diff options
author | Keith Burns (alagalah) <alagalah@gmail.com> | 2016-08-07 08:43:18 -0700 |
---|---|---|
committer | Dave Barach <openvpp@barachs.net> | 2016-08-08 15:32:42 +0000 |
commit | 06e3d075760be43dbd61385d3c7861cd5fe44af2 (patch) | |
tree | d2d5477cc0127c858a1f4b43021a4b0c635c5d00 | |
parent | ba868bb7898edc46ad5f8cd4925af6c8b92e7c89 (diff) |
VPP-263 - Coding standards cleanup - vnet/vnet/map
Change-Id: I401124320a21e30b5cdf3684d36890e51a2fcb6c
Signed-off-by: Keith Burns (alagalah) <alagalah@gmail.com>
-rw-r--r-- | vnet/vnet/map/ip4_map.c | 1040 | ||||
-rw-r--r-- | vnet/vnet/map/ip4_map_t.c | 1826 | ||||
-rw-r--r-- | vnet/vnet/map/ip6_map.c | 1697 | ||||
-rw-r--r-- | vnet/vnet/map/ip6_map_t.c | 2070 | ||||
-rw-r--r-- | vnet/vnet/map/map.c | 1979 | ||||
-rw-r--r-- | vnet/vnet/map/map.h | 65 |
6 files changed, 5035 insertions, 3642 deletions
diff --git a/vnet/vnet/map/ip4_map.c b/vnet/vnet/map/ip4_map.c index a40238f9798..1a75ca3a68b 100644 --- a/vnet/vnet/map/ip4_map.c +++ b/vnet/vnet/map/ip4_map.c @@ -22,7 +22,8 @@ vlib_node_registration_t ip4_map_reass_node; -enum ip4_map_next_e { +enum ip4_map_next_e +{ IP4_MAP_NEXT_IP6_LOOKUP, #ifdef MAP_SKIP_IP6_LOOKUP IP4_MAP_NEXT_IP6_REWRITE, @@ -35,91 +36,121 @@ enum ip4_map_next_e { IP4_MAP_N_NEXT, }; -enum ip4_map_reass_next_t { +enum ip4_map_reass_next_t +{ IP4_MAP_REASS_NEXT_IP6_LOOKUP, IP4_MAP_REASS_NEXT_IP4_FRAGMENT, IP4_MAP_REASS_NEXT_DROP, IP4_MAP_REASS_N_NEXT, }; -typedef struct { +typedef struct +{ u32 map_domain_index; u16 port; u8 cached; } map_ip4_map_reass_trace_t; u8 * -format_ip4_map_reass_trace (u8 *s, va_list *args) +format_ip4_map_reass_trace (u8 * s, va_list * args) { - CLIB_UNUSED(vlib_main_t *vm) = va_arg (*args, vlib_main_t *); - CLIB_UNUSED(vlib_node_t *node) = va_arg (*args, vlib_node_t *); + CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *); + CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *); map_ip4_map_reass_trace_t *t = va_arg (*args, map_ip4_map_reass_trace_t *); - return format(s, "MAP domain index: %d L4 port: %u Status: %s", t->map_domain_index, - t->port, t->cached?"cached":"forwarded"); + return format (s, "MAP domain index: %d L4 port: %u Status: %s", + t->map_domain_index, t->port, + t->cached ? "cached" : "forwarded"); } /* * ip4_map_get_port */ u16 -ip4_map_get_port (ip4_header_t *ip, map_dir_e dir) +ip4_map_get_port (ip4_header_t * ip, map_dir_e dir) { /* Find port information */ - if (PREDICT_TRUE((ip->protocol == IP_PROTOCOL_TCP) || - (ip->protocol == IP_PROTOCOL_UDP))) { - udp_header_t *udp = (void *)(ip + 1); - return (dir == MAP_SENDER ? udp->src_port : udp->dst_port); - } else if (ip->protocol == IP_PROTOCOL_ICMP) { - /* - * 1) ICMP Echo request or Echo reply - * 2) ICMP Error with inner packet being UDP or TCP - * 3) ICMP Error with inner packet being ICMP Echo request or Echo reply - */ - icmp46_header_t *icmp = (void *)(ip + 1); - if (icmp->type == ICMP4_echo_request || icmp->type == ICMP4_echo_reply) { - return *((u16 *)(icmp + 1)); - } else if (clib_net_to_host_u16(ip->length) >= 56) { // IP + ICMP + IP + L4 header - ip4_header_t *icmp_ip = (ip4_header_t *)(icmp + 2); - if (PREDICT_TRUE((icmp_ip->protocol == IP_PROTOCOL_TCP) || - (icmp_ip->protocol == IP_PROTOCOL_UDP))) { - udp_header_t *udp = (void *)(icmp_ip + 1); - return (dir == MAP_SENDER ? udp->dst_port : udp->src_port); - } else if (icmp_ip->protocol == IP_PROTOCOL_ICMP) { - icmp46_header_t *inner_icmp = (void *)(icmp_ip + 1); - if (inner_icmp->type == ICMP4_echo_request || inner_icmp->type == ICMP4_echo_reply) - return (*((u16 *)(inner_icmp + 1))); - } + if (PREDICT_TRUE ((ip->protocol == IP_PROTOCOL_TCP) || + (ip->protocol == IP_PROTOCOL_UDP))) + { + udp_header_t *udp = (void *) (ip + 1); + return (dir == MAP_SENDER ? udp->src_port : udp->dst_port); + } + else if (ip->protocol == IP_PROTOCOL_ICMP) + { + /* + * 1) ICMP Echo request or Echo reply + * 2) ICMP Error with inner packet being UDP or TCP + * 3) ICMP Error with inner packet being ICMP Echo request or Echo reply + */ + icmp46_header_t *icmp = (void *) (ip + 1); + if (icmp->type == ICMP4_echo_request || icmp->type == ICMP4_echo_reply) + { + return *((u16 *) (icmp + 1)); + } + else if (clib_net_to_host_u16 (ip->length) >= 56) + { // IP + ICMP + IP + L4 header + ip4_header_t *icmp_ip = (ip4_header_t *) (icmp + 2); + if (PREDICT_TRUE ((icmp_ip->protocol == IP_PROTOCOL_TCP) || + (icmp_ip->protocol == IP_PROTOCOL_UDP))) + { + udp_header_t *udp = (void *) (icmp_ip + 1); + return (dir == MAP_SENDER ? udp->dst_port : udp->src_port); + } + else if (icmp_ip->protocol == IP_PROTOCOL_ICMP) + { + icmp46_header_t *inner_icmp = (void *) (icmp_ip + 1); + if (inner_icmp->type == ICMP4_echo_request + || inner_icmp->type == ICMP4_echo_reply) + return (*((u16 *) (inner_icmp + 1))); + } + } } - } return (0); } static_always_inline u16 -ip4_map_port_and_security_check (map_domain_t *d, ip4_header_t *ip, u32 *next, u8 *error) +ip4_map_port_and_security_check (map_domain_t * d, ip4_header_t * ip, + u32 * next, u8 * error) { u16 port = 0; - if (d->psid_length > 0) { - if (ip4_get_fragment_offset(ip) == 0) { - if (PREDICT_FALSE((ip->ip_version_and_header_length != 0x45) || clib_host_to_net_u16(ip->length) < 28)) { - return 0; - } - port = ip4_map_get_port(ip, MAP_RECEIVER); - if (port) { - /* Verify that port is not among the well-known ports */ - if ((d->psid_offset > 0) && (clib_net_to_host_u16(port) < (0x1 << (16 - d->psid_offset)))) { - *error = MAP_ERROR_ENCAP_SEC_CHECK; - } else { - if (ip4_get_fragment_more(ip)) *next = IP4_MAP_NEXT_REASS; - return (port); + if (d->psid_length > 0) + { + if (ip4_get_fragment_offset (ip) == 0) + { + if (PREDICT_FALSE + ((ip->ip_version_and_header_length != 0x45) + || clib_host_to_net_u16 (ip->length) < 28)) + { + return 0; + } + port = ip4_map_get_port (ip, MAP_RECEIVER); + if (port) + { + /* Verify that port is not among the well-known ports */ + if ((d->psid_offset > 0) + && (clib_net_to_host_u16 (port) < + (0x1 << (16 - d->psid_offset)))) + { + *error = MAP_ERROR_ENCAP_SEC_CHECK; + } + else + { + if (ip4_get_fragment_more (ip)) + *next = IP4_MAP_NEXT_REASS; + return (port); + } + } + else + { + *error = MAP_ERROR_BAD_PROTOCOL; + } + } + else + { + *next = IP4_MAP_NEXT_REASS; } - } else { - *error = MAP_ERROR_BAD_PROTOCOL; - } - } else { - *next = IP4_MAP_NEXT_REASS; } - } return (0); } @@ -127,33 +158,35 @@ ip4_map_port_and_security_check (map_domain_t *d, ip4_header_t *ip, u32 *next, u * ip4_map_vtcfl */ static_always_inline u32 -ip4_map_vtcfl (ip4_header_t *ip4, vlib_buffer_t *p) +ip4_map_vtcfl (ip4_header_t * ip4, vlib_buffer_t * p) { map_main_t *mm = &map_main; u8 tc = mm->tc_copy ? ip4->tos : mm->tc; u32 vtcfl = 0x6 << 28; vtcfl |= tc << 20; - vtcfl |= vnet_buffer(p)->ip.flow_hash & 0x000fffff; + vtcfl |= vnet_buffer (p)->ip.flow_hash & 0x000fffff; - return (clib_host_to_net_u32(vtcfl)); + return (clib_host_to_net_u32 (vtcfl)); } static_always_inline bool -ip4_map_ip6_lookup_bypass (vlib_buffer_t *p0, ip4_header_t *ip) +ip4_map_ip6_lookup_bypass (vlib_buffer_t * p0, ip4_header_t * ip) { #ifdef MAP_SKIP_IP6_LOOKUP map_main_t *mm = &map_main; u32 adj_index0 = mm->adj6_index; - if (adj_index0 > 0) { - ip_lookup_main_t *lm6 = &ip6_main.lookup_main; - ip_adjacency_t *adj = ip_get_adjacency(lm6, mm->adj6_index); - if (adj->n_adj > 1) { - u32 hash_c0 = ip4_compute_flow_hash(ip, IP_FLOW_HASH_DEFAULT); - adj_index0 += (hash_c0 & (adj->n_adj - 1)); + if (adj_index0 > 0) + { + ip_lookup_main_t *lm6 = &ip6_main.lookup_main; + ip_adjacency_t *adj = ip_get_adjacency (lm6, mm->adj6_index); + if (adj->n_adj > 1) + { + u32 hash_c0 = ip4_compute_flow_hash (ip, IP_FLOW_HASH_DEFAULT); + adj_index0 += (hash_c0 & (adj->n_adj - 1)); + } + vnet_buffer (p0)->ip.adj_index[VLIB_TX] = adj_index0; + return (true); } - vnet_buffer(p0)->ip.adj_index[VLIB_TX] = adj_index0; - return (true); - } #endif return (false); } @@ -162,14 +195,14 @@ ip4_map_ip6_lookup_bypass (vlib_buffer_t *p0, ip4_header_t *ip) * ip4_map_ttl */ static inline void -ip4_map_decrement_ttl (ip4_header_t *ip, u8 *error) +ip4_map_decrement_ttl (ip4_header_t * ip, u8 * error) { i32 ttl = ip->ttl; /* Input node should have reject packets with ttl 0. */ ASSERT (ip->ttl > 0); - u32 checksum = ip->checksum + clib_host_to_net_u16(0x0100); + u32 checksum = ip->checksum + clib_host_to_net_u16 (0x0100); checksum += checksum >= 0xffff; ip->checksum = checksum; ttl -= 1; @@ -177,270 +210,356 @@ ip4_map_decrement_ttl (ip4_header_t *ip, u8 *error) *error = ttl <= 0 ? IP4_ERROR_TIME_EXPIRED : *error; /* Verify checksum. */ - ASSERT (ip->checksum == ip4_header_checksum(ip)); + ASSERT (ip->checksum == ip4_header_checksum (ip)); } static u32 -ip4_map_fragment (vlib_buffer_t *b, u16 mtu, bool df, u8 *error) +ip4_map_fragment (vlib_buffer_t * b, u16 mtu, bool df, u8 * error) { map_main_t *mm = &map_main; - if (mm->frag_inner) { - ip_frag_set_vnet_buffer(b, sizeof(ip6_header_t), mtu, IP4_FRAG_NEXT_IP6_LOOKUP, IP_FRAG_FLAG_IP6_HEADER); - return (IP4_MAP_NEXT_IP4_FRAGMENT); - } else { - if (df && !mm->frag_ignore_df) { - icmp4_error_set_vnet_buffer(b, ICMP4_destination_unreachable, - ICMP4_destination_unreachable_fragmentation_needed_and_dont_fragment_set, mtu); - vlib_buffer_advance(b, sizeof(ip6_header_t)); - *error = MAP_ERROR_DF_SET; - return (IP4_MAP_NEXT_ICMP_ERROR); + if (mm->frag_inner) + { + ip_frag_set_vnet_buffer (b, sizeof (ip6_header_t), mtu, + IP4_FRAG_NEXT_IP6_LOOKUP, + IP_FRAG_FLAG_IP6_HEADER); + return (IP4_MAP_NEXT_IP4_FRAGMENT); + } + else + { + if (df && !mm->frag_ignore_df) + { + icmp4_error_set_vnet_buffer (b, ICMP4_destination_unreachable, + ICMP4_destination_unreachable_fragmentation_needed_and_dont_fragment_set, + mtu); + vlib_buffer_advance (b, sizeof (ip6_header_t)); + *error = MAP_ERROR_DF_SET; + return (IP4_MAP_NEXT_ICMP_ERROR); + } + ip_frag_set_vnet_buffer (b, 0, mtu, IP6_FRAG_NEXT_IP6_LOOKUP, + IP_FRAG_FLAG_IP6_HEADER); + return (IP4_MAP_NEXT_IP6_FRAGMENT); } - ip_frag_set_vnet_buffer(b, 0, mtu, IP6_FRAG_NEXT_IP6_LOOKUP, IP_FRAG_FLAG_IP6_HEADER); - return (IP4_MAP_NEXT_IP6_FRAGMENT); - } } /* * ip4_map */ static uword -ip4_map (vlib_main_t *vm, - vlib_node_runtime_t *node, - vlib_frame_t *frame) +ip4_map (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) { u32 n_left_from, *from, next_index, *to_next, n_left_to_next; - vlib_node_runtime_t *error_node = vlib_node_get_runtime(vm, ip4_map_node.index); - from = vlib_frame_vector_args(frame); + vlib_node_runtime_t *error_node = + vlib_node_get_runtime (vm, ip4_map_node.index); + from = vlib_frame_vector_args (frame); n_left_from = frame->n_vectors; next_index = node->cached_next_index; map_main_t *mm = &map_main; vlib_combined_counter_main_t *cm = mm->domain_counters; - u32 cpu_index = os_get_cpu_number(); - - while (n_left_from > 0) { - vlib_get_next_frame(vm, node, next_index, to_next, n_left_to_next); - - /* Dual loop */ - while (n_left_from >= 4 && n_left_to_next >= 2) { - u32 pi0, pi1; - vlib_buffer_t *p0, *p1; - map_domain_t *d0, *d1; - u8 error0 = MAP_ERROR_NONE, error1 = MAP_ERROR_NONE; - ip4_header_t *ip40, *ip41; - u16 port0 = 0, port1 = 0; - ip6_header_t *ip6h0, *ip6h1; - u32 map_domain_index0 = ~0, map_domain_index1 = ~0; - u32 next0 = IP4_MAP_NEXT_IP6_LOOKUP, next1 = IP4_MAP_NEXT_IP6_LOOKUP; - - /* Prefetch next iteration. */ - { - vlib_buffer_t *p2, *p3; - - p2 = vlib_get_buffer(vm, from[2]); - p3 = vlib_get_buffer(vm, from[3]); - - vlib_prefetch_buffer_header(p2, STORE); - vlib_prefetch_buffer_header(p3, STORE); - /* IPv4 + 8 = 28. possibly plus -40 */ - CLIB_PREFETCH (p2->data-40, 68, STORE); - CLIB_PREFETCH (p3->data-40, 68, STORE); - } - - pi0 = to_next[0] = from[0]; - pi1 = to_next[1] = from[1]; - from += 2; - n_left_from -= 2; - to_next +=2; - n_left_to_next -= 2; - - p0 = vlib_get_buffer(vm, pi0); - p1 = vlib_get_buffer(vm, pi1); - ip40 = vlib_buffer_get_current(p0); - ip41 = vlib_buffer_get_current(p1); - d0 = ip4_map_get_domain(vnet_buffer(p0)->ip.adj_index[VLIB_TX], &map_domain_index0); - d1 = ip4_map_get_domain(vnet_buffer(p1)->ip.adj_index[VLIB_TX], &map_domain_index1); - ASSERT(d0); - ASSERT(d1); - - /* - * Shared IPv4 address - */ - port0 = ip4_map_port_and_security_check(d0, ip40, &next0, &error0); - port1 = ip4_map_port_and_security_check(d1, ip41, &next1, &error1); - - /* Decrement IPv4 TTL */ - ip4_map_decrement_ttl(ip40, &error0); - ip4_map_decrement_ttl(ip41, &error1); - bool df0 = ip40->flags_and_fragment_offset & clib_host_to_net_u16(IP4_HEADER_FLAG_DONT_FRAGMENT); - bool df1 = ip41->flags_and_fragment_offset & clib_host_to_net_u16(IP4_HEADER_FLAG_DONT_FRAGMENT); - - /* MAP calc */ - u32 da40 = clib_net_to_host_u32(ip40->dst_address.as_u32); - u32 da41 = clib_net_to_host_u32(ip41->dst_address.as_u32); - u16 dp40 = clib_net_to_host_u16(port0); - u16 dp41 = clib_net_to_host_u16(port1); - u64 dal60 = map_get_pfx(d0, da40, dp40); - u64 dal61 = map_get_pfx(d1, da41, dp41); - u64 dar60 = map_get_sfx(d0, da40, dp40); - u64 dar61 = map_get_sfx(d1, da41, dp41); - if (dal60 == 0 && dar60 == 0 && error0 == MAP_ERROR_NONE && next0 != IP4_MAP_NEXT_REASS) - error0 = MAP_ERROR_NO_BINDING; - if (dal61 == 0 && dar61 == 0 && error1 == MAP_ERROR_NONE && next1 != IP4_MAP_NEXT_REASS) - error1 = MAP_ERROR_NO_BINDING; - - /* construct ipv6 header */ - vlib_buffer_advance(p0, - sizeof(ip6_header_t)); - vlib_buffer_advance(p1, - sizeof(ip6_header_t)); - ip6h0 = vlib_buffer_get_current(p0); - ip6h1 = vlib_buffer_get_current(p1); - vnet_buffer(p0)->sw_if_index[VLIB_TX] = (u32)~0; - vnet_buffer(p1)->sw_if_index[VLIB_TX] = (u32)~0; - - ip6h0->ip_version_traffic_class_and_flow_label = ip4_map_vtcfl(ip40, p0); - ip6h1->ip_version_traffic_class_and_flow_label = ip4_map_vtcfl(ip41, p1); - ip6h0->payload_length = ip40->length; - ip6h1->payload_length = ip41->length; - ip6h0->protocol = IP_PROTOCOL_IP_IN_IP; - ip6h1->protocol = IP_PROTOCOL_IP_IN_IP; - ip6h0->hop_limit = 0x40; - ip6h1->hop_limit = 0x40; - ip6h0->src_address = d0->ip6_src; - ip6h1->src_address = d1->ip6_src; - ip6h0->dst_address.as_u64[0] = clib_host_to_net_u64(dal60); - ip6h0->dst_address.as_u64[1] = clib_host_to_net_u64(dar60); - ip6h1->dst_address.as_u64[0] = clib_host_to_net_u64(dal61); - ip6h1->dst_address.as_u64[1] = clib_host_to_net_u64(dar61); - - /* - * Determine next node. Can be one of: - * ip6-lookup, ip6-rewrite, ip4-fragment, ip4-virtreass, error-drop - */ - if (PREDICT_TRUE(error0 == MAP_ERROR_NONE)) { - if (PREDICT_FALSE(d0->mtu && (clib_net_to_host_u16(ip6h0->payload_length) + sizeof(*ip6h0) > d0->mtu))) { - next0 = ip4_map_fragment(p0, d0->mtu, df0, &error0); - } else { - next0 = ip4_map_ip6_lookup_bypass(p0, ip40) ? IP4_MAP_NEXT_IP6_REWRITE : next0; - vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_TX, cpu_index, map_domain_index0, 1, - clib_net_to_host_u16(ip6h0->payload_length) + 40); + u32 cpu_index = os_get_cpu_number (); + + while (n_left_from > 0) + { + vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); + + /* Dual loop */ + while (n_left_from >= 4 && n_left_to_next >= 2) + { + u32 pi0, pi1; + vlib_buffer_t *p0, *p1; + map_domain_t *d0, *d1; + u8 error0 = MAP_ERROR_NONE, error1 = MAP_ERROR_NONE; + ip4_header_t *ip40, *ip41; + u16 port0 = 0, port1 = 0; + ip6_header_t *ip6h0, *ip6h1; + u32 map_domain_index0 = ~0, map_domain_index1 = ~0; + u32 next0 = IP4_MAP_NEXT_IP6_LOOKUP, next1 = + IP4_MAP_NEXT_IP6_LOOKUP; + + /* Prefetch next iteration. */ + { + vlib_buffer_t *p2, *p3; + + p2 = vlib_get_buffer (vm, from[2]); + p3 = vlib_get_buffer (vm, from[3]); + + vlib_prefetch_buffer_header (p2, STORE); + vlib_prefetch_buffer_header (p3, STORE); + /* IPv4 + 8 = 28. possibly plus -40 */ + CLIB_PREFETCH (p2->data - 40, 68, STORE); + CLIB_PREFETCH (p3->data - 40, 68, STORE); + } + + pi0 = to_next[0] = from[0]; + pi1 = to_next[1] = from[1]; + from += 2; + n_left_from -= 2; + to_next += 2; + n_left_to_next -= 2; + + p0 = vlib_get_buffer (vm, pi0); + p1 = vlib_get_buffer (vm, pi1); + ip40 = vlib_buffer_get_current (p0); + ip41 = vlib_buffer_get_current (p1); + d0 = + ip4_map_get_domain (vnet_buffer (p0)->ip.adj_index[VLIB_TX], + &map_domain_index0); + d1 = + ip4_map_get_domain (vnet_buffer (p1)->ip.adj_index[VLIB_TX], + &map_domain_index1); + ASSERT (d0); + ASSERT (d1); + + /* + * Shared IPv4 address + */ + port0 = ip4_map_port_and_security_check (d0, ip40, &next0, &error0); + port1 = ip4_map_port_and_security_check (d1, ip41, &next1, &error1); + + /* Decrement IPv4 TTL */ + ip4_map_decrement_ttl (ip40, &error0); + ip4_map_decrement_ttl (ip41, &error1); + bool df0 = + ip40-> + flags_and_fragment_offset & + clib_host_to_net_u16 (IP4_HEADER_FLAG_DONT_FRAGMENT); + bool df1 = + ip41-> + flags_and_fragment_offset & + clib_host_to_net_u16 (IP4_HEADER_FLAG_DONT_FRAGMENT); + + /* MAP calc */ + u32 da40 = clib_net_to_host_u32 (ip40->dst_address.as_u32); + u32 da41 = clib_net_to_host_u32 (ip41->dst_address.as_u32); + u16 dp40 = clib_net_to_host_u16 (port0); + u16 dp41 = clib_net_to_host_u16 (port1); + u64 dal60 = map_get_pfx (d0, da40, dp40); + u64 dal61 = map_get_pfx (d1, da41, dp41); + u64 dar60 = map_get_sfx (d0, da40, dp40); + u64 dar61 = map_get_sfx (d1, da41, dp41); + if (dal60 == 0 && dar60 == 0 && error0 == MAP_ERROR_NONE + && next0 != IP4_MAP_NEXT_REASS) + error0 = MAP_ERROR_NO_BINDING; + if (dal61 == 0 && dar61 == 0 && error1 == MAP_ERROR_NONE + && next1 != IP4_MAP_NEXT_REASS) + error1 = MAP_ERROR_NO_BINDING; + + /* construct ipv6 header */ + vlib_buffer_advance (p0, -sizeof (ip6_header_t)); + vlib_buffer_advance (p1, -sizeof (ip6_header_t)); + ip6h0 = vlib_buffer_get_current (p0); + ip6h1 = vlib_buffer_get_current (p1); + vnet_buffer (p0)->sw_if_index[VLIB_TX] = (u32) ~ 0; + vnet_buffer (p1)->sw_if_index[VLIB_TX] = (u32) ~ 0; + + ip6h0->ip_version_traffic_class_and_flow_label = + ip4_map_vtcfl (ip40, p0); + ip6h1->ip_version_traffic_class_and_flow_label = + ip4_map_vtcfl (ip41, p1); + ip6h0->payload_length = ip40->length; + ip6h1->payload_length = ip41->length; + ip6h0->protocol = IP_PROTOCOL_IP_IN_IP; + ip6h1->protocol = IP_PROTOCOL_IP_IN_IP; + ip6h0->hop_limit = 0x40; + ip6h1->hop_limit = 0x40; + ip6h0->src_address = d0->ip6_src; + ip6h1->src_address = d1->ip6_src; + ip6h0->dst_address.as_u64[0] = clib_host_to_net_u64 (dal60); + ip6h0->dst_address.as_u64[1] = clib_host_to_net_u64 (dar60); + ip6h1->dst_address.as_u64[0] = clib_host_to_net_u64 (dal61); + ip6h1->dst_address.as_u64[1] = clib_host_to_net_u64 (dar61); + + /* + * Determine next node. Can be one of: + * ip6-lookup, ip6-rewrite, ip4-fragment, ip4-virtreass, error-drop + */ + if (PREDICT_TRUE (error0 == MAP_ERROR_NONE)) + { + if (PREDICT_FALSE + (d0->mtu + && (clib_net_to_host_u16 (ip6h0->payload_length) + + sizeof (*ip6h0) > d0->mtu))) + { + next0 = ip4_map_fragment (p0, d0->mtu, df0, &error0); + } + else + { + next0 = + ip4_map_ip6_lookup_bypass (p0, + ip40) ? + IP4_MAP_NEXT_IP6_REWRITE : next0; + vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_TX, + cpu_index, + map_domain_index0, 1, + clib_net_to_host_u16 + (ip6h0->payload_length) + + 40); + } + } + else + { + next0 = IP4_MAP_NEXT_DROP; + } + + /* + * Determine next node. Can be one of: + * ip6-lookup, ip6-rewrite, ip4-fragment, ip4-virtreass, error-drop + */ + if (PREDICT_TRUE (error1 == MAP_ERROR_NONE)) + { + if (PREDICT_FALSE + (d1->mtu + && (clib_net_to_host_u16 (ip6h1->payload_length) + + sizeof (*ip6h1) > d1->mtu))) + { + next1 = ip4_map_fragment (p1, d1->mtu, df1, &error1); + } + else + { + next1 = + ip4_map_ip6_lookup_bypass (p1, + ip41) ? + IP4_MAP_NEXT_IP6_REWRITE : next1; + vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_TX, + cpu_index, + map_domain_index1, 1, + clib_net_to_host_u16 + (ip6h1->payload_length) + + 40); + } + } + else + { + next1 = IP4_MAP_NEXT_DROP; + } + + if (PREDICT_FALSE (p0->flags & VLIB_BUFFER_IS_TRACED)) + { + map_trace_t *tr = vlib_add_trace (vm, node, p0, sizeof (*tr)); + tr->map_domain_index = map_domain_index0; + tr->port = port0; + } + if (PREDICT_FALSE (p1->flags & VLIB_BUFFER_IS_TRACED)) + { + map_trace_t *tr = vlib_add_trace (vm, node, p1, sizeof (*tr)); + tr->map_domain_index = map_domain_index1; + tr->port = port1; + } + + p0->error = error_node->errors[error0]; + p1->error = error_node->errors[error1]; + + vlib_validate_buffer_enqueue_x2 (vm, node, next_index, to_next, + n_left_to_next, pi0, pi1, next0, + next1); } - } else { - next0 = IP4_MAP_NEXT_DROP; - } - /* - * Determine next node. Can be one of: - * ip6-lookup, ip6-rewrite, ip4-fragment, ip4-virtreass, error-drop - */ - if (PREDICT_TRUE(error1 == MAP_ERROR_NONE)) { - if (PREDICT_FALSE(d1->mtu && (clib_net_to_host_u16(ip6h1->payload_length) + sizeof(*ip6h1) > d1->mtu))) { - next1 = ip4_map_fragment(p1, d1->mtu, df1, &error1); - } else { - next1 = ip4_map_ip6_lookup_bypass(p1, ip41) ? IP4_MAP_NEXT_IP6_REWRITE : next1; - vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_TX, cpu_index, map_domain_index1, 1, - clib_net_to_host_u16(ip6h1->payload_length) + 40); + while (n_left_from > 0 && n_left_to_next > 0) + { + u32 pi0; + vlib_buffer_t *p0; + map_domain_t *d0; + u8 error0 = MAP_ERROR_NONE; + ip4_header_t *ip40; + u16 port0 = 0; + ip6_header_t *ip6h0; + u32 next0 = IP4_MAP_NEXT_IP6_LOOKUP; + u32 map_domain_index0 = ~0; + + pi0 = to_next[0] = from[0]; + from += 1; + n_left_from -= 1; + to_next += 1; + n_left_to_next -= 1; + + p0 = vlib_get_buffer (vm, pi0); + ip40 = vlib_buffer_get_current (p0); + d0 = + ip4_map_get_domain (vnet_buffer (p0)->ip.adj_index[VLIB_TX], + &map_domain_index0); + ASSERT (d0); + + /* + * Shared IPv4 address + */ + port0 = ip4_map_port_and_security_check (d0, ip40, &next0, &error0); + + /* Decrement IPv4 TTL */ + ip4_map_decrement_ttl (ip40, &error0); + bool df0 = + ip40-> + flags_and_fragment_offset & + clib_host_to_net_u16 (IP4_HEADER_FLAG_DONT_FRAGMENT); + + /* MAP calc */ + u32 da40 = clib_net_to_host_u32 (ip40->dst_address.as_u32); + u16 dp40 = clib_net_to_host_u16 (port0); + u64 dal60 = map_get_pfx (d0, da40, dp40); + u64 dar60 = map_get_sfx (d0, da40, dp40); + if (dal60 == 0 && dar60 == 0 && error0 == MAP_ERROR_NONE + && next0 != IP4_MAP_NEXT_REASS) + error0 = MAP_ERROR_NO_BINDING; + + /* construct ipv6 header */ + vlib_buffer_advance (p0, -(sizeof (ip6_header_t))); + ip6h0 = vlib_buffer_get_current (p0); + vnet_buffer (p0)->sw_if_index[VLIB_TX] = (u32) ~ 0; + + ip6h0->ip_version_traffic_class_and_flow_label = + ip4_map_vtcfl (ip40, p0); + ip6h0->payload_length = ip40->length; + ip6h0->protocol = IP_PROTOCOL_IP_IN_IP; + ip6h0->hop_limit = 0x40; + ip6h0->src_address = d0->ip6_src; + ip6h0->dst_address.as_u64[0] = clib_host_to_net_u64 (dal60); + ip6h0->dst_address.as_u64[1] = clib_host_to_net_u64 (dar60); + + /* + * Determine next node. Can be one of: + * ip6-lookup, ip6-rewrite, ip4-fragment, ip4-virtreass, error-drop + */ + if (PREDICT_TRUE (error0 == MAP_ERROR_NONE)) + { + if (PREDICT_FALSE + (d0->mtu + && (clib_net_to_host_u16 (ip6h0->payload_length) + + sizeof (*ip6h0) > d0->mtu))) + { + next0 = ip4_map_fragment (p0, d0->mtu, df0, &error0); + } + else + { + next0 = + ip4_map_ip6_lookup_bypass (p0, + ip40) ? + IP4_MAP_NEXT_IP6_REWRITE : next0; + vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_TX, + cpu_index, + map_domain_index0, 1, + clib_net_to_host_u16 + (ip6h0->payload_length) + + 40); + } + } + else + { + next0 = IP4_MAP_NEXT_DROP; + } + + if (PREDICT_FALSE (p0->flags & VLIB_BUFFER_IS_TRACED)) + { + map_trace_t *tr = vlib_add_trace (vm, node, p0, sizeof (*tr)); + tr->map_domain_index = map_domain_index0; + tr->port = port0; + } + + p0->error = error_node->errors[error0]; + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, to_next, + n_left_to_next, pi0, next0); } - } else { - next1 = IP4_MAP_NEXT_DROP; - } - - if (PREDICT_FALSE(p0->flags & VLIB_BUFFER_IS_TRACED)) { - map_trace_t *tr = vlib_add_trace(vm, node, p0, sizeof(*tr)); - tr->map_domain_index = map_domain_index0; - tr->port = port0; - } - if (PREDICT_FALSE(p1->flags & VLIB_BUFFER_IS_TRACED)) { - map_trace_t *tr = vlib_add_trace(vm, node, p1, sizeof(*tr)); - tr->map_domain_index = map_domain_index1; - tr->port = port1; - } - - p0->error = error_node->errors[error0]; - p1->error = error_node->errors[error1]; - - vlib_validate_buffer_enqueue_x2(vm, node, next_index, to_next, n_left_to_next, pi0, pi1, next0, next1); + vlib_put_next_frame (vm, node, next_index, n_left_to_next); } - while (n_left_from > 0 && n_left_to_next > 0) { - u32 pi0; - vlib_buffer_t *p0; - map_domain_t *d0; - u8 error0 = MAP_ERROR_NONE; - ip4_header_t *ip40; - u16 port0 = 0; - ip6_header_t *ip6h0; - u32 next0 = IP4_MAP_NEXT_IP6_LOOKUP; - u32 map_domain_index0 = ~0; - - pi0 = to_next[0] = from[0]; - from += 1; - n_left_from -= 1; - to_next +=1; - n_left_to_next -= 1; - - p0 = vlib_get_buffer(vm, pi0); - ip40 = vlib_buffer_get_current(p0); - d0 = ip4_map_get_domain(vnet_buffer(p0)->ip.adj_index[VLIB_TX], &map_domain_index0); - ASSERT(d0); - - /* - * Shared IPv4 address - */ - port0 = ip4_map_port_and_security_check(d0, ip40, &next0, &error0); - - /* Decrement IPv4 TTL */ - ip4_map_decrement_ttl(ip40, &error0); - bool df0 = ip40->flags_and_fragment_offset & clib_host_to_net_u16(IP4_HEADER_FLAG_DONT_FRAGMENT); - - /* MAP calc */ - u32 da40 = clib_net_to_host_u32(ip40->dst_address.as_u32); - u16 dp40 = clib_net_to_host_u16(port0); - u64 dal60 = map_get_pfx(d0, da40, dp40); - u64 dar60 = map_get_sfx(d0, da40, dp40); - if (dal60 == 0 && dar60 == 0 && error0 == MAP_ERROR_NONE && next0 != IP4_MAP_NEXT_REASS) - error0 = MAP_ERROR_NO_BINDING; - - /* construct ipv6 header */ - vlib_buffer_advance(p0, - (sizeof(ip6_header_t))); - ip6h0 = vlib_buffer_get_current(p0); - vnet_buffer(p0)->sw_if_index[VLIB_TX] = (u32)~0; - - ip6h0->ip_version_traffic_class_and_flow_label = ip4_map_vtcfl(ip40, p0); - ip6h0->payload_length = ip40->length; - ip6h0->protocol = IP_PROTOCOL_IP_IN_IP; - ip6h0->hop_limit = 0x40; - ip6h0->src_address = d0->ip6_src; - ip6h0->dst_address.as_u64[0] = clib_host_to_net_u64(dal60); - ip6h0->dst_address.as_u64[1] = clib_host_to_net_u64(dar60); - - /* - * Determine next node. Can be one of: - * ip6-lookup, ip6-rewrite, ip4-fragment, ip4-virtreass, error-drop - */ - if (PREDICT_TRUE(error0 == MAP_ERROR_NONE)) { - if (PREDICT_FALSE(d0->mtu && (clib_net_to_host_u16(ip6h0->payload_length) + sizeof(*ip6h0) > d0->mtu))) { - next0 = ip4_map_fragment(p0, d0->mtu, df0, &error0); - } else { - next0 = ip4_map_ip6_lookup_bypass(p0, ip40) ? IP4_MAP_NEXT_IP6_REWRITE : next0; - vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_TX, cpu_index, map_domain_index0, 1, - clib_net_to_host_u16(ip6h0->payload_length) + 40); - } - } else { - next0 = IP4_MAP_NEXT_DROP; - } - - if (PREDICT_FALSE(p0->flags & VLIB_BUFFER_IS_TRACED)) { - map_trace_t *tr = vlib_add_trace(vm, node, p0, sizeof(*tr)); - tr->map_domain_index = map_domain_index0; - tr->port = port0; - } - - p0->error = error_node->errors[error0]; - vlib_validate_buffer_enqueue_x1(vm, node, next_index, to_next, n_left_to_next, pi0, next0); - } - vlib_put_next_frame(vm, node, next_index, n_left_to_next); - } - return frame->n_vectors; } @@ -448,143 +567,192 @@ ip4_map (vlib_main_t *vm, * ip4_map_reass */ static uword -ip4_map_reass (vlib_main_t *vm, - vlib_node_runtime_t *node, - vlib_frame_t *frame) +ip4_map_reass (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) { u32 n_left_from, *from, next_index, *to_next, n_left_to_next; - vlib_node_runtime_t *error_node = vlib_node_get_runtime(vm, ip4_map_reass_node.index); - from = vlib_frame_vector_args(frame); + vlib_node_runtime_t *error_node = + vlib_node_get_runtime (vm, ip4_map_reass_node.index); + from = vlib_frame_vector_args (frame); n_left_from = frame->n_vectors; next_index = node->cached_next_index; map_main_t *mm = &map_main; vlib_combined_counter_main_t *cm = mm->domain_counters; - u32 cpu_index = os_get_cpu_number(); + u32 cpu_index = os_get_cpu_number (); u32 *fragments_to_drop = NULL; u32 *fragments_to_loopback = NULL; - while (n_left_from > 0) { - vlib_get_next_frame(vm, node, next_index, to_next, n_left_to_next); - - while (n_left_from > 0 && n_left_to_next > 0) { - u32 pi0; - vlib_buffer_t *p0; - map_domain_t *d0; - u8 error0 = MAP_ERROR_NONE; - ip4_header_t *ip40; - i32 port0 = 0; - ip6_header_t *ip60; - u32 next0 = IP4_MAP_REASS_NEXT_IP6_LOOKUP; - u32 map_domain_index0; - u8 cached = 0; - - pi0 = to_next[0] = from[0]; - from += 1; - n_left_from -= 1; - to_next +=1; - n_left_to_next -= 1; - - p0 = vlib_get_buffer(vm, pi0); - ip60 = vlib_buffer_get_current(p0); - ip40 = (ip4_header_t *)(ip60 + 1); - d0 = ip4_map_get_domain(vnet_buffer(p0)->ip.adj_index[VLIB_TX], &map_domain_index0); - - map_ip4_reass_lock(); - map_ip4_reass_t *r = map_ip4_reass_get(ip40->src_address.as_u32, ip40->dst_address.as_u32, - ip40->fragment_id, ip40->protocol, &fragments_to_drop); - if (PREDICT_FALSE(!r)) { - // Could not create a caching entry - error0 = MAP_ERROR_FRAGMENT_MEMORY; - } else if (PREDICT_TRUE(ip4_get_fragment_offset(ip40))) { - if (r->port >= 0) { - // We know the port already - port0 = r->port; - } else if (map_ip4_reass_add_fragment(r, pi0)) { - // Not enough space for caching - error0 = MAP_ERROR_FRAGMENT_MEMORY; - map_ip4_reass_free(r, &fragments_to_drop); - } else { - cached = 1; - } - } else if ((port0 = ip4_get_port(ip40, MAP_RECEIVER, p0->current_length)) < 0) { - // Could not find port. We'll free the reassembly. - error0 = MAP_ERROR_BAD_PROTOCOL; - port0 = 0; - map_ip4_reass_free(r, &fragments_to_drop); - } else { - r->port = port0; - map_ip4_reass_get_fragments(r, &fragments_to_loopback); - } + while (n_left_from > 0) + { + vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); + + while (n_left_from > 0 && n_left_to_next > 0) + { + u32 pi0; + vlib_buffer_t *p0; + map_domain_t *d0; + u8 error0 = MAP_ERROR_NONE; + ip4_header_t *ip40; + i32 port0 = 0; + ip6_header_t *ip60; + u32 next0 = IP4_MAP_REASS_NEXT_IP6_LOOKUP; + u32 map_domain_index0; + u8 cached = 0; + + pi0 = to_next[0] = from[0]; + from += 1; + n_left_from -= 1; + to_next += 1; + n_left_to_next -= 1; + + p0 = vlib_get_buffer (vm, pi0); + ip60 = vlib_buffer_get_current (p0); + ip40 = (ip4_header_t *) (ip60 + 1); + d0 = + ip4_map_get_domain (vnet_buffer (p0)->ip.adj_index[VLIB_TX], + &map_domain_index0); + + map_ip4_reass_lock (); + map_ip4_reass_t *r = + map_ip4_reass_get (ip40->src_address.as_u32, + ip40->dst_address.as_u32, + ip40->fragment_id, ip40->protocol, + &fragments_to_drop); + if (PREDICT_FALSE (!r)) + { + // Could not create a caching entry + error0 = MAP_ERROR_FRAGMENT_MEMORY; + } + else if (PREDICT_TRUE (ip4_get_fragment_offset (ip40))) + { + if (r->port >= 0) + { + // We know the port already + port0 = r->port; + } + else if (map_ip4_reass_add_fragment (r, pi0)) + { + // Not enough space for caching + error0 = MAP_ERROR_FRAGMENT_MEMORY; + map_ip4_reass_free (r, &fragments_to_drop); + } + else + { + cached = 1; + } + } + else + if ((port0 = + ip4_get_port (ip40, MAP_RECEIVER, p0->current_length)) < 0) + { + // Could not find port. We'll free the reassembly. + error0 = MAP_ERROR_BAD_PROTOCOL; + port0 = 0; + map_ip4_reass_free (r, &fragments_to_drop); + } + else + { + r->port = port0; + map_ip4_reass_get_fragments (r, &fragments_to_loopback); + } #ifdef MAP_IP4_REASS_COUNT_BYTES - if (!cached && r) { - r->forwarded += clib_host_to_net_u16(ip40->length) - 20; - if (!ip4_get_fragment_more(ip40)) - r->expected_total = ip4_get_fragment_offset(ip40) * 8 + clib_host_to_net_u16(ip40->length) - 20; - if(r->forwarded >= r->expected_total) - map_ip4_reass_free(r, &fragments_to_drop); - } + if (!cached && r) + { + r->forwarded += clib_host_to_net_u16 (ip40->length) - 20; + if (!ip4_get_fragment_more (ip40)) + r->expected_total = + ip4_get_fragment_offset (ip40) * 8 + + clib_host_to_net_u16 (ip40->length) - 20; + if (r->forwarded >= r->expected_total) + map_ip4_reass_free (r, &fragments_to_drop); + } #endif - map_ip4_reass_unlock(); - - // NOTE: Most operations have already been performed by ip4_map - // All we need is the right destination address - ip60->dst_address.as_u64[0] = map_get_pfx_net(d0, ip40->dst_address.as_u32, port0); - ip60->dst_address.as_u64[1] = map_get_sfx_net(d0, ip40->dst_address.as_u32, port0); - - if (PREDICT_FALSE(d0->mtu && (clib_net_to_host_u16(ip60->payload_length) + sizeof(*ip60) > d0->mtu))) { - vnet_buffer(p0)->ip_frag.header_offset = sizeof(*ip60); - vnet_buffer(p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP6_LOOKUP; - vnet_buffer(p0)->ip_frag.mtu = d0->mtu; - vnet_buffer(p0)->ip_frag.flags = IP_FRAG_FLAG_IP6_HEADER; - next0 = IP4_MAP_REASS_NEXT_IP4_FRAGMENT; - } - - if (PREDICT_FALSE(p0->flags & VLIB_BUFFER_IS_TRACED)) { - map_ip4_map_reass_trace_t *tr = vlib_add_trace(vm, node, p0, sizeof(*tr)); - tr->map_domain_index = map_domain_index0; - tr->port = port0; - tr->cached = cached; - } - - if(cached) { - //Dequeue the packet - n_left_to_next++; - to_next--; - } else { - if (error0 == MAP_ERROR_NONE) - vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_TX, cpu_index, map_domain_index0, 1, - clib_net_to_host_u16(ip60->payload_length) + 40); - next0 = (error0 == MAP_ERROR_NONE) ? next0 : IP4_MAP_REASS_NEXT_DROP; - p0->error = error_node->errors[error0]; - vlib_validate_buffer_enqueue_x1(vm, node, next_index, to_next, n_left_to_next, pi0, next0); - } - - //Loopback when we reach the end of the inpu vector - if(n_left_from == 0 && vec_len(fragments_to_loopback)) { - from = vlib_frame_vector_args(frame); - u32 len = vec_len(fragments_to_loopback); - if(len <= VLIB_FRAME_SIZE) { - clib_memcpy(from, fragments_to_loopback, sizeof(u32)*len); - n_left_from = len; - vec_reset_length(fragments_to_loopback); - } else { - clib_memcpy(from, fragments_to_loopback + (len - VLIB_FRAME_SIZE), sizeof(u32)*VLIB_FRAME_SIZE); - n_left_from = VLIB_FRAME_SIZE; - _vec_len(fragments_to_loopback) = len - VLIB_FRAME_SIZE; - } - } + map_ip4_reass_unlock (); + + // NOTE: Most operations have already been performed by ip4_map + // All we need is the right destination address + ip60->dst_address.as_u64[0] = + map_get_pfx_net (d0, ip40->dst_address.as_u32, port0); + ip60->dst_address.as_u64[1] = + map_get_sfx_net (d0, ip40->dst_address.as_u32, port0); + + if (PREDICT_FALSE + (d0->mtu + && (clib_net_to_host_u16 (ip60->payload_length) + + sizeof (*ip60) > d0->mtu))) + { + vnet_buffer (p0)->ip_frag.header_offset = sizeof (*ip60); + vnet_buffer (p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP6_LOOKUP; + vnet_buffer (p0)->ip_frag.mtu = d0->mtu; + vnet_buffer (p0)->ip_frag.flags = IP_FRAG_FLAG_IP6_HEADER; + next0 = IP4_MAP_REASS_NEXT_IP4_FRAGMENT; + } + + if (PREDICT_FALSE (p0->flags & VLIB_BUFFER_IS_TRACED)) + { + map_ip4_map_reass_trace_t *tr = + vlib_add_trace (vm, node, p0, sizeof (*tr)); + tr->map_domain_index = map_domain_index0; + tr->port = port0; + tr->cached = cached; + } + + if (cached) + { + //Dequeue the packet + n_left_to_next++; + to_next--; + } + else + { + if (error0 == MAP_ERROR_NONE) + vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_TX, + cpu_index, map_domain_index0, + 1, + clib_net_to_host_u16 (ip60-> + payload_length) + + 40); + next0 = + (error0 == MAP_ERROR_NONE) ? next0 : IP4_MAP_REASS_NEXT_DROP; + p0->error = error_node->errors[error0]; + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, to_next, + n_left_to_next, pi0, next0); + } + + //Loopback when we reach the end of the inpu vector + if (n_left_from == 0 && vec_len (fragments_to_loopback)) + { + from = vlib_frame_vector_args (frame); + u32 len = vec_len (fragments_to_loopback); + if (len <= VLIB_FRAME_SIZE) + { + clib_memcpy (from, fragments_to_loopback, + sizeof (u32) * len); + n_left_from = len; + vec_reset_length (fragments_to_loopback); + } + else + { + clib_memcpy (from, + fragments_to_loopback + (len - + VLIB_FRAME_SIZE), + sizeof (u32) * VLIB_FRAME_SIZE); + n_left_from = VLIB_FRAME_SIZE; + _vec_len (fragments_to_loopback) = len - VLIB_FRAME_SIZE; + } + } + } + vlib_put_next_frame (vm, node, next_index, n_left_to_next); } - vlib_put_next_frame(vm, node, next_index, n_left_to_next); - } - map_send_all_to_node(vm, fragments_to_drop, node, - &error_node->errors[MAP_ERROR_FRAGMENT_DROPPED], - IP4_MAP_REASS_NEXT_DROP); + map_send_all_to_node (vm, fragments_to_drop, node, + &error_node->errors[MAP_ERROR_FRAGMENT_DROPPED], + IP4_MAP_REASS_NEXT_DROP); - vec_free(fragments_to_drop); - vec_free(fragments_to_loopback); + vec_free (fragments_to_drop); + vec_free (fragments_to_loopback); return frame->n_vectors; } @@ -594,6 +762,7 @@ static char *map_error_strings[] = { #undef _ }; +/* *INDENT-OFF* */ VLIB_REGISTER_NODE(ip4_map_node) = { .function = ip4_map, .name = "ip4-map", @@ -617,7 +786,9 @@ VLIB_REGISTER_NODE(ip4_map_node) = { [IP4_MAP_NEXT_DROP] = "error-drop", }, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_REGISTER_NODE(ip4_map_reass_node) = { .function = ip4_map_reass, .name = "ip4-map-reass", @@ -635,3 +806,12 @@ VLIB_REGISTER_NODE(ip4_map_reass_node) = { [IP4_MAP_REASS_NEXT_DROP] = "error-drop", }, }; +/* *INDENT-ON* */ + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/vnet/vnet/map/ip4_map_t.c b/vnet/vnet/map/ip4_map_t.c index 78cfae79e35..f4bae608e4d 100644 --- a/vnet/vnet/map/ip4_map_t.c +++ b/vnet/vnet/map/ip4_map_t.c @@ -18,7 +18,8 @@ #define IP4_MAP_T_DUAL_LOOP 1 -typedef enum { +typedef enum +{ IP4_MAPT_NEXT_MAPT_TCP_UDP, IP4_MAPT_NEXT_MAPT_ICMP, IP4_MAPT_NEXT_MAPT_FRAGMENTED, @@ -26,21 +27,24 @@ typedef enum { IP4_MAPT_N_NEXT } ip4_mapt_next_t; -typedef enum { +typedef enum +{ IP4_MAPT_ICMP_NEXT_IP6_LOOKUP, IP4_MAPT_ICMP_NEXT_IP6_FRAG, IP4_MAPT_ICMP_NEXT_DROP, IP4_MAPT_ICMP_N_NEXT } ip4_mapt_icmp_next_t; -typedef enum { +typedef enum +{ IP4_MAPT_TCP_UDP_NEXT_IP6_LOOKUP, IP4_MAPT_TCP_UDP_NEXT_IP6_FRAG, IP4_MAPT_TCP_UDP_NEXT_DROP, IP4_MAPT_TCP_UDP_N_NEXT } ip4_mapt_tcp_udp_next_t; -typedef enum { +typedef enum +{ IP4_MAPT_FRAGMENTED_NEXT_IP6_LOOKUP, IP4_MAPT_FRAGMENTED_NEXT_IP6_FRAG, IP4_MAPT_FRAGMENTED_NEXT_DROP, @@ -49,51 +53,60 @@ typedef enum { //This is used to pass information within the buffer data. //Buffer structure being too small to contain big structures like this. -typedef CLIB_PACKED(struct { +/* *INDENT-OFF* */ +typedef CLIB_PACKED (struct { ip6_address_t daddr; ip6_address_t saddr; //IPv6 header + Fragmentation header will be here //sizeof(ip6) + sizeof(ip_frag) - sizeof(ip4) u8 unused[28]; }) ip4_mapt_pseudo_header_t; +/* *INDENT-ON* */ #define frag_id_4to6(id) (id) //TODO: Find the right place in memory for this. +/* *INDENT-OFF* */ static u8 icmp_to_icmp6_updater_pointer_table[] = - { 0, 1, 4, 4,~0, - ~0,~0,~0, 7, 6, - ~0,~0, 8, 8, 8, - 8, 24, 24, 24, 24 }; + { 0, 1, 4, 4, ~0, + ~0, ~0, ~0, 7, 6, + ~0, ~0, 8, 8, 8, + 8, 24, 24, 24, 24 + }; +/* *INDENT-ON* */ static_always_inline int -ip4_map_fragment_cache (ip4_header_t *ip4, u16 port) +ip4_map_fragment_cache (ip4_header_t * ip4, u16 port) { u32 *ignore = NULL; - map_ip4_reass_lock(); - map_ip4_reass_t *r = map_ip4_reass_get(ip4->src_address.as_u32, ip4->dst_address.as_u32, - ip4->fragment_id, - (ip4->protocol == IP_PROTOCOL_ICMP) ? IP_PROTOCOL_ICMP6 : ip4->protocol, - &ignore); + map_ip4_reass_lock (); + map_ip4_reass_t *r = + map_ip4_reass_get (ip4->src_address.as_u32, ip4->dst_address.as_u32, + ip4->fragment_id, + (ip4->protocol == + IP_PROTOCOL_ICMP) ? IP_PROTOCOL_ICMP6 : ip4->protocol, + &ignore); if (r) r->port = port; - map_ip4_reass_unlock(); + map_ip4_reass_unlock (); return !r; } static_always_inline i32 -ip4_map_fragment_get_port (ip4_header_t *ip4) +ip4_map_fragment_get_port (ip4_header_t * ip4) { u32 *ignore = NULL; - map_ip4_reass_lock(); - map_ip4_reass_t *r = map_ip4_reass_get(ip4->src_address.as_u32, ip4->dst_address.as_u32, - ip4->fragment_id, - (ip4->protocol == IP_PROTOCOL_ICMP) ? IP_PROTOCOL_ICMP6 : ip4->protocol, - &ignore); - i32 ret = r?r->port:-1; - map_ip4_reass_unlock(); + map_ip4_reass_lock (); + map_ip4_reass_t *r = + map_ip4_reass_get (ip4->src_address.as_u32, ip4->dst_address.as_u32, + ip4->fragment_id, + (ip4->protocol == + IP_PROTOCOL_ICMP) ? IP_PROTOCOL_ICMP6 : ip4->protocol, + &ignore); + i32 ret = r ? r->port : -1; + map_ip4_reass_unlock (); return ret; } @@ -104,114 +117,121 @@ ip4_map_fragment_get_port (ip4_header_t *ip4) * */ static_always_inline int -ip4_icmp_to_icmp6_in_place (icmp46_header_t *icmp, u32 icmp_len, - i32 *receiver_port, ip4_header_t **inner_ip4) +ip4_icmp_to_icmp6_in_place (icmp46_header_t * icmp, u32 icmp_len, + i32 * receiver_port, ip4_header_t ** inner_ip4) { *inner_ip4 = NULL; - switch (icmp->type) { + switch (icmp->type) + { case ICMP4_echo_reply: - *receiver_port = ((u16 *)icmp)[2]; + *receiver_port = ((u16 *) icmp)[2]; icmp->type = ICMP6_echo_reply; break; case ICMP4_echo_request: - *receiver_port = ((u16 *)icmp)[2]; + *receiver_port = ((u16 *) icmp)[2]; icmp->type = ICMP6_echo_request; break; case ICMP4_destination_unreachable: - *inner_ip4 = (ip4_header_t *)(((u8 *) icmp) + 8); - *receiver_port = ip4_get_port(*inner_ip4, MAP_SENDER, icmp_len - 8); + *inner_ip4 = (ip4_header_t *) (((u8 *) icmp) + 8); + *receiver_port = ip4_get_port (*inner_ip4, MAP_SENDER, icmp_len - 8); - switch (icmp->code) { - case ICMP4_destination_unreachable_destination_unreachable_net: //0 - case ICMP4_destination_unreachable_destination_unreachable_host: //1 + switch (icmp->code) + { + case ICMP4_destination_unreachable_destination_unreachable_net: //0 + case ICMP4_destination_unreachable_destination_unreachable_host: //1 icmp->type = ICMP6_destination_unreachable; icmp->code = ICMP6_destination_unreachable_no_route_to_destination; break; - case ICMP4_destination_unreachable_protocol_unreachable: //2 + case ICMP4_destination_unreachable_protocol_unreachable: //2 icmp->type = ICMP6_parameter_problem; icmp->code = ICMP6_parameter_problem_unrecognized_next_header; break; - case ICMP4_destination_unreachable_port_unreachable: //3 + case ICMP4_destination_unreachable_port_unreachable: //3 icmp->type = ICMP6_destination_unreachable; icmp->code = ICMP6_destination_unreachable_port_unreachable; break; - case ICMP4_destination_unreachable_fragmentation_needed_and_dont_fragment_set: //4 - icmp->type = ICMP6_packet_too_big; + case ICMP4_destination_unreachable_fragmentation_needed_and_dont_fragment_set: //4 + icmp->type = + ICMP6_packet_too_big; icmp->code = 0; { - u32 advertised_mtu = clib_net_to_host_u32(*((u32 *)(icmp + 1))); + u32 advertised_mtu = clib_net_to_host_u32 (*((u32 *) (icmp + 1))); if (advertised_mtu) advertised_mtu += 20; else - advertised_mtu = 1000; //FIXME ! (RFC 1191 - plateau value) + advertised_mtu = 1000; //FIXME ! (RFC 1191 - plateau value) //FIXME: = minimum(advertised MTU+20, MTU_of_IPv6_nexthop, (MTU_of_IPv4_nexthop)+20) - *((u32 *)(icmp + 1)) = clib_host_to_net_u32(advertised_mtu); + *((u32 *) (icmp + 1)) = clib_host_to_net_u32 (advertised_mtu); } break; - case ICMP4_destination_unreachable_source_route_failed: //5 - case ICMP4_destination_unreachable_destination_network_unknown: //6 - case ICMP4_destination_unreachable_destination_host_unknown: //7 - case ICMP4_destination_unreachable_source_host_isolated: //8 - case ICMP4_destination_unreachable_network_unreachable_for_type_of_service: //11 - case ICMP4_destination_unreachable_host_unreachable_for_type_of_service: //12 - icmp->type = ICMP6_destination_unreachable; + case ICMP4_destination_unreachable_source_route_failed: //5 + case ICMP4_destination_unreachable_destination_network_unknown: //6 + case ICMP4_destination_unreachable_destination_host_unknown: //7 + case ICMP4_destination_unreachable_source_host_isolated: //8 + case ICMP4_destination_unreachable_network_unreachable_for_type_of_service: //11 + case ICMP4_destination_unreachable_host_unreachable_for_type_of_service: //12 + icmp->type = + ICMP6_destination_unreachable; icmp->code = ICMP6_destination_unreachable_no_route_to_destination; break; - case ICMP4_destination_unreachable_network_administratively_prohibited: //9 - case ICMP4_destination_unreachable_host_administratively_prohibited: //10 - case ICMP4_destination_unreachable_communication_administratively_prohibited: //13 - case ICMP4_destination_unreachable_precedence_cutoff_in_effect: //15 + case ICMP4_destination_unreachable_network_administratively_prohibited: //9 + case ICMP4_destination_unreachable_host_administratively_prohibited: //10 + case ICMP4_destination_unreachable_communication_administratively_prohibited: //13 + case ICMP4_destination_unreachable_precedence_cutoff_in_effect: //15 icmp->type = ICMP6_destination_unreachable; - icmp->code = ICMP6_destination_unreachable_destination_administratively_prohibited; + icmp->code = + ICMP6_destination_unreachable_destination_administratively_prohibited; break; - case ICMP4_destination_unreachable_host_precedence_violation: //14 + case ICMP4_destination_unreachable_host_precedence_violation: //14 default: return -1; - } + } break; - case ICMP4_time_exceeded: //11 - *inner_ip4 = (ip4_header_t *)(((u8 *) icmp) + 8); - *receiver_port = ip4_get_port(*inner_ip4, MAP_SENDER, icmp_len - 8); + case ICMP4_time_exceeded: //11 + *inner_ip4 = (ip4_header_t *) (((u8 *) icmp) + 8); + *receiver_port = ip4_get_port (*inner_ip4, MAP_SENDER, icmp_len - 8); icmp->type = ICMP6_time_exceeded; //icmp->code = icmp->code //unchanged break; case ICMP4_parameter_problem: - *inner_ip4 = (ip4_header_t *)(((u8 *) icmp) + 8); - *receiver_port = ip4_get_port(*inner_ip4, MAP_SENDER, icmp_len - 8); + *inner_ip4 = (ip4_header_t *) (((u8 *) icmp) + 8); + *receiver_port = ip4_get_port (*inner_ip4, MAP_SENDER, icmp_len - 8); - switch (icmp->code) { + switch (icmp->code) + { case ICMP4_parameter_problem_pointer_indicates_error: case ICMP4_parameter_problem_bad_length: icmp->type = ICMP6_parameter_problem; icmp->code = ICMP6_parameter_problem_erroneous_header_field; { - u8 ptr = icmp_to_icmp6_updater_pointer_table[*((u8 *)(icmp + 1))]; + u8 ptr = + icmp_to_icmp6_updater_pointer_table[*((u8 *) (icmp + 1))]; if (ptr == 0xff) return -1; - *((u32 *)(icmp + 1)) = clib_host_to_net_u32(ptr); + *((u32 *) (icmp + 1)) = clib_host_to_net_u32 (ptr); } break; - default: - //All other codes cause dropping the packet + default: + //All other codes cause dropping the packet return -1; - } + } break; default: //All other types cause dropping the packet return -1; break; - } + } return 0; } static_always_inline void -_ip4_map_t_icmp (map_domain_t *d, vlib_buffer_t *p, u8 *error) +_ip4_map_t_icmp (map_domain_t * d, vlib_buffer_t * p, u8 * error) { ip4_header_t *ip4, *inner_ip4; ip6_header_t *ip6, *inner_ip6; @@ -225,790 +245,1024 @@ _ip4_map_t_icmp (map_domain_t *d, vlib_buffer_t *p, u8 *error) u32 inner_frag_offset; u8 inner_frag_more; - ip4 = vlib_buffer_get_current(p); - ip_len = clib_net_to_host_u16(ip4->length); - ASSERT(ip_len <= p->current_length); - - icmp = (icmp46_header_t *)(ip4 + 1); - if (ip4_icmp_to_icmp6_in_place(icmp, ip_len - sizeof(*ip4), - &recv_port, &inner_ip4)) { - *error = MAP_ERROR_ICMP; - return; - } - - if (recv_port < 0) { - // In case of 1:1 mapping, we don't care about the port - if(d->ea_bits_len == 0 && d->rules) { - recv_port = 0; - } else { + ip4 = vlib_buffer_get_current (p); + ip_len = clib_net_to_host_u16 (ip4->length); + ASSERT (ip_len <= p->current_length); + + icmp = (icmp46_header_t *) (ip4 + 1); + if (ip4_icmp_to_icmp6_in_place (icmp, ip_len - sizeof (*ip4), + &recv_port, &inner_ip4)) + { *error = MAP_ERROR_ICMP; return; } - } - - if (inner_ip4) { - //We have 2 headers to translate. - //We need to make some room in the middle of the packet - - if (PREDICT_FALSE(ip4_is_fragment(inner_ip4))) { - //Here it starts getting really tricky - //We will add a fragmentation header in the inner packet - - if (!ip4_is_first_fragment(inner_ip4)) { - //For now we do not handle unless it is the first fragment - //Ideally we should handle the case as we are in slow path already - *error = MAP_ERROR_FRAGMENTED; - return; - } - - vlib_buffer_advance(p, - 2*(sizeof(*ip6) - sizeof(*ip4)) - sizeof(*inner_frag)); - ip6 = vlib_buffer_get_current(p); - clib_memcpy(u8_ptr_add(ip6, sizeof(*ip6) - sizeof(*ip4)), ip4, 20 + 8); - ip4 = (ip4_header_t *) u8_ptr_add(ip6, sizeof(*ip6) - sizeof(*ip4)); - icmp = (icmp46_header_t *) (ip4 + 1); - - inner_ip6 = (ip6_header_t *) u8_ptr_add(inner_ip4, sizeof(*ip4) - sizeof(*ip6) - sizeof(*inner_frag)); - inner_frag = (ip6_frag_hdr_t *) u8_ptr_add(inner_ip6, sizeof(*inner_ip6)); - ip6->payload_length = u16_net_add(ip4->length, sizeof(*ip6) - 2*sizeof(*ip4) + sizeof(*inner_frag)); - inner_frag_id = frag_id_4to6(inner_ip4->fragment_id); - inner_frag_offset = ip4_get_fragment_offset(inner_ip4); - inner_frag_more = !!(inner_ip4->flags_and_fragment_offset & clib_net_to_host_u16(IP4_HEADER_FLAG_MORE_FRAGMENTS)); - } else { - vlib_buffer_advance(p, - 2*(sizeof(*ip6) - sizeof(*ip4))); - ip6 = vlib_buffer_get_current(p); - clib_memcpy(u8_ptr_add(ip6, sizeof(*ip6) - sizeof(*ip4)), ip4, 20 + 8); - ip4 = (ip4_header_t *) u8_ptr_add(ip6, sizeof(*ip6) - sizeof(*ip4)); - icmp = (icmp46_header_t *) u8_ptr_add(ip4, sizeof(*ip4)); - inner_ip6 = (ip6_header_t *) u8_ptr_add(inner_ip4, sizeof(*ip4) - sizeof(*ip6)); - ip6->payload_length = u16_net_add(ip4->length, sizeof(*ip6) - 2*sizeof(*ip4)); - inner_frag = NULL; - } - if (PREDICT_TRUE(inner_ip4->protocol == IP_PROTOCOL_TCP)) { - inner_L4_checksum = &((tcp_header_t *) (inner_ip4 + 1))->checksum; - *inner_L4_checksum = ip_csum_fold(ip_csum_sub_even(*inner_L4_checksum, *((u64 *) (&inner_ip4->src_address)))); - } else if (PREDICT_TRUE(inner_ip4->protocol == IP_PROTOCOL_UDP)) { - inner_L4_checksum = &((udp_header_t *) (inner_ip4 + 1))->checksum; - if (!*inner_L4_checksum) { - //The inner packet was first translated, and therefore came from IPv6. - //As the packet was an IPv6 packet, the UDP checksum can't be NULL - *error = MAP_ERROR_ICMP; - return; - } - *inner_L4_checksum = ip_csum_fold(ip_csum_sub_even(*inner_L4_checksum, *((u64 *)(&inner_ip4->src_address)))); - } else if (inner_ip4->protocol == IP_PROTOCOL_ICMP) { - //We have an ICMP inside an ICMP - //It needs to be translated, but not for error ICMP messages - icmp46_header_t *inner_icmp = (icmp46_header_t *) (inner_ip4 + 1); - csum = inner_icmp->checksum; - //Only types ICMP4_echo_request and ICMP4_echo_reply are handled by ip4_icmp_to_icmp6_in_place - csum = ip_csum_sub_even(csum, *((u16 *)inner_icmp)); - inner_icmp->type = (inner_icmp->type == ICMP4_echo_request)? - ICMP6_echo_request:ICMP6_echo_reply; - csum = ip_csum_add_even(csum, *((u16 *)inner_icmp)); - csum = ip_csum_add_even(csum, clib_host_to_net_u16(IP_PROTOCOL_ICMP6)); - csum = ip_csum_add_even(csum, inner_ip4->length - sizeof(*inner_ip4)); - inner_icmp->checksum = ip_csum_fold(csum); - inner_L4_checksum = &inner_icmp->checksum; - inner_ip4->protocol = IP_PROTOCOL_ICMP6; - } else { - ASSERT(0); // We had a port from that, so it is udp or tcp or ICMP + if (recv_port < 0) + { + // In case of 1:1 mapping, we don't care about the port + if (d->ea_bits_len == 0 && d->rules) + { + recv_port = 0; + } + else + { + *error = MAP_ERROR_ICMP; + return; + } } - //FIXME: Security check with the port found in the inner packet - - csum = *inner_L4_checksum; //Initial checksum of the inner L4 header - //FIXME: Shouldn't we remove ip addresses from there ? - - inner_ip6->ip_version_traffic_class_and_flow_label = clib_host_to_net_u32((6 << 28) + (inner_ip4->tos << 20)); - inner_ip6->payload_length = u16_net_add(inner_ip4->length, - sizeof(*inner_ip4)); - inner_ip6->hop_limit = inner_ip4->ttl; - inner_ip6->protocol = inner_ip4->protocol; - - //Note that the source address is within the domain - //while the destination address is the one outside the domain - ip4_map_t_embedded_address(d, &inner_ip6->dst_address, &inner_ip4->dst_address); - inner_ip6->src_address.as_u64[0] = map_get_pfx_net(d, inner_ip4->src_address.as_u32, recv_port); - inner_ip6->src_address.as_u64[1] = map_get_sfx_net(d, inner_ip4->src_address.as_u32, recv_port); - - if (PREDICT_FALSE(inner_frag != NULL)) { - inner_frag->next_hdr = inner_ip6->protocol; - inner_frag->identification = inner_frag_id; - inner_frag->rsv = 0; - inner_frag->fragment_offset_and_more = ip6_frag_hdr_offset_and_more(inner_frag_offset, inner_frag_more); - inner_ip6->protocol = IP_PROTOCOL_IPV6_FRAGMENTATION; - inner_ip6->payload_length = clib_host_to_net_u16( - clib_net_to_host_u16(inner_ip6->payload_length) + sizeof(*inner_frag)); - } - - csum = ip_csum_add_even(csum, inner_ip6->src_address.as_u64[0]); - csum = ip_csum_add_even(csum, inner_ip6->src_address.as_u64[1]); - csum = ip_csum_add_even(csum, inner_ip6->dst_address.as_u64[0]); - csum = ip_csum_add_even(csum, inner_ip6->dst_address.as_u64[1]); - *inner_L4_checksum = ip_csum_fold(csum); + if (inner_ip4) + { + //We have 2 headers to translate. + //We need to make some room in the middle of the packet + + if (PREDICT_FALSE (ip4_is_fragment (inner_ip4))) + { + //Here it starts getting really tricky + //We will add a fragmentation header in the inner packet + + if (!ip4_is_first_fragment (inner_ip4)) + { + //For now we do not handle unless it is the first fragment + //Ideally we should handle the case as we are in slow path already + *error = MAP_ERROR_FRAGMENTED; + return; + } + + vlib_buffer_advance (p, + -2 * (sizeof (*ip6) - sizeof (*ip4)) - + sizeof (*inner_frag)); + ip6 = vlib_buffer_get_current (p); + clib_memcpy (u8_ptr_add (ip6, sizeof (*ip6) - sizeof (*ip4)), ip4, + 20 + 8); + ip4 = + (ip4_header_t *) u8_ptr_add (ip6, sizeof (*ip6) - sizeof (*ip4)); + icmp = (icmp46_header_t *) (ip4 + 1); + + inner_ip6 = + (ip6_header_t *) u8_ptr_add (inner_ip4, + sizeof (*ip4) - sizeof (*ip6) - + sizeof (*inner_frag)); + inner_frag = + (ip6_frag_hdr_t *) u8_ptr_add (inner_ip6, sizeof (*inner_ip6)); + ip6->payload_length = + u16_net_add (ip4->length, + sizeof (*ip6) - 2 * sizeof (*ip4) + + sizeof (*inner_frag)); + inner_frag_id = frag_id_4to6 (inner_ip4->fragment_id); + inner_frag_offset = ip4_get_fragment_offset (inner_ip4); + inner_frag_more = + ! !