auio.uio_resid = count; 3310 if ((error = VOP_READLINK(vp, &auio, l->l_cred)) == 0) 3311 *retval = count - auio.uio_resid; 3312 } 3313 vput(vp); 3314 return (error); 3315 } 3316 3317 int 3318 sys_readlinkat(struct lwp *l, const struct sys_readlinkat_args *uap, 3319 register_t *retval) 3320 { 3321 /* { 3322 syscallarg(int) fd; 3323 syscallarg(const char *) path; 3324 syscallarg(char *) buf; 3325 syscallarg(size_t) bufsize; 3326 } */ 3327 3328 return do_sys_readlinkat(l, SCARG(uap, fd), SCARG(uap, path), 3329 SCARG(uap, buf), SCARG(uap, bufsize), retval); 3330 } 3331 3332 /* 3333 * Change flags of a file given a path name. 3334 */ 3335 /* ARGSUSED */ 3336 int 3337 sys_chflags(struct lwp *l, const struct sys_chflags_args *uap, register_t *retval) 3338 { 3339 /* { 3340 syscallarg(const char *) path; 3341 syscallarg(u_long) flags; 3342 } */ 3343 struct vnode *vp; 3344 int error; 3345 3346 error = namei_simple_user(SCARG(uap, path), 3347 NSM_FOLLOW_TRYEMULROOT, &vp); 3348 if (error != 0) 3349 return (error); 3350 error = change_flags(vp, SCARG(uap, flags), l); 3351 vput(vp); 3352 return (error); 3353 } 3354 3355 /* 3356 * Change flags of a file given a file descriptor. 3357 */ 3358 /* ARGSUSED */ 3359 int 3360 sys_fchflags(struct lwp *l, const struct sys_fchflags_args *uap, register_t *retval) 3361 { 3362 /* { 3363 syscallarg(int) fd; 3364 syscallarg(u_long) flags; 3365 } */ 3366 struct vnode *vp; 3367 file_t *fp; 3368 int error; 3369 3370 /* fd_getvnode() will use the descriptor for us */ 3371 if ((error = fd_getvnode(SCARG(uap, fd), &fp)) != 0) 3372 return (error); 3373 vp = fp->f_vnode; 3374 error = change_flags(vp, SCARG(uap, flags), l); 3375 VOP_UNLOCK(vp); 3376 fd_putfile(SCARG(uap, fd)); 3377 return (error); 3378 } 3379 3380 /* 3381 * Change flags of a file given a path name; this version does 3382 * not follow links. 3383 */ 3384 int 3385 sys_lchflags(struct lwp *l, const struct sys_lchflags_args *uap, register_t *retval) 3386 { 3387 /* { 3388 syscallarg(const char *) path; 3389 syscallarg(u_long) flags; 3390 } */ 3391 struct vnode *vp; 3392 int error; 3393 3394 error = namei_simple_user(SCARG(uap, path), 3395 NSM_NOFOLLOW_TRYEMULROOT, &vp); 3396 if (error != 0) 3397 return (error); 3398 error = change_flags(vp, SCARG(uap, flags), l); 3399 vput(vp); 3400 return (error); 3401 } 3402 3403 /* 3404 * Common routine to change flags of a file. 3405 */ 3406 int 3407 change_flags(struct vnode *vp, u_long flags, struct lwp *l) 3408 { 3409 struct vattr vattr; 3410 int error; 3411 3412 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 3413 3414 vattr_null(&vattr); 3415 vattr.va_flags = flags; 3416 error = VOP_SETATTR(vp, &vattr, l->l_cred); 3417 3418 return (error); 3419 } 3420 3421 /* 3422 * Change mode of a file given path name; this version follows links. 3423 */ 3424 /* ARGSUSED */ 3425 int 3426 sys_chmod(struct lwp *l, const struct sys_chmod_args *uap, register_t *retval) 3427 { 3428 /* { 3429 syscallarg(const char *) path; 3430 syscallarg(int) mode; 3431 } */ 3432 return do_sys_chmodat(l, AT_FDCWD, SCARG(uap, path), 3433 SCARG(uap, mode), 0); 3434 } 3435 3436 int 3437 do_sys_chmodat(struct lwp *l, int fdat, const char *path, int mode, int flags) 3438 { 3439 int error; 3440 struct vnode *vp; 3441 namei_simple_flags_t ns_flag; 3442 3443 if (flags & AT_SYMLINK_NOFOLLOW) 3444 ns_flag = NSM_NOFOLLOW_TRYEMULROOT; 3445 else 3446 ns_flag = NSM_FOLLOW_TRYEMULROOT; 3447 3448 error = fd_nameiat_simple_user(l, fdat, path, ns_flag, &vp); 3449 if (error != 0) 3450 return error; 3451 3452 error = change_mode(vp, mode, l); 3453 3454 vrele(vp); 3455 3456 return (error); 3457 } 3458 3459 /* 3460 * Change mode of a file given a file descriptor. 3461 */ 3462 /* ARGSUSED */ 3463 int 3464 sys_fchmod(struct lwp *l, const struct sys_fchmod_args *uap, register_t *retval) 3465 { 3466 /* { 3467 syscallarg(int) fd; 3468 syscallarg(int) mode; 3469 } */ 3470 file_t *fp; 3471 int error; 3472 3473 /* fd_getvnode() will use the descriptor for us */ 3474 if ((error = fd_getvnode(SCARG(uap, fd), &fp)) != 0) 3475 return (error); 3476 error = change_mode(fp->f_vnode, SCARG(uap, mode), l); 3477 fd_putfile(SCARG(uap, fd)); 3478 return (error); 3479 } 3480 3481 int 3482 sys_fchmodat(struct lwp *l, const struct sys_fchmodat_args *uap, 3483 register_t *retval) 3484 { 3485 /* { 3486 syscallarg(int) fd; 3487 syscallarg(const char *) path; 3488 syscallarg(int) mode; 3489 syscallarg(int) flag; 3490 } */ 3491 3492 return do_sys_chmodat(l, SCARG(uap, fd), SCARG(uap, path), 3493 SCARG(uap, mode), SCARG(uap, flag)); 3494 } 3495 3496 /* 3497 * Change mode of a file given path name; this version does not follow links. 3498 */ 3499 /* ARGSUSED */ 3500 int 3501 sys_lchmod(struct lwp *l, const struct sys_lchmod_args *uap, register_t *retval) 3502 { 3503 /* { 3504 syscallarg(const char *) path; 3505 syscallarg(int) mode; 3506 } */ 3507 int error; 3508 struct vnode *vp; 3509 3510 error = namei_simple_user(SCARG(uap, path), 3511 NSM_NOFOLLOW_TRYEMULROOT, &vp); 3512 if (error != 0) 3513 return (error); 3514 3515 error = change_mode(vp, SCARG(uap, mode), l); 3516 3517 vrele(vp); 3518 return (error); 3519 } 3520 3521 /* 3522 * Common routine to set mode given a vnode. 