1 /* 2 * INET An implementation of the TCP/IP protocol suite for the LINUX 3 * operating system. INET is implemented using the BSD Socket 4 * interface as the means of communication with the user level. 5 * 6 * The IP to API glue. 7 * 8 * Version: $Id: ip_sockglue.c,v 1.62 2002/02/01 22:01:04 davem Exp $ 9 * 10 * Authors: see ip.c 11 * 12 * Fixes: 13 * Many : Split from ip.c , see ip.c for history. 14 * Martin Mares : TOS setting fixed. 15 * Alan Cox : Fixed a couple of oopses in Martin's 16 * TOS tweaks. 17 * Mike McLagan : Routing by source 18 */ 19 20 #include <linux/module.h> 21 #include <linux/types.h> 22 #include <linux/mm.h> 23 #include <linux/skbuff.h> 24 #include <linux/ip.h> 25 #include <linux/icmp.h> 26 #include <linux/inetdevice.h> 27 #include <linux/netdevice.h> 28 #include <net/sock.h> 29 #include <net/ip.h> 30 #include <net/icmp.h> 31 #include <net/tcp_states.h> 32 #include <linux/udp.h> 33 #include <linux/igmp.h> 34 #include <linux/netfilter.h> 35 #include <linux/route.h> 36 #include <linux/mroute.h> 37 #include <net/route.h> 38 #include <net/xfrm.h> 39 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) 40 #include <net/transp_v6.h> 41 #endif 42 43 #include <linux/errqueue.h> 44 #include <asm/uaccess.h> 45 46 #define IP_CMSG_PKTINFO 1 47 #define IP_CMSG_TTL 2 48 #define IP_CMSG_TOS 4 49 #define IP_CMSG_RECVOPTS 8 50 #define IP_CMSG_RETOPTS 16 51 #define IP_CMSG_PASSSEC 32 52 53 /* 54 * SOL_IP control messages. 55 */ 56 57 static void ip_cmsg_recv_pktinfo(struct msghdr *msg, struct sk_buff *skb) 58 { 59 struct in_pktinfo info; 60 struct rtable *rt = (struct rtable *)skb->dst; 61 62 info.ipi_addr.s_addr = ip_hdr(skb)->daddr; 63 if (rt) { 64 info.ipi_ifindex = rt->rt_iif; 65 info.ipi_spec_dst.s_addr = rt->rt_spec_dst; 66 } else { 67 info.ipi_ifindex = 0; 68 info.ipi_spec_dst.s_addr = 0; 69 } 70 71 put_cmsg(msg, SOL_IP, IP_PKTINFO, sizeof(info), &info); 72 } 73 74 static void ip_cmsg_recv_ttl(struct msghdr *msg, struct sk_buff *skb) 75 { 76 int ttl = ip_hdr(skb)->ttl; 77 put_cmsg(msg, SOL_IP, IP_TTL, sizeof(int), &ttl); 78 } 79 80 static void ip_cmsg_recv_tos(struct msghdr *msg, struct sk_buff *skb) 81 { 82 put_cmsg(msg, SOL_IP, IP_TOS, 1, &ip_hdr(skb)->tos); 83 } 84 85 static void ip_cmsg_recv_opts(struct msghdr *msg, struct sk_buff *skb) 86 { 87 if (IPCB(skb)->opt.optlen == 0) 88 return; 89 90 put_cmsg(msg, SOL_IP, IP_RECVOPTS, IPCB(skb)->opt.optlen, 91 ip_hdr(skb) + 1); 92 } 93 94 95 static void ip_cmsg_recv_retopts(struct msghdr *msg, struct sk_buff *skb) 96 { 97 unsigned char optbuf[sizeof(struct ip_options) + 40]; 98 struct ip_options * opt = (struct ip_options*)optbuf; 99 100 if (IPCB(skb)->opt.optlen == 0) 101 return; 102 103 if (ip_options_echo(opt, skb)) { 104 msg->msg_flags |= MSG_CTRUNC; 105 return; 106 } 107 ip_options_undo(opt); 108 109 put_cmsg(msg, SOL_IP, IP_RETOPTS, opt->optlen, opt->__data); 110 } 111 112 static void ip_cmsg_recv_security(struct msghdr *msg, struct sk_buff *skb) 113 { 114 char *secdata; 115 u32 seclen, secid; 116 int err; 117 118 err = security_socket_getpeersec_dgram(NULL, skb, &secid); 119 if (err) 120 return; 121 122 err = security_secid_to_secctx(secid, &secdata, &seclen); 123 if (err) 124 return; 125 126 put_cmsg(msg, SOL_IP, SCM_SECURITY, seclen, secdata); 127 security_release_secctx(secdata, seclen); 128 } 129 130 131 void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb) 132 { 133 struct inet_sock *inet = inet_sk(skb->sk); 134 unsigned flags = inet->cmsg_flags; 135 136 /* Ordered by supposed usage frequency */ 137 if (flags & 1) 138 ip_cmsg_recv_pktinfo(msg, skb); 139 if ((flags>>=1) == 0) 140 return; 141 142 if (flags & 1) 143 ip_cmsg_recv_ttl(msg, skb); 144 if ((flags>>=1) == 0) 145 return; 146 147 if (flags & 1) 148 ip_cmsg_recv_tos(msg, skb); 149 if ((flags>>=1) == 0) 150 return; 151 152 if (flags & 1) 153 