This commit was manufactured by cvs2svn to create branch 'vserver'.
[linux-2.6.git] / net / ipv6 / mip6.c
diff --git a/net/ipv6/mip6.c b/net/ipv6/mip6.c
new file mode 100644 (file)
index 0000000..be7dd7d
--- /dev/null
@@ -0,0 +1,518 @@
+/*
+ * Copyright (C)2003-2006 Helsinki University of Technology
+ * Copyright (C)2003-2006 USAGI/WIDE Project
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+/*
+ * Authors:
+ *     Noriaki TAKAMIYA @USAGI
+ *     Masahide NAKAMURA @USAGI
+ */
+
+#include <linux/module.h>
+#include <linux/skbuff.h>
+#include <linux/time.h>
+#include <linux/ipv6.h>
+#include <linux/icmpv6.h>
+#include <net/sock.h>
+#include <net/ipv6.h>
+#include <net/ip6_checksum.h>
+#include <net/xfrm.h>
+#include <net/mip6.h>
+
+static xfrm_address_t *mip6_xfrm_addr(struct xfrm_state *x, xfrm_address_t *addr)
+{
+       return x->coaddr;
+}
+
+static inline unsigned int calc_padlen(unsigned int len, unsigned int n)
+{
+       return (n - len + 16) & 0x7;
+}
+
+static inline void *mip6_padn(__u8 *data, __u8 padlen)
+{
+       if (!data)
+               return NULL;
+       if (padlen == 1) {
+               data[0] = MIP6_OPT_PAD_1;
+       } else if (padlen > 1) {
+               data[0] = MIP6_OPT_PAD_N;
+               data[1] = padlen - 2;
+               if (padlen > 2)
+                       memset(data+2, 0, data[1]);
+       }
+       return data + padlen;
+}
+
+static inline void mip6_param_prob(struct sk_buff *skb, int code, int pos)
+{
+       icmpv6_send(skb, ICMPV6_PARAMPROB, code, pos, skb->dev);
+}
+
+static int mip6_mh_len(int type)
+{
+       int len = 0;
+
+       switch (type) {
+       case IP6_MH_TYPE_BRR:
+               len = 0;
+               break;
+       case IP6_MH_TYPE_HOTI:
+       case IP6_MH_TYPE_COTI:
+       case IP6_MH_TYPE_BU:
+       case IP6_MH_TYPE_BACK:
+               len = 1;
+               break;
+       case IP6_MH_TYPE_HOT:
+       case IP6_MH_TYPE_COT:
+       case IP6_MH_TYPE_BERROR:
+               len = 2;
+               break;
+       }
+       return len;
+}
+
+int mip6_mh_filter(struct sock *sk, struct sk_buff *skb)
+{
+       struct ip6_mh *mh;
+       int mhlen;
+
+       if (!pskb_may_pull(skb, (skb->h.raw - skb->data) + 8) ||
+           !pskb_may_pull(skb, (skb->h.raw - skb->data) + ((skb->h.raw[1] + 1) << 3)))
+               return -1;
+
+       mh = (struct ip6_mh *)skb->h.raw;
+
+       if (mh->ip6mh_hdrlen < mip6_mh_len(mh->ip6mh_type)) {
+               LIMIT_NETDEBUG(KERN_DEBUG "mip6: MH message too short: %d vs >=%d\n",
+                              mh->ip6mh_hdrlen, mip6_mh_len(mh->ip6mh_type));
+               mip6_param_prob(skb, 0, (&mh->ip6mh_hdrlen) - skb->nh.raw);
+               return -1;
+       }
+       mhlen = (mh->ip6mh_hdrlen + 1) << 3;
+
+       if (skb->ip_summed == CHECKSUM_COMPLETE) {
+               skb->ip_summed = CHECKSUM_UNNECESSARY;
+               if (csum_ipv6_magic(&skb->nh.ipv6h->saddr,
+                                   &skb->nh.ipv6h->daddr,
+                                   mhlen, IPPROTO_MH,
+                                   skb->csum)) {
+                       LIMIT_NETDEBUG(KERN_DEBUG "mip6: MH hw checksum failed\n");
+                       skb->ip_summed = CHECKSUM_NONE;
