1 // SPDX-License-Identifier: GPL-2.0-or-later
3 * Extension Header handling for IPv6
4 * Linux INET6 implementation
7 * Pedro Roque <roque@di.fc.ul.pt>
8 * Andi Kleen <ak@muc.de>
9 * Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
13 * yoshfuji : ensure not to overrun while parsing
15 * Mitsuru KANDA @USAGI and: Remove ipv6_parse_exthdrs().
16 * YOSHIFUJI Hideaki @USAGI Register inbound extension header
17 * handlers as inet6_protocol{}.
20 #include <linux/errno.h>
21 #include <linux/types.h>
22 #include <linux/socket.h>
23 #include <linux/sockios.h>
24 #include <linux/net.h>
25 #include <linux/netdevice.h>
26 #include <linux/in6.h>
27 #include <linux/icmpv6.h>
28 #include <linux/slab.h>
29 #include <linux/export.h>
36 #include <net/protocol.h>
37 #include <net/transp_v6.h>
38 #include <net/rawv6.h>
39 #include <net/ndisc.h>
40 #include <net/ip6_route.h>
41 #include <net/addrconf.h>
42 #include <net/calipso.h>
43 #if IS_ENABLED(CONFIG_IPV6_MIP6)
46 #include <linux/seg6.h>
48 #ifdef CONFIG_IPV6_SEG6_HMAC
49 #include <net/seg6_hmac.h>
52 #include <linux/ioam6.h>
53 #include <net/ioam6.h>
54 #include <net/dst_metadata.h>
56 #include <linux/uaccess.h>
58 /*********************
60 *********************/
62 /* An unknown option is detected, decide what to do */
64 static bool ip6_tlvopt_unknown(struct sk_buff *skb, int optoff,
65 bool disallow_unknowns)
67 if (disallow_unknowns) {
68 /* If unknown TLVs are disallowed by configuration
69 * then always silently drop packet. Note this also
70 * means no ICMP parameter problem is sent which
71 * could be a good property to mitigate a reflection DOS
78 switch ((skb_network_header(skb)[optoff] & 0xC0) >> 6) {
82 case 1: /* drop packet */
85 case 3: /* Send ICMP if not a multicast address and drop packet */
86 /* Actually, it is redundant check. icmp_send
87 will recheck in any case.
89 if (ipv6_addr_is_multicast(&ipv6_hdr(skb)->daddr))
92 case 2: /* send ICMP PARM PROB regardless and drop packet */
93 icmpv6_param_prob_reason(skb, ICMPV6_UNK_OPTION, optoff,
94 SKB_DROP_REASON_UNHANDLED_PROTO);
99 kfree_skb_reason(skb, SKB_DROP_REASON_UNHANDLED_PROTO);
103 static bool ipv6_hop_ra(struct sk_buff *skb, int optoff);
104 static bool ipv6_hop_ioam(struct sk_buff *skb, int optoff);
105 static bool ipv6_hop_jumbo(struct sk_buff *skb, int optoff);
106 static bool ipv6_hop_calipso(struct sk_buff *skb, int optoff);
107 #if IS_ENABLED(CONFIG_IPV6_MIP6)
108 static bool ipv6_dest_hao(struct sk_buff *skb, int optoff);
111 /* Parse tlv encoded option header (hop-by-hop or destination) */
113 static bool ip6_parse_tlv(bool hopbyhop,
117 int len = (skb_transport_header(skb)[1] + 1) << 3;
118 const unsigned char *nh = skb_network_header(skb);
119 int off = skb_network_header_len(skb);
120 bool disallow_unknowns = false;
124 if (unlikely(max_count < 0)) {
125 disallow_unknowns = true;
126 max_count = -max_count;
135 if (nh[off] == IPV6_TLV_PAD1) {
145 optlen = nh[off + 1] + 2;
149 if (nh[off] == IPV6_TLV_PADN) {
150 /* RFC 2460 states that the purpose of PadN is
151 * to align the containing header to multiples
152 * of 8. 7 is therefore the highest valid value.
153 * See also RFC 4942, Section 2.1.9.5.
158 /* RFC 4942 recommends receiving hosts to
159 * actively check PadN payload to contain
162 for (i = 2; i < optlen; i++) {
163 if (nh[off + i] != 0)
168 if (tlv_count > max_count)
173 case IPV6_TLV_ROUTERALERT:
174 if (!ipv6_hop_ra(skb, off))
178 if (!ipv6_hop_ioam(skb, off))
181 nh = skb_network_header(skb);
184 if (!ipv6_hop_jumbo(skb, off))
187 case IPV6_TLV_CALIPSO:
188 if (!ipv6_hop_calipso(skb, off))
192 if (!ip6_tlvopt_unknown(skb, off,
199 #if IS_ENABLED(CONFIG_IPV6_MIP6)
201 if (!ipv6_dest_hao(skb, off))
206 if (!ip6_tlvopt_unknown(skb, off,
221 kfree_skb_reason(skb, SKB_DROP_REASON_IP_INHDR);