ip_cmsg_recv_opts(msg, skb); 154 if ((flags>>=1) == 0) 155 return; 156 157 if (flags & 1) 158 ip_cmsg_recv_retopts(msg, skb); 159 if ((flags>>=1) == 0) 160 return; 161 162 if (flags & 1) 163 ip_cmsg_recv_security(msg, skb); 164 } 165 166 int ip_cmsg_send(struct msghdr *msg, struct ipcm_cookie *ipc) 167 { 168 int err; 169 struct cmsghdr *cmsg; 170 171 for (cmsg = CMSG_FIRSTHDR(msg); cmsg; cmsg = CMSG_NXTHDR(msg, cmsg)) { 172 if (!CMSG_OK(msg, cmsg)) 173 return -EINVAL; 174 if (cmsg->cmsg_level != SOL_IP) 175 continue; 176 switch (cmsg->cmsg_type) { 177 case IP_RETOPTS: 178 err = cmsg->cmsg_len - CMSG_ALIGN(sizeof(struct cmsghdr)); 179 err = ip_options_get(&ipc->opt, CMSG_DATA(cmsg), err < 40 ? err : 40); 180 if (err) 181 return err; 182 break; 183 case IP_PKTINFO: 184 { 185 struct in_pktinfo *info; 186 if (cmsg->cmsg_len != CMSG_LEN(sizeof(struct in_pktinfo))) 187 return -EINVAL; 188 info = (struct in_pktinfo *)CMSG_DATA(cmsg); 189 ipc->oif = info->ipi_ifindex; 190 ipc->addr = info->ipi_spec_dst.s_addr; 191 break; 192 } 193 default: 194 return -EINVAL; 195 } 196 } 197 return 0; 198 } 199 200 201 /* Special input handler for packets caught by router alert option. 202 They are selected only by protocol field, and then processed likely 203 local ones; but only if someone wants them! Otherwise, router 204 not running rsvpd will kill RSVP. 205 206 It is user level problem, what it will make with them. 207 I have no idea, how it will masquearde or NAT them (it is joke, joke :-)), 208 but receiver should be enough clever f.e. to forward mtrace requests, 209 sent to multicast group to reach destination designated router. 210 */ 211 struct ip_ra_chain *ip_ra_chain; 212 DEFINE_RWLOCK(ip_ra_lock); 213 214 int ip_ra_control(struct sock *sk, unsigned char on, void (*destructor)(struct sock *)) 215 { 216 struct ip_ra_chain *ra, *new_ra, **rap; 217 218 if (sk->sk_type != SOCK_RAW || inet_sk(sk)->num == IPPROTO_RAW) 219 return -EINVAL; 220 221 new_ra = on ? kmalloc(sizeof(*new_ra), GFP_KERNEL) : NULL; 222 223 write_lock_bh(&ip_ra_lock); 224 for (rap = &ip_ra_chain; (ra=*rap) != NULL; rap = &ra->next) { 225 if (ra->sk == sk) { 226 if (on) { 227 write_unlock_bh(&ip_ra_lock); 228 kfree(new_ra); 229 return -EADDRINUSE; 230 } 231 *rap = ra->next; 232 write_unlock_bh(&ip_ra_lock); 233 234 if (ra->destructor) 235 ra->destructor(sk); 236 sock_put(sk); 237 kfree(ra); 238 return 0; 239 } 240 } 241 if (new_ra == NULL) { 242 write_unlock_bh(&ip_ra_lock); 243 return -ENOBUFS; 244 } 245 new_ra->sk = sk; 246 new_ra->destructor = destructor; 247 248 new_ra->next = ra; 249 *rap = new_ra; 250 sock_hold(sk); 251 write_unlock_bh(&ip_ra_lock); 252 253 return 0; 254 } 255 256 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, 257 __be16 port, u32 info, u8 *payload) 258 { 259 struct inet_sock *inet = inet_sk(sk); 260 struct sock_exterr_skb *serr; 261 262 if (!inet->recverr) 263 return; 264 265 skb = skb_clone(skb, GFP_ATOMIC); 266 if (!skb) 267 return; 268 269 serr = SKB_EXT_ERR(skb); 270 serr->ee.ee_errno = err; 271 serr->ee.ee_origin = SO_EE_ORIGIN_ICMP; 272 serr->ee.ee_type = icmp_hdr(skb)->type; 273 serr->ee.ee_code = icmp_hdr(skb)->code; 274 serr->ee.ee_pad = 0; 275 serr->ee.ee_info = info; 276 serr->ee.ee_data = 0; 277 serr->addr_offset = (u8 *)&(((struct iphdr *)(icmp_hdr(skb) + 1))->daddr) - 278 skb_network_header(skb); 279 serr->port = port; 280 281 if (skb_pull(skb, payload - skb->data) != NULL) { 282 skb_reset_transport_header(skb); 283 if (sock_queue_err_skb(sk, skb) == 0) 284 return; 285 } 286 kfree_skb(skb); 287 } 288 289 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 port, u32 info) 290 { 291 struct inet_sock *inet = inet_sk(sk); 292 struct sock_exterr_skb *serr; 293 struct iphdr *iph; 294 struct sk_buff *skb; 295 296 if (!inet->recverr) 297 return; 298 299 skb = alloc_skb(sizeof(struct iphdr), GFP_ATOMIC); 