3 * Linux INET6 implementation
6 * Pedro Roque <roque@di.fc.ul.pt>
8 * $Id: datagram.c,v 1.24 2002/02/01 22:01:04 davem Exp $
10 * This program is free software; you can redistribute it and/or
11 * modify it under the terms of the GNU General Public License
12 * as published by the Free Software Foundation; either version
13 * 2 of the License, or (at your option) any later version.
16 #include <linux/errno.h>
17 #include <linux/types.h>
18 #include <linux/kernel.h>
19 #include <linux/sched.h>
20 #include <linux/interrupt.h>
21 #include <linux/socket.h>
22 #include <linux/sockios.h>
23 #include <linux/in6.h>
24 #include <linux/ipv6.h>
25 #include <linux/route.h>
28 #include <net/ndisc.h>
29 #include <net/addrconf.h>
30 #include <net/transp_v6.h>
31 #include <net/ip6_route.h>
32 #include <net/tcp_states.h>
34 #include <linux/errqueue.h>
35 #include <asm/uaccess.h>
37 int ip6_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
39 struct sockaddr_in6 *usin = (struct sockaddr_in6 *) uaddr;
40 struct inet_sock *inet = inet_sk(sk);
41 struct ipv6_pinfo *np = inet6_sk(sk);
42 struct in6_addr *daddr, *final_p = NULL, final;
43 struct dst_entry *dst;
45 struct ip6_flowlabel *flowlabel = NULL;
49 if (usin->sin6_family == AF_INET) {
50 if (__ipv6_only_sock(sk))
52 err = ip4_datagram_connect(sk, uaddr, addr_len);
56 if (addr_len < SIN6_LEN_RFC2133)
59 if (usin->sin6_family != AF_INET6)
62 memset(&fl, 0, sizeof(fl));
64 fl.fl6_flowlabel = usin->sin6_flowinfo&IPV6_FLOWINFO_MASK;
65 if (fl.fl6_flowlabel&IPV6_FLOWLABEL_MASK) {
66 flowlabel = fl6_sock_lookup(sk, fl.fl6_flowlabel);
67 if (flowlabel == NULL)
69 ipv6_addr_copy(&usin->sin6_addr, &flowlabel->dst);
73 addr_type = ipv6_addr_type(&usin->sin6_addr);
75 if (addr_type == IPV6_ADDR_ANY) {
79 usin->sin6_addr.s6_addr[15] = 0x01;
82 daddr = &usin->sin6_addr;
84 if (addr_type == IPV6_ADDR_MAPPED) {
85 struct sockaddr_in sin;
87 if (__ipv6_only_sock(sk)) {
91 sin.sin_family = AF_INET;
92 sin.sin_addr.s_addr = daddr->s6_addr32[3];
93 sin.sin_port = usin->sin6_port;
95 err = ip4_datagram_connect(sk,
96 (struct sockaddr*) &sin,
103 ipv6_addr_set(&np->daddr, 0, 0, htonl(0x0000ffff), inet->daddr);
105 if (ipv6_addr_any(&np->saddr)) {
106 ipv6_addr_set(&np->saddr, 0, 0, htonl(0x0000ffff),
110 if (ipv6_addr_any(&np->rcv_saddr)) {
111 ipv6_addr_set(&np->rcv_saddr, 0, 0, htonl(0x0000ffff),
117 if (addr_type&IPV6_ADDR_LINKLOCAL) {
118 if (addr_len >= sizeof(struct sockaddr_in6) &&
119 usin->sin6_scope_id) {
120 if (sk->sk_bound_dev_if &&
121 sk->sk_bound_dev_if != usin->sin6_scope_id) {
125 sk->sk_bound_dev_if = usin->sin6_scope_id;
126 if (!sk->sk_bound_dev_if &&
127 (addr_type & IPV6_ADDR_MULTICAST))
128 fl.oif = np->mcast_oif;
131 /* Connect to link-local address requires an interface */
132 if (!sk->sk_bound_dev_if) {
138 ipv6_addr_copy(&np->daddr, daddr);
139 np->flow_label = fl.fl6_flowlabel;
141 inet->dport = usin->sin6_port;
144 * Check for a route to destination an obtain the
145 * destination cache for it.