225 /*****************************
226 Destination options header.
227 *****************************/
229 #if IS_ENABLED(CONFIG_IPV6_MIP6)
230 static bool ipv6_dest_hao(struct sk_buff *skb, int optoff)
232 struct ipv6_destopt_hao *hao;
233 struct inet6_skb_parm *opt = IP6CB(skb);
234 struct ipv6hdr *ipv6h = ipv6_hdr(skb);
239 net_dbg_ratelimited("hao duplicated\n");
242 opt->dsthao = opt->dst1;
245 hao = (struct ipv6_destopt_hao *)(skb_network_header(skb) + optoff);
247 if (hao->length != 16) {
248 net_dbg_ratelimited("hao invalid option length = %d\n",
250 SKB_DR_SET(reason, IP_INHDR);
254 if (!(ipv6_addr_type(&hao->addr) & IPV6_ADDR_UNICAST)) {
255 net_dbg_ratelimited("hao is not an unicast addr: %pI6\n",
257 SKB_DR_SET(reason, INVALID_PROTO);
261 ret = xfrm6_input_addr(skb, (xfrm_address_t *)&ipv6h->daddr,
262 (xfrm_address_t *)&hao->addr, IPPROTO_DSTOPTS);
263 if (unlikely(ret < 0)) {
264 SKB_DR_SET(reason, XFRM_POLICY);
268 if (skb_cloned(skb)) {
269 if (pskb_expand_head(skb, 0, 0, GFP_ATOMIC))
272 /* update all variable using below by copied skbuff */
273 hao = (struct ipv6_destopt_hao *)(skb_network_header(skb) +
275 ipv6h = ipv6_hdr(skb);
278 if (skb->ip_summed == CHECKSUM_COMPLETE)
279 skb->ip_summed = CHECKSUM_NONE;
281 swap(ipv6h->saddr, hao->addr);
283 if (skb->tstamp == 0)
284 __net_timestamp(skb);
289 kfree_skb_reason(skb, reason);
294 static int ipv6_destopt_rcv(struct sk_buff *skb)
296 struct inet6_dev *idev = __in6_dev_get(skb->dev);
297 struct inet6_skb_parm *opt = IP6CB(skb);
298 #if IS_ENABLED(CONFIG_IPV6_MIP6)
301 struct dst_entry *dst = skb_dst(skb);
302 struct net *net = dev_net(skb->dev);
305 if (!pskb_may_pull(skb, skb_transport_offset(skb) + 8) ||
306 !pskb_may_pull(skb, (skb_transport_offset(skb) +
307 ((skb_transport_header(skb)[1] + 1) << 3)))) {
308 __IP6_INC_STATS(dev_net(dst->dev), idev,
309 IPSTATS_MIB_INHDRERRORS);
315 extlen = (skb_transport_header(skb)[1] + 1) << 3;
316 if (extlen > net->ipv6.sysctl.max_dst_opts_len)
319 opt->lastopt = opt->dst1 = skb_network_header_len(skb);
320 #if IS_ENABLED(CONFIG_IPV6_MIP6)
324 if (ip6_parse_tlv(false, skb, net->ipv6.sysctl.max_dst_opts_cnt)) {
325 skb->transport_header += extlen;
327 #if IS_ENABLED(CONFIG_IPV6_MIP6)
330 opt->nhoff = opt->dst1;
335 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INHDRERRORS);
339 static void seg6_update_csum(struct sk_buff *skb)
341 struct ipv6_sr_hdr *hdr;
342 struct in6_addr *addr;
345 /* srh is at transport offset and seg_left is already decremented
346 * but daddr is not yet updated with next segment
349 hdr = (struct ipv6_sr_hdr *)skb_transport_header(skb);
350 addr = hdr->segments + hdr->segments_left;
352 hdr->segments_left++;
353 from = *(__be32 *)hdr;
355 hdr->segments_left--;
358 /* update skb csum with diff resulting from seg_left decrement */
360 update_csum_diff4(skb, from, to);
362 /* compute csum diff between current and next segment and update */
364 update_csum_diff16(skb, (__be32 *)(&ipv6_hdr(skb)->daddr),
368 static int ipv6_srh_rcv(struct sk_buff *skb)
370 struct inet6_skb_parm *opt = IP6CB(skb);
371 struct net *net = dev_net(skb->dev);
372 struct ipv6_sr_hdr *hdr;
373 struct inet6_dev *idev;
374 struct in6_addr *addr;
377 hdr = (struct ipv6_sr_hdr *)skb_transport_header(skb);
379 idev = __in6_dev_get(skb->dev);
381 accept_seg6 = net->ipv6.devconf_all->seg6_enabled;
382 if (accept_seg6 > idev->cnf.seg6_enabled)
383 accept_seg6 = idev->cnf.seg6_enabled;
390 #ifdef CONFIG_IPV6_SEG6_HMAC
391 if (!seg6_hmac_validate_skb(skb)) {
398 if (hdr->segments_left == 0) {
399 if (hdr->nexthdr == NEXTHDR_IPV6 || hdr->nexthdr == NEXTHDR_IPV4) {
400 int offset = (hdr->hdrlen + 1) << 3;