+               }
+       }
+       if (skb->ip_summed == CHECKSUM_NONE) {
+               if (csum_ipv6_magic(&skb->nh.ipv6h->saddr,
+                                   &skb->nh.ipv6h->daddr,
+                                   mhlen, IPPROTO_MH,
+                                   skb_checksum(skb, 0, mhlen, 0))) {
+                       LIMIT_NETDEBUG(KERN_DEBUG "mip6: MH checksum failed "
+                                      "[" NIP6_FMT " > " NIP6_FMT "]\n",
+                                      NIP6(skb->nh.ipv6h->saddr),
+                                      NIP6(skb->nh.ipv6h->daddr));
+                       return -1;
+               }
+               skb->ip_summed = CHECKSUM_UNNECESSARY;
+       }
+
+       if (mh->ip6mh_proto != IPPROTO_NONE) {
+               LIMIT_NETDEBUG(KERN_DEBUG "mip6: MH invalid payload proto = %d\n",
+                              mh->ip6mh_proto);
+               mip6_param_prob(skb, 0, (&mh->ip6mh_proto) - skb->nh.raw);
+               return -1;
+       }
+
+       return 0;
+}
+
+struct mip6_report_rate_limiter {
+       spinlock_t lock;
+       struct timeval stamp;
+       int iif;
+       struct in6_addr src;
+       struct in6_addr dst;
+};
+
+static struct mip6_report_rate_limiter mip6_report_rl = {
+       .lock = SPIN_LOCK_UNLOCKED
+};
+
+static int mip6_destopt_input(struct xfrm_state *x, struct sk_buff *skb)
+{
+       struct ipv6hdr *iph = skb->nh.ipv6h;
+       struct ipv6_destopt_hdr *destopt = (struct ipv6_destopt_hdr *)skb->data;
+
+       if (!ipv6_addr_equal(&iph->saddr, (struct in6_addr *)x->coaddr) &&
+           !ipv6_addr_any((struct in6_addr *)x->coaddr))
+               return -ENOENT;
+
+       return destopt->nexthdr;
+}
+
+/* Destination Option Header is inserted.
+ * IP Header's src address is replaced with Home Address Option in
+ * Destination Option Header.
+ */
+static int mip6_destopt_output(struct xfrm_state *x, struct sk_buff *skb)
+{
+       struct ipv6hdr *iph;
+       struct ipv6_destopt_hdr *dstopt;
+       struct ipv6_destopt_hao *hao;
+       u8 nexthdr;
+       int len;
+
+       iph = (struct ipv6hdr *)skb->data;
+       iph->payload_len = htons(skb->len - sizeof(*iph));
+
+       nexthdr = *skb->nh.raw;
+       *skb->nh.raw = IPPROTO_DSTOPTS;
+
+       dstopt = (struct ipv6_destopt_hdr *)skb->h.raw;
+       dstopt->nexthdr = nexthdr;
+
+       hao = mip6_padn((char *)(dstopt + 1),
+                       calc_padlen(sizeof(*dstopt), 6));
+
+       hao->type = IPV6_TLV_HAO;
+       hao->length = sizeof(*hao) - 2;
+       BUG_TRAP(hao->length == 16);
+
+       len = ((char *)hao - (char *)dstopt) + sizeof(*hao);
+
+       memcpy(&hao->addr, &iph->saddr, sizeof(hao->addr));
+       memcpy(&iph->saddr, x->coaddr, sizeof(iph->saddr));
+
+       BUG_TRAP(len == x->props.header_len);
+       dstopt->hdrlen = (x->props.header_len >> 3) - 1;
+
+       return 0;
+}
+
+static inline int mip6_report_rl_allow(struct timeval *stamp,
+                                      struct in6_addr *dst,
+                                      struct in6_addr *src, int iif)
+{
+       int allow = 0;
+
+       spin_lock_bh(&mip6_report_rl.lock);
+       if (mip6_report_rl.stamp.tv_sec != stamp->tv_sec ||
