1 /* 2 * ---------------------------------------------------------------------------- 3 * "THE BEER-WARE LICENSE" (Revision 42): 4 * <phk@FreeBSD.ORG> wrote this file. As long as you retain this notice you 5 * can do whatever you want with this stuff. If we meet some day, and you think 6 * this stuff is worth it, you can buy me a beer in return. Poul-Henning Kamp 7 * ---------------------------------------------------------------------------- 8 * 9 */ 10 /*- 11 * Copyright (c) 2006 Victor Balada Diaz <victor@bsdes.net> 12 * All rights reserved. 13 * 14 * Redistribution and use in source and binary forms, with or without 15 * modification, are permitted provided that the following conditions 16 * are met: 17 * 1. Redistributions of source code must retain the above copyright 18 * notice, this list of conditions and the following disclaimer. 19 * 2. Redistributions in binary form must reproduce the above copyright 20 * notice, this list of conditions and the following disclaimer in the 21 * documentation and/or other materials provided with the distribution. 22 * 23 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 26 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 33 * SUCH DAMAGE. 34 */ 35 36 37 /* 38 * $FreeBSD: src/sys/kern/kern_jail.c,v 1.6.2.3 2001/08/17 01:00:26 rwatson Exp $ 39 */ 40 41 #include "opt_inet6.h" 42 43 #include <sys/param.h> 44 #include <sys/types.h> 45 #include <sys/kernel.h> 46 #include <sys/systm.h> 47 #include <sys/errno.h> 48 #include <sys/sysproto.h> 49 #include <sys/malloc.h> 50 #include <sys/nlookup.h> 51 #include <sys/namecache.h> 52 #include <sys/proc.h> 53 #include <sys/priv.h> 54 #include <sys/jail.h> 55 #include <sys/socket.h> 56 #include <sys/sysctl.h> 57 #include <sys/kern_syscall.h> 58 #include <net/if.h> 59 #include <netinet/in.h> 60 #include <netinet6/in6_var.h> 61 62 static struct prison *prison_find(int); 63 static void prison_ipcache_init(struct prison *); 64 65 static prison_cap_t prison_default_caps; 66 67 MALLOC_DEFINE(M_PRISON, "prison", "Prison structures"); 68 69 SYSCTL_NODE(, OID_AUTO, jail, CTLFLAG_RW, 0, 70 "All jails settings"); 71 72 SYSCTL_NODE(_jail, OID_AUTO, defaults, CTLFLAG_RW, 0, 73 "Default options for jails"); 74 75 SYSCTL_BIT64(_jail_defaults, OID_AUTO, set_hostname_allowed, CTLFLAG_RW, 76 &prison_default_caps, 1, PRISON_CAP_SYS_SET_HOSTNAME, 77 "Processes in jail can set their hostnames"); 78 79 SYSCTL_BIT64(_jail_defaults, OID_AUTO, socket_unixiproute_only, CTLFLAG_RW, 80 &prison_default_caps, 0, PRISON_CAP_NET_UNIXIPROUTE, 81 "Processes in jail are limited to creating UNIX/IPv[46]/route sockets only"); 82 83 SYSCTL_BIT64(_jail_defaults, OID_AUTO, sysvipc_allowed, CTLFLAG_RW, 84 &prison_default_caps, 0, PRISON_CAP_SYS_SYSVIPC, 85 "Processes in jail can use System V IPC primitives"); 86 87 SYSCTL_BIT64(_jail_defaults, OID_AUTO, chflags_allowed, CTLFLAG_RW, 88 &prison_default_caps, 0, PRISON_CAP_VFS_CHFLAGS, 89 "Processes in jail can alter system file flags"); 90 91 