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/sched.h> 24 #include <linux/skbuff.h> 25 #include <linux/ip.h> 26 #include <linux/icmp.h> 27 #include <linux/inetdevice.h> 28 #include <linux/netdevice.h> 29 #include <net/sock.h> 30 #include <net/ip.h> 31 #include <net/icmp.h> 32 #include <net/tcp_states.h> 33 #include <linux/udp.h> 34 #include <linux/igmp.h> 35 #include <linux/netfilter.h> 36 #include <linux/route.h> 37 #include <linux/mroute.h> 38 #include <net/route.h> 39 #include <net/xfrm.h> 40 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) 41 #include <net/transp_v6.h> 42 #endif 43 44 #include <linux/errqueue.h> 45 #include <asm/uaccess.h> 46 47 #define IP_CMSG_PKTINFO 1 48 #define IP_CMSG_TTL 2 49 #define IP_CMSG_TOS 4 50 #define IP_CMSG_RECVOPTS 8 51 #define IP_CMSG_RETOPTS 16 52 #define IP_CMSG_PASSSEC 32 53 54 /* 55 * SOL_IP control messages. 56 */ 57 58 static void ip_cmsg_recv_pktinfo(struct msghdr *msg, struct sk_buff *skb) 59 { 60 struct in_pktinfo info; 61 struct rtable *rt = (struct rtable *)skb->dst; 62 63 info.ipi_addr.s_addr = skb->nh.iph->daddr; 64 if (rt) { 65 info.ipi_ifindex = rt->rt_iif; 66 info.ipi_spec_dst.s_addr = rt->rt_spec_dst; 67 } else { 68 info.ipi_ifindex = 0; 69 info.ipi_spec_dst.s_addr = 0; 70 } 71 72 put_cmsg(msg, SOL_IP, IP_PKTINFO, sizeof(info), &info); 73 } 74 75 static void ip_cmsg_recv_ttl(struct msghdr *msg, struct sk_buff *skb) 76 { 77 int ttl = skb->nh.iph->ttl; 78 put_cmsg(msg, SOL_IP, IP_TTL, sizeof(int), &ttl); 79 } 80 81 static void ip_cmsg_recv_tos(struct msghdr *msg, struct sk_buff *skb) 82 { 83 put_cmsg(msg, SOL_IP, IP_TOS, 1, &skb->nh.iph->tos); 84 } 85 86 static void ip_cmsg_recv_opts(struct msghdr *msg, struct sk_buff *skb) 87 { 88 if (IPCB(skb)->opt.optlen == 0) 89 return; 90 91 put_cmsg(msg, SOL_IP, IP_RECVOPTS, IPCB(skb)->opt.optlen, skb->nh.iph+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 u16 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 = skb->h.icmph->type; 273 serr->ee.ee_code = skb->h.icmph->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*)(skb->h.icmph+1))->daddr) - skb->nh.raw; 278 serr->port = port; 279 280 skb->h.raw = payload; 281 if (!skb_pull(skb, payload - skb->data) || 282 sock_queue_err_skb(sk, skb)) 283 kfree_skb(skb); 284 } 285 286 void ip_local_error(struct sock *sk, int err, u32 daddr, u16 port, u32 info) 287 { 288 struct inet_sock *inet = inet_sk(sk); 289 struct sock_exterr_skb *serr; 290 struct iphdr *iph; 291 struct sk_buff *skb; 292 293 if (!inet->recverr) 294 return; 295 296 skb = alloc_skb(sizeof(struct iphdr), GFP_ATOMIC); 297 if (!skb) 298 return; 299 300 iph = (struct iphdr*)skb_put(skb, sizeof(struct iphdr)); 301 skb->nh.iph = iph; 302 iph->daddr = daddr; 303 304 serr = SKB_EXT_ERR(skb); 305 serr->ee.ee_errno = err; 306 serr->ee.ee_origin = SO_EE_ORIGIN_LOCAL; 307 serr->ee.ee_type = 0; 308 serr->ee.ee_code = 0; 309 serr->ee.ee_pad = 0; 310 serr->ee.ee_info = info; 311 serr->ee.ee_data = 0; 312 serr->addr_offset = (u8*)&iph->daddr - skb->nh.raw; 313 serr->port = port; 314 315 skb->h.raw = skb->tail; 316 __skb_pull(skb, skb->tail - skb->data); 317 318 if (sock_queue_err_skb(sk, skb)) 319 kfree_skb(skb); 320 } 321 322 /* 