+           mip6_report_rl.stamp.tv_usec != stamp->tv_usec ||
+           mip6_report_rl.iif != iif ||
+           !ipv6_addr_equal(&mip6_report_rl.src, src) ||
+           !ipv6_addr_equal(&mip6_report_rl.dst, dst)) {
+               mip6_report_rl.stamp.tv_sec = stamp->tv_sec;
+               mip6_report_rl.stamp.tv_usec = stamp->tv_usec;
+               mip6_report_rl.iif = iif;
+               ipv6_addr_copy(&mip6_report_rl.src, src);
+               ipv6_addr_copy(&mip6_report_rl.dst, dst);
+               allow = 1;
+       }
+       spin_unlock_bh(&mip6_report_rl.lock);
+       return allow;
+}
+
+static int mip6_destopt_reject(struct xfrm_state *x, struct sk_buff *skb, struct flowi *fl)
+{
+       struct inet6_skb_parm *opt = (struct inet6_skb_parm *)skb->cb;
+       struct ipv6_destopt_hao *hao = NULL;
+       struct xfrm_selector sel;
+       int offset;
+       struct timeval stamp;
+       int err = 0;
+
+       if (unlikely(fl->proto == IPPROTO_MH &&
+                    fl->fl_mh_type <= IP6_MH_TYPE_MAX))
+               goto out;
+
+       if (likely(opt->dsthao)) {
+               offset = ipv6_find_tlv(skb, opt->dsthao, IPV6_TLV_HAO);
+               if (likely(offset >= 0))
+                       hao = (struct ipv6_destopt_hao *)(skb->nh.raw + offset);
+       }
+
+       skb_get_timestamp(skb, &stamp);
+
+       if (!mip6_report_rl_allow(&stamp, &skb->nh.ipv6h->daddr,
+                                 hao ? &hao->addr : &skb->nh.ipv6h->saddr,
+                                 opt->iif))
+               goto out;
+
+       memset(&sel, 0, sizeof(sel));
+       memcpy(&sel.daddr, (xfrm_address_t *)&skb->nh.ipv6h->daddr,
+              sizeof(sel.daddr));
+       sel.prefixlen_d = 128;
+       memcpy(&sel.saddr, (xfrm_address_t *)&skb->nh.ipv6h->saddr,
+              sizeof(sel.saddr));
+       sel.prefixlen_s = 128;
+       sel.family = AF_INET6;
+       sel.proto = fl->proto;
+       sel.dport = xfrm_flowi_dport(fl);
+       if (sel.dport)
+               sel.dport_mask = htons(~0);
+       sel.sport = xfrm_flowi_sport(fl);
+       if (sel.sport)
+               sel.sport_mask = htons(~0);
+       sel.ifindex = fl->oif;
+
+       err = km_report(IPPROTO_DSTOPTS, &sel,
+                       (hao ? (xfrm_address_t *)&hao->addr : NULL));
+
+ out:
+       return err;
+}
+
+static int mip6_destopt_offset(struct xfrm_state *x, struct sk_buff *skb,
+                              u8 **nexthdr)
+{
+       u16 offset = sizeof(struct ipv6hdr);
+       struct ipv6_opt_hdr *exthdr = (struct ipv6_opt_hdr*)(skb->nh.ipv6h + 1);
+       unsigned int packet_len = skb->tail - skb->nh.raw;
+       int found_rhdr = 0;
+
+       *nexthdr = &skb->nh.ipv6h->nexthdr;
+
+       while (offset + 1 <= packet_len) {
+
+               switch (**nexthdr) {
+               case NEXTHDR_HOP:
+                       break;
+               case NEXTHDR_ROUTING:
+                       found_rhdr = 1;
+                       break;
+               case NEXTHDR_DEST:
+                       /*
+                        * HAO MUST NOT appear more than once.
+                        * XXX: It is better to try to find by the end of
+                        * XXX: packet if HAO exists.