148 fl.proto = sk->sk_protocol;
149 ipv6_addr_copy(&fl.fl6_dst, &np->daddr);
150 ipv6_addr_copy(&fl.fl6_src, &np->saddr);
151 fl.oif = sk->sk_bound_dev_if;
152 fl.fl_ip_dport = inet->dport;
153 fl.fl_ip_sport = inet->sport;
155 if (!fl.oif && (addr_type&IPV6_ADDR_MULTICAST))
156 fl.oif = np->mcast_oif;
159 if (flowlabel->opt && flowlabel->opt->srcrt) {
160 struct rt0_hdr *rt0 = (struct rt0_hdr *) flowlabel->opt->srcrt;
161 ipv6_addr_copy(&final, &fl.fl6_dst);
162 ipv6_addr_copy(&fl.fl6_dst, rt0->addr);
165 } else if (np->opt && np->opt->srcrt) {
166 struct rt0_hdr *rt0 = (struct rt0_hdr *)np->opt->srcrt;
167 ipv6_addr_copy(&final, &fl.fl6_dst);
168 ipv6_addr_copy(&fl.fl6_dst, rt0->addr);
172 err = ip6_dst_lookup(sk, &dst, &fl);
176 ipv6_addr_copy(&fl.fl6_dst, final_p);
178 if ((err = xfrm_lookup(&dst, &fl, sk, 0)) < 0) {
183 /* source address lookup done in ip6_dst_lookup */
185 if (ipv6_addr_any(&np->saddr))
186 ipv6_addr_copy(&np->saddr, &fl.fl6_src);
188 if (ipv6_addr_any(&np->rcv_saddr)) {
189 ipv6_addr_copy(&np->rcv_saddr, &fl.fl6_src);
190 inet->rcv_saddr = LOOPBACK4_IPV6;
193 ip6_dst_store(sk, dst,
194 ipv6_addr_equal(&fl.fl6_dst, &np->daddr) ?
197 sk->sk_state = TCP_ESTABLISHED;
199 fl6_sock_release(flowlabel);
203 void ipv6_icmp_error(struct sock *sk, struct sk_buff *skb, int err,
204 u16 port, u32 info, u8 *payload)
206 struct ipv6_pinfo *np = inet6_sk(sk);
207 struct icmp6hdr *icmph = (struct icmp6hdr *)skb->h.raw;
208 struct sock_exterr_skb *serr;
213 skb = skb_clone(skb, GFP_ATOMIC);
217 serr = SKB_EXT_ERR(skb);
218 serr->ee.ee_errno = err;
219 serr->ee.ee_origin = SO_EE_ORIGIN_ICMP6;
220 serr->ee.ee_type = icmph->icmp6_type;
221 serr->ee.ee_code = icmph->icmp6_code;
223 serr->ee.ee_info = info;
224 serr->ee.ee_data = 0;
225 serr->addr_offset = (u8*)&(((struct ipv6hdr*)(icmph+1))->daddr) - skb->nh.raw;
228 skb->h.raw = payload;
229 __skb_pull(skb, payload - skb->data);
231 if (sock_queue_err_skb(sk, skb))
235 void ipv6_local_error(struct sock *sk, int err, struct flowi *fl, u32 info)
237 struct ipv6_pinfo *np = inet6_sk(sk);
238 struct sock_exterr_skb *serr;
245 skb = alloc_skb(sizeof(struct ipv6hdr), GFP_ATOMIC);
249 iph = (struct ipv6hdr*)skb_put(skb, sizeof(struct ipv6hdr));
251 ipv6_addr_copy(&iph->daddr, &fl->fl6_dst);
253 serr = SKB_EXT_ERR(skb);
254 serr->ee.ee_errno = err;
255 serr->ee.ee_origin = SO_EE_ORIGIN_LOCAL;
256 serr->ee.ee_type = 0;
257 serr->ee.ee_code = 0;
259 serr->ee.ee_info = info;
260 serr->ee.ee_data = 0;
261 serr->addr_offset = (u8*)&iph->daddr - skb->nh.raw;
262 serr->port = fl->fl_ip_dport;
264 skb->h.raw = skb->tail;
265 __skb_pull(skb, skb->tail - skb->data);
267 if (sock_queue_err_skb(sk, skb))
272 * Handle MSG_ERRQUEUE
274 int ipv6_recv_error(struct sock *sk, struct msghdr *msg, int len)
276 struct ipv6_pinfo *np = inet6_sk(sk);
277 struct sock_exterr_skb *serr;
278 struct sk_buff *skb, *skb2;