3523 */ 3524 static int 3525 change_mode(struct vnode *vp, int mode, struct lwp *l) 3526 { 3527 struct vattr vattr; 3528 int error; 3529 3530 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 3531 vattr_null(&vattr); 3532 vattr.va_mode = mode & ALLPERMS; 3533 error = VOP_SETATTR(vp, &vattr, l->l_cred); 3534 VOP_UNLOCK(vp); 3535 return (error); 3536 } 3537 3538 /* 3539 * Set ownership given a path name; this version follows links. 3540 */ 3541 /* ARGSUSED */ 3542 int 3543 sys_chown(struct lwp *l, const struct sys_chown_args *uap, register_t *retval) 3544 { 3545 /* { 3546 syscallarg(const char *) path; 3547 syscallarg(uid_t) uid; 3548 syscallarg(gid_t) gid; 3549 } */ 3550 return do_sys_chownat(l, AT_FDCWD, SCARG(uap, path), SCARG(uap,uid), 3551 SCARG(uap, gid), 0); 3552 } 3553 3554 int 3555 do_sys_chownat(struct lwp *l, int fdat, const char *path, uid_t uid, 3556 gid_t gid, int flags) 3557 { 3558 int error; 3559 struct vnode *vp; 3560 namei_simple_flags_t ns_flag; 3561 3562 if (flags & AT_SYMLINK_NOFOLLOW) 3563 ns_flag = NSM_NOFOLLOW_TRYEMULROOT; 3564 else 3565 ns_flag = NSM_FOLLOW_TRYEMULROOT; 3566 3567 error = fd_nameiat_simple_user(l, fdat, path, ns_flag, &vp); 3568 if (error != 0) 3569 return error; 3570 3571 error = change_owner(vp, uid, gid, l, 0); 3572 3573 vrele(vp); 3574 3575 return (error); 3576 } 3577 3578 /* 3579 * Set ownership given a path name; this version follows links. 3580 * Provides POSIX semantics. 3581 */ 3582 /* ARGSUSED */ 3583 int 3584 sys___posix_chown(struct lwp *l, const struct sys___posix_chown_args *uap, register_t *retval) 3585 { 3586 /* { 3587 syscallarg(const char *) path; 3588 syscallarg(uid_t) uid; 3589 syscallarg(gid_t) gid; 3590 } */ 3591 int error; 3592 struct vnode *vp; 3593 3594 error = namei_simple_user(SCARG(uap, path), 3595 NSM_FOLLOW_TRYEMULROOT, &vp); 3596 if (error != 0) 3597 return (error); 3598 3599 error = change_owner(vp, SCARG(uap, uid), SCARG(uap, gid), l, 1); 3600 3601 vrele(vp); 3602 return (error); 3603 } 3604 3605 /* 3606 * Set ownership given a file descriptor. 3607 */ 3608 /* ARGSUSED */ 3609 int 3610 sys_fchown(struct lwp *l, const struct sys_fchown_args *uap, register_t *retval) 3611 { 3612 /* { 3613 syscallarg(int) fd; 3614 syscallarg(uid_t) uid; 3615 syscallarg(gid_t) gid; 3616 } */ 3617 int error; 3618 file_t *fp; 3619 3620 /* fd_getvnode() will use the descriptor for us */ 3621 if ((error = fd_getvnode(SCARG(uap, fd), &fp)) != 0) 3622 return (error); 3623 error = change_owner(fp->f_vnode, SCARG(uap, uid), SCARG(uap, gid), 3624 l, 0); 3625 fd_putfile(SCARG(uap, fd)); 3626 return (error); 3627 } 3628 3629 int 3630 sys_fchownat(struct lwp *l, const struct sys_fchownat_args *uap, 3631 register_t *retval) 3632 { 3633 /* { 3634 syscallarg(int) fd; 3635 syscallarg(const char *) path; 3636 syscallarg(uid_t) owner; 3637 syscallarg(gid_t) group; 3638 syscallarg(int) flag; 3639 } */ 3640 3641 return do_sys_chownat(l, SCARG(uap, fd), SCARG(uap, path), 3642 SCARG(uap, owner), SCARG(uap, group), 3643 SCARG(uap, flag)); 3644 } 3645 3646 /* 3647 * Set ownership given a file descriptor, providing POSIX/XPG semantics. 3648 */ 3649 /* ARGSUSED */ 3650 int 3651 sys___posix_fchown(struct lwp *l, const struct sys___posix_fchown_args *uap, register_t *retval) 3652 { 3653 /* { 3654 syscallarg(int) fd; 3655 syscallarg(uid_t) uid; 3656 syscallarg(gid_t) gid; 3657 } */ 3658 int error; 3659 file_t *fp; 3660 3661 /* fd_getvnode() will use the descriptor for us */ 3662 if ((error = fd_getvnode(SCARG(uap, fd), &fp)) != 0) 3663 return (error); 3664 error = change_owner(fp->f_vnode, SCARG(uap, uid), SCARG(uap, gid), 3665 l, 1); 3666 fd_putfile(SCARG(uap, fd)); 3667 return (error); 3668 } 3669 3670 /* 3671 * Set ownership given a path name; this version does not follow links. 3672 */ 3673 /* ARGSUSED */ 3674 int 3675 sys_lchown(struct lwp *l, const struct sys_lchown_args *uap, register_t *retval) 3676 { 3677 /* { 3678 syscallarg(const char *) path; 3679 syscallarg(uid_t) uid; 3680 syscallarg(gid_t) gid; 3681 } */ 3682 int error; 3683 struct vnode *vp; 3684 3685 error = namei_simple_user(SCARG(uap, path), 3686 NSM_NOFOLLOW_TRYEMULROOT, &vp); 3687 if (error != 0) 3688 return (error); 3689 3690 error = change_owner(vp, SCARG(uap, uid), SCARG(uap, gid), l, 0); 3691 3692 vrele(vp); 3693 return (error); 3694 } 3695 3696 /* 3697 * Set ownership given a path name; this version does not follow links. 3698 * Provides POSIX/XPG semantics. 3699 */ 3700 /* ARGSUSED */ 3701 int 3702 sys___posix_lchown(struct lwp *l, const struct sys___posix_lchown_args *uap, register_t *retval) 3703 { 3704 /* { 3705 syscallarg(const char *) path; 3706 syscallarg(uid_t) uid; 3707 syscallarg(gid_t) gid; 3708 } */ 3709 int error; 3710 struct vnode *vp; 3711 3712 error = namei_simple_user(SCARG(uap, path), 3713 NSM_NOFOLLOW_TRYEMULROOT, &vp); 3714 if (error != 0) 3715 return (error); 3716 3717 error = change_owner(vp, SCARG(uap, uid), SCARG(uap, gid), l, 1); 3718 3719 vrele(vp); 3720 return (error); 3721 } 3722 3723 /* 3724 * Common routine to set ownership given a vnode. 