SYSCTL_BIT64(_jail_defaults, OID_AUTO, allow_raw_sockets, CTLFLAG_RW, 92 &prison_default_caps, 0, PRISON_CAP_NET_RAW_SOCKETS, 93 "Process in jail can create raw sockets"); 94 95 SYSCTL_BIT64(_jail_defaults, OID_AUTO, vfs_mount_nullfs, CTLFLAG_RW, 96 &prison_default_caps, 0, PRISON_CAP_VFS_MOUNT_NULLFS, 97 "Process in jail can mount nullfs(5) filesystems"); 98 99 SYSCTL_BIT64(_jail_defaults, OID_AUTO, vfs_mount_tmpfs, CTLFLAG_RW, 100 &prison_default_caps, 0, PRISON_CAP_VFS_MOUNT_TMPFS, 101 "Process in jail can mount tmpfs(5) filesystems"); 102 103 int lastprid = 0; 104 int prisoncount = 0; 105 106 static struct lock jail_lock = 107 LOCK_INITIALIZER("jail", 0, LK_CANRECURSE); 108 109 LIST_HEAD(prisonlist, prison); 110 struct prisonlist allprison = LIST_HEAD_INITIALIZER(&allprison); 111 112 static int 113 kern_jail_attach(int jid) 114 { 115 struct proc *p = curthread->td_proc; 116 struct prison *pr; 117 struct ucred *cr; 118 int error; 119 120 pr = prison_find(jid); 121 if (pr == NULL) 122 return(EINVAL); 123 124 error = kern_chroot(&pr->pr_root); 125 if (error) 126 return(error); 127 128 prison_hold(pr); 129 lwkt_gettoken(&p->p_token); 130 cr = cratom_proc(p); 131 cr->cr_prison = pr; 132 p->p_flags |= P_JAILED; 133 lwkt_reltoken(&p->p_token); 134 135 return(0); 136 } 137 138 static int 139 assign_prison_id(struct prison *pr) 140 { 141 int tryprid; 142 struct prison *tpr; 143 144 tryprid = lastprid + 1; 145 if (tryprid == JAIL_MAX) 146 tryprid = 1; 147 148 lockmgr(&jail_lock, LK_EXCLUSIVE); 149 next: 150 LIST_FOREACH(tpr, &allprison, pr_list) { 151 if (tpr->pr_id != tryprid) 152 continue; 153 tryprid++; 154 if (tryprid == JAIL_MAX) { 155 lockmgr(&jail_lock, LK_RELEASE); 156 return (ERANGE); 157 } 158 goto next; 159 } 160 pr->pr_id = lastprid = tryprid; 161 lockmgr(&jail_lock, LK_RELEASE); 162 163 return (0); 164 } 165 166 static int 167 kern_jail(struct prison *pr, struct jail *j) 168 { 169 int error; 170 struct nlookupdata nd; 171 172 error = nlookup_init(&nd, j->path, UIO_USERSPACE, NLC_FOLLOW); 173 if (error) { 174 nlookup_done(&nd); 175 return (error); 176 } 177 error = nlookup(&nd); 178 if (error) { 179 nlookup_done(&nd); 180 return (error); 181 } 182 cache_copy(&nd.nl_nch, &pr->pr_root); 183 184 varsymset_init(&pr->pr_varsymset, NULL); 185 prison_ipcache_init(pr); 186 187 error = assign_prison_id(pr); 188 if (error) { 189 varsymset_clean(&pr->pr_varsymset); 190 nlookup_done(&nd); 191 return (error); 192 } 193 194 lockmgr(&jail_lock, LK_EXCLUSIVE); 195 LIST_INSERT_HEAD(&allprison, pr, pr_list); 196 ++prisoncount; 197 lockmgr(&jail_lock, LK_RELEASE); 198 199 error = prison_sysctl_create(pr); 200 if (error) 201 goto out; 202 203 error = kern_jail_attach(pr->pr_id); 204 if (error) 205 goto out2; 206 207 nlookup_done(&nd); 208 return 0; 209 210 out2: 211 prison_sysctl_done(pr); 212 213 out: 214 lockmgr(&jail_lock, LK_EXCLUSIVE); 215 LIST_REMOVE(pr, pr_list); 216 --prisoncount; 217 lockmgr(&jail_lock, LK_RELEASE); 218 varsymset_clean(&pr->pr_varsymset); 219 nlookup_done(&nd); 220 return (error); 221 } 