279 struct sockaddr_in6 *sin;
281 struct sock_extended_err ee;
282 struct sockaddr_in6 offender;
288 skb = skb_dequeue(&sk->sk_error_queue);
294 msg->msg_flags |= MSG_TRUNC;
297 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
301 sock_recv_timestamp(msg, sk, skb);
303 serr = SKB_EXT_ERR(skb);
305 sin = (struct sockaddr_in6 *)msg->msg_name;
307 sin->sin6_family = AF_INET6;
308 sin->sin6_flowinfo = 0;
309 sin->sin6_port = serr->port;
310 sin->sin6_scope_id = 0;
311 if (serr->ee.ee_origin == SO_EE_ORIGIN_ICMP6) {
312 ipv6_addr_copy(&sin->sin6_addr,
313 (struct in6_addr *)(skb->nh.raw + serr->addr_offset));
315 sin->sin6_flowinfo = *(u32*)(skb->nh.raw + serr->addr_offset - 24) & IPV6_FLOWINFO_MASK;
316 if (ipv6_addr_type(&sin->sin6_addr) & IPV6_ADDR_LINKLOCAL)
317 sin->sin6_scope_id = IP6CB(skb)->iif;
319 ipv6_addr_set(&sin->sin6_addr, 0, 0,
321 *(u32*)(skb->nh.raw + serr->addr_offset));
325 memcpy(&errhdr.ee, &serr->ee, sizeof(struct sock_extended_err));
326 sin = &errhdr.offender;
327 sin->sin6_family = AF_UNSPEC;
328 if (serr->ee.ee_origin != SO_EE_ORIGIN_LOCAL) {
329 sin->sin6_family = AF_INET6;
330 sin->sin6_flowinfo = 0;
331 sin->sin6_scope_id = 0;
332 if (serr->ee.ee_origin == SO_EE_ORIGIN_ICMP6) {
333 ipv6_addr_copy(&sin->sin6_addr, &skb->nh.ipv6h->saddr);
335 datagram_recv_ctl(sk, msg, skb);
336 if (ipv6_addr_type(&sin->sin6_addr) & IPV6_ADDR_LINKLOCAL)
337 sin->sin6_scope_id = IP6CB(skb)->iif;
339 struct inet_sock *inet = inet_sk(sk);
341 ipv6_addr_set(&sin->sin6_addr, 0, 0,
344 if (inet->cmsg_flags)
345 ip_cmsg_recv(msg, skb);
349 put_cmsg(msg, SOL_IPV6, IPV6_RECVERR, sizeof(errhdr), &errhdr);
351 /* Now we could try to dump offended packet options */
353 msg->msg_flags |= MSG_ERRQUEUE;
356 /* Reset and regenerate socket error */
357 spin_lock_bh(&sk->sk_error_queue.lock);
359 if ((skb2 = skb_peek(&sk->sk_error_queue)) != NULL) {
360 sk->sk_err = SKB_EXT_ERR(skb2)->ee.ee_errno;
361 spin_unlock_bh(&sk->sk_error_queue.lock);
362 sk->sk_error_report(sk);
364 spin_unlock_bh(&sk->sk_error_queue.lock);
375 int datagram_recv_ctl(struct sock *sk, struct msghdr *msg, struct sk_buff *skb)
377 struct ipv6_pinfo *np = inet6_sk(sk);
378 struct inet6_skb_parm *opt = IP6CB(skb);
380 if (np->rxopt.bits.rxinfo) {
381 struct in6_pktinfo src_info;
383 src_info.ipi6_ifindex = opt->iif;
384 ipv6_addr_copy(&src_info.ipi6_addr, &skb->nh.ipv6h->daddr);
385 put_cmsg(msg, SOL_IPV6, IPV6_PKTINFO, sizeof(src_info), &src_info);
388 if (np->rxopt.bits.rxhlim) {
389 int hlim = skb->nh.ipv6h->hop_limit;
390 put_cmsg(msg, SOL_IPV6, IPV6_HOPLIMIT, sizeof(hlim), &hlim);
393 if (np->rxopt.bits.rxflow && (*(u32*)skb->nh.raw & IPV6_FLOWINFO_MASK)) {
394 u32 flowinfo = *(u32*)skb->nh.raw & IPV6_FLOWINFO_MASK;
395 put_cmsg(msg, SOL_IPV6, IPV6_FLOWINFO, sizeof(flowinfo), &flowinfo);
397 if (np->rxopt.bits.hopopts && opt->hop) {
398 u8 *ptr = skb->nh.raw + opt->hop;
