From: Steven Barth Subject: Add support for MAP-E FMRs (mesh mode) MAP-E FMRs (draft-ietf-softwire-map-10) are rules for IPv4-communication between MAP CEs (mesh mode) without the need to forward such data to a border relay. This is similar to how 6rd works but for IPv4 over IPv6. Signed-off-by: Steven Barth --- include/net/ip6_tunnel.h | 13 ++ include/uapi/linux/if_tunnel.h | 13 ++ net/ipv6/ip6_tunnel.c | 276 +++++++++++++++++++++++++++++++++++++++-- 3 files changed, 291 insertions(+), 11 deletions(-) --- a/include/net/ip6_tunnel.h +++ b/include/net/ip6_tunnel.h @@ -18,6 +18,18 @@ /* determine capability on a per-packet basis */ #define IP6_TNL_F_CAP_PER_PACKET 0x40000 +/* IPv6 tunnel FMR */ +struct __ip6_tnl_fmr { + struct __ip6_tnl_fmr *next; /* next fmr in list */ + struct in6_addr ip6_prefix; + struct in_addr ip4_prefix; + + __u8 ip6_prefix_len; + __u8 ip4_prefix_len; + __u8 ea_len; + __u8 offset; +}; + struct __ip6_tnl_parm { char name[IFNAMSIZ]; /* name of tunnel device */ int link; /* ifindex of underlying L2 interface */ @@ -29,6 +41,7 @@ struct __ip6_tnl_parm { __u32 flags; /* tunnel flags */ struct in6_addr laddr; /* local tunnel end-point address */ struct in6_addr raddr; /* remote tunnel end-point address */ + struct __ip6_tnl_fmr *fmrs; /* FMRs */ __be16 i_flags; __be16 o_flags; --- a/include/uapi/linux/if_tunnel.h +++ b/include/uapi/linux/if_tunnel.h @@ -77,10 +77,23 @@ enum { IFLA_IPTUN_ENCAP_DPORT, IFLA_IPTUN_COLLECT_METADATA, IFLA_IPTUN_FWMARK, + IFLA_IPTUN_FMRS, __IFLA_IPTUN_MAX, }; #define IFLA_IPTUN_MAX (__IFLA_IPTUN_MAX - 1) +enum { + IFLA_IPTUN_FMR_UNSPEC, + IFLA_IPTUN_FMR_IP6_PREFIX, + IFLA_IPTUN_FMR_IP4_PREFIX, + IFLA_IPTUN_FMR_IP6_PREFIX_LEN, + IFLA_IPTUN_FMR_IP4_PREFIX_LEN, + IFLA_IPTUN_FMR_EA_LEN, + IFLA_IPTUN_FMR_OFFSET, + __IFLA_IPTUN_FMR_MAX, +}; +#define IFLA_IPTUN_FMR_MAX (__IFLA_IPTUN_FMR_MAX - 1) + enum tunnel_encap_types { TUNNEL_ENCAP_NONE, TUNNEL_ENCAP_FOU, --- a/net/ipv6/ip6_tunnel.c +++ b/net/ipv6/ip6_tunnel.c @@ -11,6 +11,9 @@ * linux/net/ipv6/sit.c and linux/net/ipv4/ipip.c * * RFC 2473 + * + * Changes: + * Steven Barth : MAP-E FMR support */ #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt @@ -67,9 +70,9 @@ static bool log_ecn_error = true; module_param(log_ecn_error, bool, 0644); MODULE_PARM_DESC(log_ecn_error, "Log packets received with corrupted ECN"); -static u32 HASH(const struct in6_addr *addr1, const struct in6_addr *addr2) +static u32 HASH(const struct in6_addr *addr) { - u32 hash = ipv6_addr_hash(addr1) ^ ipv6_addr_hash(addr2); + u32 hash = ipv6_addr_hash(addr); return hash_32(hash, IP6_TUNNEL_HASH_SIZE_SHIFT); } @@ -144,17 +147,33 @@ static struct ip6_tnl * ip6_tnl_lookup(struct net *net, int link, const struct in6_addr *remote, const struct in6_addr *local) { - unsigned int hash = HASH(remote, local); + unsigned int hash = HASH(local); struct ip6_tnl *t, *cand = NULL; struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id); struct in6_addr any; for_each_ip6_tunnel_rcu(ip6n->tnls_r_l[hash]) { if (!ipv6_addr_equal(local, &t->parms.laddr) || - !ipv6_addr_equal(remote, &t->parms.raddr) || !