402 skb_postpull_rcsum(skb, skb_network_header(skb),
403 skb_network_header_len(skb));
404 skb_pull(skb, offset);
405 skb_postpull_rcsum(skb, skb_transport_header(skb),
408 skb_reset_network_header(skb);
409 skb_reset_transport_header(skb);
410 skb->encapsulation = 0;
411 if (hdr->nexthdr == NEXTHDR_IPV4)
412 skb->protocol = htons(ETH_P_IP);
413 __skb_tunnel_rx(skb, skb->dev, net);
419 opt->srcrt = skb_network_header_len(skb);
420 opt->lastopt = opt->srcrt;
421 skb->transport_header += (hdr->hdrlen + 1) << 3;
422 opt->nhoff = (&hdr->nexthdr) - skb_network_header(skb);
427 if (hdr->segments_left >= (hdr->hdrlen >> 1)) {
428 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INHDRERRORS);
429 icmpv6_param_prob(skb, ICMPV6_HDR_FIELD,
430 ((&hdr->segments_left) -
431 skb_network_header(skb)));
435 if (skb_cloned(skb)) {
436 if (pskb_expand_head(skb, 0, 0, GFP_ATOMIC)) {
437 __IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
438 IPSTATS_MIB_OUTDISCARDS);
443 hdr = (struct ipv6_sr_hdr *)skb_transport_header(skb);
446 hdr->segments_left--;
447 addr = hdr->segments + hdr->segments_left;
449 skb_push(skb, sizeof(struct ipv6hdr));
451 if (skb->ip_summed == CHECKSUM_COMPLETE)
452 seg6_update_csum(skb);
454 ipv6_hdr(skb)->daddr = *addr;
456 ip6_route_input(skb);
458 if (skb_dst(skb)->error) {
463 if (skb_dst(skb)->dev->flags & IFF_LOOPBACK) {
464 if (ipv6_hdr(skb)->hop_limit <= 1) {
465 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INHDRERRORS);
466 icmpv6_send(skb, ICMPV6_TIME_EXCEED,
467 ICMPV6_EXC_HOPLIMIT, 0);
471 ipv6_hdr(skb)->hop_limit--;
473 skb_pull(skb, sizeof(struct ipv6hdr));
482 static int ipv6_rpl_srh_rcv(struct sk_buff *skb)
484 struct ipv6_rpl_sr_hdr *hdr, *ohdr, *chdr;
485 struct inet6_skb_parm *opt = IP6CB(skb);
486 struct net *net = dev_net(skb->dev);
487 struct inet6_dev *idev;
488 struct ipv6hdr *oldhdr;
495 idev = __in6_dev_get(skb->dev);
497 accept_rpl_seg = net->ipv6.devconf_all->rpl_seg_enabled;
498 if (accept_rpl_seg > idev->cnf.rpl_seg_enabled)
499 accept_rpl_seg = idev->cnf.rpl_seg_enabled;
501 if (!accept_rpl_seg) {
507 hdr = (struct ipv6_rpl_sr_hdr *)skb_transport_header(skb);
509 if (hdr->segments_left == 0) {
510 if (hdr->nexthdr == NEXTHDR_IPV6) {
511 int offset = (hdr->hdrlen + 1) << 3;
513 skb_postpull_rcsum(skb, skb_network_header(skb),
514 skb_network_header_len(skb));
515 skb_pull(skb, offset);
516 skb_postpull_rcsum(skb, skb_transport_header(skb),
519 skb_reset_network_header(skb);
520 skb_reset_transport_header(skb);
521 skb->encapsulation = 0;
523 __skb_tunnel_rx(skb, skb->dev, net);
529 opt->srcrt = skb_network_header_len(skb);
530 opt->lastopt = opt->srcrt;
531 skb->transport_header += (hdr->hdrlen + 1) << 3;
532 opt->nhoff = (&hdr->nexthdr) - skb_network_header(skb);
537 n = (hdr->hdrlen << 3) - hdr->pad - (16 - hdr->cmpre);
538 r = do_div(n, (16 - hdr->cmpri));
539 /* checks if calculation was without remainder and n fits into
540 * unsigned char which is segments_left field. Should not be
543 if (r || (n + 1) > 255) {
548 if (hdr->segments_left > n + 1) {
549 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INHDRERRORS);
550 icmpv6_param_prob(skb, ICMPV6_HDR_FIELD,
551 ((&hdr->segments_left) -
552 skb_network_header(skb)));
556 hdr->segments_left--;
557 i = n - hdr->segments_left;
559 buf = kcalloc(struct_size(hdr, segments.addr, n + 2), 2, GFP_ATOMIC);
560 if (unlikely(!buf)) {
565 ohdr = (struct ipv6_rpl_sr_hdr *)buf;
566 ipv6_rpl_srh_decompress(ohdr, hdr, &ipv6_hdr(skb)->daddr, n);
567 chdr = (struct ipv6_rpl_sr_hdr *)(buf + ((ohdr->hdrlen + 1) << 3));
569 if (ipv6_addr_is_multicast(&ohdr->rpl_segaddr[i])) {
575 err = ipv6_chk_rpl_srh_loop(net, ohdr->rpl_segaddr, n + 1);
