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(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 if (optlen < sizeof(struct ip_mreq)) 629 goto e_inval; 630 err = -EFAULT; 631 if (optlen >= sizeof(struct ip_mreqn)) { 632 if (copy_from_user(&mreq,optval,sizeof(mreq))) 633 break; 634 } else { 635 memset(&mreq, 0, sizeof(mreq)); 636 if (copy_from_user(&mreq,optval,sizeof(struct ip_mreq))) 637 break; 638 } 639 640 if (optname == IP_ADD_MEMBERSHIP) 641 err = ip_mc_join_group(sk, &mreq); 642 else 643 err = ip_mc_leave_group(sk, &mreq); 644 break; 645 } 646 case IP_MSFILTER: 647 { 648 extern int sysctl_igmp_max_msf; 649 struct ip_msfilter *msf; 650 651 if (optlen < IP_MSFILTER_SIZE(0)) 652 goto e_inval; 653 if (optlen > sysctl_optmem_max) { 654 err = -ENOBUFS; 655 break; 656 } 657 msf = kmalloc(optlen, GFP_KERNEL); 658 if (msf == 0) { 659 err = -ENOBUFS; 660 break; 661 } 662 err = -EFAULT; 663 if (copy_from_user(msf, optval, optlen)) { 664 kfree(msf); 665 break; 666 } 667 /* numsrc >= (1G-4) overflow in 32 bits */ 668 if (msf->imsf_numsrc >= 0x3ffffffcU || 669 msf->imsf_numsrc > sysctl_igmp_max_msf) { 670 kfree(msf); 671 err = -ENOBUFS; 672 break; 673 } 674 if (IP_MSFILTER_SIZE(msf->imsf_numsrc) > optlen) { 675 kfree(msf); 676 err = -EINVAL; 677 break; 678 } 679 err = ip_mc_msfilter(sk, msf, 0); 680 kfree(msf); 681 break; 682 } 683 case IP_BLOCK_SOURCE: 684 case IP_UNBLOCK_SOURCE: 685 case IP_ADD_SOURCE_MEMBERSHIP: 686 case IP_DROP_SOURCE_MEMBERSHIP: 687 { 688 struct ip_mreq_source mreqs; 689 int omode, add; 690 691 if (optlen != sizeof(struct ip_mreq_source)) 692 goto e_inval; 693 if (copy_from_user(&mreqs, optval, sizeof(mreqs))) { 694 err = -EFAULT; 695 break; 696 } 697 if (optname == IP_BLOCK_SOURCE) { 698 omode = MCAST_EXCLUDE; 699 add = 1; 700 } else if (optname == IP_UNBLOCK_SOURCE) { 701 omode = MCAST_EXCLUDE; 702 add = 0; 703 } else if (optname == IP_ADD_SOURCE_MEMBERSHIP) { 704 struct ip_mreqn mreq; 705 706 mreq.imr_multiaddr.s_addr = mreqs.imr_multiaddr; 707 mreq.imr_address.s_addr = mreqs.imr_interface; 708 mreq.imr_ifindex = 0; 709 err = ip_mc_join_group(sk, &mreq); 710 if (err && err != -EADDRINUSE) 711 break; 712 omode = MCAST_INCLUDE; 713 add = 1; 714 } else /* IP_DROP_SOURCE_MEMBERSHIP */ { 715 omode = MCAST_INCLUDE; 716 add = 0; 717 } 718 err = ip_mc_source(add, omode, sk, &mreqs, 0); 719 break; 720 } 721 case MCAST_JOIN_GROUP: 722 case MCAST_LEAVE_GROUP: 723 { 724 struct group_req greq; 725 struct sockaddr_in *psin; 726 struct ip_mreqn mreq; 727 728 if (optlen < sizeof(struct group_req)) 729 goto e_inval; 730 err = -EFAULT; 731 if (copy_from_user(&greq, optval, sizeof(greq))) 732 break; 733 psin = (struct sockaddr_in *)&greq.gr_group; 734 if (psin->sin_family != AF_INET) 735 goto e_inval; 736 memset(&mreq, 0, sizeof(mreq)); 737 mreq.imr_multiaddr = psin->sin_addr; 738 mreq.imr_ifindex = greq.gr_interface; 739 740 if (optname == MCAST_JOIN_GROUP) 741 err = ip_mc_join_group(sk, &mreq); 742 else 743 err = ip_mc_leave_group(sk, &mreq); 744 break; 745 } 746 case MCAST_JOIN_SOURCE_GROUP: 747 case MCAST_LEAVE_SOURCE_GROUP: 748 case MCAST_BLOCK_SOURCE: 749 case MCAST_UNBLOCK_SOURCE: 750 { 751 struct group_source_req greqs; 752 struct ip_mreq_source mreqs; 753 struct sockaddr_in *psin; 754 int omode, add; 755 756 if (optlen != sizeof(struct group_source_req)) 757 goto e_inval; 758 if (copy_from_user(&greqs, optval, sizeof(greqs))) { 759 err = -EFAULT; 760 break; 761 } 762 if (greqs.gsr_group.ss_family != AF_INET || 763 greqs.gsr_source.ss_family != AF_INET) { 764 err = -EADDRNOTAVAIL; 765 break; 766 } 767 psin = (struct sockaddr_in *)&greqs.gsr_group; 768 mreqs.imr_multiaddr = psin->sin_addr.s_addr; 769 psin = (struct sockaddr_in *)&greqs.gsr_source; 770 mreqs.imr_sourceaddr = psin->sin_addr.s_addr; 771 mreqs.imr_interface = 0; /* use index for mc_source */ 772 773 if (optname == MCAST_BLOCK_SOURCE) { 774 omode = MCAST_EXCLUDE; 775 add = 1; 776 } else if (optname == MCAST_UNBLOCK_SOURCE) { 777 omode = MCAST_EXCLUDE; 778 add = 0; 779 } else if (optname == MCAST_JOIN_SOURCE_GROUP) { 780 struct ip_mreqn mreq; 781 782 psin = (struct sockaddr_in *)&greqs.gsr_group; 783 mreq.imr_multiaddr = psin->sin_addr; 784 mreq.imr_address.s_addr = 0; 785 mreq.imr_ifindex = greqs.gsr_interface; 786 err = ip_mc_join_group(sk, &mreq); 787 if (err && err != -EADDRINUSE) 788 break; 789 greqs.gsr_interface = mreq.imr_ifindex; 790 omode = MCAST_INCLUDE; 791 add = 1; 792 } else /* MCAST_LEAVE_SOURCE_GROUP */ { 793 omode = MCAST_INCLUDE; 794 add = 0; 795 } 796 err = ip_mc_source(add, omode, sk, &mreqs, 797 greqs.gsr_interface); 798 break; 799 } 800 case MCAST_MSFILTER: 801 { 802 extern int sysctl_igmp_max_msf; 803 struct sockaddr_in *psin; 804 struct ip_msfilter *msf = NULL; 805 struct group_filter *gsf = NULL; 806 int msize, i, ifindex; 807 808 if (optlen < GROUP_FILTER_SIZE(0)) 809 goto e_inval; 810 if (optlen > sysctl_optmem_max) { 811 err = -ENOBUFS; 812 break; 813 } 814 gsf = kmalloc(optlen,GFP_KERNEL); 815 if (gsf == 0) { 816 err = -ENOBUFS; 817 break; 818 } 819 err = -EFAULT; 820 if (copy_from_user(gsf, optval, optlen)) { 821 goto mc_msf_out; 822 } 823 /* numsrc >= (4G-140)/128 overflow in 32 bits */ 824 if (gsf->gf_numsrc >= 0x1ffffff || 825 gsf->gf_numsrc > sysctl_igmp_max_msf) { 826 err = -ENOBUFS; 827 goto mc_msf_out; 828 } 829 if (GROUP_FILTER_SIZE(gsf->gf_numsrc) > optlen) { 830 err = -EINVAL; 831 goto mc_msf_out; 832 } 833 msize = IP_MSFILTER_SIZE(gsf->gf_numsrc); 834 msf = kmalloc(msize,GFP_KERNEL); 835 if (msf == 0) { 836 err = -ENOBUFS; 837 goto mc_msf_out; 838 } 839 ifindex = gsf->gf_interface; 840 psin = (struct sockaddr_in *)&gsf->gf_group; 841 if (psin->sin_family != AF_INET) { 842 err = -EADDRNOTAVAIL; 843 goto mc_msf_out; 844 } 845 msf->imsf_multiaddr = psin->sin_addr.s_addr; 846 msf->imsf_interface = 0; 847 msf->imsf_fmode = gsf->gf_fmode; 848 msf->imsf_numsrc = gsf->gf_numsrc; 849 err = -EADDRNOTAVAIL; 850 for (i=0; i<gsf->gf_numsrc; ++i) { 851 psin = (struct sockaddr_in *)&gsf->gf_slist[i]; 852 853 if (psin->sin_family != AF_INET) 854 goto mc_msf_out; 855 msf->imsf_slist[i] = psin->sin_addr.s_addr; 856 } 857 kfree(gsf); 858 gsf = NULL; 859 860 err = ip_mc_msfilter(sk, msf, ifindex); 861 mc_msf_out: 862 kfree(msf); 863 kfree(gsf); 864 break; 865 } 866 case IP_ROUTER_ALERT: 867 err = ip_ra_control(sk, val ? 