(t->dev->flags & IFF_UP)) continue; + if (!ipv6_addr_equal(remote, &t->parms.raddr)) { + struct __ip6_tnl_fmr *fmr; + bool found = false; + + for (fmr = t->parms.fmrs; fmr; fmr = fmr->next) { + if (!ipv6_prefix_equal(remote, &fmr->ip6_prefix, + fmr->ip6_prefix_len)) + continue; + + found = true; + break; + } + + if (!found) + continue; + } + if (link == t->parms.link) return t; else @@ -162,7 +181,7 @@ ip6_tnl_lookup(struct net *net, int link } memset(&any, 0, sizeof(any)); - hash = HASH(&any, local); + hash = HASH(local); for_each_ip6_tunnel_rcu(ip6n->tnls_r_l[hash]) { if (!ipv6_addr_equal(local, &t->parms.laddr) || !ipv6_addr_any(&t->parms.raddr) || @@ -175,7 +194,7 @@ ip6_tnl_lookup(struct net *net, int link cand = t; } - hash = HASH(remote, &any); + hash = HASH(&any); for_each_ip6_tunnel_rcu(ip6n->tnls_r_l[hash]) { if (!ipv6_addr_equal(remote, &t->parms.raddr) || !ipv6_addr_any(&t->parms.laddr) || @@ -223,7 +242,7 @@ ip6_tnl_bucket(struct ip6_tnl_net *ip6n, if (!ipv6_addr_any(remote) || !ipv6_addr_any(local)) { prio = 1; - h = HASH(remote, local); + h = HASH(local); } return &ip6n->tnls[prio][h]; } @@ -405,6 +424,12 @@ ip6_tnl_dev_uninit(struct net_device *de struct net *net = t->net; struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id); + while (t->parms.fmrs) { + struct __ip6_tnl_fmr *next = t->parms.fmrs->next; + kfree(t->parms.fmrs); + t->parms.fmrs = next; + } + if (dev == ip6n->fb_tnl_dev) RCU_INIT_POINTER(ip6n->tnls_wc[0], NULL); else @@ -821,6 +846,107 @@ int ip6_tnl_rcv_ctl(struct ip6_tnl *t, } EXPORT_SYMBOL_GPL(ip6_tnl_rcv_ctl); +/** + * ip4ip6_fmr_calc - calculate target / source IPv6-address based on FMR + * @dest: destination IPv6 address buffer + * @skb: received socket buffer + * @fmr: MAP FMR + * @xmit: Calculate for xmit or rcv + **/ +static void ip4ip6_fmr_calc(struct in6_addr *dest, + const struct iphdr *iph, const uint8_t *end, + const struct __ip6_tnl_fmr *fmr, bool xmit) +{ + int psidlen = fmr->ea_len - (32 - fmr->ip4_prefix_len); + u8 *portp = NULL; + bool use_dest_addr; + const struct iphdr *dsth = iph; + + if ((u8*)dsth >= end) + return; + + /* find significant IP header */ + if (iph->protocol == IPPROTO_ICMP) { + struct icmphdr *ih = (struct icmphdr*)(((u8*)dsth) + dsth->ihl * 4); + if (ih && ((u8*)&ih[1]) <= end && ( + ih->type == ICMP_DEST_UNREACH || + ih->type == ICMP_SOURCE_QUENCH || + ih->type == ICMP_TIME_EXCEEDED || + ih->type == ICMP_PARAMETERPROB || + ih->type == ICMP_REDIRECT)) + dsth = (const struct iphdr*)&ih[1]; + } + + /* in xmit-path use dest port by default and source port only if + this is an ICMP reply to something else; vice versa in rcv-path */ + use_dest_addr = (xmit && dsth == iph) || (!xmit && dsth != iph); + + /* get dst port */ + if (((u8*)&dsth[1]) <= end && ( + dsth->protocol == IPPROTO_UDP || + dsth->protocol == IPPROTO_TCP || + dsth->protocol == IPPROTO_SCTP || + dsth->protocol == IPPROTO_DCCP)) { + /* for