577 icmpv6_send(skb, ICMPV6_PARAMPROB, 0, 0);
583 swap(ipv6_hdr(skb)->daddr, ohdr->rpl_segaddr[i]);
585 ipv6_rpl_srh_compress(chdr, ohdr, &ipv6_hdr(skb)->daddr, n);
587 oldhdr = ipv6_hdr(skb);
589 skb_pull(skb, ((hdr->hdrlen + 1) << 3));
590 skb_postpull_rcsum(skb, oldhdr,
591 sizeof(struct ipv6hdr) + ((hdr->hdrlen + 1) << 3));
592 if (unlikely(!hdr->segments_left)) {
593 if (pskb_expand_head(skb, sizeof(struct ipv6hdr) + ((chdr->hdrlen + 1) << 3), 0,
595 __IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)), IPSTATS_MIB_OUTDISCARDS);
601 oldhdr = ipv6_hdr(skb);
603 skb_push(skb, ((chdr->hdrlen + 1) << 3) + sizeof(struct ipv6hdr));
604 skb_reset_network_header(skb);
605 skb_mac_header_rebuild(skb);
606 skb_set_transport_header(skb, sizeof(struct ipv6hdr));
608 memmove(ipv6_hdr(skb), oldhdr, sizeof(struct ipv6hdr));
609 memcpy(skb_transport_header(skb), chdr, (chdr->hdrlen + 1) << 3);
611 ipv6_hdr(skb)->payload_len = htons(skb->len - sizeof(struct ipv6hdr));
612 skb_postpush_rcsum(skb, ipv6_hdr(skb),
613 sizeof(struct ipv6hdr) + ((chdr->hdrlen + 1) << 3));
617 ip6_route_input(skb);
619 if (skb_dst(skb)->error) {
624 if (skb_dst(skb)->dev->flags & IFF_LOOPBACK) {
625 if (ipv6_hdr(skb)->hop_limit <= 1) {
626 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INHDRERRORS);
627 icmpv6_send(skb, ICMPV6_TIME_EXCEED,
628 ICMPV6_EXC_HOPLIMIT, 0);
632 ipv6_hdr(skb)->hop_limit--;
634 skb_pull(skb, sizeof(struct ipv6hdr));
643 /********************************
645 ********************************/
647 /* called with rcu_read_lock() */
648 static int ipv6_rthdr_rcv(struct sk_buff *skb)
650 struct inet6_dev *idev = __in6_dev_get(skb->dev);
651 struct inet6_skb_parm *opt = IP6CB(skb);
652 struct in6_addr *addr = NULL;
654 struct ipv6_rt_hdr *hdr;
655 struct rt0_hdr *rthdr;
656 struct net *net = dev_net(skb->dev);
657 int accept_source_route = net->ipv6.devconf_all->accept_source_route;
659 if (idev && accept_source_route > idev->cnf.accept_source_route)
660 accept_source_route = idev->cnf.accept_source_route;
662 if (!pskb_may_pull(skb, skb_transport_offset(skb) + 8) ||
663 !pskb_may_pull(skb, (skb_transport_offset(skb) +
664 ((skb_transport_header(skb)[1] + 1) << 3)))) {
665 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INHDRERRORS);
670 hdr = (struct ipv6_rt_hdr *)skb_transport_header(skb);
672 if (ipv6_addr_is_multicast(&ipv6_hdr(skb)->daddr) ||
673 skb->pkt_type != PACKET_HOST) {
674 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INADDRERRORS);
680 case IPV6_SRCRT_TYPE_4:
681 /* segment routing */
682 return ipv6_srh_rcv(skb);
683 case IPV6_SRCRT_TYPE_3:
684 /* rpl segment routing */
685 return ipv6_rpl_srh_rcv(skb);
691 if (hdr->segments_left == 0) {
693 #if IS_ENABLED(CONFIG_IPV6_MIP6)
694 case IPV6_SRCRT_TYPE_2:
695 /* Silently discard type 2 header unless it was
699 __IP6_INC_STATS(net, idev,
700 IPSTATS_MIB_INADDRERRORS);
710 opt->lastopt = opt->srcrt = skb_network_header_len(skb);
711 skb->transport_header += (hdr->hdrlen + 1) << 3;
712 opt->dst0 = opt->dst1;
714 opt->nhoff = (&hdr->nexthdr) - skb_network_header(skb);
719 #if IS_ENABLED(CONFIG_IPV6_MIP6)
720 case IPV6_SRCRT_TYPE_2:
721 if (accept_source_route < 0)
723 /* Silently discard invalid RTH type 2 */
724 if (hdr->hdrlen != 2 || hdr->segments_left != 1) {
725 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INHDRERRORS);
736 * This is the routing header forwarding algorithm from
740 n = hdr->hdrlen >> 1;
742 if (hdr->segments_left > n) {
743 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INHDRERRORS);
744 icmpv6_param_prob(skb, ICMPV6_HDR_FIELD,
745 ((&hdr->segments_left) -
746 skb_network_header(skb)));
750 /* We are about to mangle packet header. Be careful!
