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.
11 * Many : Split from ip.c , see ip.c for history.
12 * Martin Mares : TOS setting fixed.
13 * Alan Cox : Fixed a couple of oopses in Martin's
15 * Mike McLagan : Routing by source
18 #include <linux/module.h>
19 #include <linux/types.h>
21 #include <linux/skbuff.h>
23 #include <linux/icmp.h>
24 #include <linux/inetdevice.h>
25 #include <linux/netdevice.h>
26 #include <linux/slab.h>
30 #include <net/tcp_states.h>
31 #include <linux/udp.h>
32 #include <linux/igmp.h>
33 #include <linux/netfilter.h>
34 #include <linux/route.h>
35 #include <linux/mroute.h>
36 #include <net/route.h>
38 #include <net/compat.h>
39 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
40 #include <net/transp_v6.h>
43 #include <linux/errqueue.h>
44 #include <asm/uaccess.h>
46 #define IP_CMSG_PKTINFO 1
49 #define IP_CMSG_RECVOPTS 8
50 #define IP_CMSG_RETOPTS 16
51 #define IP_CMSG_PASSSEC 32
52 #define IP_CMSG_ORIGDSTADDR 64
55 * SOL_IP control messages.
58 static void ip_cmsg_recv_pktinfo(struct msghdr *msg, struct sk_buff *skb)
60 struct in_pktinfo info;
61 struct rtable *rt = skb_rtable(skb);
63 info.ipi_addr.s_addr = ip_hdr(skb)->daddr;
65 info.ipi_ifindex = rt->rt_iif;
66 info.ipi_spec_dst.s_addr = rt->rt_spec_dst;
69 info.ipi_spec_dst.s_addr = 0;
72 put_cmsg(msg, SOL_IP, IP_PKTINFO, sizeof(info), &info);
75 static void ip_cmsg_recv_ttl(struct msghdr *msg, struct sk_buff *skb)
77 int ttl = ip_hdr(skb)->ttl;
78 put_cmsg(msg, SOL_IP, IP_TTL, sizeof(int), &ttl);
81 static void ip_cmsg_recv_tos(struct msghdr *msg, struct sk_buff *skb)
83 put_cmsg(msg, SOL_IP, IP_TOS, 1, &ip_hdr(skb)->tos);
86 static void ip_cmsg_recv_opts(struct msghdr *msg, struct sk_buff *skb)
88 if (IPCB(skb)->opt.optlen == 0)
91 put_cmsg(msg, SOL_IP, IP_RECVOPTS, IPCB(skb)->opt.optlen,
96 static void ip_cmsg_recv_retopts(struct msghdr *msg, struct sk_buff *skb)
98 unsigned char optbuf[sizeof(struct ip_options) + 40];
99 struct ip_options * opt = (struct ip_options *)optbuf;
101 if (IPCB(skb)->opt.optlen == 0)
104 if (ip_options_echo(opt, skb)) {
105 msg->msg_flags |= MSG_CTRUNC;
108 ip_options_undo(opt);
110 put_cmsg(msg, SOL_IP, IP_RETOPTS, opt->optlen, opt->__data);
113 static void ip_cmsg_recv_security(struct msghdr *msg, struct sk_buff *skb)
119 err = security_socket_getpeersec_dgram(NULL, skb, &secid);
123 err = security_secid_to_secctx(secid, &secdata, &seclen);
127 put_cmsg(msg, SOL_IP, SCM_SECURITY, seclen, secdata);
128 security_release_secctx(secdata, seclen);
131 static void ip_cmsg_recv_dstaddr(struct msghdr *msg, struct sk_buff *skb)
133 struct sockaddr_in sin;
134 struct iphdr *iph = ip_hdr(skb);
135 __be16 *ports = (__be16 *)skb_transport_header(skb);
137 if (skb_transport_offset(skb) + 4 > skb->len)
140 /* All current transport protocols have the port numbers in the
141 * first four bytes of the transport header and this function is
142 * written with this assumption in mind.