222 223 /* 224 * jail() 225 * 226 * jail_args(syscallarg(struct jail *) jail) 227 * 228 * MPALMOSTSAFE 229 */ 230 int 231 sys_jail(struct jail_args *uap) 232 { 233 struct thread *td = curthread; 234 struct prison *pr; 235 struct jail_ip_storage *jip; 236 struct jail j; 237 int error; 238 uint32_t jversion; 239 240 uap->sysmsg_result = -1; 241 242 error = priv_check(td, PRIV_JAIL_CREATE); 243 if (error) 244 return (error); 245 246 error = copyin(uap->jail, &jversion, sizeof(jversion)); 247 if (error) 248 return (error); 249 250 pr = kmalloc(sizeof(*pr), M_PRISON, M_WAITOK | M_ZERO); 251 SLIST_INIT(&pr->pr_ips); 252 lockmgr(&jail_lock, LK_EXCLUSIVE); 253 254 switch (jversion) { 255 case 0: 256 /* Single IPv4 jails. */ 257 { 258 struct jail_v0 jv0; 259 struct sockaddr_in ip4addr; 260 261 error = copyin(uap->jail, &jv0, sizeof(jv0)); 262 if (error) 263 goto out; 264 265 j.path = jv0.path; 266 j.hostname = jv0.hostname; 267 268 jip = kmalloc(sizeof(*jip), M_PRISON, M_WAITOK | M_ZERO); 269 ip4addr.sin_family = AF_INET; 270 ip4addr.sin_addr.s_addr = htonl(jv0.ip_number); 271 memcpy(&jip->ip, &ip4addr, sizeof(ip4addr)); 272 SLIST_INSERT_HEAD(&pr->pr_ips, jip, entries); 273 break; 274 } 275 276 case 1: 277 /* 278 * DragonFly multi noIP/IPv4/IPv6 jails 279 * 280 * NOTE: This version is unsupported by FreeBSD 281 * (which uses version 2 instead). 282 */ 283 284 error = copyin(uap->jail, &j, sizeof(j)); 285 if (error) 286 goto out; 287 288 for (int i = 0; i < j.n_ips; i++) { 289 jip = kmalloc(sizeof(*jip), M_PRISON, 290 M_WAITOK | M_ZERO); 291 SLIST_INSERT_HEAD(&pr->pr_ips, jip, entries); 292 error = copyin(&j.ips[i], &jip->ip, 293 sizeof(struct sockaddr_storage)); 294 if (error) 295 goto out; 296 } 297 break; 298 default: 299 error = EINVAL; 300 goto out; 301 } 302 303 error = copyinstr(j.hostname, &pr->pr_host, sizeof(pr->pr_host), 0); 304 if (error) 305 goto out; 306 307 /* Use default capabilities as a template */ 308 pr->pr_caps = prison_default_caps; 309 310 error = kern_jail(pr, &j); 311 if (error) 312 goto out; 313 314 uap->sysmsg_result = pr->pr_id; 315 lockmgr(&jail_lock, LK_RELEASE); 316 317 return (0); 318 319 out: 320 /* Delete all ips */ 321 while (!SLIST_EMPTY(&pr->pr_ips)) { 322 jip = SLIST_FIRST(&pr->pr_ips); 323 SLIST_REMOVE_HEAD(&pr->pr_ips, entries); 324 kfree(jip, M_PRISON); 325 } 326 lockmgr(&jail_lock, LK_RELEASE); 327 kfree(pr, M_PRISON); 328 329 return (error); 330 } 331 332 /* 333 * int jail_attach(int jid); 334 * 335 * MPALMOSTSAFE 336 */ 337 int 338 sys_jail_attach(struct jail_attach_args *uap) 339 { 340 struct thread *td = curthread; 341 int error; 342 343 error = priv_check(td, PRIV_JAIL_ATTACH); 344 if (error) 345 return(error); 346 lockmgr(&jail_lock, LK_EXCLUSIVE); 347 error = kern_jail_attach(uap->jid); 348 lockmgr(&jail_lock, LK_RELEASE); 349 return (error); 350 } 351 352 static void 353 prison_ipcache_init(struct prison *pr) 354 { 355 struct jail_ip_storage *jis; 356 struct sockaddr_in *ip4; 357 struct sockaddr_in6 *ip6; 358 359 