323 * Handle MSG_ERRQUEUE 324 */ 325 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len) 326 { 327 struct sock_exterr_skb *serr; 328 struct sk_buff *skb, *skb2; 329 struct sockaddr_in *sin; 330 struct { 331 struct sock_extended_err ee; 332 struct sockaddr_in offender; 333 } errhdr; 334 int err; 335 int copied; 336 337 err = -EAGAIN; 338 skb = skb_dequeue(&sk->sk_error_queue); 339 if (skb == NULL) 340 goto out; 341 342 copied = skb->len; 343 if (copied > len) { 344 msg->msg_flags |= MSG_TRUNC; 345 copied = len; 346 } 347 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied); 348 if (err) 349 goto out_free_skb; 350 351 sock_recv_timestamp(msg, sk, skb); 352 353 serr = SKB_EXT_ERR(skb); 354 355 sin = (struct sockaddr_in *)msg->msg_name; 356 if (sin) { 357 sin->sin_family = AF_INET; 358 sin->sin_addr.s_addr = *(u32*)(skb->nh.raw + serr->addr_offset); 359 sin->sin_port = serr->port; 360 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero)); 361 } 362 363 memcpy(&errhdr.ee, &serr->ee, sizeof(struct sock_extended_err)); 364 sin = &errhdr.offender; 365 sin->sin_family = AF_UNSPEC; 366 if (serr->ee.ee_origin == SO_EE_ORIGIN_ICMP) { 367 struct inet_sock *inet = inet_sk(sk); 368 369 sin->sin_family = AF_INET; 370 sin->sin_addr.s_addr = skb->nh.iph->saddr; 371 sin->sin_port = 0; 372 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero)); 373 if (inet->cmsg_flags) 374 ip_cmsg_recv(msg, skb); 375 } 376 377 put_cmsg(msg, SOL_IP, IP_RECVERR, sizeof(errhdr), &errhdr); 378 379 /* Now we could try to dump offended packet options */ 380 381 msg->msg_flags |= MSG_ERRQUEUE; 382 err = copied; 383 384 /* Reset and regenerate socket error */ 385 spin_lock_bh(&sk->sk_error_queue.lock); 386 sk->sk_err = 0; 387 if ((skb2 = skb_peek(&sk->sk_error_queue)) != NULL) { 388 sk->sk_err = SKB_EXT_ERR(skb2)->ee.ee_errno; 389 spin_unlock_bh(&sk->sk_error_queue.lock); 390 sk->sk_error_report(sk); 391 } else 392 spin_unlock_bh(&sk->sk_error_queue.lock); 393 394 out_free_skb: 395 kfree_skb(skb); 396 out: 397 return err; 398 } 399 400 401 /* 402 * Socket option code for IP. This is the end of the line after any TCP,UDP etc options on 403 * an IP socket. 404 */ 405 406 static int do_ip_setsockopt(struct sock *sk, int level, 407 int optname, char __user *optval, int optlen) 408 { 409 struct inet_sock *inet = inet_sk(sk); 410 int val=0,err; 411 412 if (((1<<optname) & ((1<<IP_PKTINFO) | (1<<IP_RECVTTL) | 413 (1<<IP_RECVOPTS) | (1<<IP_RECVTOS) | 414 (1<<IP_RETOPTS) | (1<<IP_TOS) | 415 (1<<IP_TTL) | (1<<IP_HDRINCL) | 416 (1<<IP_MTU_DISCOVER) | (1<<IP_RECVERR) | 417 (1<<IP_ROUTER_ALERT) | (1<<IP_FREEBIND) | 418 (1<<IP_PASSSEC))) || 419 optname == IP_MULTICAST_TTL || 420 optname == IP_MULTICAST_LOOP) { 421 if (optlen >= sizeof(int)) { 422 if (get_user(val, (int __user *) optval)) 423 return -EFAULT; 424 } else if (optlen >= sizeof(char)) { 425 unsigned char ucval; 426 427 if (get_user(ucval, (unsigned char __user *) optval)) 428 return -EFAULT; 429 val = (int) ucval; 430 } 431 } 432 433 /* If optlen==0, it is equivalent to val == 0 */ 434 435 #ifdef CONFIG_IP_MROUTE 436 if (optname >= MRT_BASE && optname <= (MRT_BASE + 10)) 437 return ip_mroute_setsockopt(sk,optname,optval,optlen); 438 #endif 439 440 err = 0; 441 lock_sock(sk); 442 443 switch (optname) { 444 case IP_OPTIONS: 445 { 446 struct ip_options * opt = NULL; 447 if (optlen > 40 || optlen < 0) 448 goto e_inval; 449 err = ip_options_get_from_user(&opt, optval, optlen); 450 if (err) 451 break; 452 if (inet->is_icsk) { 453 struct inet_connection_sock *icsk = inet_csk(sk); 454 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) 455 if (sk->sk_family == PF_INET || 456 (!