+                        */
+                       if (ipv6_find_tlv(skb, offset, IPV6_TLV_HAO) >= 0) {
+                               LIMIT_NETDEBUG(KERN_WARNING "mip6: hao exists already, override\n");
+                               return offset;
+                       }
+
+                       if (found_rhdr)
+                               return offset;
+
+                       break;
+               default:
+                       return offset;
+               }
+
+               offset += ipv6_optlen(exthdr);
+               *nexthdr = &exthdr->nexthdr;
+               exthdr = (struct ipv6_opt_hdr*)(skb->nh.raw + offset);
+       }
+
+       return offset;
+}
+
+static int mip6_destopt_init_state(struct xfrm_state *x)
+{
+       if (x->id.spi) {
+               printk(KERN_INFO "%s: spi is not 0: %u\n", __FUNCTION__,
+                      x->id.spi);
+               return -EINVAL;
+       }
+       if (x->props.mode != XFRM_MODE_ROUTEOPTIMIZATION) {
+               printk(KERN_INFO "%s: state's mode is not %u: %u\n",
+                      __FUNCTION__, XFRM_MODE_ROUTEOPTIMIZATION, x->props.mode);
+               return -EINVAL;
+       }
+
+       x->props.header_len = sizeof(struct ipv6_destopt_hdr) +
+               calc_padlen(sizeof(struct ipv6_destopt_hdr), 6) +
+               sizeof(struct ipv6_destopt_hao);
+       BUG_TRAP(x->props.header_len == 24);
+
+       return 0;
+}
+
+/*
+ * Do nothing about destroying since it has no specific operation for
+ * destination options header unlike IPsec protocols.
+ */
+static void mip6_destopt_destroy(struct xfrm_state *x)
+{
+}
+
+static struct xfrm_type mip6_destopt_type =
+{
+       .description    = "MIP6DESTOPT",
+       .owner          = THIS_MODULE,
+       .proto          = IPPROTO_DSTOPTS,
+       .flags          = XFRM_TYPE_NON_FRAGMENT,
+       .init_state     = mip6_destopt_init_state,
+       .destructor     = mip6_destopt_destroy,
+       .input          = mip6_destopt_input,
+       .output         = mip6_destopt_output,
+       .reject         = mip6_destopt_reject,
+       .hdr_offset     = mip6_destopt_offset,
+       .local_addr     = mip6_xfrm_addr,
+};
+
+static int mip6_rthdr_input(struct xfrm_state *x, struct sk_buff *skb)
+{
+       struct rt2_hdr *rt2 = (struct rt2_hdr *)skb->data;
+
+       if (!ipv6_addr_equal(&rt2->addr, (struct in6_addr *)x->coaddr) &&
+           !ipv6_addr_any((struct in6_addr *)x->coaddr))
+               return -ENOENT;
+
+       return rt2->rt_hdr.nexthdr;
+}
+
+/* Routing Header type 2 is inserted.
+ * IP Header's dst address is replaced with Routing Header's Home Address.
+ */
+static int mip6_rthdr_output(struct xfrm_state *x, struct sk_buff *skb)
+{
+       struct ipv6hdr *iph;
+       struct rt2_hdr *rt2;
+       u8 nexthdr;
+
+       iph = (struct ipv6hdr *)skb->data;
+       iph->payload_len = htons(skb->len - sizeof(*iph));
+
+       nexthdr = *skb->nh.raw;
+       *skb->nh.raw = IPPROTO_ROUTING;
+
+       rt2 = (struct rt2_hdr *)skb->h.raw;
+       rt2->rt_hdr.nexthdr = nexthdr;
+       rt2->rt_hdr.hdrlen = (x->props.header_len >> 3) - 1;
+       rt2->rt_hdr.type = IPV6_SRCRT_TYPE_2;
+       rt2->rt_hdr.segments_left = 1;
+       memset(&rt2->reserved, 0, sizeof(rt2->reserved));
+
+       BUG_TRAP(rt2->rt_hdr.hdrlen == 2);
+
+       memcpy(&rt2->addr, &iph->daddr, sizeof(rt2->addr));
+       memcpy(&iph->daddr, x->coaddr, sizeof(iph->daddr));
+
+       return 0;
+}
+
+static int mip6_rthdr_offset(struct xfrm_state *x, struct sk_buff *skb,
+                            u8 **nexthdr)
+{
+       u16 offset = sizeof(struct ipv6hdr);