lockmgr(&jail_lock, LK_EXCLUSIVE); 360 SLIST_FOREACH(jis, &pr->pr_ips, entries) { 361 switch (jis->ip.ss_family) { 362 case AF_INET: 363 ip4 = (struct sockaddr_in *)&jis->ip; 364 if ((ntohl(ip4->sin_addr.s_addr) >> IN_CLASSA_NSHIFT) == 365 IN_LOOPBACKNET) { 366 /* loopback address */ 367 if (pr->local_ip4 == NULL) 368 pr->local_ip4 = ip4; 369 } else { 370 /* public address */ 371 if (pr->nonlocal_ip4 == NULL) 372 pr->nonlocal_ip4 = ip4; 373 } 374 break; 375 376 case AF_INET6: 377 ip6 = (struct sockaddr_in6 *)&jis->ip; 378 if (IN6_IS_ADDR_LOOPBACK(&ip6->sin6_addr)) { 379 /* loopback address */ 380 if (pr->local_ip6 == NULL) 381 pr->local_ip6 = ip6; 382 } else { 383 /* public address */ 384 if (pr->nonlocal_ip6 == NULL) 385 pr->nonlocal_ip6 = ip6; 386 } 387 break; 388 } 389 } 390 lockmgr(&jail_lock, LK_RELEASE); 391 } 392 393 /* 394 * Changes INADDR_LOOPBACK for a valid jail address. 395 * ip is in network byte order. 396 * Returns 1 if the ip is among jail valid ips. 397 * Returns 0 if is not among jail valid ips or 398 * if couldn't replace INADDR_LOOPBACK for a valid 399 * IP. 400 */ 401 int 402 prison_replace_wildcards(struct thread *td, struct sockaddr *ip) 403 { 404 struct sockaddr_in *ip4 = (struct sockaddr_in *)ip; 405 struct sockaddr_in6 *ip6 = (struct sockaddr_in6 *)ip; 406 struct prison *pr; 407 408 if (td->td_proc == NULL || td->td_ucred == NULL) 409 return (1); 410 if ((pr = td->td_ucred->cr_prison) == NULL) 411 return (1); 412 413 if ((ip->sa_family == AF_INET && 414 ip4->sin_addr.s_addr == htonl(INADDR_ANY)) || 415 (ip->sa_family == AF_INET6 && 416 IN6_IS_ADDR_UNSPECIFIED(&ip6->sin6_addr))) 417 return (1); 418 if ((ip->sa_family == AF_INET && 419 ip4->sin_addr.s_addr == htonl(INADDR_LOOPBACK)) || 420 (ip->sa_family == AF_INET6 && 421 IN6_IS_ADDR_LOOPBACK(&ip6->sin6_addr))) { 422 if (!prison_get_local(pr, ip->sa_family, ip) && 423 !prison_get_nonlocal(pr, ip->sa_family, ip)) 424 return(0); 425 else 426 return(1); 427 } 428 if (jailed_ip(pr, ip)) 429 return(1); 430 return(0); 431 } 432 433 int 434 prison_remote_ip(struct thread *td, struct sockaddr *ip) 435 { 436 struct sockaddr_in *ip4 = (struct sockaddr_in *)ip; 437 struct sockaddr_in6 *ip6 = (struct sockaddr_in6 *)ip; 438 struct prison *pr; 439 440 if (td == NULL || td->td_proc == NULL || td->td_ucred == NULL) 441 return(1); 442 if ((pr = td->td_ucred->cr_prison) == NULL) 443 return(1); 444 if ((ip->sa_family == AF_INET && 445 ip4->sin_addr.s_addr == htonl(INADDR_LOOPBACK)) || 446 (ip->sa_family == AF_INET6 && 447 IN6_IS_ADDR_LOOPBACK(&ip6->sin6_addr))) { 448 if (!prison_get_local(pr, ip->sa_family, ip) && 449 !prison_get_nonlocal(pr, ip->sa_family, ip)) 450 return(0); 451 else 452 return(1); 453 } 454 return(1); 455 } 456 457 /* 458 * Prison get non loopback ip: 459 * - af is the address family of the ip we want (AF_INET|AF_INET6). 460 * - If ip != NULL, put the first IP address that is not a loopback address 461 * into *ip. 462 * 463 * ip is in network by order and we don't touch it unless we find a valid ip. 464 * No matter if ip == NULL or not, we return either a valid struct sockaddr *, 465 * or NULL. This struct may not be modified. 