3725 */ 3726 static int 3727 change_owner(struct vnode *vp, uid_t uid, gid_t gid, struct lwp *l, 3728 int posix_semantics) 3729 { 3730 struct vattr vattr; 3731 mode_t newmode; 3732 int error; 3733 3734 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 3735 if ((error = VOP_GETATTR(vp, &vattr, l->l_cred)) != 0) 3736 goto out; 3737 3738 #define CHANGED(x) ((int)(x) != -1) 3739 newmode = vattr.va_mode; 3740 if (posix_semantics) { 3741 /* 3742 * POSIX/XPG semantics: if the caller is not the super-user, 3743 * clear set-user-id and set-group-id bits. Both POSIX and 3744 * the XPG consider the behaviour for calls by the super-user 3745 * implementation-defined; we leave the set-user-id and set- 3746 * group-id settings intact in that case. 3747 */ 3748 if (vattr.va_mode & S_ISUID) { 3749 if (kauth_authorize_vnode(l->l_cred, 3750 KAUTH_VNODE_RETAIN_SUID, vp, NULL, EPERM) != 0) 3751 newmode &= ~S_ISUID; 3752 } 3753 if (vattr.va_mode & S_ISGID) { 3754 if (kauth_authorize_vnode(l->l_cred, 3755 KAUTH_VNODE_RETAIN_SGID, vp, NULL, EPERM) != 0) 3756 newmode &= ~S_ISGID; 3757 } 3758 } else { 3759 /* 3760 * NetBSD semantics: when changing owner and/or group, 3761 * clear the respective bit(s). 3762 */ 3763 if (CHANGED(uid)) 3764 newmode &= ~S_ISUID; 3765 if (CHANGED(gid)) 3766 newmode &= ~S_ISGID; 3767 } 3768 /* Update va_mode iff altered. */ 3769 if (vattr.va_mode == newmode) 3770 newmode = VNOVAL; 3771 3772 vattr_null(&vattr); 3773 vattr.va_uid = CHANGED(uid) ? uid : (uid_t)VNOVAL; 3774 vattr.va_gid = CHANGED(gid) ? gid : (gid_t)VNOVAL; 3775 vattr.va_mode = newmode; 3776 error = VOP_SETATTR(vp, &vattr, l->l_cred); 3777 #undef CHANGED 3778 3779 out: 3780 VOP_UNLOCK(vp); 3781 return (error); 3782 } 3783 3784 /* 3785 * Set the access and modification times given a path name; this 3786 * version follows links. 3787 */ 3788 /* ARGSUSED */ 3789 int 3790 sys___utimes50(struct lwp *l, const struct sys___utimes50_args *uap, 3791 register_t *retval) 3792 { 3793 /* { 3794 syscallarg(const char *) path; 3795 syscallarg(const struct timeval *) tptr; 3796 } */ 3797 3798 return do_sys_utimes(l, NULL, SCARG(uap, path), FOLLOW, 3799 SCARG(uap, tptr), UIO_USERSPACE); 3800 } 3801 3802 /* 3803 * Set the access and modification times given a file descriptor. 3804 */ 3805 /* ARGSUSED */ 3806 int 3807 sys___futimes50(struct lwp *l, const struct sys___futimes50_args *uap, 3808 register_t *retval) 3809 { 3810 /* { 3811 syscallarg(int) fd; 3812 syscallarg(const struct timeval *) tptr; 3813 } */ 3814 int error; 3815 file_t *fp; 3816 3817 /* fd_getvnode() will use the descriptor for us */ 3818 if ((error = fd_getvnode(SCARG(uap, fd), &fp)) != 0) 3819 return (error); 3820 error = do_sys_utimes(l, fp->f_vnode, NULL, 0, SCARG(uap, tptr), 3821 UIO_USERSPACE); 3822 fd_putfile(SCARG(uap, fd)); 3823 return (error); 3824 } 3825 3826 int 3827 sys_futimens(struct lwp *l, const struct sys_futimens_args *uap, 3828 register_t *retval) 3829 { 3830 /* { 3831 syscallarg(int) fd; 3832 syscallarg(const struct timespec *) tptr; 3833 } */ 3834 int error; 3835 file_t *fp; 3836 3837 /* fd_getvnode() will use the descriptor for us */ 3838 if ((error = fd_getvnode(SCARG(uap, fd), &fp)) != 0) 3839 return (error); 3840 error = do_sys_utimensat(l, AT_FDCWD, fp->f_vnode, NULL, 0, 3841 SCARG(uap, tptr), UIO_USERSPACE); 3842 fd_putfile(SCARG(uap, fd)); 3843 return (error); 3844 } 3845 3846 /* 3847 * Set the access and modification times given a path name; this 3848 * version does not follow links. 3849 */ 3850 int 3851 sys___lutimes50(struct lwp *l, const struct sys___lutimes50_args *uap, 3852 register_t *retval) 3853 { 3854 /* { 3855 syscallarg(const char *) path; 3856 syscallarg(const struct timeval *) tptr; 3857 } */ 3858 3859 return do_sys_utimes(l, NULL, SCARG(uap, path), NOFOLLOW, 3860 SCARG(uap, tptr), UIO_USERSPACE); 3861 } 3862 3863 int 3864 sys_utimensat(struct lwp *l, const struct sys_utimensat_args *uap, 3865 register_t *retval) 3866 { 3867 /* { 3868 syscallarg(int) fd; 3869 syscallarg(const char *) path; 3870 syscallarg(const struct timespec *) tptr; 3871 syscallarg(int) flag; 3872 } */ 3873 int follow; 3874 const struct timespec *tptr; 3875 int error; 3876 3877 tptr = SCARG(uap, tptr); 3878 follow = (SCARG(uap, flag) & AT_SYMLINK_NOFOLLOW) ? NOFOLLOW : FOLLOW; 3879 3880 error = do_sys_utimensat(l, SCARG(uap, fd), NULL, 3881 SCARG(uap, path), follow, tptr, UIO_USERSPACE); 3882 3883 return error; 3884 } 3885 3886 /* 3887 * Common routine to set access and modification times given a vnode. 3888 */ 3889 int 3890 do_sys_utimens(struct lwp *l, struct vnode *vp, const char *path, int flag, 3891 const struct timespec *tptr, enum uio_seg seg) 3892 { 3893 return do_sys_utimensat(l, AT_FDCWD, vp, path, flag, tptr, seg); 3894 } 3895 3896 int 3897 do_sys_utimensat(struct lwp *l, int fdat, struct vnode *vp, 3898 const char *path, int flag, const struct timespec *tptr, enum uio_seg seg) 3899 { 3900 struct vattr vattr; 3901 int error, dorele = 0; 3902 namei_simple_flags_t sflags; 3903 bool vanull, setbirthtime; 3904 struct timespec ts[2]; 3905 3906 KASSERT(l != NULL || fdat == AT_FDCWD); 3907 3908 /* 3909 * I have checked all callers and they pass either FOLLOW, 3910 * NOFOLLOW, or 0 (when they don't pass a path), and NOFOLLOW 3911 * is 0. More to the point, they don't pass anything else. 3912 * Let's keep it that way at least until the namei interfaces 3913 * are fully sanitized. 3914 */ 3915 KASSERT(flag == NOFOLLOW || flag == FOLLOW); 3916 sflags = (flag == FOLLOW) ? 