145 sin.sin_family = AF_INET;
146 sin.sin_addr.s_addr = iph->daddr;
147 sin.sin_port = ports[1];
148 memset(sin.sin_zero, 0, sizeof(sin.sin_zero));
150 put_cmsg(msg, SOL_IP, IP_ORIGDSTADDR, sizeof(sin), &sin);
153 void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
155 struct inet_sock *inet = inet_sk(skb->sk);
156 unsigned flags = inet->cmsg_flags;
158 /* Ordered by supposed usage frequency */
160 ip_cmsg_recv_pktinfo(msg, skb);
161 if ((flags >>= 1) == 0)
165 ip_cmsg_recv_ttl(msg, skb);
166 if ((flags >>= 1) == 0)
170 ip_cmsg_recv_tos(msg, skb);
171 if ((flags >>= 1) == 0)
175 ip_cmsg_recv_opts(msg, skb);
176 if ((flags >>= 1) == 0)
180 ip_cmsg_recv_retopts(msg, skb);
181 if ((flags >>= 1) == 0)
185 ip_cmsg_recv_security(msg, skb);
187 if ((flags >>= 1) == 0)
190 ip_cmsg_recv_dstaddr(msg, skb);
193 EXPORT_SYMBOL(ip_cmsg_recv);
195 int ip_cmsg_send(struct net *net, struct msghdr *msg, struct ipcm_cookie *ipc)
198 struct cmsghdr *cmsg;
200 for (cmsg = CMSG_FIRSTHDR(msg); cmsg; cmsg = CMSG_NXTHDR(msg, cmsg)) {
201 if (!CMSG_OK(msg, cmsg))
203 if (cmsg->cmsg_level != SOL_IP)
205 switch (cmsg->cmsg_type) {
207 err = cmsg->cmsg_len - CMSG_ALIGN(sizeof(struct cmsghdr));
208 err = ip_options_get(net, &ipc->opt, CMSG_DATA(cmsg),
209 err < 40 ? err : 40);
215 struct in_pktinfo *info;
216 if (cmsg->cmsg_len != CMSG_LEN(sizeof(struct in_pktinfo)))
218 info = (struct in_pktinfo *)CMSG_DATA(cmsg);
219 ipc->oif = info->ipi_ifindex;
220 ipc->addr = info->ipi_spec_dst.s_addr;
231 /* Special input handler for packets caught by router alert option.
232 They are selected only by protocol field, and then processed likely
233 local ones; but only if someone wants them! Otherwise, router
234 not running rsvpd will kill RSVP.
236 It is user level problem, what it will make with them.
237 I have no idea, how it will masquearde or NAT them (it is joke, joke :-)),
238 but receiver should be enough clever f.e. to forward mtrace requests,
239 sent to multicast group to reach destination designated router.
241 struct ip_ra_chain *ip_ra_chain;
242 static DEFINE_SPINLOCK(ip_ra_lock);
245 static void ip_ra_destroy_rcu(struct rcu_head *head)
247 struct ip_ra_chain *ra = container_of(head, struct ip_ra_chain, rcu);
249 sock_put(ra->saved_sk);
253 int ip_ra_control(struct sock *sk, unsigned char on,
254 void (*destructor)(struct sock *))
256 struct ip_ra_chain *ra, *new_ra, **rap;
258 if (sk->sk_type != SOCK_RAW || inet_sk(sk)->inet_num == IPPROTO_RAW)
261 new_ra = on ? kmalloc(sizeof(*new_ra), GFP_KERNEL) : NULL;
263 spin_lock_bh(&ip_ra_lock);
264 for (rap = &ip_ra_chain; (ra = *rap) != NULL; rap = &ra->next) {
267 spin_unlock_bh(&ip_ra_lock);
271 /* dont let ip_call_ra_chain() use sk again */
273 rcu_assign_pointer(*rap, ra->next);
274 spin_unlock_bh(&ip_ra_lock);
279 * Delay sock_put(sk) and kfree(ra) after one rcu grace
280 * period. This guarantee ip_call_ra_chain() dont need
281 * to mess with socket refcounts.