399 put_cmsg(msg, SOL_IPV6, IPV6_HOPOPTS, (ptr[1]+1)<<3, ptr);
401 if (np->rxopt.bits.dstopts && opt->dst0) {
402 u8 *ptr = skb->nh.raw + opt->dst0;
403 put_cmsg(msg, SOL_IPV6, IPV6_DSTOPTS, (ptr[1]+1)<<3, ptr);
405 if (np->rxopt.bits.srcrt && opt->srcrt) {
406 struct ipv6_rt_hdr *rthdr = (struct ipv6_rt_hdr *)(skb->nh.raw + opt->srcrt);
407 put_cmsg(msg, SOL_IPV6, IPV6_RTHDR, (rthdr->hdrlen+1) << 3, rthdr);
409 if (np->rxopt.bits.dstopts && opt->dst1) {
410 u8 *ptr = skb->nh.raw + opt->dst1;
411 put_cmsg(msg, SOL_IPV6, IPV6_DSTOPTS, (ptr[1]+1)<<3, ptr);
416 int datagram_send_ctl(struct msghdr *msg, struct flowi *fl,
417 struct ipv6_txoptions *opt,
420 struct in6_pktinfo *src_info;
421 struct cmsghdr *cmsg;
422 struct ipv6_rt_hdr *rthdr;
423 struct ipv6_opt_hdr *hdr;
427 for (cmsg = CMSG_FIRSTHDR(msg); cmsg; cmsg = CMSG_NXTHDR(msg, cmsg)) {
429 struct net_device *dev = NULL;
431 if (!CMSG_OK(msg, cmsg)) {
436 if (cmsg->cmsg_level != SOL_IPV6)
439 switch (cmsg->cmsg_type) {
441 if (cmsg->cmsg_len < CMSG_LEN(sizeof(struct in6_pktinfo))) {
446 src_info = (struct in6_pktinfo *)CMSG_DATA(cmsg);
448 if (src_info->ipi6_ifindex) {
449 if (fl->oif && src_info->ipi6_ifindex != fl->oif)
451 fl->oif = src_info->ipi6_ifindex;
454 addr_type = ipv6_addr_type(&src_info->ipi6_addr);
456 if (addr_type == IPV6_ADDR_ANY)
459 if (addr_type & IPV6_ADDR_LINKLOCAL) {
460 if (!src_info->ipi6_ifindex)
463 dev = dev_get_by_index(src_info->ipi6_ifindex);
468 if (!ipv6_chk_addr(&src_info->ipi6_addr, dev, 0)) {
477 ipv6_addr_copy(&fl->fl6_src, &src_info->ipi6_addr);
481 if (cmsg->cmsg_len < CMSG_LEN(4)) {
486 if (fl->fl6_flowlabel&IPV6_FLOWINFO_MASK) {
487 if ((fl->fl6_flowlabel^*(u32 *)CMSG_DATA(cmsg))&~IPV6_FLOWINFO_MASK) {
492 fl->fl6_flowlabel = IPV6_FLOWINFO_MASK & *(u32 *)CMSG_DATA(cmsg);
496 if (opt->hopopt || cmsg->cmsg_len < CMSG_LEN(sizeof(struct ipv6_opt_hdr))) {
501 hdr = (struct ipv6_opt_hdr *)CMSG_DATA(cmsg);
502 len = ((hdr->hdrlen + 1) << 3);
503 if (cmsg->cmsg_len < CMSG_LEN(len)) {
507 if (!capable(CAP_NET_RAW)) {
511 opt->opt_nflen += len;
516 if (cmsg->cmsg_len < CMSG_LEN(sizeof(struct ipv6_opt_hdr))) {
521 hdr = (struct ipv6_opt_hdr *)CMSG_DATA(cmsg);
522 len = ((hdr->hdrlen + 1) << 3);
523 if (cmsg->cmsg_len < CMSG_LEN(len)) {
527 if (!capable(CAP_NET_RAW)) {
535 opt->opt_flen += len;
540 if (cmsg->cmsg_len < CMSG_LEN(sizeof(struct ipv6_rt_hdr))) {
545 rthdr = (struct ipv6_rt_hdr *)CMSG_DATA(cmsg);
555 len = ((rthdr->hdrlen + 1) << 3);
557 if (cmsg->cmsg_len < CMSG_LEN(len)) {
562 /* segments left must also match */
563 if ((rthdr->hdrlen >> 1) != rthdr->segments_left) {
568 opt->opt_nflen += len;
572 int dsthdrlen = ((opt->dst1opt->hdrlen+1)<<3);
574 opt->opt_nflen += dsthdrlen;
575 opt->dst0opt = opt->dst1opt;
577 opt->opt_flen -= dsthdrlen;
583 if (cmsg->cmsg_len != CMSG_LEN(sizeof(int))) {
588 *hlimit = *(int *)CMSG_DATA(cmsg);
592 LIMIT_NETDEBUG(KERN_DEBUG "invalid cmsg type: %d\n",