466 */ 467 struct sockaddr * 468 prison_get_nonlocal(struct prison *pr, sa_family_t af, struct sockaddr *ip) 469 { 470 struct sockaddr_in *ip4 = (struct sockaddr_in *)ip; 471 struct sockaddr_in6 *ip6 = (struct sockaddr_in6 *)ip; 472 473 /* Check if it is cached */ 474 switch(af) { 475 case AF_INET: 476 if (ip4 != NULL && pr->nonlocal_ip4 != NULL) 477 ip4->sin_addr.s_addr = pr->nonlocal_ip4->sin_addr.s_addr; 478 return (struct sockaddr *)pr->nonlocal_ip4; 479 480 case AF_INET6: 481 if (ip6 != NULL && pr->nonlocal_ip6 != NULL) 482 ip6->sin6_addr = pr->nonlocal_ip6->sin6_addr; 483 return (struct sockaddr *)pr->nonlocal_ip6; 484 } 485 486 /* NOTREACHED */ 487 return NULL; 488 } 489 490 /* 491 * Prison get loopback ip. 492 * - af is the address family of the ip we want (AF_INET|AF_INET6). 493 * - If ip != NULL, put the first IP address that is not a loopback address 494 * into *ip. 495 * 496 * ip is in network by order and we don't touch it unless we find a valid ip. 497 * No matter if ip == NULL or not, we return either a valid struct sockaddr *, 498 * or NULL. This struct may not be modified. 499 */ 500 struct sockaddr * 501 prison_get_local(struct prison *pr, sa_family_t af, struct sockaddr *ip) 502 { 503 struct sockaddr_in *ip4 = (struct sockaddr_in *)ip; 504 struct sockaddr_in6 *ip6 = (struct sockaddr_in6 *)ip; 505 506 /* Check if it is cached */ 507 switch(af) { 508 case AF_INET: 509 if (ip4 != NULL && pr->local_ip4 != NULL) 510 ip4->sin_addr.s_addr = pr->local_ip4->sin_addr.s_addr; 511 return (struct sockaddr *)pr->local_ip4; 512 513 case AF_INET6: 514 if (ip6 != NULL && pr->local_ip6 != NULL) 515 ip6->sin6_addr = pr->local_ip6->sin6_addr; 516 return (struct sockaddr *)pr->local_ip6; 517 } 518 519 /* NOTREACHED */ 520 return NULL; 521 } 522 523 /* Check if the IP is among ours, if it is return 1, else 0 */ 524 int 525 jailed_ip(struct prison *pr, struct sockaddr *ip) 526 { 527 struct jail_ip_storage *jis; 528 struct sockaddr_in *jip4, *ip4; 529 struct sockaddr_in6 *jip6, *ip6; 530 531 if (pr == NULL) 532 return(0); 533 ip4 = (struct sockaddr_in *)ip; 534 ip6 = (struct sockaddr_in6 *)ip; 535 536 lockmgr(&jail_lock, LK_EXCLUSIVE); 537 SLIST_FOREACH(jis, &pr->pr_ips, entries) { 538 switch (ip->sa_family) { 539 case AF_INET: 540 jip4 = (struct sockaddr_in *) &jis->ip; 541 if (jip4->sin_family == AF_INET && 542 ip4->sin_addr.s_addr == jip4->sin_addr.s_addr) { 543 lockmgr(&jail_lock, LK_RELEASE); 544 return(1); 545 } 546 break; 547 case AF_INET6: 548 jip6 = (struct sockaddr_in6 *) &jis->ip; 549 if (jip6->sin6_family == AF_INET6 && 550 IN6_ARE_ADDR_EQUAL(&ip6->sin6_addr, 551 &jip6->sin6_addr)) { 552 lockmgr(&jail_lock, LK_RELEASE); 553 return(1); 554 } 555 break; 556 } 557 } 558 lockmgr(&jail_lock, LK_RELEASE); 559 /* Ip not in list */ 560 return(0); 561 } 562 563 int 564 prison_if(struct ucred *cred, struct sockaddr *sa) 565 { 566 struct prison *pr; 567 