300 if (!skb) 301 return; 302 303 skb_put(skb, sizeof(struct iphdr)); 304 skb_reset_network_header(skb); 305 iph = ip_hdr(skb); 306 iph->daddr = daddr; 307 308 serr = SKB_EXT_ERR(skb); 309 serr->ee.ee_errno = err; 310 serr->ee.ee_origin = SO_EE_ORIGIN_LOCAL; 311 serr->ee.ee_type = 0; 312 serr->ee.ee_code = 0; 313 serr->ee.ee_pad = 0; 314 serr->ee.ee_info = info; 315 serr->ee.ee_data = 0; 316 serr->addr_offset = (u8 *)&iph->daddr - skb_network_header(skb); 317 serr->port = port; 318 319 __skb_pull(skb, skb_tail_pointer(skb) - skb->data); 320 skb_reset_transport_header(skb); 321 322 if (sock_queue_err_skb(sk, skb)) 323 kfree_skb(skb); 324 } 325 326 /* 327 * Handle MSG_ERRQUEUE 328 */ 329 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len) 330 { 331 struct sock_exterr_skb *serr; 332 struct sk_buff *skb, *skb2; 333 struct sockaddr_in *sin; 334 struct { 335 struct sock_extended_err ee; 336 struct sockaddr_in offender; 337 } errhdr; 338 int err; 339 int copied; 340 341 err = -EAGAIN; 342 skb = skb_dequeue(&sk->sk_error_queue); 343 if (skb == NULL) 344 goto out; 345 346 copied = skb->len; 347 if (copied > len) { 348 msg->msg_flags |= MSG_TRUNC; 349 copied = len; 350 } 351 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied); 352 if (err) 353 goto out_free_skb; 354 355 sock_recv_timestamp(msg, sk, skb); 356 357 serr = SKB_EXT_ERR(skb); 358 359 sin = (struct sockaddr_in *)msg->msg_name; 360 if (sin) { 361 sin->sin_family = AF_INET; 362 sin->sin_addr.s_addr = *(__be32 *)(skb_network_header(skb) + 363 serr->addr_offset); 364 sin->sin_port = serr->port; 365 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero)); 366 } 367 368 memcpy(&errhdr.ee, &serr->ee, sizeof(struct sock_extended_err)); 369 sin = &errhdr.offender; 370 sin->sin_family = AF_UNSPEC; 371 if (serr->ee.ee_origin == SO_EE_ORIGIN_ICMP) { 372 struct inet_sock *inet = inet_sk(sk); 373 374 sin->sin_family = AF_INET; 375 sin->sin_addr.s_addr = ip_hdr(skb)->saddr; 376 sin->sin_port = 0; 377 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero)); 378 if (inet->cmsg_flags) 379 ip_cmsg_recv(msg, skb); 380 } 381 382 put_cmsg(msg, SOL_IP, IP_RECVERR, sizeof(errhdr), &errhdr); 383 384 /* Now we could try to dump offended packet options */ 385 386 msg->msg_flags |= MSG_ERRQUEUE; 387 err = copied; 388 389 /* Reset and regenerate socket error */ 390 spin_lock_bh(&sk->sk_error_queue.lock); 391 sk->sk_err = 0; 392 if ((skb2 = skb_peek(&sk->sk_error_queue)) != NULL) { 393 sk->sk_err = SKB_EXT_ERR(skb2)->ee.ee_errno; 394 spin_unlock_bh(&sk->sk_error_queue.lock); 395 sk->sk_error_report(sk); 396 } else 397 spin_unlock_bh(&sk->sk_error_queue.lock); 398 399 out_free_skb: 400 kfree_skb(skb); 401 out: 402 return err; 403 } 404 405 406 /* 407 * Socket option code for IP. This is the end of the line after any TCP,UDP etc options on 408 * an IP socket. 409 */ 410 411 static int do_ip_setsockopt(struct sock *sk, int level, 412 int optname, char __user *optval, int optlen) 413 { 414 struct inet_sock *inet = inet_sk(sk); 415 int val=0,err; 416 417 if (((1<<optname) & ((1<<IP_PKTINFO) | (1<<IP_RECVTTL) | 418 (1<<IP_RECVOPTS) | (1<<IP_RECVTOS) | 419 (1<<IP_RETOPTS) | (1<<IP_TOS) | 420 (1<<IP_TTL) | (1<<IP_HDRINCL) | 421 (1<<IP_MTU_DISCOVER) | (1<<IP_RECVERR) | 422 (1<<IP_ROUTER_ALERT) | (1<<IP_FREEBIND) | 423 (1<<IP_PASSSEC))) || 424 optname == IP_MULTICAST_TTL || 425 optname == IP_MULTICAST_LOOP) { 426 if (optlen >= sizeof(int)) { 427 if (get_user(val, (int __user *) optval)) 428 return -EFAULT; 429 } else if (optlen >= sizeof(char)) { 430 unsigned char ucval; 431 432 if (get_user(ucval, (unsigned char __user *) optval)) 433 return -EFAULT; 434 val = (int) ucval; 435 } 436 } 437 438 /* If optlen==0, it is equivalent to val == 0 */ 439 440 #ifdef CONFIG_IP_MROUTE 441 if (optname >= MRT_BASE && optname <= (MRT_BASE + 10)) 442 return ip_mroute_setsockopt(sk,optname,optval,optlen); 443 #endif 444 445 err = 0; 446 lock_sock(sk); 447 448 switch (optname) { 449 case IP_OPTIONS: 450 { 451 struct ip_options * opt = NULL; 452 if (optlen > 40 || optlen < 0) 453 goto e_inval; 454 err = ip_options_get_from_user(&opt, optval, optlen); 455 if (err) 456 break; 457 if (inet->is_icsk) { 458 struct inet_connection_sock *icsk = inet_csk(sk); 459 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) 460 if (sk->sk_family == PF_INET || 461 (!