284 call_rcu(&ra->rcu, ip_ra_destroy_rcu);
288 if (new_ra == NULL) {
289 spin_unlock_bh(&ip_ra_lock);
293 new_ra->destructor = destructor;
296 rcu_assign_pointer(*rap, new_ra);
298 spin_unlock_bh(&ip_ra_lock);
303 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err,
304 __be16 port, u32 info, u8 *payload)
306 struct sock_exterr_skb *serr;
308 skb = skb_clone(skb, GFP_ATOMIC);
312 serr = SKB_EXT_ERR(skb);
313 serr->ee.ee_errno = err;
314 serr->ee.ee_origin = SO_EE_ORIGIN_ICMP;
315 serr->ee.ee_type = icmp_hdr(skb)->type;
316 serr->ee.ee_code = icmp_hdr(skb)->code;
318 serr->ee.ee_info = info;
319 serr->ee.ee_data = 0;
320 serr->addr_offset = (u8 *)&(((struct iphdr *)(icmp_hdr(skb) + 1))->daddr) -
321 skb_network_header(skb);
324 if (skb_pull(skb, payload - skb->data) != NULL) {
325 skb_reset_transport_header(skb);
326 if (sock_queue_err_skb(sk, skb) == 0)
332 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 port, u32 info)
334 struct inet_sock *inet = inet_sk(sk);
335 struct sock_exterr_skb *serr;
342 skb = alloc_skb(sizeof(struct iphdr), GFP_ATOMIC);
346 skb_put(skb, sizeof(struct iphdr));
347 skb_reset_network_header(skb);
351 serr = SKB_EXT_ERR(skb);
352 serr->ee.ee_errno = err;
353 serr->ee.ee_origin = SO_EE_ORIGIN_LOCAL;
354 serr->ee.ee_type = 0;
355 serr->ee.ee_code = 0;
357 serr->ee.ee_info = info;
358 serr->ee.ee_data = 0;
359 serr->addr_offset = (u8 *)&iph->daddr - skb_network_header(skb);
362 __skb_pull(skb, skb_tail_pointer(skb) - skb->data);
363 skb_reset_transport_header(skb);
365 if (sock_queue_err_skb(sk, skb))
370 * Handle MSG_ERRQUEUE
372 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len)
374 struct sock_exterr_skb *serr;
375 struct sk_buff *skb, *skb2;
376 struct sockaddr_in *sin;
378 struct sock_extended_err ee;
379 struct sockaddr_in offender;
385 skb = skb_dequeue(&sk->sk_error_queue);
391 msg->msg_flags |= MSG_TRUNC;
394 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
398 sock_recv_timestamp(msg, sk, skb);
400 serr = SKB_EXT_ERR(skb);
402 sin = (struct sockaddr_in *)msg->msg_name;
404 sin->sin_family = AF_INET;
405 sin->sin_addr.s_addr = *(__be32 *)(skb_network_header(skb) +
407 sin->sin_port = serr->port;
408 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
411 memcpy(&errhdr.ee, &serr->ee, sizeof(struct sock_extended_err));
412 sin = &errhdr.offender;
413 sin->sin_family = AF_UNSPEC;
414 if (serr->ee.ee_origin == SO_EE_ORIGIN_ICMP) {
415 struct inet_sock *inet = inet_sk(sk);
417 sin->sin_family = AF_INET;
418 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
420 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
421 if (inet->cmsg_flags)
422 ip_cmsg_recv(msg, skb);
425 put_cmsg(msg, SOL_IP, IP_RECVERR, sizeof(errhdr), &errhdr);
427 /* Now we could try to dump offended packet options */
429 msg->msg_flags |= MSG_ERRQUEUE;
432 /* Reset and regenerate socket error */
433 spin_lock_bh(&sk->sk_error_queue.lock);
435 skb2 = skb_peek(&sk->sk_error_queue);
437 sk->sk_err = SKB_EXT_ERR(skb2)->ee.ee_errno;
438 spin_unlock_bh(&sk->sk_error_queue.lock);
439 sk->sk_error_report(sk);
441 spin_unlock_bh(&sk->sk_error_queue.lock);
451 * Socket option code for IP. This is the end of the line after any
452 * TCP,UDP etc options on an IP socket.