((1 << sk->sk_state) & 457 (TCPF_LISTEN | TCPF_CLOSE)) && 458 inet->daddr != LOOPBACK4_IPV6)) { 459 #endif 460 if (inet->opt) 461 icsk->icsk_ext_hdr_len -= inet->opt->optlen; 462 if (opt) 463 icsk->icsk_ext_hdr_len += opt->optlen; 464 icsk->icsk_sync_mss(sk, icsk->icsk_pmtu_cookie); 465 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) 466 } 467 #endif 468 } 469 opt = xchg(&inet->opt, opt); 470 kfree(opt); 471 break; 472 } 473 case IP_PKTINFO: 474 if (val) 475 inet->cmsg_flags |= IP_CMSG_PKTINFO; 476 else 477 inet->cmsg_flags &= ~IP_CMSG_PKTINFO; 478 break; 479 case IP_RECVTTL: 480 if (val) 481 inet->cmsg_flags |= IP_CMSG_TTL; 482 else 483 inet->cmsg_flags &= ~IP_CMSG_TTL; 484 break; 485 case IP_RECVTOS: 486 if (val) 487 inet->cmsg_flags |= IP_CMSG_TOS; 488 else 489 inet->cmsg_flags &= ~IP_CMSG_TOS; 490 break; 491 case IP_RECVOPTS: 492 if (val) 493 inet->cmsg_flags |= IP_CMSG_RECVOPTS; 494 else 495 inet->cmsg_flags &= ~IP_CMSG_RECVOPTS; 496 break; 497 case IP_RETOPTS: 498 if (val) 499 inet->cmsg_flags |= IP_CMSG_RETOPTS; 500 else 501 inet->cmsg_flags &= ~IP_CMSG_RETOPTS; 502 break; 503 case IP_PASSSEC: 504 if (val) 505 inet->cmsg_flags |= IP_CMSG_PASSSEC; 506 else 507 inet->cmsg_flags &= ~IP_CMSG_PASSSEC; 508 break; 509 case IP_TOS: /* This sets both TOS and Precedence */ 510 if (sk->sk_type == SOCK_STREAM) { 511 val &= ~3; 512 val |= inet->tos & 3; 513 } 514 if (IPTOS_PREC(val) >= IPTOS_PREC_CRITIC_ECP && 515 !capable(CAP_NET_ADMIN)) { 516 err = -EPERM; 517 break; 518 } 519 if (inet->tos != val) { 520 inet->tos = val; 521 sk->sk_priority = rt_tos2priority(val); 522 sk_dst_reset(sk); 523 } 524 break; 525 case IP_TTL: 526 if (optlen<1) 527 goto e_inval; 528 if (val != -1 && (val < 1 || val>255)) 529 goto e_inval; 530 inet->uc_ttl = val; 531 break; 532 case IP_HDRINCL: 533 if (sk->sk_type != SOCK_RAW) { 534 err = -ENOPROTOOPT; 535 break; 536 } 537 inet->hdrincl = val ? 1 : 0; 538 break; 539 case IP_MTU_DISCOVER: 540 if (val<0 || val>2) 541 goto e_inval; 542 inet->pmtudisc = val; 543 break; 544 case IP_RECVERR: 545 inet->recverr = !!val; 546 if (!val) 547 skb_queue_purge(&sk->sk_error_queue); 548 break; 549 case IP_MULTICAST_TTL: 550 if (sk->sk_type == SOCK_STREAM) 551 goto e_inval; 552 if (optlen<1) 553 goto e_inval; 554 if (val==-1) 555 val = 1; 556 if (val < 0 || val > 255) 557 goto e_inval; 558 inet->mc_ttl = val; 559 break; 560 case IP_MULTICAST_LOOP: 561 if (optlen<1) 562 goto e_inval; 563 inet->mc_loop = !!val; 564 break; 565 case IP_MULTICAST_IF: 566 { 567 struct ip_mreqn mreq; 568 struct net_device *dev = NULL; 569 570 if (sk->sk_type == SOCK_STREAM) 571 goto e_inval; 572 /* 573 * Check the arguments are allowable 574 */ 575 576 err = -EFAULT; 577 if (optlen >= sizeof(struct