UDP, TCP, SCTP and DCCP source and dest port + follow IPv4 header directly */ + portp = ((u8*)dsth) + dsth->ihl * 4; + + if (use_dest_addr) + portp += sizeof(u16); + } else if (iph->protocol == IPPROTO_ICMP) { + struct icmphdr *ih = (struct icmphdr*)(((u8*)dsth) + dsth->ihl * 4); + + /* use icmp identifier as port */ + if (((u8*)&ih) <= end && ( + (use_dest_addr && ( + ih->type == ICMP_ECHOREPLY || + ih->type == ICMP_TIMESTAMPREPLY || + ih->type == ICMP_INFO_REPLY || + ih->type == ICMP_ADDRESSREPLY)) || + (!use_dest_addr && ( + ih->type == ICMP_ECHO || + ih->type == ICMP_TIMESTAMP || + ih->type == ICMP_INFO_REQUEST || + ih->type == ICMP_ADDRESS) + ))) + portp = (u8*)&ih->un.echo.id; + } + + if ((portp && &portp[2] <= end) || psidlen == 0) { + int frombyte = fmr->ip6_prefix_len / 8; + int fromrem = fmr->ip6_prefix_len % 8; + int bytes = sizeof(struct in6_addr) - frombyte; + const u32 *addr = (use_dest_addr) ? &iph->daddr : &iph->saddr; + u64 eabits = ((u64)ntohl(*addr)) << (32 + fmr->ip4_prefix_len); + u64 t = 0; + + /* extract PSID from port and add it to eabits */ + u16 psidbits = 0; + if (psidlen > 0) { + psidbits = ((u16)portp[0]) << 8 | ((u16)portp[1]); + psidbits >>= 16 - psidlen - fmr->offset; + psidbits = (u16)(psidbits << (16 - psidlen)); + eabits |= ((u64)psidbits) << (48 - (fmr->ea_len - psidlen)); + } + + /* rewrite destination address */ + *dest = fmr->ip6_prefix; + memcpy(&dest->s6_addr[10], addr, sizeof(*addr)); + dest->s6_addr16[7] = htons(psidbits >> (16 - psidlen)); + + if (bytes > sizeof(u64)) + bytes = sizeof(u64); + + /* insert eabits */ + memcpy(&t, &dest->s6_addr[frombyte], bytes); + t = be64_to_cpu(t) & ~(((((u64)1) << fmr->ea_len) - 1) + << (64 - fmr->ea_len - fromrem)); + t = cpu_to_be64(t | (eabits >> fromrem)); + memcpy(&dest->s6_addr[frombyte], &t, bytes); + } +} + + static int __ip6_tnl_rcv(struct ip6_tnl *tunnel, struct sk_buff *skb, const struct tnl_ptk_info *tpi, struct metadata_dst *tun_dst, @@ -873,6 +999,27 @@ static int __ip6_tnl_rcv(struct ip6_tnl skb_reset_network_header(skb); memset(skb->cb, 0, sizeof(struct inet6_skb_parm)); + if (tpi->proto == htons(ETH_P_IP) && tunnel->parms.fmrs && + !ipv6_addr_equal(&ipv6h->saddr, &tunnel->parms.raddr)) { + /* Packet didn't come from BR, so lookup FMR */ + struct __ip6_tnl_fmr *fmr; + struct in6_addr expected = tunnel->parms.raddr; + for (fmr = tunnel->parms.fmrs; fmr; fmr = fmr->next) + if (ipv6_prefix_equal(&ipv6h->saddr, + &fmr->ip6_prefix, fmr->ip6_prefix_len)) + break; + + /* Check that IPv6 matches IPv4 source to prevent spoofing */ + if (fmr) + ip4ip6_fmr_calc(&expected, ip_hdr(skb), + skb_tail_pointer(skb), fmr, false); + + if (!ipv6_addr_equal(&ipv6h->saddr, &expected)) { + rcu_read_unlock(); + goto drop; + } + } + __skb_tunnel_rx(skb, tunnel->dev, tunnel->net); err = dscp_ecn_decapsulate(tunnel, ipv6h, skb); @@ -1024,6 +1171,7 @@ static void init_tel_txopt(struct ipv6_t opt->ops.opt_nflen = 8; } + /** * ip6_tnl_addr_conflict - compare packet addresses to tunnel's own * @t: the outgoing