455 static int do_ip_setsockopt(struct sock *sk, int level,
456 int optname, char __user *optval, unsigned int optlen)
458 struct inet_sock *inet = inet_sk(sk);
461 if (((1<<optname) & ((1<<IP_PKTINFO) | (1<<IP_RECVTTL) |
462 (1<<IP_RECVOPTS) | (1<<IP_RECVTOS) |
463 (1<<IP_RETOPTS) | (1<<IP_TOS) |
464 (1<<IP_TTL) | (1<<IP_HDRINCL) |
465 (1<<IP_MTU_DISCOVER) | (1<<IP_RECVERR) |
466 (1<<IP_ROUTER_ALERT) | (1<<IP_FREEBIND) |
467 (1<<IP_PASSSEC) | (1<<IP_TRANSPARENT) |
468 (1<<IP_MINTTL) | (1<<IP_NODEFRAG))) ||
469 optname == IP_MULTICAST_TTL ||
470 optname == IP_MULTICAST_ALL ||
471 optname == IP_MULTICAST_LOOP ||
472 optname == IP_RECVORIGDSTADDR) {
473 if (optlen >= sizeof(int)) {
474 if (get_user(val, (int __user *) optval))
476 } else if (optlen >= sizeof(char)) {
479 if (get_user(ucval, (unsigned char __user *) optval))
485 /* If optlen==0, it is equivalent to val == 0 */
487 if (ip_mroute_opt(optname))
488 return ip_mroute_setsockopt(sk, optname, optval, optlen);
496 struct ip_options *opt = NULL;
499 err = ip_options_get_from_user(sock_net(sk), &opt,
504 struct inet_connection_sock *icsk = inet_csk(sk);
505 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
506 if (sk->sk_family == PF_INET ||
507 (!((1 << sk->sk_state) &
508 (TCPF_LISTEN | TCPF_CLOSE)) &&
509 inet->inet_daddr != LOOPBACK4_IPV6)) {
512 icsk->icsk_ext_hdr_len -= inet->opt->optlen;
514 icsk->icsk_ext_hdr_len += opt->optlen;
515 icsk->icsk_sync_mss(sk, icsk->icsk_pmtu_cookie);
516 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
520 opt = xchg(&inet->opt, opt);
526 inet->cmsg_flags |= IP_CMSG_PKTINFO;
528 inet->cmsg_flags &= ~IP_CMSG_PKTINFO;
532 inet->cmsg_flags |= IP_CMSG_TTL;
534 inet->cmsg_flags &= ~IP_CMSG_TTL;
538 inet->cmsg_flags |= IP_CMSG_TOS;
540 inet->cmsg_flags &= ~IP_CMSG_TOS;
544 inet->cmsg_flags |= IP_CMSG_RECVOPTS;
546 inet->cmsg_flags &= ~IP_CMSG_RECVOPTS;
550 inet->cmsg_flags |= IP_CMSG_RETOPTS;
552 inet->cmsg_flags &= ~IP_CMSG_RETOPTS;
556 inet->cmsg_flags |= IP_CMSG_PASSSEC;
558 inet->cmsg_flags &= ~IP_CMSG_PASSSEC;
560 case IP_RECVORIGDSTADDR:
562 inet->cmsg_flags |= IP_CMSG_ORIGDSTADDR;
564 inet->cmsg_flags &= ~IP_CMSG_ORIGDSTADDR;
566 case IP_TOS: /* This sets both TOS and Precedence */
567 if (sk->sk_type == SOCK_STREAM) {
569 val |= inet->tos & 3;
571 if (inet->tos != val) {
573 sk->sk_priority = rt_tos2priority(val);
580 if (val != -1 && (val < 0 || val > 255))
585 if (sk->sk_type != SOCK_RAW) {
589 inet->hdrincl = val ? 1 : 0;
592 if (sk->sk_type != SOCK_RAW) {
596 inet->nodefrag = val ? 1 : 0;
598 case IP_MTU_DISCOVER:
599 if (val < IP_PMTUDISC_DONT || val > IP_PMTUDISC_PROBE)
601 inet->pmtudisc = val;
604 inet->recverr = !!val;
606 skb_queue_purge(&sk->sk_error_queue);
608 case IP_MULTICAST_TTL:
609 if (sk->sk_type == SOCK_STREAM)
615 if (val < 0 || val > 255)
619 case IP_MULTICAST_LOOP:
622 inet->mc_loop = !!val;
624 case IP_MULTICAST_IF:
626 struct ip_mreqn mreq;
627 struct net_device *dev = NULL;
629 if (sk->sk_type == SOCK_STREAM)
632 * Check the arguments are allowable