751 Do not damage packets queued somewhere.
753 if (skb_cloned(skb)) {
754 /* the copy is a forwarded packet */
755 if (pskb_expand_head(skb, 0, 0, GFP_ATOMIC)) {
756 __IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
757 IPSTATS_MIB_OUTDISCARDS);
761 hdr = (struct ipv6_rt_hdr *)skb_transport_header(skb);
764 if (skb->ip_summed == CHECKSUM_COMPLETE)
765 skb->ip_summed = CHECKSUM_NONE;
767 i = n - --hdr->segments_left;
769 rthdr = (struct rt0_hdr *) hdr;
774 #if IS_ENABLED(CONFIG_IPV6_MIP6)
775 case IPV6_SRCRT_TYPE_2:
776 if (xfrm6_input_addr(skb, (xfrm_address_t *)addr,
777 (xfrm_address_t *)&ipv6_hdr(skb)->saddr,
778 IPPROTO_ROUTING) < 0) {
779 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INADDRERRORS);
783 if (!ipv6_chk_home_addr(dev_net(skb_dst(skb)->dev), addr)) {
784 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INADDRERRORS);
794 if (ipv6_addr_is_multicast(addr)) {
795 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INADDRERRORS);
800 swap(*addr, ipv6_hdr(skb)->daddr);
802 ip6_route_input(skb);
803 if (skb_dst(skb)->error) {
804 skb_push(skb, skb->data - skb_network_header(skb));
809 if (skb_dst(skb)->dev->flags&IFF_LOOPBACK) {
810 if (ipv6_hdr(skb)->hop_limit <= 1) {
811 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INHDRERRORS);
812 icmpv6_send(skb, ICMPV6_TIME_EXCEED, ICMPV6_EXC_HOPLIMIT,
817 ipv6_hdr(skb)->hop_limit--;
821 skb_push(skb, skb->data - skb_network_header(skb));
826 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INHDRERRORS);
827 icmpv6_param_prob(skb, ICMPV6_HDR_FIELD,
828 (&hdr->type) - skb_network_header(skb));
832 static const struct inet6_protocol rthdr_protocol = {
833 .handler = ipv6_rthdr_rcv,
834 .flags = INET6_PROTO_NOPOLICY,
837 static const struct inet6_protocol destopt_protocol = {
838 .handler = ipv6_destopt_rcv,
839 .flags = INET6_PROTO_NOPOLICY,
842 static const struct inet6_protocol nodata_protocol = {
843 .handler = dst_discard,
844 .flags = INET6_PROTO_NOPOLICY,
847 int __init ipv6_exthdrs_init(void)
851 ret = inet6_add_protocol(&rthdr_protocol, IPPROTO_ROUTING);
855 ret = inet6_add_protocol(&destopt_protocol, IPPROTO_DSTOPTS);
859 ret = inet6_add_protocol(&nodata_protocol, IPPROTO_NONE);
866 inet6_del_protocol(&destopt_protocol, IPPROTO_DSTOPTS);
868 inet6_del_protocol(&rthdr_protocol, IPPROTO_ROUTING);
872 void ipv6_exthdrs_exit(void)
874 inet6_del_protocol(&nodata_protocol, IPPROTO_NONE);
875 inet6_del_protocol(&destopt_protocol, IPPROTO_DSTOPTS);
876 inet6_del_protocol(&rthdr_protocol, IPPROTO_ROUTING);
879 /**********************************
881 **********************************/
884 * Note: we cannot rely on skb_dst(skb) before we assign it in ip6_route_input().