((1 << sk->sk_state) & 462 (TCPF_LISTEN | TCPF_CLOSE)) && 463 inet->daddr != LOOPBACK4_IPV6)) { 464 #endif 465 if (inet->opt) 466 icsk->icsk_ext_hdr_len -= inet->opt->optlen; 467 if (opt) 468 icsk->icsk_ext_hdr_len += opt->optlen; 469 icsk->icsk_sync_mss(sk, icsk->icsk_pmtu_cookie); 470 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) 471 } 472 #endif 473 } 474 opt = xchg(&inet->opt, opt); 475 kfree(opt); 476 break; 477 } 478 case IP_PKTINFO: 479 if (val) 480 inet->cmsg_flags |= IP_CMSG_PKTINFO; 481 else 482 inet->cmsg_flags &= ~IP_CMSG_PKTINFO; 483 break; 484 case IP_RECVTTL: 485 if (val) 486 inet->cmsg_flags |= IP_CMSG_TTL; 487 else 488 inet->cmsg_flags &= ~IP_CMSG_TTL; 489 break; 490 case IP_RECVTOS: 491 if (val) 492 inet->cmsg_flags |= IP_CMSG_TOS; 493 else 494 inet->cmsg_flags &= ~IP_CMSG_TOS; 495 break; 496 case IP_RECVOPTS: 497 if (val) 498 inet->cmsg_flags |= IP_CMSG_RECVOPTS; 499 else 500 inet->cmsg_flags &= ~IP_CMSG_RECVOPTS; 501 break; 502 case IP_RETOPTS: 503 if (val) 504 inet->cmsg_flags |= IP_CMSG_RETOPTS; 505 else 506 inet->cmsg_flags &= ~IP_CMSG_RETOPTS; 507 break; 508 case IP_PASSSEC: 509 if (val) 510 inet->cmsg_flags |= IP_CMSG_PASSSEC; 511 else 512 inet->cmsg_flags &= ~IP_CMSG_PASSSEC; 513 break; 514 case IP_TOS: /* This sets both TOS and Precedence */ 515 if (sk->sk_type == SOCK_STREAM) { 516 val &= ~3; 517 val |= inet->tos & 3; 518 } 519 if (IPTOS_PREC(val) >= IPTOS_PREC_CRITIC_ECP && 520 !capable(CAP_NET_ADMIN)) { 521 err = -EPERM; 522 break; 523 } 524 if (inet->tos != val) { 525 inet->tos = val; 526 sk->sk_priority = rt_tos2priority(val); 527 sk_dst_reset(sk); 528 } 529 break; 530 case IP_TTL: 531 if (optlen<1) 532 goto e_inval; 533 if (val != -1 && (val < 1 || val>255)) 534 goto e_inval; 535 inet->uc_ttl = val; 536 break; 537 case IP_HDRINCL: 538 if (sk->sk_type != SOCK_RAW) { 539 err = -ENOPROTOOPT; 540 break; 541 } 542 inet->hdrincl = val ? 1 : 0; 543 break; 544 case IP_MTU_DISCOVER: 545 if (val<0 || val>3) 546 goto e_inval; 547 inet->pmtudisc = val; 548 break; 549 case IP_RECVERR: 550 inet->recverr = !!val; 551 if (!val) 552 skb_queue_purge(&sk->sk_error_queue); 553 break; 554 case IP_MULTICAST_TTL: 555 if (sk->sk_type == SOCK_STREAM) 556 goto e_inval; 557 if (optlen<1) 558 goto e_inval; 559 if (val==-1) 560 val = 1; 561 if (val < 0 || val > 255) 562 goto e_inval; 563 inet->mc_ttl = val; 564 break; 565 case IP_MULTICAST_LOOP: 566 if (optlen<1) 567 goto e_inval; 568 inet->mc_loop = !!val; 569 break; 570 case IP_MULTICAST_IF: 571 { 572 struct ip_mreqn mreq; 573 struct net_device *dev = NULL; 574 575 if (sk->sk_type == SOCK_STREAM) 576 goto e_inval; 577 /* 578 * Check the arguments are allowable 579 */ 580 581 err = -EFAULT; 582 if (optlen >= sizeof(struct ip_mreqn)) { 583 if (copy_from_user(&mreq,optval,sizeof(mreq))) 584 break; 585 } else { 586 memset(&mreq, 0, sizeof(mreq)); 587 if (optlen >= sizeof(struct in_addr) && 588 copy_from_user(&mreq.imr_address,optval,sizeof(struct in_addr))) 589 break; 590 } 591 592 if (!mreq.imr_ifindex) { 593 if (mreq.imr_address.s_addr == INADDR_ANY) { 594 inet->mc_index = 0; 595 inet->mc_addr = 0; 596 err = 0; 597 break; 598 } 599 dev = ip_dev_find(mreq.imr_address.s_addr); 600 if (dev) { 601 mreq.imr_ifindex = dev->ifindex; 602 dev_put(dev); 603 } 604 } else 605 dev = __dev_get_by_index(&init_net, mreq.imr_ifindex); 606 607 608 err = -EADDRNOTAVAIL; 