tunnel device @@ -1304,6 +1452,7 @@ ipxip6_tnl_xmit(struct sk_buff *skb, str u8 protocol) { struct ip6_tnl *t = netdev_priv(dev); + struct __ip6_tnl_fmr *fmr; struct ipv6hdr *ipv6h; const struct iphdr *iph; int encap_limit = -1; @@ -1403,6 +1552,18 @@ ipxip6_tnl_xmit(struct sk_buff *skb, str fl6.flowi6_uid = sock_net_uid(dev_net(dev), NULL); dsfield = INET_ECN_encapsulate(dsfield, orig_dsfield); + /* try to find matching FMR */ + for (fmr = t->parms.fmrs; fmr; fmr = fmr->next) { + unsigned mshift = 32 - fmr->ip4_prefix_len; + if (ntohl(fmr->ip4_prefix.s_addr) >> mshift == + ntohl(ip_hdr(skb)->daddr) >> mshift) + break; + } + + /* change dstaddr according to FMR */ + if (fmr) + ip4ip6_fmr_calc(&fl6.daddr, ip_hdr(skb), skb_tail_pointer(skb), fmr, true); + if (iptunnel_handle_offloads(skb, SKB_GSO_IPXIP6)) return -1; @@ -1555,6 +1716,14 @@ ip6_tnl_change(struct ip6_tnl *t, const t->parms.link = p->link; t->parms.proto = p->proto; t->parms.fwmark = p->fwmark; + + while (t->parms.fmrs) { + struct __ip6_tnl_fmr *next = t->parms.fmrs->next; + kfree(t->parms.fmrs); + t->parms.fmrs = next; + } + t->parms.fmrs = p->fmrs; + dst_cache_reset(&t->dst_cache); ip6_tnl_link_config(t); return 0; @@ -1593,6 +1762,7 @@ ip6_tnl_parm_from_user(struct __ip6_tnl_ p->flowinfo = u->flowinfo; p->link = u->link; p->proto = u->proto; + p->fmrs = NULL; memcpy(p->name, u->name, sizeof(u->name)); } @@ -1978,6 +2148,15 @@ static int ip6_tnl_validate(struct nlatt return 0; } +static const struct nla_policy ip6_tnl_fmr_policy[IFLA_IPTUN_FMR_MAX + 1] = { + [IFLA_IPTUN_FMR_IP6_PREFIX] = { .len = sizeof(struct in6_addr) }, + [IFLA_IPTUN_FMR_IP4_PREFIX] = { .len = sizeof(struct in_addr) }, + [IFLA_IPTUN_FMR_IP6_PREFIX_LEN] = { .type = NLA_U8 }, + [IFLA_IPTUN_FMR_IP4_PREFIX_LEN] = { .type = NLA_U8 }, + [IFLA_IPTUN_FMR_EA_LEN] = { .type = NLA_U8 }, + [IFLA_IPTUN_FMR_OFFSET] = { .type = NLA_U8 } +}; + static void ip6_tnl_netlink_parms(struct nlattr *data[], struct __ip6_tnl_parm *parms) { @@ -2015,6 +2194,46 @@ static void ip6_tnl_netlink_parms(struct if (data[IFLA_IPTUN_FWMARK]) parms->fwmark = nla_get_u32(data[IFLA_IPTUN_FWMARK]); + + if (data[IFLA_IPTUN_FMRS]) { + unsigned rem; + struct nlattr *fmr; + nla_for_each_nested(fmr, data[IFLA_IPTUN_FMRS], rem) { + struct nlattr *fmrd[IFLA_IPTUN_FMR_MAX + 1], *c; + struct __ip6_tnl_fmr *nfmr; + + nla_parse_nested(fmrd, IFLA_IPTUN_FMR_MAX, + fmr, ip6_tnl_fmr_policy, NULL); + + if (!(nfmr = kzalloc(sizeof(*nfmr), GFP_KERNEL))) + continue; + + nfmr->offset = 6; + + if ((c = fmrd[IFLA_IPTUN_FMR_IP6_PREFIX])) + nla_memcpy(&nfmr->ip6_prefix, fmrd[IFLA_IPTUN_FMR_IP6_PREFIX], + sizeof(nfmr->ip6_prefix)); + + if ((c = fmrd[IFLA_IPTUN_FMR_IP4_PREFIX])) + nla_memcpy(&nfmr->ip4_prefix, fmrd[IFLA_IPTUN_FMR_IP4_PREFIX], + sizeof(nfmr->ip4_prefix)); + + if ((c = fmrd[IFLA_IPTUN_FMR_IP6_PREFIX_LEN])) + nfmr->ip6_prefix_len = nla_get_u8(c); + + if ((c = fmrd[IFLA_IPTUN_FMR_IP4_PREFIX_LEN])) + nfmr->ip4_prefix_len = nla_get_u8(c); + + if ((c = fmrd[IFLA_IPTUN_FMR_EA_LEN])) + nfmr->ea_len = nla_get_u8(c); + + if ((c = fmrd[IFLA_IPTUN_FMR_OFFSET])) + nfmr->offset = nla_get_u8(c); + + nfmr->next = parms->fmrs; + parms->fmrs = nfmr; + } + } } static bool ip6_tnl_netlink_encap_parms(struct nlattr *data[], @@ -2130,6 +2349,12 @@ static void ip6_tnl_dellink(struct net_d static size_t ip6_tnl_get_size(const struct net_device *dev) { + const struct ip6_tnl *t = netdev_priv(dev); + struct __ip6_tnl_fmr *c; + int fmrs = 0; + for (c = t->parms.fmrs; c; c = c->next) + ++fmrs; + return /* IFLA_IPTUN_LINK */ nla_total_size(4) + @@ -2159,6 +2384,24 @@ static size_t ip6_tnl_get_size(const str nla_total_size(0) + /* IFLA_IPTUN_FWMARK */ nla_total_size(4) + + /* IFLA_IPTUN_FMRS */ + nla_total_size(0) + + ( + /* nest */ + nla_total_size(0) + + /* IFLA_IPTUN_FMR_IP6_PREFIX */ + nla_total_size(sizeof(struct in6_addr)) + + /* IFLA_IPTUN_FMR_IP4_PREFIX */ + nla_total_size(sizeof(struct in_addr)) + + /* IFLA_IPTUN_FMR_EA_LEN */ + nla_total_size(1) + + /* IFLA_IPTUN_FMR_IP6_PREFIX_LEN */ + nla_total_size(1) + + /* IFLA_IPTUN_FMR_IP4_PREFIX_LEN */ + nla_total_size(1) + + /* IFLA_IPTUN_FMR_OFFSET */ + nla_total_size(1) + ) * fmrs + 0; } @@ -2166,6 +2409,9 @@ static int ip6_tnl_fill_info(struct sk_b { struct ip6_tnl *tunnel = netdev_priv(dev); struct __ip6_tnl_parm *parm = &tunnel->parms; + struct __ip6_tnl_fmr *c; + int fmrcnt = 0; + struct nlattr *fmrs; if (nla_put_u32(skb, IFLA_IPTUN_LINK, parm->link) || nla_put_in6_addr(skb, IFLA_IPTUN_LOCAL, &parm->laddr) || @@ -2175,9 +2421,27 @@ static int ip6_tnl_fill_info(struct sk_b nla_put_be32(skb, IFLA_IPTUN_FLOWINFO, parm->flowinfo) || nla_put_u32(skb, IFLA_IPTUN_FLAGS, parm->flags) || nla_put_u8(skb, IFLA_IPTUN_PROTO, parm->proto) || - nla_put_u32(skb, IFLA_IPTUN_FWMARK, parm->fwmark)) + nla_put_u32(skb, IFLA_IPTUN_FWMARK, parm->fwmark) || + !(fmrs = nla_nest_start(skb, IFLA_IPTUN_FMRS))) goto nla_put_failure; + for (c = parm->fmrs; c; c = c->next) { + struct nlattr *fmr = nla_nest_start(skb, ++fmrcnt); + if (!fmr || + nla_put(skb, IFLA_IPTUN_FMR_IP6_PREFIX, + sizeof(c->ip6_prefix), &c->ip6_prefix) || + nla_put(skb, IFLA_IPTUN_FMR_IP4_PREFIX, + sizeof(c->ip4_prefix), &c->ip4_prefix) || + nla_put_u8(skb, IFLA_IPTUN_FMR_IP6_PREFIX_LEN, c->ip6_prefix_len) || + nla_put_u8(skb, IFLA_IPTUN_FMR_IP4_PREFIX_LEN, c->ip4_prefix_len) || + nla_put_u8(skb, IFLA_IPTUN_FMR_EA_LEN, c->ea_len) || + nla_put_u8(skb, IFLA_IPTUN_FMR_OFFSET, c->offset)) + goto nla_put_failure; + + nla_nest_end(skb, fmr); + } + nla_nest_end(skb, fmrs); + if (nla_put_u16(skb, IFLA_IPTUN_ENCAP_TYPE, tunnel->encap.type) || nla_put_be16(skb, IFLA_IPTUN_ENCAP_SPORT, tunnel->encap.sport) || nla_put_be16(skb, IFLA_IPTUN_ENCAP_DPORT, tunnel->encap.dport) || @@ -2217,6 +2481,7 @@ static const struct nla_policy ip6_tnl_p [IFLA_IPTUN_ENCAP_DPORT] = { .type = NLA_U16 }, [IFLA_IPTUN_COLLECT_METADATA] = { .type = NLA_FLAG }, [IFLA_IPTUN_FWMARK] = { .type = NLA_U32 }, + [IFLA_IPTUN_FMRS] = { .type = NLA_NESTED }, }; static struct rtnl_link_ops ip6_link_ops __read_mostly = {