struct sockaddr_in *sai = (struct sockaddr_in*) sa; 568 569 pr = cred->cr_prison; 570 571 if (((sai->sin_family != AF_INET) && (sai->sin_family != AF_INET6)) 572 && PRISON_CAP_ISSET(pr->pr_caps, PRISON_CAP_NET_UNIXIPROUTE)) 573 return(1); 574 else if ((sai->sin_family != AF_INET) && (sai->sin_family != AF_INET6)) 575 return(0); 576 else if (jailed_ip(pr, sa)) 577 return(0); 578 return(1); 579 } 580 581 /* 582 * Returns a prison instance, or NULL on failure. 583 */ 584 static struct prison * 585 prison_find(int prid) 586 { 587 struct prison *pr; 588 589 lockmgr(&jail_lock, LK_EXCLUSIVE); 590 LIST_FOREACH(pr, &allprison, pr_list) { 591 if (pr->pr_id == prid) 592 break; 593 } 594 lockmgr(&jail_lock, LK_RELEASE); 595 596 return(pr); 597 } 598 599 static int 600 sysctl_jail_list(SYSCTL_HANDLER_ARGS) 601 { 602 struct thread *td = curthread; 603 struct jail_ip_storage *jip; 604 #ifdef INET6 605 struct sockaddr_in6 *jsin6; 606 #endif 607 struct sockaddr_in *jsin; 608 struct lwp *lp; 609 struct prison *pr; 610 unsigned int jlssize, jlsused; 611 int count, error; 612 char *jls; /* Jail list */ 613 char *oip; /* Output ip */ 614 char *fullpath, *freepath; 615 616 jlsused = 0; 617 618 if (jailed(td->td_ucred)) 619 return (0); 620 lp = td->td_lwp; 621 retry: 622 count = prisoncount; 623 624 if (count == 0) 625 return(0); 626 627 jlssize = (count * 1024); 628 jls = kmalloc(jlssize + 1, M_TEMP, M_WAITOK | M_ZERO); 629 if (count < prisoncount) { 630 kfree(jls, M_TEMP); 631 goto retry; 632 } 633 count = prisoncount; 634 635 lockmgr(&jail_lock, LK_EXCLUSIVE); 636 LIST_FOREACH(pr, &allprison, pr_list) { 637 error = cache_fullpath(lp->lwp_proc, &pr->pr_root, NULL, 638 &fullpath, &freepath, 0); 639 if (error) 640 continue; 641 if (jlsused && jlsused < jlssize) 642 jls[jlsused++] = '\n'; 643 count = ksnprintf(jls + jlsused, (jlssize - jlsused), 644 "%d %s %s", 645 pr->pr_id, pr->pr_host, fullpath); 646 kfree(freepath, M_TEMP); 647 if (count < 0) 648 goto end; 649 jlsused += count; 650 651 /* Copy the IPS */ 652 SLIST_FOREACH(jip, &pr->pr_ips, entries) { 653 char buf[INET_ADDRSTRLEN]; 654 655 jsin = (struct sockaddr_in *)&jip->ip; 656 657 switch(jsin->sin_family) { 658 case AF_INET: 659 oip = kinet_ntoa(jsin->sin_addr, buf); 660 break; 661 #ifdef INET6 662 case AF_INET6: 663 jsin6 = (struct sockaddr_in6 *)&jip->ip; 664 oip = ip6_sprintf(&jsin6->sin6_addr); 665 break; 666 #endif 667 default: 668 oip = "?family?"; 669 break; 670 } 671 672 if ((jlssize - jlsused) < (strlen(oip) + 1)) { 673 error = ERANGE; 674 goto end; 675 } 676 count = ksnprintf(jls + jlsused, (jlssize - jlsused), 677 " %s", oip); 678 if (count < 0) 679 goto end; 680 jlsused += count; 681 } 682 } 683 684 /* 685 * The format is: 686 * pr_id <SPC> hostname1 <SPC> PATH1 <SPC> IP1 <SPC> IP2\npr_id... 687 */ 688 error = SYSCTL_OUT(req, jls, jlsused); 689 end: 690 lockmgr(&jail_lock, LK_RELEASE); 691 kfree(jls, M_TEMP); 692 693 return(error); 694 } 695 696 SYSCTL_OID(_jail, OID_AUTO, list, CTLTYPE_STRING | CTLFLAG_RD, NULL, 0, 697 