886 static inline struct net *ipv6_skb_net(struct sk_buff *skb)
888 return skb_dst(skb) ? dev_net(skb_dst(skb)->dev) : dev_net(skb->dev);
891 /* Router Alert as of RFC 2711 */
893 static bool ipv6_hop_ra(struct sk_buff *skb, int optoff)
895 const unsigned char *nh = skb_network_header(skb);
897 if (nh[optoff + 1] == 2) {
898 IP6CB(skb)->flags |= IP6SKB_ROUTERALERT;
899 memcpy(&IP6CB(skb)->ra, nh + optoff + 2, sizeof(IP6CB(skb)->ra));
902 net_dbg_ratelimited("ipv6_hop_ra: wrong RA length %d\n",
904 kfree_skb_reason(skb, SKB_DROP_REASON_IP_INHDR);
910 static bool ipv6_hop_ioam(struct sk_buff *skb, int optoff)
912 struct ioam6_trace_hdr *trace;
913 struct ioam6_namespace *ns;
914 struct ioam6_hdr *hdr;
916 /* Bad alignment (must be 4n-aligned) */
920 /* Ignore if IOAM is not enabled on ingress */
921 if (!__in6_dev_get(skb->dev)->cnf.ioam6_enabled)
924 /* Truncated Option header */
925 hdr = (struct ioam6_hdr *)(skb_network_header(skb) + optoff);
926 if (hdr->opt_len < 2)
930 case IOAM6_TYPE_PREALLOC:
931 /* Truncated Pre-allocated Trace header */
932 if (hdr->opt_len < 2 + sizeof(*trace))
935 /* Malformed Pre-allocated Trace header */
936 trace = (struct ioam6_trace_hdr *)((u8 *)hdr + sizeof(*hdr));
937 if (hdr->opt_len < 2 + sizeof(*trace) + trace->remlen * 4)
940 /* Ignore if the IOAM namespace is unknown */
941 ns = ioam6_namespace(ipv6_skb_net(skb), trace->namespace_id);
945 if (!skb_valid_dst(skb))
946 ip6_route_input(skb);
948 /* About to mangle packet header */
949 if (skb_ensure_writable(skb, optoff + 2 + hdr->opt_len))
952 /* Trace pointer may have changed */
953 trace = (struct ioam6_trace_hdr *)(skb_network_header(skb)
954 + optoff + sizeof(*hdr));
956 ioam6_fill_trace_data(skb, ns, trace, true);
966 kfree_skb_reason(skb, SKB_DROP_REASON_IP_INHDR);
972 static bool ipv6_hop_jumbo(struct sk_buff *skb, int optoff)
974 const unsigned char *nh = skb_network_header(skb);
978 if (nh[optoff + 1] != 4 || (optoff & 3) != 2) {
979 net_dbg_ratelimited("ipv6_hop_jumbo: wrong jumbo opt length/alignment %d\n",
981 SKB_DR_SET(reason, IP_INHDR);
985 pkt_len = ntohl(*(__be32 *)(nh + optoff + 2));
986 if (pkt_len <= IPV6_MAXPLEN) {
987 icmpv6_param_prob_reason(skb, ICMPV6_HDR_FIELD, optoff + 2,
988 SKB_DROP_REASON_IP_INHDR);
991 if (ipv6_hdr(skb)->payload_len) {
992 icmpv6_param_prob_reason(skb, ICMPV6_HDR_FIELD, optoff,
993 SKB_DROP_REASON_IP_INHDR);
997 if (pkt_len > skb->len - sizeof(struct ipv6hdr)) {
998 SKB_DR_SET(reason, PKT_TOO_SMALL);
1002 if (pskb_trim_rcsum(skb, pkt_len + sizeof(struct ipv6hdr)))
1005 IP6CB(skb)->flags |= IP6SKB_JUMBOGRAM;
1009 kfree_skb_reason(skb, reason);
1013 /* CALIPSO RFC 5570 */
1015 static bool ipv6_hop_calipso(struct sk_buff *skb, int optoff)
1017 const unsigned char *nh = skb_network_header(skb);
1019 if (nh[optoff + 1] < 8)
1022 if (nh[optoff + 6] * 4 + 8 > nh[optoff + 1])
1025 if (!calipso_validate(skb, nh + optoff))
1031 kfree_skb_reason(skb, SKB_DROP_REASON_IP_INHDR);
1035 int ipv6_parse_hopopts(struct sk_buff *skb)
1037 struct inet6_skb_parm *opt = IP6CB(skb);
1038 struct net *net = dev_net(skb->dev);
1042 * skb_network_header(skb) is equal to skb->data, and
1043 * skb_network_header_len(skb) is always equal to
1044 * sizeof(struct ipv6hdr) by definition of
1045 * hop-by-hop options.
1047 if (!pskb_may_pull(skb, sizeof(struct ipv6hdr) + 8) ||
1048 !pskb_may_pull(skb, (sizeof(struct ipv6hdr) +
1049 ((skb_transport_header(skb)[1] + 1) << 3)))) {
1055 extlen = (skb_transport_header(skb)[1] + 1) << 3;
1056 if (extlen > net->ipv6.sysctl.max_hbh_opts_len)
1059 opt->flags |= IP6SKB_HOPBYHOP;
1060 if (ip6_parse_tlv(true, skb, net->ipv6.sysctl.max_hbh_opts_cnt)) {
1061 skb->transport_header += extlen;
1063 opt->nhoff = sizeof(struct ipv6hdr);
1070 * Creating outbound headers.