ip_mreqn)) { 578 if (copy_from_user(&mreq,optval,sizeof(mreq))) 579 break; 580 } else { 581 memset(&mreq, 0, sizeof(mreq)); 582 if (optlen >= sizeof(struct in_addr) && 583 copy_from_user(&mreq.imr_address,optval,sizeof(struct in_addr))) 584 break; 585 } 586 587 if (!mreq.imr_ifindex) { 588 if (mreq.imr_address.s_addr == INADDR_ANY) { 589 inet->mc_index = 0; 590 inet->mc_addr = 0; 591 err = 0; 592 break; 593 } 594 dev = ip_dev_find(mreq.imr_address.s_addr); 595 if (dev) { 596 mreq.imr_ifindex = dev->ifindex; 597 dev_put(dev); 598 } 599 } else 600 dev = __dev_get_by_index(mreq.imr_ifindex); 601 602 603 err = -EADDRNOTAVAIL; 604 if (!dev) 605 break; 606 607 err = -EINVAL; 608 if (sk->sk_bound_dev_if && 609 mreq.imr_ifindex != sk->sk_bound_dev_if) 610 break; 611 612 inet->mc_index = mreq.imr_ifindex; 613 inet->mc_addr = mreq.imr_address.s_addr; 614 err = 0; 615 break; 616 } 617 618 case IP_ADD_MEMBERSHIP: 619 case IP_DROP_MEMBERSHIP: 620 { 621 struct ip_mreqn mreq; 622 623 if (optlen < sizeof(struct ip_mreq)) 624 goto e_inval; 625 err = -EFAULT; 626 if (optlen >= sizeof(struct ip_mreqn)) { 627 if(copy_from_user(&mreq,optval,sizeof(mreq))) 628 break; 629 } else { 630 memset(&mreq, 0, sizeof(mreq)); 631 if (copy_from_user(&mreq,optval,sizeof(struct ip_mreq))) 632 break; 633 } 634 635 if (optname == IP_ADD_MEMBERSHIP) 636 err = ip_mc_join_group(sk, &mreq); 637 else 638 err = ip_mc_leave_group(sk, &mreq); 639 break; 640 } 641 case IP_MSFILTER: 642 { 643 extern int sysctl_igmp_max_msf; 644 struct ip_msfilter *msf; 645 646 if (optlen < IP_MSFILTER_SIZE(0)) 647 goto e_inval; 648 if (optlen > sysctl_optmem_max) { 649 err = -ENOBUFS; 650 break; 651 } 652 msf = kmalloc(optlen, GFP_KERNEL); 653 if (msf == 0) { 654 err = -ENOBUFS; 655 break; 656 } 657 err = -EFAULT; 658 if (copy_from_user(msf, optval, optlen)) { 659 kfree(msf); 660 break; 661 } 662 /* numsrc >= (1G-4) overflow in 32 bits */ 663 if (msf->imsf_numsrc >= 0x3ffffffcU || 664 msf->imsf_numsrc > sysctl_igmp_max_msf) { 665 kfree(msf); 666 err = -ENOBUFS; 667 break; 668 } 669 if (IP_MSFILTER_SIZE(msf->imsf_numsrc) > optlen) { 670 kfree(msf); 671 err = -EINVAL; 672 break; 673 } 674 err = ip_mc_msfilter(sk, msf, 0); 675 kfree(msf); 676 break; 677 } 678 case IP_BLOCK_SOURCE: 679 case IP_UNBLOCK_SOURCE: 680 case IP_ADD_SOURCE_MEMBERSHIP: 681 case IP_DROP_SOURCE_MEMBERSHIP: 682 { 683 struct ip_mreq_source mreqs; 684 int omode, add; 685 686 if (optlen != sizeof(struct ip_mreq_source)) 687 goto e_inval; 688 if (copy_from_user(&mreqs, optval, sizeof(mreqs))) { 689 err = -EFAULT; 690 break; 691 } 692 if (optname == IP_BLOCK_SOURCE) { 693 omode = MCAST_EXCLUDE; 694 add = 1; 695 } else if (optname == IP_UNBLOCK_SOURCE) { 696 omode = MCAST_EXCLUDE; 697 add = 0; 698 } else if (optname == IP_ADD_SOURCE_MEMBERSHIP) { 699 struct ip_mreqn mreq; 700 701 mreq.imr_multiaddr.s_addr = mreqs.imr_multiaddr; 702 mreq.imr_address.s_addr = mreqs.imr_interface; 703 mreq.imr_ifindex = 0; 704 err = ip_mc_join_group(sk, &mreq); 705 if (err && err != -EADDRINUSE) 706 break; 707 omode = MCAST_INCLUDE; 708 add = 1; 709 } else /* IP_DROP_SOURCE_MEMBERSHIP */ { 710 omode = MCAST_INCLUDE; 711 add = 0; 712 } 713 err = ip_mc_source(add, omode, sk, &mreqs, 0); 714 break; 715 } 716 case MCAST_JOIN_GROUP: 717 case MCAST_LEAVE_GROUP: 718 { 719 struct group_req greq; 720 struct sockaddr_in *psin; 721 struct ip_mreqn mreq; 722 723 if (optlen < sizeof(struct group_req)) 724 goto e_inval; 725 err = -EFAULT; 726 if(copy_from_user(&greq, optval, sizeof(greq))) 727 break; 728 psin = (struct sockaddr_in *)&greq.gr_group; 729 if (psin->sin_family != AF_INET) 730 goto e_inval; 731 memset(&mreq, 0, sizeof(mreq)); 732 mreq.imr_multiaddr = psin->sin_addr; 733 mreq.imr_ifindex = greq.gr_interface; 734 735 if (optname == MCAST_JOIN_GROUP) 736 err = ip_mc_join_group(sk, &mreq); 737 else 738 err = ip_mc_leave_group(sk, &mreq); 739 break; 740 } 741 case MCAST_JOIN_SOURCE_GROUP: 742 case MCAST_LEAVE_SOURCE_GROUP: 743 case MCAST_BLOCK_SOURCE: 744 case MCAST_UNBLOCK_SOURCE: 745 { 746 struct group_source_req greqs; 747 struct ip_mreq_source mreqs; 748 struct sockaddr_in *psin; 749 int omode, add; 750 751 if (optlen != sizeof(struct group_source_req)) 752 goto e_inval; 753 if (copy_from_user(&greqs, optval, sizeof(greqs))) { 754 err = -EFAULT; 755 break; 756 } 757 if (greqs.gsr_group.ss_family != AF_INET || 758 greqs.gsr_source.ss_family != AF_INET) { 759 err = -EADDRNOTAVAIL; 760 break; 761 } 762 psin = (struct sockaddr_in *)&greqs.gsr_group; 763 mreqs.imr_multiaddr = psin->sin_addr.s_addr; 764 psin = (struct sockaddr_in *)&greqs.gsr_source; 765 mreqs.imr_sourceaddr = psin->sin_addr.s_addr; 766 mreqs.imr_interface = 0; /* use index for mc_source */ 767 768 if (optname == MCAST_BLOCK_SOURCE) { 769 omode = MCAST_EXCLUDE; 770 add = 1; 771 } else if (optname == MCAST_UNBLOCK_SOURCE) { 772 omode = MCAST_EXCLUDE; 773 add = 0; 774 } else if (optname == MCAST_JOIN_SOURCE_GROUP) { 775 struct ip_mreqn mreq; 776 777 psin = (struct sockaddr_in *)&greqs.gsr_group; 778 mreq.imr_multiaddr = psin->sin_addr; 779 mreq.imr_address.s_addr = 0; 780 mreq.imr_ifindex = greqs.gsr_interface; 781 err = ip_mc_join_group(sk, &mreq); 782 if (err && err != -EADDRINUSE) 783 break; 784 greqs.gsr_interface = mreq.imr_ifindex; 785 omode = MCAST_INCLUDE; 786 add = 1; 787 } else /* MCAST_LEAVE_SOURCE_GROUP */ { 788 omode = MCAST_INCLUDE; 789 add = 0; 790 } 791 err = ip_mc_source(add, omode, sk, &mreqs, 792 greqs.gsr_interface); 793 break; 794 } 795 case MCAST_MSFILTER: 796 { 797 extern int sysctl_igmp_max_msf; 798 struct sockaddr_in *psin; 799 struct ip_msfilter *msf = NULL; 800 struct group_filter *gsf = NULL; 801 int msize, i, ifindex; 802 803 if (optlen < GROUP_FILTER_SIZE(0)) 804 goto e_inval; 805 if (optlen > sysctl_optmem_max) { 806 err = -ENOBUFS; 807 break; 808 } 809 gsf = kmalloc(optlen,GFP_KERNEL); 810 if (gsf == 0) { 811 err = -ENOBUFS; 812 break; 813 } 814 err = -EFAULT; 815 if (copy_from_user(gsf, optval, optlen)) { 816 goto mc_msf_out; 817 } 818 /* numsrc >= (4G-140)/128 overflow in 32 bits */ 819 if (gsf->gf_numsrc >= 0x1ffffff || 820 gsf->gf_numsrc > sysctl_igmp_max_msf) { 821 err = -ENOBUFS; 822 goto mc_msf_out; 823 } 824 if (GROUP_FILTER_SIZE(gsf->gf_numsrc) > optlen) { 825 err = -EINVAL; 826 goto mc_msf_out; 827 } 828 msize = IP_MSFILTER_SIZE(gsf->gf_numsrc); 829 msf = kmalloc(msize,GFP_KERNEL); 830 if (msf == 0) { 831 err = -ENOBUFS; 832 goto mc_msf_out; 833 } 834 ifindex = gsf->gf_interface; 835 psin = (struct sockaddr_in *)&gsf->gf_group; 836 if (psin->sin_family != AF_INET) { 837 err = -EADDRNOTAVAIL; 838 goto mc_msf_out; 839 } 840 msf->imsf_multiaddr = psin->sin_addr.s_addr; 841 msf->imsf_interface = 0; 842 msf->imsf_fmode = gsf->gf_fmode; 843 msf->imsf_numsrc = gsf->gf_numsrc; 844 err = -EADDRNOTAVAIL; 845 for (i=0; i<gsf->gf_numsrc; ++i) { 846 psin = (struct sockaddr_in *)&gsf->gf_slist[i]; 847 848 if (psin->sin_family != AF_INET) 849 goto mc_msf_out; 850 msf->imsf_slist[i] = psin->sin_addr.s_addr; 851 } 852 kfree(gsf); 853 gsf = NULL; 854 855 err = ip_mc_msfilter(sk, msf, ifindex); 856 mc_msf_out: 857 kfree(msf); 858 kfree(gsf); 859 break; 860 } 861 case IP_ROUTER_ALERT: 862 err = ip_ra_control(sk, val ? 1 : 0, NULL); 863 break; 864 865 case IP_FREEBIND: 866 if (optlen<1) 867 goto e_inval; 868 inet->freebind = !!val; 869 break; 870 871 case IP_IPSEC_POLICY: 872 case IP_XFRM_POLICY: 873 err = -EPERM; 874 if (!capable(CAP_NET_ADMIN)) 875 break; 876 err = xfrm_user_policy(sk, optname, optval, optlen); 877 break; 878 879 default: 880 err = -ENOPROTOOPT; 881 break; 882 } 883 release_sock(sk); 884 return err; 885 886 e_inval: 887 release_sock(sk); 888 return -EINVAL; 889 } 890 891 int ip_setsockopt(struct sock *sk, int level, 892 int optname, char __user *optval, int optlen) 893 { 894 int err; 895 896 if (level != SOL_IP) 897 return -ENOPROTOOPT; 898 899 err = do_ip_setsockopt(sk, level, optname, optval, optlen); 900 #ifdef CONFIG_NETFILTER 901 /* we need to exclude all possible ENOPROTOOPTs except default case */ 902 if (err == -ENOPROTOOPT && optname != IP_HDRINCL && 903 optname != IP_IPSEC_POLICY && optname != IP_XFRM_POLICY 904 #ifdef CONFIG_IP_MROUTE 905 && (optname < MRT_BASE || optname > (MRT_BASE + 10)) 906 #endif 907 ) { 908 lock_sock(sk); 909 err = nf_setsockopt(sk, PF_INET, optname, optval, optlen); 910 release_sock(sk); 911 } 912 #endif 913 return err; 914 } 915 916 #ifdef CONFIG_COMPAT 917 int compat_ip_setsockopt(struct sock *sk, int level, int optname, 918 char __user *optval, int optlen) 919 { 920 int err; 921 922 if (level != SOL_IP) 923 return -ENOPROTOOPT; 924 925 err = do_ip_setsockopt(sk, level, optname, optval, optlen); 926 #ifdef CONFIG_NETFILTER 927 /* we need to exclude all possible ENOPROTOOPTs except default case */ 928 if (err == -ENOPROTOOPT && optname != IP_HDRINCL && 929 optname != IP_IPSEC_POLICY && optname != IP_XFRM_POLICY 930 #ifdef CONFIG_IP_MROUTE 931 && (optname < MRT_BASE || optname > (MRT_BASE + 10)) 932 #endif 933 ) { 934 lock_sock(sk); 935 err = compat_nf_setsockopt(sk, PF_INET, optname, 936 optval, optlen); 937 release_sock(sk); 938 } 939 #endif 940 return err; 941 } 942 943 EXPORT_SYMBOL(compat_ip_setsockopt); 944 #endif 945 946 /* 947 * Get the options. Note for future reference. The GET of IP options gets the 948 * _received_ ones. The set sets the _sent_ ones. 949 */ 950 951 static int do_ip_getsockopt(struct sock *sk, int level, int optname, 952 char __user *optval, int __user *optlen) 953 { 954 struct inet_sock *inet = inet_sk(sk); 955 int val; 956 int len; 957 958 if(level!