609 if (!dev) 610 break; 611 612 err = -EINVAL; 613 if (sk->sk_bound_dev_if && 614 mreq.imr_ifindex != sk->sk_bound_dev_if) 615 break; 616 617 inet->mc_index = mreq.imr_ifindex; 618 inet->mc_addr = mreq.imr_address.s_addr; 619 err = 0; 620 break; 621 } 622 623 case IP_ADD_MEMBERSHIP: 624 case IP_DROP_MEMBERSHIP: 625 { 626 struct ip_mreqn mreq; 627 628 err = -EPROTO; 629 if (inet_sk(sk)->is_icsk) 630 break; 631 632 if (optlen < sizeof(struct ip_mreq)) 633 goto e_inval; 634 err = -EFAULT; 635 if (optlen >= sizeof(struct ip_mreqn)) { 636 if (copy_from_user(&mreq,optval,sizeof(mreq))) 637 break; 638 } else { 639 memset(&mreq, 0, sizeof(mreq)); 640 if (copy_from_user(&mreq,optval,sizeof(struct ip_mreq))) 641 break; 642 } 643 644 if (optname == IP_ADD_MEMBERSHIP) 645 err = ip_mc_join_group(sk, &mreq); 646 else 647 err = ip_mc_leave_group(sk, &mreq); 648 break; 649 } 650 case IP_MSFILTER: 651 { 652 extern int sysctl_igmp_max_msf; 653 struct ip_msfilter *msf; 654 655 if (optlen < IP_MSFILTER_SIZE(0)) 656 goto e_inval; 657 if (optlen > sysctl_optmem_max) { 658 err = -ENOBUFS; 659 break; 660 } 661 msf = kmalloc(optlen, GFP_KERNEL); 662 if (!msf) { 663 err = -ENOBUFS; 664 break; 665 } 666 err = -EFAULT; 667 if (copy_from_user(msf, optval, optlen)) { 668 kfree(msf); 669 break; 670 } 671 /* numsrc >= (1G-4) overflow in 32 bits */ 672 if (msf->imsf_numsrc >= 0x3ffffffcU || 673 msf->imsf_numsrc > sysctl_igmp_max_msf) { 674 kfree(msf); 675 err = -ENOBUFS; 676 break; 677 } 678 if (IP_MSFILTER_SIZE(msf->imsf_numsrc) > optlen) { 679 kfree(msf); 680 err = -EINVAL; 681 break; 682 } 683 err = ip_mc_msfilter(sk, msf, 0); 684 kfree(msf); 685 break; 686 } 687 case IP_BLOCK_SOURCE: 688 case IP_UNBLOCK_SOURCE: 689 case IP_ADD_SOURCE_MEMBERSHIP: 690 case IP_DROP_SOURCE_MEMBERSHIP: 691 { 692 struct ip_mreq_source mreqs; 693 int omode, add; 694 695 if (optlen != sizeof(struct ip_mreq_source)) 696 goto e_inval; 697 if (copy_from_user(&mreqs, optval, sizeof(mreqs))) { 698 err = -EFAULT; 699 break; 700 } 701 if (optname == IP_BLOCK_SOURCE) { 702 omode = MCAST_EXCLUDE; 703 add = 1; 704 } else if (optname == IP_UNBLOCK_SOURCE) { 705 omode = MCAST_EXCLUDE; 706 add = 0; 707 } else if (optname == IP_ADD_SOURCE_MEMBERSHIP) { 708 struct ip_mreqn mreq; 709 710 mreq.imr_multiaddr.s_addr = mreqs.imr_multiaddr; 711 mreq.imr_address.s_addr = mreqs.imr_interface; 712 mreq.imr_ifindex = 0; 713 err = ip_mc_join_group(sk, &mreq); 714 if (err && err != -EADDRINUSE) 715 break; 716 omode = MCAST_INCLUDE; 717 add = 1; 718 } else /* IP_DROP_SOURCE_MEMBERSHIP */ { 719 omode = MCAST_INCLUDE; 720 add = 0; 721 } 722 err = ip_mc_source(add, omode, sk, &mreqs, 0); 723 break; 724 } 725 case MCAST_JOIN_GROUP: 726 case MCAST_LEAVE_GROUP: 727 { 728 struct group_req greq; 729 struct sockaddr_in *psin; 730 struct ip_mreqn mreq; 731 732 if (optlen < sizeof(struct group_req)) 733 goto e_inval; 734 err = -EFAULT; 735 if (copy_from_user(&greq, optval, sizeof(greq))) 736 break; 737 psin = (struct sockaddr_in *)&greq.gr_group; 738 if (psin->sin_family != AF_INET) 739 goto e_inval; 740 memset(&mreq, 0, sizeof(mreq)); 741 mreq.imr_multiaddr = psin->sin_addr; 742 mreq.imr_ifindex = greq.gr_interface; 743 744 if (optname == MCAST_JOIN_GROUP) 745 err = ip_mc_join_group(sk, &mreq); 746 else 747 err = ip_mc_leave_group(sk, &mreq); 748 break; 749 } 750 case MCAST_JOIN_SOURCE_GROUP: 751 case MCAST_LEAVE_SOURCE_GROUP: 752 case MCAST_BLOCK_SOURCE: 753 case MCAST_UNBLOCK_SOURCE: 754 { 755 struct group_source_req greqs; 756 struct ip_mreq_source mreqs; 757 struct sockaddr_in *psin; 758 int omode, add; 759 760 if (optlen != sizeof(struct group_source_req)) 761 goto e_inval; 762 if (copy_from_user(&greqs, optval, sizeof(greqs))) { 763 err = -EFAULT; 764 break; 765 } 766 if (greqs.gsr_group.ss_family != AF_INET || 767 greqs.gsr_source.ss_family != AF_INET) { 768 err = -EADDRNOTAVAIL; 769 break; 770 } 771 psin = (struct sockaddr_in *)&greqs.gsr_group; 772 mreqs.imr_multiaddr = psin->sin_addr.s_addr; 773 psin = (struct sockaddr_in *)&greqs.gsr_source; 774 mreqs.imr_sourceaddr = psin->sin_addr.s_addr; 775 mreqs.imr_interface = 0; /* use index for mc_source */ 776 777 if (optname == MCAST_BLOCK_SOURCE) { 778 omode = MCAST_EXCLUDE; 779 add = 1; 780 } else if (optname == MCAST_UNBLOCK_SOURCE) { 781 omode = MCAST_EXCLUDE; 782 add = 0; 783 } else if (optname == MCAST_JOIN_SOURCE_GROUP) { 784 struct ip_mreqn mreq; 785 786 psin = (struct sockaddr_in *)&greqs.gsr_group; 787 mreq.imr_multiaddr = psin->sin_addr; 788 mreq.imr_address.s_addr = 0; 789 mreq.imr_ifindex = greqs.gsr_interface; 790 err = ip_mc_join_group(sk, &mreq); 791 if (err && err != -EADDRINUSE) 792 break; 793 greqs.gsr_interface = mreq.imr_ifindex; 794 omode = MCAST_INCLUDE; 795 add = 1; 796 } else /* MCAST_LEAVE_SOURCE_GROUP */ { 797 omode = MCAST_INCLUDE; 798 add = 0; 799 } 800 err = ip_mc_source(add, omode, sk, &mreqs, 801 greqs.gsr_interface); 802 break; 803 } 804 case MCAST_MSFILTER: 805 { 806 extern int sysctl_igmp_max_msf; 807 struct sockaddr_in *psin; 808 struct ip_msfilter *msf = NULL; 809 struct group_filter *gsf = NULL; 810 int msize, i, ifindex; 811 812 if (optlen < GROUP_FILTER_SIZE(0)) 813 goto e_inval; 814 if (optlen > sysctl_optmem_max) { 815 err = -ENOBUFS; 816 break; 817 } 818 gsf = kmalloc(optlen,GFP_KERNEL); 819 if (!gsf) { 820 err = -ENOBUFS; 821 break; 822 } 823 err = -EFAULT; 824 if (copy_from_user(gsf, optval, optlen)) { 825 goto mc_msf_out; 826 } 827 /* numsrc >= (4G-140)/128 overflow in 32 bits */ 828 if (gsf->gf_numsrc >= 0x1ffffff || 829 gsf->gf_numsrc > sysctl_igmp_max_msf) { 830 err = -ENOBUFS; 831 goto mc_msf_out; 832 } 833 if (GROUP_FILTER_SIZE(gsf->gf_numsrc) > optlen) { 834 err = -EINVAL; 835 goto mc_msf_out; 836 } 837 msize = IP_MSFILTER_SIZE(gsf->gf_numsrc); 838 msf = kmalloc(msize,GFP_KERNEL); 839 if (!msf) { 840 err = -ENOBUFS; 841 goto mc_msf_out; 842 } 843 ifindex = gsf->gf_interface; 844 psin = (struct sockaddr_in *)&gsf->gf_group; 845 if (psin->sin_family != AF_INET) { 846 err = -EADDRNOTAVAIL; 847 goto mc_msf_out; 848 } 849 msf->imsf_multiaddr = psin->sin_addr.s_addr; 850 msf->imsf_interface = 0; 851 msf->imsf_fmode = gsf->gf_fmode; 852 msf->imsf_numsrc = gsf->gf_numsrc; 853 err = -EADDRNOTAVAIL; 854 for (i=0; i<gsf->gf_numsrc; ++i) { 855 psin = (struct sockaddr_in *)&gsf->gf_slist[i]; 856 857 if (psin->sin_family != AF_INET) 858 goto mc_msf_out; 859 msf->imsf_slist[i] = psin->sin_addr.s_addr; 860 } 861 kfree(gsf); 862 gsf = NULL; 863 864 err = ip_mc_msfilter(sk, msf, ifindex); 865 mc_msf_out: 866 kfree(msf); 867 kfree(gsf); 868 break; 869 } 870 case IP_ROUTER_ALERT: 871 err = ip_ra_control(sk, val ? 1 : 0, NULL); 872 break; 873 874 case IP_FREEBIND: 875 if (optlen<1) 876 goto e_inval; 877 inet->freebind = !!val; 878 break; 879 880 case IP_IPSEC_POLICY: 881 case IP_XFRM_POLICY: 882 err = -EPERM; 883 if (!capable(CAP_NET_ADMIN)) 884 break; 885 err = xfrm_user_policy(sk, optname, optval, optlen); 886 break; 887 888 default: 889 err = -ENOPROTOOPT; 890 break; 891 } 892 release_sock(sk); 893 return err; 894 895 e_inval: 896 release_sock(sk); 897 return -EINVAL; 898 } 899 900 int ip_setsockopt(struct sock *sk, int level, 901 int optname, char __user *optval, int optlen) 902 { 903 int err; 904 905 if (level != SOL_IP) 906 return -ENOPROTOOPT; 907 908 err = do_ip_setsockopt(sk, level, optname, optval, optlen); 909 #ifdef CONFIG_NETFILTER 910 /* we need to exclude all possible ENOPROTOOPTs except default case */ 911 if (err == -ENOPROTOOPT && optname != IP_HDRINCL && 912 optname != IP_IPSEC_POLICY && optname != IP_XFRM_POLICY 913 #ifdef CONFIG_IP_MROUTE 914 && (optname < MRT_BASE || optname > (MRT_BASE + 10)) 915 #endif 916 ) { 917 lock_sock(sk); 918 err = nf_setsockopt(sk, PF_INET, optname, optval, optlen); 919 release_sock(sk); 920 } 921 #endif 922 return err; 923 } 924 925 #ifdef CONFIG_COMPAT 926 int compat_ip_setsockopt(struct sock *sk, int level, int optname, 927 char __user *optval, int optlen) 928 { 929 int err; 930 931 if (level != SOL_IP) 932 return -ENOPROTOOPT; 933 934 err = do_ip_setsockopt(sk, level, optname, optval, optlen); 935 #ifdef CONFIG_NETFILTER 936 /* we need to exclude all possible ENOPROTOOPTs except default case */ 937 if (err == -ENOPROTOOPT && optname != IP_HDRINCL && 938 optname != IP_IPSEC_POLICY && optname != IP_XFRM_POLICY 939 #ifdef CONFIG_IP_MROUTE 940 && (optname < MRT_BASE || optname > (MRT_BASE + 10)) 941 #endif 942 ) { 943 lock_sock(sk); 944 err = compat_nf_setsockopt(sk, PF_INET, optname, 945 optval, optlen); 946 release_sock(sk); 947 } 948 #endif 949 return err; 950 } 951 952 EXPORT_SYMBOL(compat_ip_setsockopt); 953 #endif 954 955 /* 956 * Get the options. Note for future reference. The GET of IP options gets the 957 * _received_ ones. The set sets the _sent_ ones. 958 */ 959 960 static int do_ip_getsockopt(struct sock *sk, int level, int optname, 961 char __user *optval, int __user *optlen) 962 { 963 struct inet_sock *inet = inet_sk(sk); 964 int val; 965 int len; 966 967 if (level != SOL_IP) 968 return -EOPNOTSUPP; 969 970 #ifdef CONFIG_IP_MROUTE 971 if (optname >= MRT_BASE && optname <= MRT_BASE+10) { 972 return ip_mroute_getsockopt(sk,optname,optval,optlen); 973 } 974 #endif 975 976 if (get_user(len,optlen)) 977 return -EFAULT; 978 if (len < 0) 979 return -EINVAL; 980 981 lock_sock(sk); 982 983 switch (optname) { 984 case IP_OPTIONS: 985 { 986 unsigned char optbuf[sizeof(struct ip_options)+40]; 987 struct ip_options * opt = (struct ip_options*)optbuf; 988 opt->optlen = 0; 989 if (inet->opt) 990 memcpy(optbuf, inet->opt, 991 sizeof(struct ip_options)+ 992 inet->opt->optlen); 993 release_sock(sk); 994 995 if (opt->optlen == 0) 996 return put_user(0, optlen); 997 998 ip_options_undo(opt); 999 1000 len = min_t(unsigned int, len, opt->optlen); 1001 if (put_user(len, optlen)) 1002 return -EFAULT; 1003 if (copy_to_user(optval, opt->__data, len)) 1004 return -EFAULT; 1005 return 0; 1006 } 1007 case IP_PKTINFO: 1008 val = (inet->cmsg_flags & IP_CMSG_PKTINFO) != 0; 1009 break; 1010 case IP_RECVTTL: 1011 val = (inet->cmsg_flags & IP_CMSG_TTL) != 0; 1012 break; 1013 case IP_RECVTOS: 1014 val = (inet->cmsg_flags & IP_CMSG_TOS) != 0; 1015 break; 1016 case IP_RECVOPTS: 1017 val = (inet->cmsg_flags & IP_CMSG_RECVOPTS) != 0; 1018 break; 1019 case IP_RETOPTS: 1020 val = (inet->cmsg_flags & IP_CMSG_RETOPTS) != 0; 1021 break; 1022 case IP_PASSSEC: 1023 val = (inet->cmsg_flags & IP_CMSG_PASSSEC) != 0; 1024 break; 1025 case IP_TOS: 1026 val = inet->tos; 1027 break; 1028 case IP_TTL: 1029 val = (inet->uc_ttl == -1 ? 1030 sysctl_ip_default_ttl : 1031 inet->uc_ttl); 1032 break; 1033 case IP_HDRINCL: 1034 val = inet->hdrincl; 1035 break; 1036 case IP_MTU_DISCOVER: 1037 val = inet->pmtudisc; 1038 break; 1039 case IP_MTU: 1040 { 1041 struct dst_entry *dst; 1042 val = 0; 1043 dst = sk_dst_get(sk); 1044 if (dst) { 1045 val = dst_mtu(dst); 1046 dst_release(dst); 1047 } 1048 if (!val) { 1049 release_sock(sk); 1050 return -ENOTCONN; 1051 } 1052 break; 1053 } 1054 case IP_RECVERR: 1055 val = inet->recverr; 1056 break; 1057 case IP_MULTICAST_TTL: 1058 val = inet->mc_ttl; 1059 break; 1060 case IP_MULTICAST_LOOP: 1061 val = inet->mc_loop; 1062 break; 1063 case IP_MULTICAST_IF: 1064 { 1065 struct in_addr addr; 1066 len = min_t(unsigned int, len, sizeof(struct in_addr)); 1067 addr.s_addr = inet->mc_addr; 1068 release_sock(sk); 1069 1070 if (put_user(len, optlen)) 1071 return -EFAULT; 1072 if (copy_to_user(optval, &addr, len)) 1073 return -EFAULT; 1074 return 0; 1075 } 1076 case IP_MSFILTER: 1077 { 1078 struct ip_msfilter msf; 1079 int