1072 * "build" functions work when skb is filled from head to tail (datagram)
1073 * "push" functions work when headers are added from tail to head (tcp)
1075 * In both cases we assume, that caller reserved enough room
1079 static void ipv6_push_rthdr0(struct sk_buff *skb, u8 *proto,
1080 struct ipv6_rt_hdr *opt,
1081 struct in6_addr **addr_p, struct in6_addr *saddr)
1083 struct rt0_hdr *phdr, *ihdr;
1086 ihdr = (struct rt0_hdr *) opt;
1088 phdr = skb_push(skb, (ihdr->rt_hdr.hdrlen + 1) << 3);
1089 memcpy(phdr, ihdr, sizeof(struct rt0_hdr));
1091 hops = ihdr->rt_hdr.hdrlen >> 1;
1094 memcpy(phdr->addr, ihdr->addr + 1,
1095 (hops - 1) * sizeof(struct in6_addr));
1097 phdr->addr[hops - 1] = **addr_p;
1098 *addr_p = ihdr->addr;
1100 phdr->rt_hdr.nexthdr = *proto;
1101 *proto = NEXTHDR_ROUTING;
1104 static void ipv6_push_rthdr4(struct sk_buff *skb, u8 *proto,
1105 struct ipv6_rt_hdr *opt,
1106 struct in6_addr **addr_p, struct in6_addr *saddr)
1108 struct ipv6_sr_hdr *sr_phdr, *sr_ihdr;
1111 sr_ihdr = (struct ipv6_sr_hdr *)opt;
1112 plen = (sr_ihdr->hdrlen + 1) << 3;
1114 sr_phdr = skb_push(skb, plen);
1115 memcpy(sr_phdr, sr_ihdr, sizeof(struct ipv6_sr_hdr));
1117 hops = sr_ihdr->first_segment + 1;
1118 memcpy(sr_phdr->segments + 1, sr_ihdr->segments + 1,
1119 (hops - 1) * sizeof(struct in6_addr));
1121 sr_phdr->segments[0] = **addr_p;
1122 *addr_p = &sr_ihdr->segments[sr_ihdr->segments_left];
1124 if (sr_ihdr->hdrlen > hops * 2) {
1125 int tlvs_offset, tlvs_length;
1127 tlvs_offset = (1 + hops * 2) << 3;
1128 tlvs_length = (sr_ihdr->hdrlen - hops * 2) << 3;
1129 memcpy((char *)sr_phdr + tlvs_offset,
1130 (char *)sr_ihdr + tlvs_offset, tlvs_length);
1133 #ifdef CONFIG_IPV6_SEG6_HMAC
1134 if (sr_has_hmac(sr_phdr)) {
1135 struct net *net = NULL;
1138 net = dev_net(skb->dev);
1140 net = sock_net(skb->sk);
1145 seg6_push_hmac(net, saddr, sr_phdr);
1149 sr_phdr->nexthdr = *proto;
1150 *proto = NEXTHDR_ROUTING;
1153 static void ipv6_push_rthdr(struct sk_buff *skb, u8 *proto,
1154 struct ipv6_rt_hdr *opt,
1155 struct in6_addr **addr_p, struct in6_addr *saddr)
1157 switch (opt->type) {
1158 case IPV6_SRCRT_TYPE_0:
1159 case IPV6_SRCRT_STRICT:
1160 case IPV6_SRCRT_TYPE_2:
1161 ipv6_push_rthdr0(skb, proto, opt, addr_p, saddr);
1163 case IPV6_SRCRT_TYPE_4:
1164 ipv6_push_rthdr4(skb, proto, opt, addr_p, saddr);
1171 static void ipv6_push_exthdr(struct sk_buff *skb, u8 *proto, u8 type, struct ipv6_opt_hdr *opt)
1173 struct ipv6_opt_hdr *h = skb_push(skb, ipv6_optlen(opt));
1175 memcpy(h, opt, ipv6_optlen(opt));
1176 h->nexthdr = *proto;
1180 void ipv6_push_nfrag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
1182 struct in6_addr **daddr, struct in6_addr *saddr)
1185 ipv6_push_rthdr(skb, proto, opt->srcrt, daddr, saddr);
1187 * IPV6_RTHDRDSTOPTS is ignored
1188 * unless IPV6_RTHDR is set (RFC3542).
1191 ipv6_push_exthdr(skb, proto, NEXTHDR_DEST, opt->dst0opt);
1194 ipv6_push_exthdr(skb, proto, NEXTHDR_HOP, opt->hopopt);
1197 void ipv6_push_frag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt, u8 *proto)
1200 ipv6_push_exthdr(skb, proto, NEXTHDR_DEST, opt->dst1opt);
1202 EXPORT_SYMBOL(ipv6_push_frag_opts);
1204 struct ipv6_txoptions *
1205 ipv6_dup_options(struct sock *sk, struct ipv6_txoptions *opt)
1207 struct ipv6_txoptions *opt2;
1209 opt2 = sock_kmalloc(sk, opt->tot_len, GFP_ATOMIC);
1211 long dif = (char *)opt2 - (char *)opt;
1212 memcpy(opt2, opt, opt->tot_len);
1214 *((char **)&opt2->hopopt) += dif;
1216 *((char **)&opt2->dst0opt) += dif;
1218 *((char **)&opt2->dst1opt) += dif;
1220 *((char **)&opt2->srcrt) += dif;
1221 refcount_set(&opt2->refcnt, 1);
1225 EXPORT_SYMBOL_GPL(ipv6_dup_options);
1227 static void ipv6_renew_option(int renewtype,
1228 struct ipv6_opt_hdr **dest,
1229 struct ipv6_opt_hdr *old,
1230 struct ipv6_opt_hdr *new,
1231 int newtype, char **p)
1233 struct ipv6_opt_hdr *src;
1235 src = (renewtype == newtype ? new : old);
1239 memcpy(*p, src, ipv6_optlen(src));
1240 *dest = (struct ipv6_opt_hdr *)*p;
1241 *p += CMSG_ALIGN(ipv6_optlen(*dest));
1245 * ipv6_renew_options - replace a specific ext hdr with a new one.