=SOL_IP) 959 return -EOPNOTSUPP; 960 961 #ifdef CONFIG_IP_MROUTE 962 if(optname>=MRT_BASE && optname <=MRT_BASE+10) 963 { 964 return ip_mroute_getsockopt(sk,optname,optval,optlen); 965 } 966 #endif 967 968 if(get_user(len,optlen)) 969 return -EFAULT; 970 if(len < 0) 971 return -EINVAL; 972 973 lock_sock(sk); 974 975 switch(optname) { 976 case IP_OPTIONS: 977 { 978 unsigned char optbuf[sizeof(struct ip_options)+40]; 979 struct ip_options * opt = (struct ip_options*)optbuf; 980 opt->optlen = 0; 981 if (inet->opt) 982 memcpy(optbuf, inet->opt, 983 sizeof(struct ip_options)+ 984 inet->opt->optlen); 985 release_sock(sk); 986 987 if (opt->optlen == 0) 988 return put_user(0, optlen); 989 990 ip_options_undo(opt); 991 992 len = min_t(unsigned int, len, opt->optlen); 993 if(put_user(len, optlen)) 994 return -EFAULT; 995 if(copy_to_user(optval, opt->__data, len)) 996 return -EFAULT; 997 return 0; 998 } 999 case IP_PKTINFO: 1000 val = (inet->cmsg_flags & IP_CMSG_PKTINFO) != 0; 1001 break; 1002 case IP_RECVTTL: 1003 val = (inet->cmsg_flags & IP_CMSG_TTL) != 0; 1004 break; 1005 case IP_RECVTOS: 1006 val = (inet->cmsg_flags & IP_CMSG_TOS) != 0; 1007 break; 1008 case IP_RECVOPTS: 1009 val = (inet->cmsg_flags & IP_CMSG_RECVOPTS) != 0; 1010 break; 1011 case IP_RETOPTS: 1012 val = (inet->cmsg_flags & IP_CMSG_RETOPTS) != 0; 1013 break; 1014 case IP_PASSSEC: 1015 val = (inet->cmsg_flags & IP_CMSG_PASSSEC) != 0; 1016 break; 1017 case IP_TOS: 1018 val = inet->tos; 1019 break; 1020 case IP_TTL: 1021 val = (inet->uc_ttl == -1 ? 1022 sysctl_ip_default_ttl : 1023 inet->uc_ttl); 1024 break; 1025 case IP_HDRINCL: 1026 val = inet->hdrincl; 1027 break; 1028 case IP_MTU_DISCOVER: 1029 val = inet->pmtudisc; 1030 break; 1031 case IP_MTU: 1032 { 1033 struct dst_entry *dst; 1034 val = 0; 1035 dst = sk_dst_get(sk); 1036 if (dst) { 1037 val = dst_mtu(dst); 1038 dst_release(dst); 1039 } 1040 if (!val) { 1041 release_sock(sk); 1042 return -ENOTCONN; 1043 } 1044 break; 1045 } 1046 case IP_RECVERR: 1047 val = inet->recverr; 1048 break; 1049 case IP_MULTICAST_TTL: 1050 val = inet->mc_ttl; 1051 break; 1052 case IP_MULTICAST_LOOP: 1053 val = inet->mc_loop; 1054 break; 1055 case IP_MULTICAST_IF: 1056 { 1057 struct in_addr addr; 1058 len = min_t(unsigned int, len, sizeof(struct in_addr)); 1059 addr.s_addr = inet->mc_addr; 1060 release_sock(sk); 1061 1062 if(put_user(len, optlen)) 1063 return -EFAULT; 1064 if(copy_to_user(optval, &addr, len)) 1065 return -EFAULT; 1066 return 0; 1067 } 1068 case IP_MSFILTER: 1069 { 1070 struct ip_msfilter msf; 1071 int err; 1072 1073 if (len < IP_MSFILTER_SIZE(0)) { 1074 release_sock(sk); 1075 return -EINVAL; 1076 } 1077 if (copy_from_user(&msf, optval, IP_MSFILTER_SIZE(0))) { 1078 release_sock(sk); 1079 return -EFAULT; 1080 } 1081 err = ip_mc_msfget(sk, &msf, 1082 (struct ip_msfilter __user *)optval, optlen); 1083 release_sock(sk); 1084 return err; 1085 } 1086 case MCAST_MSFILTER: 1087 { 1088 struct group_filter gsf; 1089 int err; 1090 1091 if (len < GROUP_FILTER_SIZE(0)) { 1092 release_sock(sk); 1093 return -EINVAL; 1094 } 1095 if (copy_from_user(&gsf, optval, GROUP_FILTER_SIZE(0))) { 1096 release_sock(sk); 1097 return -EFAULT; 1098 } 1099 err = ip_mc_gsfget(sk, &gsf, 