sysctl_jail_list, "A", "List of active jails"); 698 699 static int 700 sysctl_jail_jailed(SYSCTL_HANDLER_ARGS) 701 { 702 int error, injail; 703 704 injail = jailed(req->td->td_ucred); 705 error = SYSCTL_OUT(req, &injail, sizeof(injail)); 706 707 return (error); 708 } 709 710 SYSCTL_PROC(_jail, OID_AUTO, jailed, 711 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_NOLOCK, NULL, 0, 712 sysctl_jail_jailed, "I", "Process in jail?"); 713 714 /* 715 * MPSAFE 716 */ 717 void 718 prison_hold(struct prison *pr) 719 { 720 atomic_add_int(&pr->pr_ref, 1); 721 } 722 723 /* 724 * MPALMOSTSAFE 725 */ 726 void 727 prison_free(struct prison *pr) 728 { 729 struct jail_ip_storage *jls; 730 731 KKASSERT(pr->pr_ref > 0); 732 if (atomic_fetchadd_int(&pr->pr_ref, -1) != 1) 733 return; 734 735 /* 736 * The MP lock is needed on the last ref to adjust 737 * the list. 738 */ 739 lockmgr(&jail_lock, LK_EXCLUSIVE); 740 if (pr->pr_ref) { 741 lockmgr(&jail_lock, LK_RELEASE); 742 return; 743 } 744 LIST_REMOVE(pr, pr_list); 745 --prisoncount; 746 747 /* 748 * Clean up 749 */ 750 while (!SLIST_EMPTY(&pr->pr_ips)) { 751 jls = SLIST_FIRST(&pr->pr_ips); 752 SLIST_REMOVE_HEAD(&pr->pr_ips, entries); 753 kfree(jls, M_PRISON); 754 } 755 lockmgr(&jail_lock, LK_RELEASE); 756 757 if (pr->pr_linux != NULL) 758 kfree(pr->pr_linux, M_PRISON); 759 varsymset_clean(&pr->pr_varsymset); 760 761 /* Release the sysctl tree */ 762 prison_sysctl_done(pr); 763 764 cache_drop(&pr->pr_root); 765 kfree(pr, M_PRISON); 766 } 767 768 /* 769 * Check if permisson for a specific privilege is granted within jail. 770 * 771 * MPSAFE 772 */ 773 int 774 prison_priv_check(struct ucred *cred, int priv) 775 { 776 struct prison *pr = cred->cr_prison; 777 778 if (!jailed(cred)) 779 return (0); 780 781 switch (priv) { 782 case PRIV_CRED_SETUID: 783 case PRIV_CRED_SETEUID: 784 case PRIV_CRED_SETGID: 785 case PRIV_CRED_SETEGID: 786 case PRIV_CRED_SETGROUPS: 787 case PRIV_CRED_SETREUID: 788 case PRIV_CRED_SETREGID: 789 case PRIV_CRED_SETRESUID: 790 case PRIV_CRED_SETRESGID: 791 792 case PRIV_VFS_SYSFLAGS: 793 case PRIV_VFS_CHOWN: 794 case PRIV_VFS_CHMOD: 795 case PRIV_VFS_CHROOT: 796 case PRIV_VFS_LINK: 797 case PRIV_VFS_CHFLAGS_DEV: 798 case PRIV_VFS_REVOKE: 799 case PRIV_VFS_MKNOD_BAD: 800 case PRIV_VFS_MKNOD_WHT: 801 case PRIV_VFS_MKNOD_DIR: 802 return (0); 803 804 case PRIV_VFS_MOUNT_NULLFS: 805 if (PRISON_CAP_ISSET(pr->pr_caps, PRISON_CAP_VFS_MOUNT_NULLFS)) 806 return (0); 807 else 808 return (EPERM); 809 case PRIV_VFS_MOUNT_DEVFS: 810 return (EPERM); 811 case PRIV_VFS_MOUNT_TMPFS: 812 if (PRISON_CAP_ISSET(pr->pr_caps, PRISON_CAP_VFS_MOUNT_TMPFS)) 813 return (0); 814 else 815 return (EPERM); 816 817 case PRIV_VFS_SETATTR: 818 case PRIV_VFS_SETGID: 819 820 case PRIV_PROC_SETRLIMIT: 821 case PRIV_PROC_SETLOGIN: 822 823 case PRIV_SYSCTL_WRITEJAIL: 824 825 case PRIV_VARSYM_SYS: 826 827 case PRIV_SETHOSTNAME: 828 829 case PRIV_PROC_TRESPASS: 830 831 return (0); 832 833 case PRIV_UFS_QUOTAON: 834 case PRIV_UFS_QUOTAOFF: 835 case PRIV_VFS_SETQUOTA: 836 case PRIV_UFS_SETUSE: 837 case PRIV_VFS_GETQUOTA: 838 return (0); 839 840 841 case PRIV_DEBUG_UNPRIV: 842 return (0); 843 844 845 /* 846 * Allow jailed root to bind reserved ports. 