1247 * @sk: sock from which to allocate memory
1248 * @opt: original options
1249 * @newtype: option type to replace in @opt
1250 * @newopt: new option of type @newtype to replace (user-mem)
1252 * Returns a new set of options which is a copy of @opt with the
1253 * option type @newtype replaced with @newopt.
1255 * @opt may be NULL, in which case a new set of options is returned
1256 * containing just @newopt.
1258 * @newopt may be NULL, in which case the specified option type is
1259 * not copied into the new set of options.
1261 * The new set of options is allocated from the socket option memory
1264 struct ipv6_txoptions *
1265 ipv6_renew_options(struct sock *sk, struct ipv6_txoptions *opt,
1266 int newtype, struct ipv6_opt_hdr *newopt)
1270 struct ipv6_txoptions *opt2;
1273 if (newtype != IPV6_HOPOPTS && opt->hopopt)
1274 tot_len += CMSG_ALIGN(ipv6_optlen(opt->hopopt));
1275 if (newtype != IPV6_RTHDRDSTOPTS && opt->dst0opt)
1276 tot_len += CMSG_ALIGN(ipv6_optlen(opt->dst0opt));
1277 if (newtype != IPV6_RTHDR && opt->srcrt)
1278 tot_len += CMSG_ALIGN(ipv6_optlen(opt->srcrt));
1279 if (newtype != IPV6_DSTOPTS && opt->dst1opt)
1280 tot_len += CMSG_ALIGN(ipv6_optlen(opt->dst1opt));
1284 tot_len += CMSG_ALIGN(ipv6_optlen(newopt));
1289 tot_len += sizeof(*opt2);
1290 opt2 = sock_kmalloc(sk, tot_len, GFP_ATOMIC);
1292 return ERR_PTR(-ENOBUFS);
1294 memset(opt2, 0, tot_len);
1295 refcount_set(&opt2->refcnt, 1);
1296 opt2->tot_len = tot_len;
1297 p = (char *)(opt2 + 1);
1299 ipv6_renew_option(IPV6_HOPOPTS, &opt2->hopopt,
1300 (opt ? opt->hopopt : NULL),
1301 newopt, newtype, &p);
1302 ipv6_renew_option(IPV6_RTHDRDSTOPTS, &opt2->dst0opt,
1303 (opt ? opt->dst0opt : NULL),
1304 newopt, newtype, &p);
1305 ipv6_renew_option(IPV6_RTHDR,
1306 (struct ipv6_opt_hdr **)&opt2->srcrt,
1307 (opt ? (struct ipv6_opt_hdr *)opt->srcrt : NULL),
1308 newopt, newtype, &p);
1309 ipv6_renew_option(IPV6_DSTOPTS, &opt2->dst1opt,
1310 (opt ? opt->dst1opt : NULL),
1311 newopt, newtype, &p);
1313 opt2->opt_nflen = (opt2->hopopt ? ipv6_optlen(opt2->hopopt) : 0) +
1314 (opt2->dst0opt ? ipv6_optlen(opt2->dst0opt) : 0) +
1315 (opt2->srcrt ? ipv6_optlen(opt2->srcrt) : 0);
1316 opt2->opt_flen = (opt2->dst1opt ? ipv6_optlen(opt2->dst1opt) : 0);
1321 struct ipv6_txoptions *__ipv6_fixup_options(struct ipv6_txoptions *opt_space,
1322 struct ipv6_txoptions *opt)
1325 * ignore the dest before srcrt unless srcrt is being included.
1328 if (opt->dst0opt && !opt->srcrt) {
1329 if (opt_space != opt) {
1330 memcpy(opt_space, opt, sizeof(*opt_space));
1333 opt->opt_nflen -= ipv6_optlen(opt->dst0opt);
1334 opt->dst0opt = NULL;
1339 EXPORT_SYMBOL_GPL(__ipv6_fixup_options);
1342 * fl6_update_dst - update flowi destination address with info given
1343 * by srcrt option, if any.
1345 * @fl6: flowi6 for which daddr is to be updated
1346 * @opt: struct ipv6_txoptions in which to look for srcrt opt
1347 * @orig: copy of original daddr address if modified
1349 * Returns NULL if no txoptions or no srcrt, otherwise returns orig
1350 * and initial value of fl6->daddr set in orig
1352 struct in6_addr *fl6_update_dst(struct flowi6 *fl6,
1353 const struct ipv6_txoptions *opt,
1354 struct in6_addr *orig)
1356 if (!opt || !opt->srcrt)
1361 switch (opt->srcrt->type) {
1362 case IPV6_SRCRT_TYPE_0:
1363 case IPV6_SRCRT_STRICT:
1364 case IPV6_SRCRT_TYPE_2:
1365 fl6->daddr = *((struct rt0_hdr *)opt->srcrt)->addr;
1367 case IPV6_SRCRT_TYPE_4:
1369 struct ipv6_sr_hdr *srh = (struct ipv6_sr_hdr *)opt->srcrt;
1371 fl6->daddr = srh->segments[srh->segments_left];
1380 EXPORT_SYMBOL_GPL(fl6_update_dst);