1100 (struct group_filter __user *)optval, optlen); 1101 release_sock(sk); 1102 return err; 1103 } 1104 case IP_PKTOPTIONS: 1105 { 1106 struct msghdr msg; 1107 1108 release_sock(sk); 1109 1110 if (sk->sk_type != SOCK_STREAM) 1111 return -ENOPROTOOPT; 1112 1113 msg.msg_control = optval; 1114 msg.msg_controllen = len; 1115 msg.msg_flags = 0; 1116 1117 if (inet->cmsg_flags & IP_CMSG_PKTINFO) { 1118 struct in_pktinfo info; 1119 1120 info.ipi_addr.s_addr = inet->rcv_saddr; 1121 info.ipi_spec_dst.s_addr = inet->rcv_saddr; 1122 info.ipi_ifindex = inet->mc_index; 1123 put_cmsg(&msg, SOL_IP, IP_PKTINFO, sizeof(info), &info); 1124 } 1125 if (inet->cmsg_flags & IP_CMSG_TTL) { 1126 int hlim = inet->mc_ttl; 1127 put_cmsg(&msg, SOL_IP, IP_TTL, sizeof(hlim), &hlim); 1128 } 1129 len -= msg.msg_controllen; 1130 return put_user(len, optlen); 1131 } 1132 case IP_FREEBIND: 1133 val = inet->freebind; 1134 break; 1135 default: 1136 release_sock(sk); 1137 return -ENOPROTOOPT; 1138 } 1139 release_sock(sk); 1140 1141 if (len < sizeof(int) && len > 0 && val>=0 && val<255) { 1142 unsigned char ucval = (unsigned char)val; 1143 len = 1; 1144 if(put_user(len, optlen)) 1145 return -EFAULT; 1146 if(copy_to_user(optval,&ucval,1)) 1147 return -EFAULT; 1148 } else { 1149 len = min_t(unsigned int, sizeof(int), len); 1150 if(put_user(len, optlen)) 1151 return -EFAULT; 1152 if(copy_to_user(optval,&val,len)) 1153 return -EFAULT; 1154 } 1155 return 0; 1156 } 1157 1158 int ip_getsockopt(struct sock *sk, int level, 1159 int optname, char __user *optval, int __user *optlen) 1160 { 1161 int err; 1162 1163 err = do_ip_getsockopt(sk, level, optname, optval, optlen); 1164 #ifdef CONFIG_NETFILTER 1165 /* we need to exclude all possible ENOPROTOOPTs except default case */ 1166 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS 1167 #ifdef CONFIG_IP_MROUTE 1168 && (optname < MRT_BASE || optname > MRT_BASE+10) 1169 #endif 1170 ) { 1171 int len; 1172 1173 if(get_user(len,optlen)) 1174 return -EFAULT; 1175 1176 lock_sock(sk); 1177 err = nf_getsockopt(sk, PF_INET, optname, optval, 1178 &len); 1179 release_sock(sk); 1180 if (err >= 0) 1181 err = put_user(len, optlen); 1182 return err; 1183 } 1184 #endif 1185 return err; 1186 } 1187 1188 #ifdef CONFIG_COMPAT 1189 int compat_ip_getsockopt(struct sock *sk, int level, int optname, 1190 char __user *optval, int __user *optlen) 1191 { 1192 int err = do_ip_getsockopt(sk, level, optname, optval, optlen); 1193 #ifdef CONFIG_NETFILTER 1194 /* we need to exclude all possible ENOPROTOOPTs except default case */ 1195 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS 1196 #ifdef CONFIG_IP_MROUTE 1197 && (optname < MRT_BASE || optname > MRT_BASE+10) 1198 #endif 1199 ) { 1200 int len; 1201 1202 if (get_user(len, optlen)) 1203 return -EFAULT; 1204 1205 lock_sock(sk); 1206 err = compat_nf_getsockopt(sk, PF_INET, optname, optval, &len); 1207 release_sock(sk); 1208 if (err >= 0) 1209 err = put_user(len, optlen); 1210 return err; 1211 } 1212 #endif 1213 return err; 1214 } 1215 1216 EXPORT_SYMBOL(compat_ip_getsockopt); 1217 #endif 1218 1219 EXPORT_SYMBOL(ip_cmsg_recv); 1220 1221 EXPORT_SYMBOL(ip_getsockopt); 1222 EXPORT_SYMBOL(ip_setsockopt); 1223