847 */ 848 case PRIV_NETINET_RESERVEDPORT: 849 return (0); 850 851 852 /* 853 * Conditionally allow creating raw sockets in jail. 854 */ 855 case PRIV_NETINET_RAW: 856 if (PRISON_CAP_ISSET(pr->pr_caps, 857 PRISON_CAP_NET_RAW_SOCKETS)) 858 return (0); 859 else 860 return (EPERM); 861 862 case PRIV_HAMMER_IOCTL: 863 return (0); 864 865 default: 866 867 return (EPERM); 868 } 869 } 870 871 872 /* 873 * Create a per-jail sysctl tree to control the prison 874 */ 875 int 876 prison_sysctl_create(struct prison *pr) 877 { 878 char id_str[7]; 879 880 ksnprintf(id_str, 6, "%d", pr->pr_id); 881 882 pr->pr_sysctl_ctx = (struct sysctl_ctx_list *) kmalloc( 883 sizeof(struct sysctl_ctx_list), M_PRISON, M_WAITOK | M_ZERO); 884 885 sysctl_ctx_init(pr->pr_sysctl_ctx); 886 887 /* Main jail node */ 888 pr->pr_sysctl_tree = SYSCTL_ADD_NODE(pr->pr_sysctl_ctx, 889 SYSCTL_STATIC_CHILDREN(_jail), 890 OID_AUTO, id_str, CTLFLAG_RD, 0, 891 "Jail specific settings"); 892 893 SYSCTL_ADD_BIT64(pr->pr_sysctl_ctx, SYSCTL_CHILDREN(pr->pr_sysctl_tree), 894 OID_AUTO, "sys_set_hostname", CTLFLAG_RW, 895 &pr->pr_caps, 0, PRISON_CAP_SYS_SET_HOSTNAME, 896 "Processes in jail can set their hostnames"); 897 898 SYSCTL_ADD_BIT64(pr->pr_sysctl_ctx, SYSCTL_CHILDREN(pr->pr_sysctl_tree), 899 OID_AUTO, "sys_sysvipc", CTLFLAG_RW, 900 &pr->pr_caps, 0, PRISON_CAP_SYS_SYSVIPC, 901 "Processes in jail can use System V IPC primitives"); 902 903 SYSCTL_ADD_BIT64(pr->pr_sysctl_ctx, SYSCTL_CHILDREN(pr->pr_sysctl_tree), 904 OID_AUTO, "net_unixiproute", CTLFLAG_RW, 905 &pr->pr_caps, 0, PRISON_CAP_NET_UNIXIPROUTE, 906 "Processes in jail are limited to creating UNIX/IPv[46]/route sockets only"); 907 908 SYSCTL_ADD_BIT64(pr->pr_sysctl_ctx, SYSCTL_CHILDREN(pr->pr_sysctl_tree), 909 OID_AUTO, "net_raw_sockets", CTLFLAG_RW, 910 &pr->pr_caps, 0, PRISON_CAP_NET_RAW_SOCKETS, 911 "Process in jail can create raw sockets"); 912 913 SYSCTL_ADD_BIT64(pr->pr_sysctl_ctx, SYSCTL_CHILDREN(pr->pr_sysctl_tree), 914 OID_AUTO, "vfs_chflags", CTLFLAG_RW, 915 &pr->pr_caps, 0, PRISON_CAP_VFS_CHFLAGS, 916 "Processes in jail can alter system file flags"); 917 918 SYSCTL_ADD_BIT64(pr->pr_sysctl_ctx, SYSCTL_CHILDREN(pr->pr_sysctl_tree), 919 OID_AUTO, "vfs_mount_nullfs", CTLFLAG_RW, 920 &pr->pr_caps, 0, PRISON_CAP_VFS_MOUNT_NULLFS, 921 "Processes in jail can mount nullfs(5) filesystems"); 922 923 SYSCTL_ADD_BIT64(pr->pr_sysctl_ctx, SYSCTL_CHILDREN(pr->pr_sysctl_tree), 924 OID_AUTO, "vfs_mount_tmpfs", CTLFLAG_RW, 925 &pr->pr_caps, 0, PRISON_CAP_VFS_MOUNT_TMPFS, 926 "Processes in jail can mount tmpfs(5) filesystems"); 927 928 return 0; 929 } 930 931 int 932 prison_sysctl_done(struct prison *pr) 933 { 934 if (pr->pr_sysctl_tree) { 935 sysctl_ctx_free(pr->pr_sysctl_ctx); 936 kfree(pr->pr_sysctl_ctx, M_PRISON); 937 pr->pr_sysctl_tree = NULL; 938 } 939 940 return 0; 941 } 942