1 : 0, NULL); 868 break; 869 870 case IP_FREEBIND: 871 if (optlen<1) 872 goto e_inval; 873 inet->freebind = !!val; 874 break; 875 876 case IP_IPSEC_POLICY: 877 case IP_XFRM_POLICY: 878 err = -EPERM; 879 if (!capable(CAP_NET_ADMIN)) 880 break; 881 err = xfrm_user_policy(sk, optname, optval, optlen); 882 break; 883 884 default: 885 err = -ENOPROTOOPT; 886 break; 887 } 888 release_sock(sk); 889 return err; 890 891 e_inval: 892 release_sock(sk); 893 return -EINVAL; 894 } 895 896 int ip_setsockopt(struct sock *sk, int level, 897 int optname, char __user *optval, int optlen) 898 { 899 int err; 900 901 if (level != SOL_IP) 902 return -ENOPROTOOPT; 903 904 err = do_ip_setsockopt(sk, level, optname, optval, optlen); 905 #ifdef CONFIG_NETFILTER 906 /* we need to exclude all possible ENOPROTOOPTs except default case */ 907 if (err == -ENOPROTOOPT && optname != IP_HDRINCL && 908 optname != IP_IPSEC_POLICY && optname != IP_XFRM_POLICY 909 #ifdef CONFIG_IP_MROUTE 910 && (optname < MRT_BASE || optname > (MRT_BASE + 10)) 911 #endif 912 ) { 913 lock_sock(sk); 914 err = nf_setsockopt(sk, PF_INET, optname, optval, optlen); 915 release_sock(sk); 916 } 917 #endif 918 return err; 919 } 920 921 #ifdef CONFIG_COMPAT 922 int compat_ip_setsockopt(struct sock *sk, int level, int optname, 923 char __user *optval, int optlen) 924 { 925 int err; 926 927 if (level != SOL_IP) 928 return -ENOPROTOOPT; 929 930 err = do_ip_setsockopt(sk, level, optname, optval, optlen); 931 #ifdef CONFIG_NETFILTER 932 /* we need to exclude all possible ENOPROTOOPTs except default case */ 933 if (err == -ENOPROTOOPT && optname != IP_HDRINCL && 934 optname != IP_IPSEC_POLICY && optname != IP_XFRM_POLICY 935 #ifdef CONFIG_IP_MROUTE 936 && (optname < MRT_BASE || optname > (MRT_BASE + 10)) 937 #endif 938 ) { 939 lock_sock(sk); 940 err = compat_nf_setsockopt(sk, PF_INET, optname, 941 optval, optlen); 942 release_sock(sk); 943 } 944 #endif 945 return err; 946 } 947 948 EXPORT_SYMBOL(compat_ip_setsockopt); 949 #endif 950 951 /* 952 * Get the options. Note for future reference. The GET of IP options gets the 953 * _received_ ones. The set sets the _sent_ ones. 954 */ 955 956 static int do_ip_getsockopt(struct sock *sk, int level, int optname, 957 char __user *optval, int __user *optlen) 958 { 959 struct inet_sock *inet = inet_sk(sk); 960 int val; 961 int len; 962 963 if (level != SOL_IP) 964 return -EOPNOTSUPP; 965 966 #ifdef CONFIG_IP_MROUTE 967 if (optname >= MRT_BASE && optname <= MRT_BASE+10) { 968 return ip_mroute_getsockopt(sk,optname,optval,optlen); 969 } 970 #endif 971 972 if (get_user(len,optlen)) 973 return -EFAULT; 974 if (len < 0) 975 return -EINVAL; 976 977 lock_sock(sk); 978 979 switch (optname) { 980 case IP_OPTIONS: 981 { 982 