635 if (optlen < sizeof(struct in_addr))
639 if (optlen >= sizeof(struct ip_mreqn)) {
640 if (copy_from_user(&mreq, optval, sizeof(mreq)))
643 memset(&mreq, 0, sizeof(mreq));
644 if (optlen >= sizeof(struct in_addr) &&
645 copy_from_user(&mreq.imr_address, optval,
646 sizeof(struct in_addr)))
650 if (!mreq.imr_ifindex) {
651 if (mreq.imr_address.s_addr == htonl(INADDR_ANY)) {
657 dev = ip_dev_find(sock_net(sk), mreq.imr_address.s_addr);
659 mreq.imr_ifindex = dev->ifindex;
661 dev = dev_get_by_index(sock_net(sk), mreq.imr_ifindex);
664 err = -EADDRNOTAVAIL;
670 if (sk->sk_bound_dev_if &&
671 mreq.imr_ifindex != sk->sk_bound_dev_if)
674 inet->mc_index = mreq.imr_ifindex;
675 inet->mc_addr = mreq.imr_address.s_addr;
680 case IP_ADD_MEMBERSHIP:
681 case IP_DROP_MEMBERSHIP:
683 struct ip_mreqn mreq;
686 if (inet_sk(sk)->is_icsk)
689 if (optlen < sizeof(struct ip_mreq))
692 if (optlen >= sizeof(struct ip_mreqn)) {
693 if (copy_from_user(&mreq, optval, sizeof(mreq)))
696 memset(&mreq, 0, sizeof(mreq));
697 if (copy_from_user(&mreq, optval, sizeof(struct ip_mreq)))
701 if (optname == IP_ADD_MEMBERSHIP)
702 err = ip_mc_join_group(sk, &mreq);
704 err = ip_mc_leave_group(sk, &mreq);
709 struct ip_msfilter *msf;
711 if (optlen < IP_MSFILTER_SIZE(0))
713 if (optlen > sysctl_optmem_max) {
717 msf = kmalloc(optlen, GFP_KERNEL);
723 if (copy_from_user(msf, optval, optlen)) {
727 /* numsrc >= (1G-4) overflow in 32 bits */
728 if (msf->imsf_numsrc >= 0x3ffffffcU ||
729 msf->imsf_numsrc > sysctl_igmp_max_msf) {
734 if (IP_MSFILTER_SIZE(msf->imsf_numsrc) > optlen) {
739 err = ip_mc_msfilter(sk, msf, 0);
743 case IP_BLOCK_SOURCE:
744 case IP_UNBLOCK_SOURCE:
745 case IP_ADD_SOURCE_MEMBERSHIP:
746 case IP_DROP_SOURCE_MEMBERSHIP:
748 struct ip_mreq_source mreqs;
751 if (optlen != sizeof(struct ip_mreq_source))
753 if (copy_from_user(&mreqs, optval, sizeof(mreqs))) {
757 if (optname == IP_BLOCK_SOURCE) {
758 omode = MCAST_EXCLUDE;
760 } else if (optname == IP_UNBLOCK_SOURCE) {
761 omode = MCAST_EXCLUDE;
763 } else if (optname == IP_ADD_SOURCE_MEMBERSHIP) {
764 struct ip_mreqn mreq;
766 mreq.imr_multiaddr.s_addr = mreqs.imr_multiaddr;
767 mreq.imr_address.s_addr = mreqs.imr_interface;
768 mreq.imr_ifindex = 0;
769 err = ip_mc_join_group(sk, &mreq);
770 if (err && err != -EADDRINUSE)
772 omode = MCAST_INCLUDE;
774 } else /* IP_DROP_SOURCE_MEMBERSHIP */ {
775 omode = MCAST_INCLUDE;
778 err = ip_mc_source(add, omode, sk, &mreqs, 0);
781 case MCAST_JOIN_GROUP:
782 case MCAST_LEAVE_GROUP:
784 struct group_req greq;
785 struct sockaddr_in *psin;
786 struct ip_mreqn mreq;
788 if (optlen < sizeof(struct group_req))
791 if (copy_from_user(&greq, optval, sizeof(greq)))
793 psin = (struct sockaddr_in *)&greq.gr_group;
794 if (psin->sin_family != AF_INET)
796 memset(&mreq, 0, sizeof(mreq));
797 mreq.imr_multiaddr = psin->sin_addr;
798 mreq.imr_ifindex = greq.gr_interface;
800 if (optname == MCAST_JOIN_GROUP)
801 err = ip_mc_join_group(sk, &mreq);
803 err = ip_mc_leave_group(sk, &mreq);
806 case MCAST_JOIN_SOURCE_GROUP:
807 case MCAST_LEAVE_SOURCE_GROUP:
808 case MCAST_BLOCK_SOURCE:
809 case MCAST_UNBLOCK_SOURCE:
811 struct group_source_req greqs;
812 struct ip_mreq_source mreqs;
813 struct sockaddr_in *psin;
816 if (optlen != sizeof(struct group_source_req))
818 if (copy_from_user(&greqs, optval, sizeof(greqs))) {
822 if (greqs.gsr_group.ss_family != AF_INET ||
823 greqs.gsr_source.ss_family != AF_INET) {
824 err = -EADDRNOTAVAIL;
827 psin = (struct sockaddr_in *)&greqs.gsr_group;
828 mreqs.imr_multiaddr = psin->sin_addr.s_addr;
829 psin = (struct sockaddr_in *)&greqs.gsr_source;
830 mreqs.imr_sourceaddr = psin->sin_addr.s_addr;
831 mreqs.imr_interface = 0; /* use index for mc_source */
833 if (optname == MCAST_BLOCK_SOURCE) {
834 omode = MCAST_EXCLUDE;
836 } else if (optname == MCAST_UNBLOCK_SOURCE) {
837 omode = MCAST_EXCLUDE;
839 } else if (optname == MCAST_JOIN_SOURCE_GROUP) {
840 struct ip_mreqn mreq;
842 psin = (struct sockaddr_in *)&greqs.gsr_group;
843 mreq.imr_multiaddr = psin->sin_addr;
844 mreq.imr_address.s_addr = 0;
845 mreq.imr_ifindex = greqs.gsr_interface;
846 err = ip_mc_join_group(sk, &mreq);
847 if (err && err != -EADDRINUSE)
849 greqs.gsr_interface = mreq.imr_ifindex;
850 omode = MCAST_INCLUDE;
852 } else /* MCAST_LEAVE_SOURCE_GROUP */ {
853 omode = MCAST_INCLUDE;
856 err = ip_mc_source(add, omode, sk, &mreqs,
857 greqs.gsr_interface);
862 struct sockaddr_in *psin;
863 struct ip_msfilter *msf = NULL;
864 struct group_filter *gsf = NULL;
865 int msize, i, ifindex;
867 if (optlen < GROUP_FILTER_SIZE(0))
869 if (optlen > sysctl_optmem_max) {
873 gsf = kmalloc(optlen, GFP_KERNEL);
879 if (copy_from_user(gsf, optval, optlen))
882 /* numsrc >= (4G-140)/128 overflow in 32 bits */
883 if (gsf->gf_numsrc >= 0x1ffffff ||
884 gsf->gf_numsrc > sysctl_igmp_max_msf) {
888 if (GROUP_FILTER_SIZE(gsf->gf_numsrc) > optlen) {
892 msize = IP_MSFILTER_SIZE(gsf->gf_numsrc);
893 msf = kmalloc(msize, GFP_KERNEL);
898 ifindex = gsf->gf_interface;
899 psin = (struct sockaddr_in *)&gsf->gf_group;
900 if (psin->sin_family != AF_INET) {
901 err = -EADDRNOTAVAIL;
904 msf->imsf_multiaddr = psin->sin_addr.s_addr;
905 msf->imsf_interface = 0;
906 msf->imsf_fmode = gsf->gf_fmode;
907 msf->imsf_numsrc = gsf->gf_numsrc;
908 err = -EADDRNOTAVAIL;
909 for (i = 0; i < gsf->gf_numsrc; ++i) {
910 psin = (struct sockaddr_in *)&gsf->gf_slist[i];
912 if (psin->sin_family != AF_INET)
914 msf->imsf_slist[i] = psin->sin_addr.s_addr;
919 err = ip_mc_msfilter(sk, msf, ifindex);
925 case IP_MULTICAST_ALL:
928 if (val != 0 && val != 1)
932 case IP_ROUTER_ALERT:
933 err = ip_ra_control(sk, val ? 1 : 0, NULL);
939 inet->freebind = !!val;
942 case IP_IPSEC_POLICY:
945 if (!capable(CAP_NET_ADMIN))
947 err = xfrm_user_policy(sk, optname, optval, optlen);
951 if (!capable(CAP_NET_ADMIN)) {
957 inet->transparent = !!val;
963 if (val < 0 || val > 255)
981 * ip_queue_rcv_skb - Queue an skb into sock receive queue
985 * Queues an skb into socket receive queue. If IP_CMSG_PKTINFO option
986 * is not set, we drop skb dst entry now, while dst cache line is hot.