3917 NSM_FOLLOW_TRYEMULROOT : NSM_NOFOLLOW_TRYEMULROOT; 3918 3919 if (tptr == NULL) { 3920 vanull = true; 3921 nanotime(&ts[0]); 3922 ts[1] = ts[0]; 3923 } else { 3924 vanull = false; 3925 if (seg != UIO_SYSSPACE) { 3926 error = copyin(tptr, ts, sizeof (ts)); 3927 if (error != 0) 3928 return error; 3929 } else { 3930 ts[0] = tptr[0]; 3931 ts[1] = tptr[1]; 3932 } 3933 } 3934 3935 if (ts[0].tv_nsec == UTIME_NOW) { 3936 nanotime(&ts[0]); 3937 if (ts[1].tv_nsec == UTIME_NOW) { 3938 vanull = true; 3939 ts[1] = ts[0]; 3940 } 3941 } else if (ts[1].tv_nsec == UTIME_NOW) 3942 nanotime(&ts[1]); 3943 3944 if (vp == NULL) { 3945 /* note: SEG describes TPTR, not PATH; PATH is always user */ 3946 error = fd_nameiat_simple_user(l, fdat, path, sflags, &vp); 3947 if (error != 0) 3948 return error; 3949 dorele = 1; 3950 } 3951 3952 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 3953 setbirthtime = (VOP_GETATTR(vp, &vattr, l->l_cred) == 0 && 3954 timespeccmp(&ts[1], &vattr.va_birthtime, <)); 3955 vattr_null(&vattr); 3956 3957 if (ts[0].tv_nsec != UTIME_OMIT) 3958 vattr.va_atime = ts[0]; 3959 3960 if (ts[1].tv_nsec != UTIME_OMIT) { 3961 vattr.va_mtime = ts[1]; 3962 if (setbirthtime) 3963 vattr.va_birthtime = ts[1]; 3964 } 3965 3966 if (vanull) 3967 vattr.va_vaflags |= VA_UTIMES_NULL; 3968 error = VOP_SETATTR(vp, &vattr, l->l_cred); 3969 VOP_UNLOCK(vp); 3970 3971 if (dorele != 0) 3972 vrele(vp); 3973 3974 return error; 3975 } 3976 3977 int 3978 do_sys_utimes(struct lwp *l, struct vnode *vp, const char *path, int flag, 3979 const struct timeval *tptr, enum uio_seg seg) 3980 { 3981 struct timespec ts[2]; 3982 struct timespec *tsptr = NULL; 3983 int error; 3984 3985 if (tptr != NULL) { 3986 struct timeval tv[2]; 3987 3988 if (seg != UIO_SYSSPACE) { 3989 error = copyin(tptr, tv, sizeof(tv)); 3990 if (error != 0) 3991 return error; 3992 tptr = tv; 3993 } 3994 3995 if ((tptr[0].tv_usec == UTIME_NOW) || 3996 (tptr[0].tv_usec == UTIME_OMIT)) 3997 ts[0].tv_nsec = tptr[0].tv_usec; 3998 else { 3999 if (tptr[0].tv_usec < 0 || tptr[0].tv_usec >= 1000000) 4000 return EINVAL; 4001 4002 TIMEVAL_TO_TIMESPEC(&tptr[0], &ts[0]); 4003 } 4004 4005 if ((tptr[1].tv_usec == UTIME_NOW) || 4006 (tptr[1].tv_usec == UTIME_OMIT)) 4007 ts[1].tv_nsec = tptr[1].tv_usec; 4008 else { 4009 if (tptr[1].tv_usec < 0 || tptr[1].tv_usec >= 1000000) 4010 return EINVAL; 4011 4012 TIMEVAL_TO_TIMESPEC(&tptr[1], &ts[1]); 4013 } 4014 4015 tsptr = &ts[0]; 4016 } 4017 4018 return do_sys_utimens(l, vp, path, flag, tsptr, UIO_SYSSPACE); 4019 } 4020 4021 /* 4022 * Truncate a file given its path name. 4023 */ 4024 /* ARGSUSED */ 4025 int 4026 sys_truncate(struct lwp *l, const struct sys_truncate_args *uap, register_t *retval) 4027 { 4028 /* { 4029 syscallarg(const char *) path; 4030 syscallarg(int) pad; 4031 syscallarg(off_t) length; 4032 } */ 4033 struct vnode *vp; 4034 struct vattr vattr; 4035 int error; 4036 4037 if (SCARG(uap, length) < 0) 4038 return EINVAL; 4039 4040 error = namei_simple_user(SCARG(uap, path), 4041 NSM_FOLLOW_TRYEMULROOT, &vp); 4042 if (error != 0) 4043 return (error); 4044 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 4045 if (vp->v_type == VDIR) 4046 error = EISDIR; 4047 else if ((error = vn_writechk(vp)) == 0 && 4048 (error = VOP_ACCESS(vp, VWRITE, l->l_cred)) == 0) { 4049 vattr_null(&vattr); 4050 vattr.va_size = SCARG(uap, length); 4051 error = VOP_SETATTR(vp, &vattr, l->l_cred); 4052 } 4053 vput(vp); 4054 return (error); 4055 } 4056 4057 /* 4058 * Truncate a file given a file descriptor. 4059 */ 4060 /* ARGSUSED */ 4061 int 4062 sys_ftruncate(struct lwp *l, const struct sys_ftruncate_args *uap, register_t *retval) 4063 { 4064 /* { 4065 syscallarg(int) fd; 4066 syscallarg(int) pad; 4067 syscallarg(off_t) length; 4068 } */ 4069 struct vattr vattr; 4070 struct vnode *vp; 4071 file_t *fp; 4072 int error; 4073 4074 if (SCARG(uap, length) < 0) 4075 return EINVAL; 4076 4077 /* fd_getvnode() will use the descriptor for us */ 4078 if ((error = fd_getvnode(SCARG(uap, fd), &fp)) != 0) 4079 return (error); 4080 if ((fp->f_flag & FWRITE) == 0) { 4081 error = EINVAL; 4082 goto out; 4083 } 4084 vp = fp->f_vnode; 4085 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 4086 if (vp->v_type == VDIR) 4087 error = EISDIR; 4088 else if ((error = vn_writechk(vp)) == 0) { 4089 vattr_null(&vattr); 4090 vattr.va_size = SCARG(uap, length); 4091 error = VOP_SETATTR(vp, &vattr, fp->f_cred); 4092 } 4093 VOP_UNLOCK(vp); 4094 out: 4095 fd_putfile(SCARG(uap, fd)); 4096 return (error); 4097 } 4098 4099 /* 4100 * Sync an open file. 4101 */ 4102 /* ARGSUSED */ 4103 int 4104 sys_fsync(struct lwp *l, const struct sys_fsync_args *uap, register_t *retval) 4105 { 4106 /* { 4107 syscallarg(int) fd; 4108 } */ 4109 struct vnode *vp; 4110 file_t *fp; 4111 int error; 4112 4113 /* fd_getvnode() will use the descriptor for us */ 4114 if ((error = fd_getvnode(SCARG(uap, fd), &fp)) != 0) 4115 return (error); 4116 vp = fp->f_vnode; 4117 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 4118 error = VOP_FSYNC(vp, fp->f_cred, FSYNC_WAIT, 0, 0); 4119 VOP_UNLOCK(vp); 4120 fd_putfile(SCARG(uap, fd)); 4121 return (error); 4122 } 4123 4124 /* 4125 * Sync a range of file data. API modeled after that found in AIX. 4126 * 4127 * FDATASYNC indicates that we need only save enough metadata to be able 4128 * to re-read the written data. 4129 */ 4130 /* ARGSUSED */ 4131 int 4132 sys_fsync_range(struct lwp *l, const struct sys_fsync_range_args *uap, register_t *retval) 4133 { 4134 /* { 4135 syscallarg(int) fd; 4136 syscallarg(int) flags; 4137 syscallarg(off_t) start; 4138 syscallarg(off_t) length; 