unsigned char optbuf[sizeof(struct ip_options)+40]; 983 struct ip_options * opt = (struct ip_options*)optbuf; 984 opt->optlen = 0; 985 if (inet->opt) 986 memcpy(optbuf, inet->opt, 987 sizeof(struct ip_options)+ 988 inet->opt->optlen); 989 release_sock(sk); 990 991 if (opt->optlen == 0) 992 return put_user(0, optlen); 993 994 ip_options_undo(opt); 995 996 len = min_t(unsigned int, len, opt->optlen); 997 if (put_user(len, optlen)) 998 return -EFAULT; 999 if (copy_to_user(optval, opt->__data, len)) 1000 return -EFAULT; 1001 return 0; 1002 } 1003 case IP_PKTINFO: 1004 val = (inet->cmsg_flags & IP_CMSG_PKTINFO) != 0; 1005 break; 1006 case IP_RECVTTL: 1007 val = (inet->cmsg_flags & IP_CMSG_TTL) != 0; 1008 break; 1009 case IP_RECVTOS: 1010 val = (inet->cmsg_flags & IP_CMSG_TOS) != 0; 1011 break; 1012 case IP_RECVOPTS: 1013 val = (inet->cmsg_flags & IP_CMSG_RECVOPTS) != 0; 1014 break; 1015 case IP_RETOPTS: 1016 val = (inet->cmsg_flags & IP_CMSG_RETOPTS) != 0; 1017 break; 1018 case IP_PASSSEC: 1019 val = (inet->cmsg_flags & IP_CMSG_PASSSEC) != 0; 1020 break; 1021 case IP_TOS: 1022 val = inet->tos; 1023 break; 1024 case IP_TTL: 1025 val = (inet->uc_ttl == -1 ? 1026 sysctl_ip_default_ttl : 1027 inet->uc_ttl); 1028 break; 1029 case IP_HDRINCL: 1030 val = inet->hdrincl; 1031 break; 1032 case IP_MTU_DISCOVER: 1033 val = inet->pmtudisc; 1034 break; 1035 case IP_MTU: 1036 { 1037 struct dst_entry *dst; 1038 val = 0; 1039 dst = sk_dst_get(sk); 1040 if (dst) { 1041 val = dst_mtu(dst); 1042 dst_release(dst); 1043 } 1044 if (!val) { 1045 release_sock(sk); 1046 return -ENOTCONN; 1047 } 1048 break; 1049 } 1050 case IP_RECVERR: 1051 val = inet->recverr; 1052 break; 1053 case IP_MULTICAST_TTL: 1054 val = inet->mc_ttl; 1055 break; 1056 case IP_MULTICAST_LOOP: 1057 val = inet->mc_loop; 1058 break; 1059 case IP_MULTICAST_IF: 1060 { 1061 struct in_addr addr; 1062 len = min_t(unsigned int, len, sizeof(struct in_addr)); 1063 addr.s_addr = inet->mc_addr; 1064 release_sock(sk); 1065 1066 if (put_user(len, optlen)) 1067 return -EFAULT; 1068 if (copy_to_user(optval, &addr, len)) 1069 return -EFAULT; 1070 return 0; 1071 } 1072 case IP_MSFILTER: 1073 { 1074 struct ip_msfilter msf; 1075 int err; 1076 1077 if (len < IP_MSFILTER_SIZE(0)) { 1078 release_sock(sk); 1079 return -EINVAL; 1080 } 1081 if (copy_from_user(&msf, optval, IP_MSFILTER_SIZE(0))) { 1082 release_sock(sk); 1083 return -EFAULT; 1084 } 1085 err = ip_mc_msfget(sk, &msf, 1086 (struct ip_msfilter __user *)optval, optlen); 1087 release_sock(sk); 1088 return err; 1089 } 1090 case MCAST_MSFILTER: 1091 { 1092 struct group_filter gsf; 1093 int err; 1094 1095 if (len < GROUP_FILTER_SIZE(0)) { 1096 release_sock(sk); 1097 return -EINVAL; 1098 } 1099 if (copy_from_user(&gsf, optval, GROUP_FILTER_SIZE(0))) { 1100 release_sock(sk); 1101 return -EFAULT; 1102 } 1103 err = ip_mc_gsfget(sk, &gsf, 1104 (struct group_filter __user *)optval, optlen); 1105 release_sock(sk); 1106 return err; 1107 } 1108 case IP_PKTOPTIONS: 1109 { 1110 struct msghdr msg; 1111 1112 