988 int ip_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
990 if (!(inet_sk(sk)->cmsg_flags & IP_CMSG_PKTINFO))
992 return sock_queue_rcv_skb(sk, skb);
994 EXPORT_SYMBOL(ip_queue_rcv_skb);
996 int ip_setsockopt(struct sock *sk, int level,
997 int optname, char __user *optval, unsigned int optlen)
1001 if (level != SOL_IP)
1002 return -ENOPROTOOPT;
1004 err = do_ip_setsockopt(sk, level, optname, optval, optlen);
1005 #ifdef CONFIG_NETFILTER
1006 /* we need to exclude all possible ENOPROTOOPTs except default case */
1007 if (err == -ENOPROTOOPT && optname != IP_HDRINCL &&
1008 optname != IP_IPSEC_POLICY &&
1009 optname != IP_XFRM_POLICY &&
1010 !ip_mroute_opt(optname)) {
1012 err = nf_setsockopt(sk, PF_INET, optname, optval, optlen);
1018 EXPORT_SYMBOL(ip_setsockopt);
1020 #ifdef CONFIG_COMPAT
1021 int compat_ip_setsockopt(struct sock *sk, int level, int optname,
1022 char __user *optval, unsigned int optlen)
1026 if (level != SOL_IP)
1027 return -ENOPROTOOPT;
1029 if (optname >= MCAST_JOIN_GROUP && optname <= MCAST_MSFILTER)
1030 return compat_mc_setsockopt(sk, level, optname, optval, optlen,
1033 err = do_ip_setsockopt(sk, level, optname, optval, optlen);
1034 #ifdef CONFIG_NETFILTER
1035 /* we need to exclude all possible ENOPROTOOPTs except default case */
1036 if (err == -ENOPROTOOPT && optname != IP_HDRINCL &&
1037 optname != IP_IPSEC_POLICY &&
1038 optname != IP_XFRM_POLICY &&
1039 !ip_mroute_opt(optname)) {
1041 err = compat_nf_setsockopt(sk, PF_INET, optname,
1048 EXPORT_SYMBOL(compat_ip_setsockopt);
1052 * Get the options. Note for future reference. The GET of IP options gets
1053 * the _received_ ones. The set sets the _sent_ ones.
1056 static int do_ip_getsockopt(struct sock *sk, int level, int optname,
1057 char __user *optval, int __user *optlen)
1059 struct inet_sock *inet = inet_sk(sk);
1063 if (level != SOL_IP)
1066 if (ip_mroute_opt(optname))
1067 return ip_mroute_getsockopt(sk, optname, optval, optlen);
1069 if (get_user(len, optlen))
1079 unsigned char optbuf[sizeof(struct ip_options)+40];
1080 struct ip_options * opt = (struct ip_options *)optbuf;
1083 memcpy(optbuf, inet->opt,
1084 sizeof(struct ip_options)+
1088 if (opt->optlen == 0)
1089 return put_user(0, optlen);
1091 ip_options_undo(opt);
1093 len = min_t(unsigned int, len, opt->optlen);
1094 if (put_user(len, optlen))
1096 if (copy_to_user(optval, opt->__data, len))
1101 val = (inet->cmsg_flags & IP_CMSG_PKTINFO) != 0;
1104 val = (inet->cmsg_flags & IP_CMSG_TTL) != 0;
1107 val = (inet->cmsg_flags & IP_CMSG_TOS) != 0;
1110 val = (inet->cmsg_flags & IP_CMSG_RECVOPTS) != 0;
1113 val = (inet->cmsg_flags & IP_CMSG_RETOPTS) != 0;
1116 val = (inet->cmsg_flags & IP_CMSG_PASSSEC) != 0;
1118 case IP_RECVORIGDSTADDR:
1119 val = (inet->cmsg_flags & IP_CMSG_ORIGDSTADDR) != 0;
1125 val = (inet->uc_ttl == -1 ?