4139 } */ 4140 struct vnode *vp; 4141 file_t *fp; 4142 int flags, nflags; 4143 off_t s, e, len; 4144 int error; 4145 4146 /* fd_getvnode() will use the descriptor for us */ 4147 if ((error = fd_getvnode(SCARG(uap, fd), &fp)) != 0) 4148 return (error); 4149 4150 if ((fp->f_flag & FWRITE) == 0) { 4151 error = EBADF; 4152 goto out; 4153 } 4154 4155 flags = SCARG(uap, flags); 4156 if (((flags & (FDATASYNC | FFILESYNC)) == 0) || 4157 ((~flags & (FDATASYNC | FFILESYNC)) == 0)) { 4158 error = EINVAL; 4159 goto out; 4160 } 4161 /* Now set up the flags for value(s) to pass to VOP_FSYNC() */ 4162 if (flags & FDATASYNC) 4163 nflags = FSYNC_DATAONLY | FSYNC_WAIT; 4164 else 4165 nflags = FSYNC_WAIT; 4166 if (flags & FDISKSYNC) 4167 nflags |= FSYNC_CACHE; 4168 4169 len = SCARG(uap, length); 4170 /* If length == 0, we do the whole file, and s = e = 0 will do that */ 4171 if (len) { 4172 s = SCARG(uap, start); 4173 e = s + len; 4174 if (e < s) { 4175 error = EINVAL; 4176 goto out; 4177 } 4178 } else { 4179 e = 0; 4180 s = 0; 4181 } 4182 4183 vp = fp->f_vnode; 4184 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 4185 error = VOP_FSYNC(vp, fp->f_cred, nflags, s, e); 4186 VOP_UNLOCK(vp); 4187 out: 4188 fd_putfile(SCARG(uap, fd)); 4189 return (error); 4190 } 4191 4192 /* 4193 * Sync the data of an open file. 4194 */ 4195 /* ARGSUSED */ 4196 int 4197 sys_fdatasync(struct lwp *l, const struct sys_fdatasync_args *uap, register_t *retval) 4198 { 4199 /* { 4200 syscallarg(int) fd; 4201 } */ 4202 struct vnode *vp; 4203 file_t *fp; 4204 int error; 4205 4206 /* fd_getvnode() will use the descriptor for us */ 4207 if ((error = fd_getvnode(SCARG(uap, fd), &fp)) != 0) 4208 return (error); 4209 vp = fp->f_vnode; 4210 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 4211 error = VOP_FSYNC(vp, fp->f_cred, FSYNC_WAIT|FSYNC_DATAONLY, 0, 0); 4212 VOP_UNLOCK(vp); 4213 fd_putfile(SCARG(uap, fd)); 4214 return (error); 4215 } 4216 4217 /* 4218 * Rename files, (standard) BSD semantics frontend. 4219 */ 4220 /* ARGSUSED */ 4221 int 4222 sys_rename(struct lwp *l, const struct sys_rename_args *uap, register_t *retval) 4223 { 4224 /* { 4225 syscallarg(const char *) from; 4226 syscallarg(const char *) to; 4227 } */ 4228 4229 return (do_sys_renameat(l, AT_FDCWD, SCARG(uap, from), AT_FDCWD, 4230 SCARG(uap, to), UIO_USERSPACE, 0)); 4231 } 4232 4233 int 4234 sys_renameat(struct lwp *l, const struct sys_renameat_args *uap, 4235 register_t *retval) 4236 { 4237 /* { 4238 syscallarg(int) fromfd; 4239 syscallarg(const char *) from; 4240 syscallarg(int) tofd; 4241 syscallarg(const char *) to; 4242 } */ 4243 4244 return (do_sys_renameat(l, SCARG(uap, fromfd), SCARG(uap, from), 4245 SCARG(uap, tofd), SCARG(uap, to), UIO_USERSPACE, 0)); 4246 } 4247 4248 /* 4249 * Rename files, POSIX semantics frontend. 4250 */ 4251 /* ARGSUSED */ 4252 int 4253 sys___posix_rename(struct lwp *l, const struct sys___posix_rename_args *uap, register_t *retval) 4254 { 4255 /* { 4256 syscallarg(const char *) from; 4257 syscallarg(const char *) to; 4258 } */ 4259 4260 return (do_sys_renameat(l, AT_FDCWD, SCARG(uap, from), AT_FDCWD, 4261 SCARG(uap, to), UIO_USERSPACE, 1)); 4262 } 4263 4264 /* 4265 * Rename files. Source and destination must either both be directories, 4266 * or both not be directories. If target is a directory, it must be empty. 4267 * If `from' and `to' refer to the same object, the value of the `retain' 4268 * argument is used to determine whether `from' will be 4269 * 4270 * (retain == 0) deleted unless `from' and `to' refer to the same 4271 * object in the file system's name space (BSD). 4272 * (retain == 1) always retained (POSIX). 4273 * 4274 * XXX Synchronize with nfsrv_rename in nfs_serv.c. 4275 */ 4276 int 4277 do_sys_rename(const char *from, const char *to, enum uio_seg seg, int retain) 4278 { 4279 return do_sys_renameat(NULL, AT_FDCWD, from, AT_FDCWD, to, seg, retain); 4280 } 4281 4282 static int 4283 do_sys_renameat(struct lwp *l, int fromfd, const char *from, int tofd, 4284 const char *to, enum uio_seg seg, int retain) 4285 { 4286 struct pathbuf *fpb, *tpb; 4287 struct nameidata fnd, tnd; 4288 struct vnode *fdvp, *fvp; 4289 struct vnode *tdvp, *tvp; 4290 struct mount *mp, *tmp; 4291 int error; 4292 4293 KASSERT(l != NULL || (fromfd == AT_FDCWD && tofd == AT_FDCWD)); 4294 4295 error = pathbuf_maybe_copyin(from, seg, &fpb); 4296 if (error) 4297 goto out0; 4298 KASSERT(fpb != NULL); 4299 4300 error = pathbuf_maybe_copyin(to, seg, &tpb); 4301 if (error) 4302 goto out1; 4303 KASSERT(tpb != NULL); 4304 4305 /* 4306 * Lookup from. 4307 * 4308 * XXX LOCKPARENT is wrong because we don't actually want it 4309 * locked yet, but (a) namei is insane, and (b) VOP_RENAME is 4310 * insane, so for the time being we need to leave it like this. 4311 */ 4312 NDINIT(&fnd, DELETE, (LOCKPARENT | TRYEMULROOT), fpb); 4313 if ((error = fd_nameiat(l, fromfd, &fnd)) != 0) 4314 goto out2; 4315 4316 /* 4317 * Pull out the important results of the lookup, fdvp and fvp. 4318 * Of course, fvp is bogus because we're about to unlock fdvp. 4319 */ 4320 fdvp = fnd.ni_dvp; 4321 fvp = fnd.ni_vp; 4322 mp = fdvp->v_mount; 4323 KASSERT(fdvp != NULL); 4324 KASSERT(fvp != NULL); 4325 KASSERT((fdvp == fvp) || (VOP_ISLOCKED(fdvp) == LK_EXCLUSIVE)); 4326 /* 4327 * Bracket the operation with fstrans_start()/fstrans_done(). 