release_sock(sk); 1113 1114 if (sk->sk_type != SOCK_STREAM) 1115 return -ENOPROTOOPT; 1116 1117 msg.msg_control = optval; 1118 msg.msg_controllen = len; 1119 msg.msg_flags = 0; 1120 1121 if (inet->cmsg_flags & IP_CMSG_PKTINFO) { 1122 struct in_pktinfo info; 1123 1124 info.ipi_addr.s_addr = inet->rcv_saddr; 1125 info.ipi_spec_dst.s_addr = inet->rcv_saddr; 1126 info.ipi_ifindex = inet->mc_index; 1127 put_cmsg(&msg, SOL_IP, IP_PKTINFO, sizeof(info), &info); 1128 } 1129 if (inet->cmsg_flags & IP_CMSG_TTL) { 1130 int hlim = inet->mc_ttl; 1131 put_cmsg(&msg, SOL_IP, IP_TTL, sizeof(hlim), &hlim); 1132 } 1133 len -= msg.msg_controllen; 1134 return put_user(len, optlen); 1135 } 1136 case IP_FREEBIND: 1137 val = inet->freebind; 1138 break; 1139 default: 1140 release_sock(sk); 1141 return -ENOPROTOOPT; 1142 } 1143 release_sock(sk); 1144 1145 if (len < sizeof(int) && len > 0 && val>=0 && val<255) { 1146 unsigned char ucval = (unsigned char)val; 1147 len = 1; 1148 if (put_user(len, optlen)) 1149 return -EFAULT; 1150 if (copy_to_user(optval,&ucval,1)) 1151 return -EFAULT; 1152 } else { 1153 len = min_t(unsigned int, sizeof(int), len); 1154 if (put_user(len, optlen)) 1155 return -EFAULT; 1156 if (copy_to_user(optval,&val,len)) 1157 return -EFAULT; 1158 } 1159 return 0; 1160 } 1161 1162 int ip_getsockopt(struct sock *sk, int level, 1163 int optname, char __user *optval, int __user *optlen) 1164 { 1165 int err; 1166 1167 err = do_ip_getsockopt(sk, level, optname, optval, optlen); 1168 #ifdef CONFIG_NETFILTER 1169 /* we need to exclude all possible ENOPROTOOPTs except default case */ 1170 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS 1171 #ifdef CONFIG_IP_MROUTE 1172 && (optname < MRT_BASE || optname > MRT_BASE+10) 1173 #endif 1174 ) { 1175 int len; 1176 1177 if (get_user(len,optlen)) 1178 return -EFAULT; 1179 1180 lock_sock(sk); 1181 err = nf_getsockopt(sk, PF_INET, optname, optval, 1182 &len); 1183 release_sock(sk); 1184 if (err >= 0) 1185 err = put_user(len, optlen); 1186 return err; 1187 } 1188 #endif 1189 return err; 1190 } 1191 1192 #ifdef CONFIG_COMPAT 1193 int compat_ip_getsockopt(struct sock *sk, int level, int optname, 1194 char __user *optval, int __user *optlen) 1195 { 1196 int err = do_ip_getsockopt(sk, level, optname, optval, optlen); 1197 #ifdef CONFIG_NETFILTER 1198 /* we need to exclude all possible ENOPROTOOPTs except default case */ 1199 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS 1200 #ifdef CONFIG_IP_MROUTE 1201 && (optname < MRT_BASE || optname > MRT_BASE+10) 1202 #endif 1203 ) { 1204 int len; 1205 1206 if (get_user(len, optlen)) 1207 return -EFAULT; 1208 1209 lock_sock(sk); 1210 err = compat_nf_getsockopt(sk, PF_INET, optname, optval, &len); 1211 release_sock(sk); 1212 if (err >= 0) 1213 err = put_user(len, optlen); 1214 return err; 1215 } 1216 #endif 1217 return err; 1218 } 1219 1220 EXPORT_SYMBOL(compat_ip_getsockopt); 1221 #endif 1222 1223 EXPORT_SYMBOL(ip_cmsg_recv); 1224 1225 EXPORT_SYMBOL(ip_getsockopt); 1226 EXPORT_SYMBOL(ip_setsockopt); 1227