err; 1080 1081 if (len < IP_MSFILTER_SIZE(0)) { 1082 release_sock(sk); 1083 return -EINVAL; 1084 } 1085 if (copy_from_user(&msf, optval, IP_MSFILTER_SIZE(0))) { 1086 release_sock(sk); 1087 return -EFAULT; 1088 } 1089 err = ip_mc_msfget(sk, &msf, 1090 (struct ip_msfilter __user *)optval, optlen); 1091 release_sock(sk); 1092 return err; 1093 } 1094 case MCAST_MSFILTER: 1095 { 1096 struct group_filter gsf; 1097 int err; 1098 1099 if (len < GROUP_FILTER_SIZE(0)) { 1100 release_sock(sk); 1101 return -EINVAL; 1102 } 1103 if (copy_from_user(&gsf, optval, GROUP_FILTER_SIZE(0))) { 1104 release_sock(sk); 1105 return -EFAULT; 1106 } 1107 err = ip_mc_gsfget(sk, &gsf, 1108 (struct group_filter __user *)optval, optlen); 1109 release_sock(sk); 1110 return err; 1111 } 1112 case IP_PKTOPTIONS: 1113 { 1114 struct msghdr msg; 1115 1116 release_sock(sk); 1117 1118 if (sk->sk_type != SOCK_STREAM) 1119 return -ENOPROTOOPT; 1120 1121 msg.msg_control = optval; 1122 msg.msg_controllen = len; 1123 msg.msg_flags = 0; 1124 1125 if (inet->cmsg_flags & IP_CMSG_PKTINFO) { 1126 struct in_pktinfo info; 1127 1128 info.ipi_addr.s_addr = inet->rcv_saddr; 1129 info.ipi_spec_dst.s_addr = inet->rcv_saddr; 1130 info.ipi_ifindex = inet->mc_index; 1131 put_cmsg(&msg, SOL_IP, IP_PKTINFO, sizeof(info), &info); 1132 } 1133 if (inet->cmsg_flags & IP_CMSG_TTL) { 1134 int hlim = inet->mc_ttl; 1135 put_cmsg(&msg, SOL_IP, IP_TTL, sizeof(hlim), &hlim); 1136 } 1137 len -= msg.msg_controllen; 1138 return put_user(len, optlen); 1139 } 1140 case IP_FREEBIND: 1141 val = inet->freebind; 1142 break; 1143 default: 1144 release_sock(sk); 1145 return -ENOPROTOOPT; 1146 } 1147 release_sock(sk); 1148 1149 if (len < sizeof(int) && len > 0 && val>=0 && val<255) { 1150 unsigned char ucval = (unsigned char)val; 1151 len = 1; 1152 if (put_user(len, optlen)) 1153 return -EFAULT; 1154 if (copy_to_user(optval,&ucval,1)) 1155 return -EFAULT; 1156 } else { 1157 len = min_t(unsigned int, sizeof(int), len); 1158 if (put_user(len, optlen)) 1159 return -EFAULT; 1160 if (copy_to_user(optval,&val,len)) 1161 return -EFAULT; 1162 } 1163 return 0; 1164 } 1165 1166 int ip_getsockopt(struct sock *sk, int level, 1167 int optname, char __user *optval, int __user *optlen) 1168 { 1169 int err; 1170 1171 err = do_ip_getsockopt(sk, level, optname, optval, optlen); 1172 #ifdef CONFIG_NETFILTER 1173 /* we need to exclude all possible ENOPROTOOPTs except default case */ 1174 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS 1175 #ifdef CONFIG_IP_MROUTE 1176 && (optname < MRT_BASE || optname > MRT_BASE+10) 1177 #endif 1178 ) { 1179 int len; 1180 1181 if (get_user(len,optlen)) 1182 return -EFAULT; 1183 1184 lock_sock(sk); 1185 err = nf_getsockopt(sk, PF_INET, optname, optval, 1186 &len); 1187 release_sock(sk); 1188 if (err >= 0) 1189 err = put_user(len, optlen); 1190 return err; 1191 } 1192 #endif 1193 return err; 1194 } 1195 1196 #ifdef CONFIG_COMPAT 1197 int compat_ip_getsockopt(struct sock *sk, int level, int optname, 1198 char __user *optval, int __user *optlen) 1199 { 1200 int err = do_ip_getsockopt(sk, level, optname, optval, optlen); 1201 #ifdef CONFIG_NETFILTER 1202 /* we need to exclude all possible ENOPROTOOPTs except default case */ 1203 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS 1204 #ifdef CONFIG_IP_MROUTE 1205 && (optname < MRT_BASE || optname > MRT_BASE+10) 1206 #endif 1207 ) { 1208 int len; 1209 1210 if (get_user(len, optlen)) 1211 return -EFAULT; 1212 1213 lock_sock(sk); 1214 err = compat_nf_getsockopt(sk, PF_INET, optname, optval, &len); 1215 release_sock(sk); 1216 if (err >= 0) 1217 err = put_user(len, optlen); 1218 return err; 1219 } 1220 #endif 1221 return err; 1222 } 1223 1224 EXPORT_SYMBOL(compat_ip_getsockopt); 1225 #endif 1226 1227 EXPORT_SYMBOL(ip_cmsg_recv); 1228 1229 EXPORT_SYMBOL(ip_getsockopt); 1230 EXPORT_SYMBOL(ip_setsockopt); 1231