4328 * 4329 * Inside the bracket this file system cannot be unmounted so 4330 * a vnode on this file system cannot change its v_mount. 4331 * A vnode on another file system may still change to dead mount. 4332 */ 4333 fstrans_start(mp); 4334 4335 /* 4336 * Make sure neither fdvp nor fvp is locked. 4337 */ 4338 if (fdvp != fvp) 4339 VOP_UNLOCK(fdvp); 4340 /* XXX KASSERT(VOP_ISLOCKED(fdvp) != LK_EXCLUSIVE); */ 4341 /* XXX KASSERT(VOP_ISLOCKED(fvp) != LK_EXCLUSIVE); */ 4342 4343 /* 4344 * Reject renaming `.' and `..'. Can't do this until after 4345 * namei because we need namei's parsing to find the final 4346 * component name. (namei should just leave us with the final 4347 * component name and not look it up itself, but anyway...) 4348 * 4349 * This was here before because we used to relookup from 4350 * instead of to and relookup requires the caller to check 4351 * this, but now file systems may depend on this check, so we 4352 * must retain it until the file systems are all rototilled. 4353 */ 4354 if (((fnd.ni_cnd.cn_namelen == 1) && 4355 (fnd.ni_cnd.cn_nameptr[0] == '.')) || 4356 ((fnd.ni_cnd.cn_namelen == 2) && 4357 (fnd.ni_cnd.cn_nameptr[0] == '.') && 4358 (fnd.ni_cnd.cn_nameptr[1] == '.'))) { 4359 error = EINVAL; /* XXX EISDIR? */ 4360 goto abort0; 4361 } 4362 4363 /* 4364 * Lookup to. 4365 * 4366 * XXX LOCKPARENT is wrong, but...insanity, &c. Also, using 4367 * fvp here to decide whether to add CREATEDIR is a load of 4368 * bollocks because fvp might be the wrong node by now, since 4369 * fdvp is unlocked. 4370 * 4371 * XXX Why not pass CREATEDIR always? 4372 */ 4373 NDINIT(&tnd, RENAME, 4374 (LOCKPARENT | NOCACHE | TRYEMULROOT | 4375 ((fvp->v_type == VDIR)? CREATEDIR : 0)), 4376 tpb); 4377 if ((error = fd_nameiat(l, tofd, &tnd)) != 0) 4378 goto abort0; 4379 4380 /* 4381 * Pull out the important results of the lookup, tdvp and tvp. 4382 * Of course, tvp is bogus because we're about to unlock tdvp. 4383 */ 4384 tdvp = tnd.ni_dvp; 4385 tvp = tnd.ni_vp; 4386 KASSERT(tdvp != NULL); 4387 KASSERT((tdvp == tvp) || (VOP_ISLOCKED(tdvp) == LK_EXCLUSIVE)); 4388 4389 /* 4390 * Make sure neither tdvp nor tvp is locked. 4391 */ 4392 if (tdvp != tvp) 4393 VOP_UNLOCK(tdvp); 4394 /* XXX KASSERT(VOP_ISLOCKED(tdvp) != LK_EXCLUSIVE); */ 4395 /* XXX KASSERT((tvp == NULL) || (VOP_ISLOCKED(tvp) != LK_EXCLUSIVE)); */ 4396 4397 /* 4398 * Reject renaming onto `.' or `..'. relookup is unhappy with 4399 * these, which is why we must do this here. Once upon a time 4400 * we relooked up from instead of to, and consequently didn't 4401 * need this check, but now that we relookup to instead of 4402 * from, we need this; and we shall need it forever forward 4403 * until the VOP_RENAME protocol changes, because file systems 4404 * will no doubt begin to depend on this check. 4405 */ 4406 if ((tnd.ni_cnd.cn_namelen == 1) && (tnd.ni_cnd.cn_nameptr[0] == '.')) { 4407 error = EISDIR; 4408 goto abort1; 4409 } 4410 if ((tnd.ni_cnd.cn_namelen == 2) && 4411 (tnd.ni_cnd.cn_nameptr[0] == '.') && 4412 (tnd.ni_cnd.cn_nameptr[1] == '.')) { 4413 error = EINVAL; 4414 goto abort1; 4415 } 4416 4417 /* 4418 * Make sure the mount points match. Although we don't hold 4419 * any vnode locks, the v_mount on fdvp file system are stable. 4420 * 4421 * Unmounting another file system at an inopportune moment may 4422 * cause tdvp to disappear and change its v_mount to dead. 4423 * 4424 * So in either case different v_mount means cross-device rename. 4425 */ 4426 KASSERT(mp != NULL); 4427 tmp = tdvp->v_mount; 4428 4429 if (mp != tmp) { 4430 error = EXDEV; 4431 goto abort1; 4432 } 4433 4434 /* 4435 * Take the vfs rename lock to avoid cross-directory screw cases. 4436 * Nothing is locked currently, so taking this lock is safe. 4437 */ 4438 error = VFS_RENAMELOCK_ENTER(mp); 4439 if (error) 4440 goto abort1; 4441 4442 /* 4443 * Now fdvp, fvp, tdvp, and (if nonnull) tvp are referenced, 4444 * and nothing is locked except for the vfs rename lock. 4445 * 4446 * The next step is a little rain dance to conform to the 4447 * insane lock protocol, even though it does nothing to ward 4448 * off race conditions. 4449 * 4450 * We need tdvp and tvp to be locked. However, because we have 4451 * unlocked tdvp in order to hold no locks while we take the 4452 * vfs rename lock, tvp may be wrong here, and we can't safely 4453 * lock it even if the sensible file systems will just unlock 4454 * it straight away. Consequently, we must lock tdvp and then 4455 * relookup tvp to get it locked. 4456 * 4457 * Finally, because the VOP_RENAME protocol is brain-damaged 4458 * and various file systems insanely depend on the semantics of 4459 * this brain damage, the lookup of to must be the last lookup 4460 * before VOP_RENAME. 4461 */ 4462 vn_lock(tdvp, LK_EXCLUSIVE | LK_RETRY); 4463 error = relookup(tdvp, &tnd.ni_vp, &tnd.ni_cnd, 0); 4464 if (error) 4465 goto abort2; 4466 4467 /* 4468 * Drop the old tvp and pick up the new one -- which might be 4469 * the same, but that doesn't matter to us. After this, tdvp 4470 * and tvp should both be locked. 4471 */ 4472 if (tvp != NULL) 4473 vrele(tvp); 4474 tvp = tnd.ni_vp; 4475 KASSERT(VOP_ISLOCKED(tdvp) == LK_EXCLUSIVE); 4476 KASSERT((tvp == NULL) || (VOP_ISLOCKED(tvp) == LK_EXCLUSIVE)); 4477 4478 /* 4479 * The old do_sys_rename had various consistency checks here 4480 * involving fvp and tvp. fvp is bogus already here, and tvp 4481 * will become bogus soon in any sensible file system, so the 4482 * only purpose in putting these checks here is to give lip 4483 * service to these screw cases and to acknowledge that they 4484 * exist, not actually to handle them, but here you go 4485 * anyway... 