1126 sysctl_ip_default_ttl :
1130 val = inet->hdrincl;
1132 case IP_MTU_DISCOVER:
1133 val = inet->pmtudisc;
1137 struct dst_entry *dst;
1139 dst = sk_dst_get(sk);
1151 val = inet->recverr;
1153 case IP_MULTICAST_TTL:
1156 case IP_MULTICAST_LOOP:
1157 val = inet->mc_loop;
1159 case IP_MULTICAST_IF:
1161 struct in_addr addr;
1162 len = min_t(unsigned int, len, sizeof(struct in_addr));
1163 addr.s_addr = inet->mc_addr;
1166 if (put_user(len, optlen))
1168 if (copy_to_user(optval, &addr, len))
1174 struct ip_msfilter msf;
1177 if (len < IP_MSFILTER_SIZE(0)) {
1181 if (copy_from_user(&msf, optval, IP_MSFILTER_SIZE(0))) {
1185 err = ip_mc_msfget(sk, &msf,
1186 (struct ip_msfilter __user *)optval, optlen);
1190 case MCAST_MSFILTER:
1192 struct group_filter gsf;
1195 if (len < GROUP_FILTER_SIZE(0)) {
1199 if (copy_from_user(&gsf, optval, GROUP_FILTER_SIZE(0))) {
1203 err = ip_mc_gsfget(sk, &gsf,
1204 (struct group_filter __user *)optval,
1209 case IP_MULTICAST_ALL:
1218 if (sk->sk_type != SOCK_STREAM)
1219 return -ENOPROTOOPT;
1221 msg.msg_control = optval;
1222 msg.msg_controllen = len;
1225 if (inet->cmsg_flags & IP_CMSG_PKTINFO) {
1226 struct in_pktinfo info;
1228 info.ipi_addr.s_addr = inet->inet_rcv_saddr;
1229 info.ipi_spec_dst.s_addr = inet->inet_rcv_saddr;
1230 info.ipi_ifindex = inet->mc_index;
1231 put_cmsg(&msg, SOL_IP, IP_PKTINFO, sizeof(info), &info);
1233 if (inet->cmsg_flags & IP_CMSG_TTL) {
1234 int hlim = inet->mc_ttl;
1235 put_cmsg(&msg, SOL_IP, IP_TTL, sizeof(hlim), &hlim);
1237 len -= msg.msg_controllen;
1238 return put_user(len, optlen);
1241 val = inet->freebind;
1243 case IP_TRANSPARENT:
1244 val = inet->transparent;
1247 val = inet->min_ttl;
1251 return -ENOPROTOOPT;
1255 if (len < sizeof(int) && len > 0 && val >= 0 && val <= 255) {
1256 unsigned char ucval = (unsigned char)val;
1258 if (put_user(len, optlen))
1260 if (copy_to_user(optval, &ucval, 1))
1263 len = min_t(unsigned int, sizeof(int), len);
1264 if (put_user(len, optlen))
1266 if (copy_to_user(optval, &val, len))
1272 int ip_getsockopt(struct sock *sk, int level,
1273 int optname, char __user *optval, int __user *optlen)
1277 err = do_ip_getsockopt(sk, level, optname, optval, optlen);
1278 #ifdef CONFIG_NETFILTER
1279 /* we need to exclude all possible ENOPROTOOPTs except default case */
1280 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS &&
1281 !ip_mroute_opt(optname)) {
1284 if (get_user(len, optlen))
1288 err = nf_getsockopt(sk, PF_INET, optname, optval,
1292 err = put_user(len, optlen);
1298 EXPORT_SYMBOL(ip_getsockopt);
1300 #ifdef CONFIG_COMPAT
1301 int compat_ip_getsockopt(struct sock *sk, int level, int optname,
1302 char __user *optval, int __user *optlen)
1306 if (optname == MCAST_MSFILTER)
1307 return compat_mc_getsockopt(sk, level, optname, optval, optlen,
1310 err = do_ip_getsockopt(sk, level, optname, optval, optlen);
1312 #ifdef CONFIG_NETFILTER
1313 /* we need to exclude all possible ENOPROTOOPTs except default case */
1314 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS &&
1315 !ip_mroute_opt(optname)) {
1318 if (get_user(len, optlen))
1322 err = compat_nf_getsockopt(sk, PF_INET, optname, optval, &len);
1325 err = put_user(len, optlen);
1331 EXPORT_SYMBOL(compat_ip_getsockopt);