+       struct ipv6_opt_hdr *exthdr = (struct ipv6_opt_hdr*)(skb->nh.ipv6h + 1);
+       unsigned int packet_len = skb->tail - skb->nh.raw;
+       int found_rhdr = 0;
+
+       *nexthdr = &skb->nh.ipv6h->nexthdr;
+
+       while (offset + 1 <= packet_len) {
+
+               switch (**nexthdr) {
+               case NEXTHDR_HOP:
+                       break;
+               case NEXTHDR_ROUTING:
+                       if (offset + 3 <= packet_len) {
+                               struct ipv6_rt_hdr *rt;
+                               rt = (struct ipv6_rt_hdr *)(skb->nh.raw + offset);
+                               if (rt->type != 0)
+                                       return offset;
+                       }
+                       found_rhdr = 1;
+                       break;
+               case NEXTHDR_DEST:
+                       if (ipv6_find_tlv(skb, offset, IPV6_TLV_HAO) >= 0)
+                               return offset;
+
+                       if (found_rhdr)
+                               return offset;
+
+                       break;
+               default:
+                       return offset;
+               }
+
+               offset += ipv6_optlen(exthdr);
+               *nexthdr = &exthdr->nexthdr;
+               exthdr = (struct ipv6_opt_hdr*)(skb->nh.raw + offset);
+       }
+
+       return offset;
+}
+
+static int mip6_rthdr_init_state(struct xfrm_state *x)
+{
+       if (x->id.spi) {
+               printk(KERN_INFO "%s: spi is not 0: %u\n", __FUNCTION__,
+                      x->id.spi);
+               return -EINVAL;
+       }
+       if (x->props.mode != XFRM_MODE_ROUTEOPTIMIZATION) {
+               printk(KERN_INFO "%s: state's mode is not %u: %u\n",
+                      __FUNCTION__, XFRM_MODE_ROUTEOPTIMIZATION, x->props.mode);
+               return -EINVAL;
+       }
+
+       x->props.header_len = sizeof(struct rt2_hdr);
+
+       return 0;
+}
+
+/*
+ * Do nothing about destroying since it has no specific operation for routing
+ * header type 2 unlike IPsec protocols.
+ */
+static void mip6_rthdr_destroy(struct xfrm_state *x)
+{
+}
+
+static struct xfrm_type mip6_rthdr_type =
+{
+       .description    = "MIP6RT",
+       .owner          = THIS_MODULE,
+       .proto          = IPPROTO_ROUTING,
+       .flags          = XFRM_TYPE_NON_FRAGMENT,
+       .init_state     = mip6_rthdr_init_state,
+       .destructor     = mip6_rthdr_destroy,
+       .input          = mip6_rthdr_input,
+       .output         = mip6_rthdr_output,
+       .hdr_offset     = mip6_rthdr_offset,
+       .remote_addr    = mip6_xfrm_addr,
+};
+
+int __init mip6_init(void)
+{
+       printk(KERN_INFO "Mobile IPv6\n");
+
+       if (xfrm_register_type(&mip6_destopt_type, AF_INET6) < 0) {
+               printk(KERN_INFO "%s: can't add xfrm type(destopt)\n", __FUNCTION__);
+               goto mip6_destopt_xfrm_fail;
+       }
+       if (xfrm_register_type(&mip6_rthdr_type, AF_INET6) < 0) {
+               printk(KERN_INFO "%s: can't add xfrm type(rthdr)\n", __FUNCTION__);
+               goto mip6_rthdr_xfrm_fail;
+       }
+       return 0;
+
+ mip6_rthdr_xfrm_fail:
+       xfrm_unregister_type(&mip6_destopt_type, AF_INET6);
+ mip6_destopt_xfrm_fail:
+       return -EAGAIN;
+}
+
+void __exit mip6_fini(void)
+{
+       if (xfrm_unregister_type(&mip6_rthdr_type, AF_INET6) < 0)
+               printk(KERN_INFO "%s: can't remove xfrm type(rthdr)\n", __FUNCTION__);
+       if (xfrm_unregister_type(&mip6_destopt_type, AF_INET6) < 0)
+               printk(KERN_INFO "%s: can't remove xfrm type(destopt)\n", __FUNCTION__);
+}