4486 */ 4487 4488 /* 4489 * Acknowledge that directories and non-directories aren't 4490 * suposed to mix. 4491 */ 4492 if (tvp != NULL) { 4493 if ((fvp->v_type == VDIR) && (tvp->v_type != VDIR)) { 4494 error = ENOTDIR; 4495 goto abort3; 4496 } else if ((fvp->v_type != VDIR) && (tvp->v_type == VDIR)) { 4497 error = EISDIR; 4498 goto abort3; 4499 } 4500 } 4501 4502 /* 4503 * Acknowledge some random screw case, among the dozens that 4504 * might arise. 4505 */ 4506 if (fvp == tdvp) { 4507 error = EINVAL; 4508 goto abort3; 4509 } 4510 4511 /* 4512 * Acknowledge that POSIX has a wacky screw case. 4513 * 4514 * XXX Eventually the retain flag needs to be passed on to 4515 * VOP_RENAME. 4516 */ 4517 if (fvp == tvp) { 4518 if (retain) { 4519 error = 0; 4520 goto abort3; 4521 } else if ((fdvp == tdvp) && 4522 (fnd.ni_cnd.cn_namelen == tnd.ni_cnd.cn_namelen) && 4523 (0 == memcmp(fnd.ni_cnd.cn_nameptr, tnd.ni_cnd.cn_nameptr, 4524 fnd.ni_cnd.cn_namelen))) { 4525 error = 0; 4526 goto abort3; 4527 } 4528 } 4529 4530 /* 4531 * Make sure veriexec can screw us up. (But a race can screw 4532 * up veriexec, of course -- remember, fvp and (soon) tvp are 4533 * bogus.) 4534 */ 4535 #if NVERIEXEC > 0 4536 { 4537 char *f1, *f2; 4538 size_t f1_len; 4539 size_t f2_len; 4540 4541 f1_len = fnd.ni_cnd.cn_namelen + 1; 4542 f1 = kmem_alloc(f1_len, KM_SLEEP); 4543 strlcpy(f1, fnd.ni_cnd.cn_nameptr, f1_len); 4544 4545 f2_len = tnd.ni_cnd.cn_namelen + 1; 4546 f2 = kmem_alloc(f2_len, KM_SLEEP); 4547 strlcpy(f2, tnd.ni_cnd.cn_nameptr, f2_len); 4548 4549 error = veriexec_renamechk(curlwp, fvp, f1, tvp, f2); 4550 4551 kmem_free(f1, f1_len); 4552 kmem_free(f2, f2_len); 4553 4554 if (error) 4555 goto abort3; 4556 } 4557 #endif /* NVERIEXEC > 0 */ 4558 4559 /* 4560 * All ready. Incant the rename vop. 4561 */ 4562 /* XXX KASSERT(VOP_ISLOCKED(fdvp) != LK_EXCLUSIVE); */ 4563 /* XXX KASSERT(VOP_ISLOCKED(fvp) != LK_EXCLUSIVE); */ 4564 KASSERT(VOP_ISLOCKED(tdvp) == LK_EXCLUSIVE); 4565 KASSERT((tvp == NULL) || (VOP_ISLOCKED(tvp) == LK_EXCLUSIVE)); 4566 error = VOP_RENAME(fdvp, fvp, &fnd.ni_cnd, tdvp, tvp, &tnd.ni_cnd); 4567 4568 /* 4569 * VOP_RENAME releases fdvp, fvp, tdvp, and tvp, and unlocks 4570 * tdvp and tvp. But we can't assert any of that. 4571 */ 4572 /* XXX KASSERT(VOP_ISLOCKED(fdvp) != LK_EXCLUSIVE); */ 4573 /* XXX KASSERT(VOP_ISLOCKED(fvp) != LK_EXCLUSIVE); */ 4574 /* XXX KASSERT(VOP_ISLOCKED(tdvp) != LK_EXCLUSIVE); */ 4575 /* XXX KASSERT((tvp == NULL) || (VOP_ISLOCKED(tvp) != LK_EXCLUSIVE)); */ 4576 4577 /* 4578 * So all we have left to do is to drop the rename lock and 4579 * destroy the pathbufs. 4580 */ 4581 VFS_RENAMELOCK_EXIT(mp); 4582 fstrans_done(mp); 4583 goto out2; 4584 4585 abort3: if ((tvp != NULL) && (tvp != tdvp)) 4586 VOP_UNLOCK(tvp); 4587 abort2: VOP_UNLOCK(tdvp); 4588 VFS_RENAMELOCK_EXIT(mp); 4589 abort1: VOP_ABORTOP(tdvp, &tnd.ni_cnd); 4590 vrele(tdvp); 4591 if (tvp != NULL) 4592 vrele(tvp); 4593 abort0: VOP_ABORTOP(fdvp, &fnd.ni_cnd); 4594 vrele(fdvp); 4595 vrele(fvp); 4596 fstrans_done(mp); 4597 out2: pathbuf_destroy(tpb); 4598 out1: pathbuf_destroy(fpb); 4599 out0: return error; 4600 } 4601 4602 /* 4603 * Make a directory file. 4604 */ 4605 /* ARGSUSED */ 4606 int 4607 sys_mkdir(struct lwp *l, const struct sys_mkdir_args *uap, register_t *retval) 4608 { 4609 /* { 4610 syscallarg(const char *) path; 4611 syscallarg(int) mode; 4612 } */ 4613 4614 return do_sys_mkdirat(l, AT_FDCWD, SCARG(uap, path), 4615 SCARG(uap, mode), UIO_USERSPACE); 4616 } 4617 4618 int 4619 sys_mkdirat(struct lwp *l, const struct sys_mkdirat_args *uap, 4620 register_t *retval) 4621 { 4622 /* { 4623 syscallarg(int) fd; 4624 syscallarg(const char *) path; 4625 syscallarg(int) mode; 4626 } */ 4627 4628 return do_sys_mkdirat(l, SCARG(uap, fd), SCARG(uap, path), 4629 SCARG(uap, mode), UIO_USERSPACE); 4630 } 4631 4632 4633 int 4634 do_sys_mkdir(const char *path, mode_t mode, enum uio_seg seg) 4635 { 4636 return do_sys_mkdirat(NULL, AT_FDCWD, path, mode, seg); 4637 } 4638 4639 static int 4640 do_sys_mkdirat(struct lwp *l, int fdat, const char *path, mode_t mode, 4641 enum uio_seg seg) 4642 { 4643 struct proc *p = curlwp->l_proc; 4644 struct vnode *vp; 4645 struct vattr vattr; 4646 int error; 4647 struct pathbuf *pb; 4648 struct nameidata nd; 4649 4650 KASSERT(l != NULL || fdat == AT_FDCWD); 4651 4652 /* XXX bollocks, should pass in a pathbuf */ 4653 error = pathbuf_maybe_copyin(path, seg, &pb); 4654 if (error) { 4655 return error; 4656 } 4657 4658 NDINIT(&nd, CREATE, LOCKPARENT | CREATEDIR | TRYEMULROOT, pb); 4659 4660 if ((error = fd_nameiat(l, fdat, &nd)) != 0) { 4661 pathbuf_destroy(pb); 4662 return (error); 4663 } 4664 vp = nd.ni_vp; 4665 if (vp != NULL) { 4666 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd); 4667 if (nd.ni_dvp == vp) 4668 vrele(nd.ni_dvp); 4669 else 4670 vput(nd.ni_dvp); 4671 vrele(vp); 4672 pathbuf_destroy(pb); 4673 return (EEXIST); 4674 } 4675 vattr_null(&vattr); 4676 vattr.va_type = VDIR; 4677 /* We will read cwdi->cwdi_cmask unlocked. */ 4678 vattr.va_mode = (mode & ACCESSPERMS) &~ p->p_cwdi->cwdi_cmask; 4679 error = VOP_MKDIR(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr); 4680 if (!error) 4681 vrele(nd.ni_vp); 4682 vput(nd.ni_dvp); 4683 pathbuf_destroy(pb); 4684 return (error); 4685 } 4686 4687 /* 4688 * Remove a directory file. 4689 */ 4690 /* ARGSUSED */ 4691 int 4692 sys_rmdir(struct lwp *l, const struct sys_rmdir_args *uap, register_t *retval) 4693 { 4694 return do_sys_unlinkat(l, AT_FDCWD, SCARG(uap, path), 4695 AT_REMOVEDIR, UIO_USERSPACE); 4696 } 4697 4698 /* 4699 * Read a block of directory entries in a file system independent format. 4700 */ 4701 int 4702 sys___getdents30(struct lwp *l, const struct sys___getdents30_args *uap, register_t *retval) 4703 { 4704 /* { 4705 syscallarg(int) fd; 4706 syscallarg(char *) buf; 4707 syscallarg(size_t) count; 4708 } */ 4709 file_t *fp; 4710 int error, done; 4711 4712 /* fd_getvnode() will use the descriptor for us */ 4713 if ((error = fd_getvnode(SCARG(uap, fd), &fp)) != 0) 4714 return (error); 4715 if ((fp->f_flag & FREAD) == 0) { 4716 error = EBADF; 4717 goto out; 4718 } 4719 error = vn_readdir(fp, SCARG(uap, buf), UIO_USERSPACE, 4720 SCARG(uap, count), &done, l, 0, 0); 4721 ktrgenio(SCARG(uap, fd), UIO_READ, SCARG(uap, buf), done, error); 4722 *retval = done; 4723 out: 4724 fd_putfile(SCARG(uap, fd)); 4725 return (error); 4726 } 4727 4728 /* 4729 * Set the mode mask for creation of filesystem nodes. 4730 */ 4731 int 4732 sys_umask(struct lwp *l, const struct sys_umask_args *uap, register_t *retval) 4733 { 4734 /* { 4735 syscallarg(mode_t) newmask; 4736 } */ 4737 4738 /* 4739 * cwdi->cwdi_cmask will be read unlocked elsewhere, and no kind of 4740 * serialization with those reads is required. It's important to 4741 * return a coherent answer for the caller of umask() though, and 4742 * the atomic operation accomplishes that. 4743 */ 4744 *retval = atomic_swap_uint(&curproc->p_cwdi->cwdi_cmask, 4745 SCARG(uap, newmask) & ALLPERMS); 4746 4747 return (0); 4748 } 4749 4750 int 4751 dorevoke(struct vnode *vp, kauth_cred_t cred) 4752 { 4753 struct vattr vattr; 4754 int error, fs_decision; 4755 4756 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 4757 error = VOP_GETATTR(vp, &vattr, cred); 4758 VOP_UNLOCK(vp); 4759 if (error != 0) 4760 return error; 4761 fs_decision = (kauth_cred_geteuid(cred) == vattr.va_uid) ? 0 : EPERM; 4762 error = kauth_authorize_vnode(cred, KAUTH_VNODE_REVOKE, vp, NULL, 4763 fs_decision); 4764 if (!error) 4765 VOP_REVOKE(vp, REVOKEALL); 4766 return (error); 4767 } 4768 4769 /* 4770 * Void all references to file by ripping underlying filesystem 4771 * away from vnode. 4772 */ 4773 /* ARGSUSED */ 4774 int 4775 sys_revoke(struct lwp *l, const struct sys_revoke_args *uap, register_t *retval) 4776 { 4777 /* { 4778 syscallarg(const char *) path; 4779 } */ 4780 struct vnode *vp; 4781 int error; 4782 4783 error = namei_simple_user(SCARG(uap, path), 4784 NSM_FOLLOW_TRYEMULROOT, &vp); 4785 if (error != 0) 4786 return (error); 4787 error = dorevoke(vp, l->l_cred); 4788 vrele(vp); 4789 return (error); 4790 } 4791 4792 /* 4793 * Allocate backing store for a file, filling a hole without having to 4794 * explicitly write anything out. 4795 */ 4796 /* ARGSUSED */ 4797 int 4798 sys_posix_fallocate(struct lwp *l, const struct sys_posix_fallocate_args *uap, 4799 register_t *retval) 4800 { 4801 /* { 4802 syscallarg(int) fd; 4803 syscallarg(off_t) pos; 4804 syscallarg(off_t) len; 4805 } */ 4806 int fd; 4807 off_t pos, len; 4808 struct file *fp; 4809 struct vnode *vp; 4810 int error; 4811 4812 fd = SCARG(uap, fd); 4813 pos = SCARG(uap, pos); 4814 len = SCARG(uap, len); 4815 4816 if (pos < 0 || len < 0 || len > OFF_T_MAX - pos) { 4817 *retval = EINVAL; 4818 return 0; 4819 } 4820 4821 error = fd_getvnode(fd, &fp); 4822 if (error) { 4823 *retval = error; 4824 return 0; 4825 } 4826 if ((fp->f_flag & FWRITE) == 0) { 4827 error = EBADF; 4828 goto fail; 4829 } 4830 vp = fp->f_vnode; 4831 4832 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 4833 if (vp->v_type == VDIR) { 4834 error = EISDIR; 4835 } else { 4836 error = VOP_FALLOCATE(vp, pos, len); 4837 } 4838 VOP_UNLOCK(vp); 4839 4840 fail: 4841 fd_putfile(fd); 4842 *retval = error; 4843 return 0; 4844 } 4845 4846 /* 4847 * Deallocate backing store for a file, creating a hole. Also used for 4848 * invoking TRIM on disks. 4849 */ 4850 /* ARGSUSED */ 4851 int 4852 sys_fdiscard(struct lwp *l, const struct sys_fdiscard_args *uap, 4853 register_t *retval) 4854 { 4855 /* { 4856 syscallarg(int) fd; 4857 syscallarg(off_t) pos; 4858 syscallarg(off_t) len; 4859 } */ 4860 int fd; 4861 off_t pos, len; 4862 struct file *fp; 4863 struct vnode *vp; 4864 int error; 4865 4866 fd = SCARG(uap, fd); 4867 pos = SCARG(uap, pos); 4868 len = SCARG(uap, len); 4869 4870 if (pos < 0 || len < 0 || len > OFF_T_MAX - pos) { 4871 return EINVAL; 4872 } 4873 4874 error = fd_getvnode(fd, &fp); 4875 if (error) { 4876 return error; 4877 } 4878 if ((fp->f_flag & FWRITE) == 0) { 4879 error = EBADF; 4880 goto fail; 4881 } 4882 vp = fp->f_vnode; 4883 4884 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 4885 if (vp->v_type == VDIR) { 4886 error = EISDIR; 4887 } else { 4888 error = VOP_FDISCARD(vp, pos, len); 4889 } 4890 VOP_UNLOCK(vp); 4891 4892 fail: 4893 fd_putfile(fd); 4894 return error; 4895 } 4896
Indexes created Fri Oct 03 02:09:57 GMT 2025