(inner_ip4->flags_and_fragment_offset & + clib_net_to_host_u16 (IP4_HEADER_FLAG_MORE_FRAGMENTS)); + } + else + { + vlib_buffer_advance (p, -2 * (sizeof (*ip6) - sizeof (*ip4))); + ip6 = vlib_buffer_get_current (p); + clib_memcpy (u8_ptr_add (ip6, sizeof (*ip6) - sizeof (*ip4)), ip4, + 20 + 8); + ip4 = + (ip4_header_t *) u8_ptr_add (ip6, sizeof (*ip6) - sizeof (*ip4)); + icmp = (icmp46_header_t *) u8_ptr_add (ip4, sizeof (*ip4)); + inner_ip6 = + (ip6_header_t *) u8_ptr_add (inner_ip4, + sizeof (*ip4) - sizeof (*ip6)); + ip6->payload_length = + u16_net_add (ip4->length, sizeof (*ip6) - 2 * sizeof (*ip4)); + inner_frag = NULL; + } + + if (PREDICT_TRUE (inner_ip4->protocol == IP_PROTOCOL_TCP)) + { + inner_L4_checksum = &((tcp_header_t *) (inner_ip4 + 1))->checksum; + *inner_L4_checksum = + ip_csum_fold (ip_csum_sub_even + (*inner_L4_checksum, + *((u64 *) (&inner_ip4->src_address)))); + } + else if (PREDICT_TRUE (inner_ip4->protocol == IP_PROTOCOL_UDP)) + { + inner_L4_checksum = &((udp_header_t *) (inner_ip4 + 1))->checksum; + if (!*inner_L4_checksum) + { + //The inner packet was first translated, and therefore came from IPv6. + //As the packet was an IPv6 packet, the UDP checksum can't be NULL + *error = MAP_ERROR_ICMP; + return; + } + *inner_L4_checksum = + ip_csum_fold (ip_csum_sub_even + (*inner_L4_checksum, + *((u64 *) (&inner_ip4->src_address)))); + } + else if (inner_ip4->protocol == IP_PROTOCOL_ICMP) + { + //We have an ICMP inside an ICMP + //It needs to be translated, but not for error ICMP messages + icmp46_header_t *inner_icmp = (icmp46_header_t *) (inner_ip4 + 1); + csum = inner_icmp->checksum; + //Only types ICMP4_echo_request and ICMP4_echo_reply are handled by ip4_icmp_to_icmp6_in_place + csum = ip_csum_sub_even (csum, *((u16 *) inner_icmp)); + inner_icmp->type = (inner_icmp->type == ICMP4_echo_request) ? + ICMP6_echo_request : ICMP6_echo_reply; + csum = ip_csum_add_even (csum, *((u16 *) inner_icmp)); + csum = + ip_csum_add_even (csum, clib_host_to_net_u16 (IP_PROTOCOL_ICMP6)); + csum = + ip_csum_add_even (csum, inner_ip4->length - sizeof (*inner_ip4)); + inner_icmp->checksum = ip_csum_fold (csum); + inner_L4_checksum = &inner_icmp->checksum; + inner_ip4->protocol = IP_PROTOCOL_ICMP6; + } + else + { + ASSERT (0); // We had a port from that, so it is udp or tcp or ICMP + } + + //FIXME: Security check with the port found in the inner packet + + csum = *inner_L4_checksum; //Initial checksum of the inner L4 header + //FIXME: Shouldn't we remove ip addresses from there ? + + inner_ip6->ip_version_traffic_class_and_flow_label = + clib_host_to_net_u32 ((6 << 28) + (inner_ip4->tos << 20)); + inner_ip6->payload_length = + u16_net_add (inner_ip4->length, -sizeof (*inner_ip4)); + inner_ip6->hop_limit = inner_ip4->ttl; + inner_ip6->protocol = inner_ip4->protocol; + + //Note that the source address is within the domain + //while the destination address is the one outside the domain + ip4_map_t_embedded_address (d, &inner_ip6->dst_address, + &inner_ip4->dst_address); + inner_ip6->src_address.as_u64[0] = + map_get_pfx_net (d, inner_ip4->src_address.as_u32, recv_port); + inner_ip6->src_address.as_u64[1] = + map_get_sfx_net (d, inner_ip4->src_address.as_u32, recv_port); + + if (PREDICT_FALSE (inner_frag != NULL)) + { + inner_frag->next_hdr = inner_ip6->protocol; + inner_frag->identification = inner_frag_id; + inner_frag->rsv = 0; + inner_frag->fragment_offset_and_more = + ip6_frag_hdr_offset_and_more (inner_frag_offset, inner_frag_more); + inner_ip6->protocol = IP_PROTOCOL_IPV6_FRAGMENTATION; + inner_ip6->payload_length = + clib_host_to_net_u16 (clib_net_to_host_u16 + (inner_ip6->payload_length) + + sizeof (*inner_frag)); + } + + csum = ip_csum_add_even (csum, inner_ip6->src_address.as_u64[0]); + csum = ip_csum_add_even (csum, inner_ip6->src_address.as_u64[1]); + csum = ip_csum_add_even (csum, inner_ip6->dst_address.as_u64[0]); + csum = ip_csum_add_even (csum, inner_ip6->dst_address.as_u64[1]); + *inner_L4_checksum = ip_csum_fold (csum); - } else { - vlib_buffer_advance(p, sizeof(*ip4) - sizeof(*ip6)); - ip6 = vlib_buffer_get_current(p); - ip6->payload_length = clib_host_to_net_u16(clib_net_to_host_u16(ip4->length) - sizeof(*ip4)); - } + } + else + { + vlib_buffer_advance (p, sizeof (*ip4) - sizeof (*ip6)); + ip6 = vlib_buffer_get_current (p); + ip6->payload_length = + clib_host_to_net_u16 (clib_net_to_host_u16 (ip4->length) - + sizeof (*ip4)); + } //Translate outer IPv6 - ip6->ip_version_traffic_class_and_flow_label = clib_host_to_net_u32((6 << 28) + (ip4->tos << 20)); + ip6->ip_version_traffic_class_and_flow_label = + clib_host_to_net_u32 ((6 << 28) + (ip4->tos << 20)); ip6->hop_limit = ip4->ttl; ip6->protocol = IP_PROTOCOL_ICMP6; - ip4_map_t_embedded_address(d, &ip6->src_address, &ip4->src_address); - ip6->dst_address.as_u64[0] = map_get_pfx_net(d, ip4->dst_address.as_u32, recv_port); - ip6->dst_address.as_u64[1] = map_get_sfx_net(d, ip4->dst_address.as_u32, recv_port); + ip4_map_t_embedded_address (d, &ip6->src_address, &ip4->src_address); + ip6->dst_address.as_u64[0] = + map_get_pfx_net (d, ip4->dst_address.as_u32, recv_port); + ip6->dst_address.as_u64[1] = + map_get_sfx_net (d, ip4->dst_address.as_u32, recv_port); //Truncate when the packet exceeds the minimal IPv6 MTU - if (p->current_length > 1280) { - ip6->payload_length = clib_host_to_net_u16(1280 - sizeof(*ip6)); - p->current_length = 1280; //Looks too simple to be correct... - } + if (p->current_length > 1280) + { + ip6->payload_length = clib_host_to_net_u16 (1280 - sizeof (*ip6)); + p->current_length = 1280; //Looks too simple to be correct... + } //TODO: We could do an easy diff-checksum for echo requests/replies //Recompute ICMP checksum icmp->checksum = 0; - csum = ip_csum_with_carry(0, ip6->payload_length); - csum = ip_csum_with_carry(csum, clib_host_to_net_u16(ip6->protocol)); - csum = ip_csum_with_carry(csum, ip6->src_address.as_u64[0]); - csum = ip_csum_with_carry(csum, ip6->src_address.as_u64[1]); - csum = ip_csum_with_carry(csum, ip6->dst_address.as_u64[0]); - csum = ip_csum_with_carry(csum, ip6->dst_address.as_u64[1]); - csum = ip_incremental_checksum(csum, icmp, clib_net_to_host_u16(ip6->payload_length)); + csum = ip_csum_with_carry (0, ip6->payload_length); + csum = ip_csum_with_carry (csum, clib_host_to_net_u16 (ip6->protocol)); + csum = ip_csum_with_carry (csum, ip6->src_address.as_u64[0]); + csum = ip_csum_with_carry (csum, ip6->src_address.as_u64[1]); + csum = ip_csum_with_carry (csum, ip6->dst_address.as_u64[0]); + csum = ip_csum_with_carry (csum, ip6->dst_address.as_u64[1]); + csum = + ip_incremental_checksum (csum, icmp, + clib_net_to_host_u16 (ip6->payload_length)); icmp->checksum = ~ip_csum_fold (csum); } static uword -ip4_map_t_icmp (vlib_main_t *vm, - vlib_node_runtime_t *node, - vlib_frame_t *frame) +ip4_map_t_icmp (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) { u32 n_left_from, *from, next_index, *to_next, n_left_to_next; - vlib_node_runtime_t *error_node = vlib_node_get_runtime(vm, ip4_map_t_icmp_node.index); - from = vlib_frame_vector_args(frame); + vlib_node_runtime_t *error_node = + vlib_node_get_runtime (vm, ip4_map_t_icmp_node.index); + from = vlib_frame_vector_args (frame); n_left_from = frame->n_vectors; next_index = node->cached_next_index; vlib_combined_counter_main_t *cm = map_main.domain_counters; - u32 cpu_index = os_get_cpu_number(); - - while (n_left_from > 0) { - vlib_get_next_frame(vm, node, next_index, to_next, n_left_to_next); - - while (n_left_from > 0 && n_left_to_next > 0) { - u32 pi0; - vlib_buffer_t *p0; - ip4_mapt_icmp_next_t next0; - u8 error0; - map_domain_t *d0; - u16 len0; - - next0 = IP4_MAPT_ICMP_NEXT_IP6_LOOKUP; - pi0 = to_next[0] = from[0]; - from += 1; - n_left_from -= 1; - to_next +=1; - n_left_to_next -= 1; - error0 = MAP_ERROR_NONE; - - p0 = vlib_get_buffer(vm, pi0); - vlib_buffer_advance(p0, sizeof(ip4_mapt_pseudo_header_t)); //The pseudo-header is not used - len0 = clib_net_to_host_u16(((ip4_header_t *)vlib_buffer_get_current(p0))->length); - d0 = pool_elt_at_index(map_main.domains, vnet_buffer(p0)->map_t.map_domain_index); - _ip4_map_t_icmp(d0, p0, &error0); - - if(vnet_buffer(p0)->map_t.mtu < p0->current_length) { - vnet_buffer(p0)->ip_frag.header_offset = 0; - vnet_buffer(p0)->ip_frag.mtu = vnet_buffer(p0)->map_t.mtu; - vnet_buffer(p0)->ip_frag.next_index = IP6_FRAG_NEXT_IP6_LOOKUP; - next0 = IP4_MAPT_ICMP_NEXT_IP6_FRAG; - } - if (PREDICT_TRUE(error0 == MAP_ERROR_NONE)) { - vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_TX, cpu_index, - vnet_buffer(p0)->map_t.map_domain_index, 1, - len0); - } else { - next0 = IP4_MAPT_ICMP_NEXT_DROP; - } - p0->error = error_node->errors[error0]; - vlib_validate_buffer_enqueue_x1(vm, node, next_index, - to_next, n_left_to_next, pi0, - next0); + u32 cpu_index = os_get_cpu_number (); + + while (n_left_from > 0) + { + vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); + + while (n_left_from > 0 && n_left_to_next > 0) + { + u32 pi0; + vlib_buffer_t *p0; + ip4_mapt_icmp_next_t next0; + u8 error0; + map_domain_t *d0; + u16 len0; + + next0 = IP4_MAPT_ICMP_NEXT_IP6_LOOKUP; + pi0 = to_next[0] = from[0]; + from += 1; + n_left_from -= 1; + to_next += 1; + n_left_to_next -= 1; + error0 = MAP_ERROR_NONE; + + p0 = vlib_get_buffer (vm, pi0); + vlib_buffer_advance (p0, sizeof (ip4_mapt_pseudo_header_t)); //The pseudo-header is not used + len0 = + clib_net_to_host_u16 (((ip4_header_t *) + vlib_buffer_get_current (p0))->length); + d0 = + pool_elt_at_index (map_main.domains, + vnet_buffer (p0)->map_t.map_domain_index); + _ip4_map_t_icmp (d0, p0, &error0); + + if (vnet_buffer (p0)->map_t.mtu < p0->current_length) + { + vnet_buffer (p0)->ip_frag.header_offset = 0; + vnet_buffer (p0)->ip_frag.mtu = vnet_buffer (p0)->map_t.mtu; + vnet_buffer (p0)->ip_frag.next_index = IP6_FRAG_NEXT_IP6_LOOKUP; + next0 = IP4_MAPT_ICMP_NEXT_IP6_FRAG; + } + if (PREDICT_TRUE (error0 == MAP_ERROR_NONE)) + { + vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_TX, + cpu_index, + vnet_buffer (p0)->map_t. + map_domain_index, 1, len0); + } + else + { + next0 = IP4_MAPT_ICMP_NEXT_DROP; + } + p0->error = error_node->errors[error0]; + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, + to_next, n_left_to_next, pi0, + next0); + } + vlib_put_next_frame (vm, node, next_index, n_left_to_next); } - vlib_put_next_frame(vm, node, next_index, n_left_to_next); - } return frame->n_vectors; } static uword -ip4_map_t_fragmented (vlib_main_t *vm, - vlib_node_runtime_t *node, - vlib_frame_t *frame) +ip4_map_t_fragmented (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) { u32 n_left_from, *from, next_index, *to_next, n_left_to_next; - from = vlib_frame_vector_args(frame); + from = vlib_frame_vector_args (frame); n_left_from = frame->n_vectors; next_index = node->cached_next_index; - while (n_left_from > 0) { - vlib_get_next_frame(vm, node, next_index, to_next, n_left_to_next); - - while (n_left_from > 0 && n_left_to_next > 0) { - u32 pi0; - vlib_buffer_t *p0; - ip4_header_t *ip40; - ip6_header_t *ip60; - ip6_frag_hdr_t *frag0; - ip4_mapt_pseudo_header_t *pheader0; - ip4_mapt_fragmented_next_t next0; - - next0 = IP4_MAPT_FRAGMENTED_NEXT_IP6_LOOKUP; - pi0 = to_next[0] = from[0]; - from += 1; - n_left_from -= 1; - to_next +=1; - n_left_to_next -= 1; - - p0 = vlib_get_buffer(vm, pi0); - - //Accessing pseudo header - pheader0 = vlib_buffer_get_current(p0); - vlib_buffer_advance(p0, sizeof(*pheader0)); - - //Accessing ip4 header - ip40 = vlib_buffer_get_current(p0); - frag0 = (ip6_frag_hdr_t *) u8_ptr_add(ip40, sizeof(*ip40) - sizeof(*frag0)); - ip60 = (ip6_header_t *) u8_ptr_add(ip40, sizeof(*ip40) - sizeof(*frag0) - sizeof(*ip60)); - vlib_buffer_advance(p0, sizeof(*ip40) - sizeof(*ip60) - sizeof(*frag0)); - - //We know that the protocol was one of ICMP, TCP or UDP - //because the first fragment was found and cached - frag0->next_hdr = (ip40->protocol == IP_PROTOCOL_ICMP) ? IP_PROTOCOL_ICMP6 : ip40->protocol; - frag0->identification = frag_id_4to6(ip40->fragment_id); - frag0->rsv = 0; - frag0->fragment_offset_and_more = ip6_frag_hdr_offset_and_more( - ip4_get_fragment_offset(ip40), - clib_net_to_host_u16(ip40->flags_and_fragment_offset) & IP4_HEADER_FLAG_MORE_FRAGMENTS); - - ip60->ip_version_traffic_class_and_flow_label = clib_host_to_net_u32((6 << 28) + (ip40->tos << 20)); - ip60->payload_length = clib_host_to_net_u16(clib_net_to_host_u16(ip40->length) - sizeof(*ip40) + sizeof(*frag0)); - ip60->hop_limit = ip40->ttl; - ip60->protocol = IP_PROTOCOL_IPV6_FRAGMENTATION; - ip60->dst_address.as_u64[0] = pheader0->daddr.as_u64[0]; - ip60->dst_address.as_u64[1] = pheader0->daddr.as_u64[1]; - ip60->src_address.as_u64[0] = pheader0->saddr.as_u64[0]; - ip60->src_address.as_u64[1] = pheader0->saddr.as_u64[1]; - - if(vnet_buffer(p0)->map_t.mtu < p0->current_length) { - vnet_buffer(p0)->ip_frag.header_offset = 0; - vnet_buffer(p0)->ip_frag.mtu = vnet_buffer(p0)->map_t.mtu; - vnet_buffer(p0)->ip_frag.next_index = IP6_FRAG_NEXT_IP6_LOOKUP; - next0 = IP4_MAPT_FRAGMENTED_NEXT_IP6_FRAG; - } - - vlib_validate_buffer_enqueue_x1(vm, node, next_index, - to_next, n_left_to_next, pi0, - next0); + while (n_left_from > 0) + { + vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); + + while (n_left_from > 0 && n_left_to_next > 0) + { + u32 pi0; + vlib_buffer_t *p0; + ip4_header_t *ip40; + ip6_header_t *ip60; + ip6_frag_hdr_t *frag0; + ip4_mapt_pseudo_header_t *pheader0; + ip4_mapt_fragmented_next_t next0; + + next0 = IP4_MAPT_FRAGMENTED_NEXT_IP6_LOOKUP; + pi0 = to_next[0] = from[0]; + from += 1; + n_left_from -= 1; + to_next += 1; + n_left_to_next -= 1; + + p0 = vlib_get_buffer (vm, pi0); + + //Accessing pseudo header + pheader0 = vlib_buffer_get_current (p0); + vlib_buffer_advance (p0, sizeof (*pheader0)); + + //Accessing ip4 header + ip40 = vlib_buffer_get_current (p0); + frag0 = + (ip6_frag_hdr_t *) u8_ptr_add (ip40, + sizeof (*ip40) - sizeof (*frag0)); + ip60 = + (ip6_header_t *) u8_ptr_add (ip40, + sizeof (*ip40) - sizeof (*frag0) - + sizeof (*ip60)); + vlib_buffer_advance (p0, + sizeof (*ip40) - sizeof (*ip60) - + sizeof (*frag0)); + + //We know that the protocol was one of ICMP, TCP or UDP + //because the first fragment was found and cached + frag0->next_hdr = + (ip40->protocol == + IP_PROTOCOL_ICMP) ? IP_PROTOCOL_ICMP6 : ip40->protocol; + frag0->identification = frag_id_4to6 (ip40->fragment_id); + frag0->rsv = 0; + frag0->fragment_offset_and_more = + ip6_frag_hdr_offset_and_more (ip4_get_fragment_offset (ip40), + clib_net_to_host_u16 + (ip40->flags_and_fragment_offset) & + IP4_HEADER_FLAG_MORE_FRAGMENTS); + + ip60->ip_version_traffic_class_and_flow_label = + clib_host_to_net_u32 ((6 << 28) + (ip40->tos << 20)); + ip60->payload_length = + clib_host_to_net_u16 (clib_net_to_host_u16 (ip40->length) - + sizeof (*ip40) + sizeof (*frag0)); + ip60->hop_limit = ip40->ttl; + ip60->protocol = IP_PROTOCOL_IPV6_FRAGMENTATION; + ip60->dst_address.as_u64[0] = pheader0->daddr.as_u64[0]; + ip60->dst_address.as_u64[1] = pheader0->daddr.as_u64[1]; + ip60->src_address.as_u64[0] = pheader0->saddr.as_u64[0]; + ip60->src_address.as_u64[1] = pheader0->saddr.as_u64[1]; + + if (vnet_buffer (p0)->map_t.mtu < p0->current_length) + { + vnet_buffer (p0)->ip_frag.header_offset = 0; + vnet_buffer (p0)->ip_frag.mtu = vnet_buffer (p0)->map_t.mtu; + vnet_buffer (p0)->ip_frag.next_index = IP6_FRAG_NEXT_IP6_LOOKUP; + next0 = IP4_MAPT_FRAGMENTED_NEXT_IP6_FRAG; + } + + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, + to_next, n_left_to_next, pi0, + next0); + } + vlib_put_next_frame (vm, node, next_index, n_left_to_next); } - vlib_put_next_frame(vm, node, next_index, n_left_to_next); - } return frame->n_vectors; } static uword -ip4_map_t_tcp_udp(vlib_main_t *vm, - vlib_node_runtime_t *node, - vlib_frame_t *frame) +ip4_map_t_tcp_udp (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) { u32 n_left_from, *from, next_index, *to_next, n_left_to_next; - from = vlib_frame_vector_args(frame); + from = vlib_frame_vector_args (frame); n_left_from = frame->n_vectors; next_index = node->cached_next_index; - while (n_left_from > 0) { - vlib_get_next_frame(vm, node, next_index, to_next, n_left_to_next); + while (n_left_from > 0) + { + vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); #ifdef IP4_MAP_T_DUAL_LOOP - while (n_left_from >= 4 && n_left_to_next >= 2) { - u32 pi0, pi1; - vlib_buffer_t *p0, *p1; - ip4_header_t *ip40, *ip41; - ip6_header_t *ip60, *ip61; - ip_csum_t csum0, csum1; - u16 *checksum0, *checksum1; - ip6_frag_hdr_t *frag0, *frag1; - u32 frag_id0, frag_id1; - ip4_mapt_pseudo_header_t *pheader0, *pheader1; - ip4_mapt_tcp_udp_next_t next0, next1; - - pi0 = to_next[0] = from[0]; - pi1 = to_next[1] = from[1]; - from += 2; - n_left_from -= 2; - to_next +=2; - n_left_to_next -= 2; - - next0 = IP4_MAPT_TCP_UDP_NEXT_IP6_LOOKUP; - next1 = IP4_MAPT_TCP_UDP_NEXT_IP6_LOOKUP; - p0 = vlib_get_buffer(vm, pi0); - p1 = vlib_get_buffer(vm, pi1); - - //Accessing pseudo header - pheader0 = vlib_buffer_get_current(p0); - pheader1 = vlib_buffer_get_current(p1); - vlib_buffer_advance(p0, sizeof(*pheader0)); - vlib_buffer_advance(p1, sizeof(*pheader1)); - - //Accessing ip4 header - ip40 = vlib_buffer_get_current(p0); - ip41 = vlib_buffer_get_current(p1); - checksum0 = (u16 *) u8_ptr_add(ip40, vnet_buffer(p0)->map_t.checksum_offset); - checksum1 = (u16 *) u8_ptr_add(ip41, vnet_buffer(p1)->map_t.checksum_offset); - - //UDP checksum is optional over IPv4 but mandatory for IPv6 - //We do not check udp->length sanity but use our safe computed value instead - if (PREDICT_FALSE(!*checksum0 && ip40->protocol == IP_PROTOCOL_UDP)) { - u16 udp_len = clib_host_to_net_u16(ip40->length) - sizeof(*ip40); - udp_header_t *udp = (udp_header_t *) u8_ptr_add(ip40, sizeof(*ip40)); - ip_csum_t csum; - csum = ip_incremental_checksum(0, udp, udp_len); - csum = ip_csum_with_carry(csum, clib_host_to_net_u16(udp_len)); - csum = ip_csum_with_carry(csum, clib_host_to_net_u16(IP_PROTOCOL_UDP)); - csum = ip_csum_with_carry(csum, *((u64 *)(&ip40->src_address))); - *checksum0 = ~ip_csum_fold(csum); - } - if (PREDICT_FALSE(!*checksum1 && ip41->protocol == IP_PROTOCOL_UDP)) { - u16 udp_len = clib_host_to_net_u16(ip41->length) - sizeof(*ip40); - udp_header_t *udp = (udp_header_t *) u8_ptr_add(ip41, sizeof(*ip40)); - ip_csum_t csum; - csum = ip_incremental_checksum(0, udp, udp_len); - csum = ip_csum_with_carry(csum, clib_host_to_net_u16(udp_len)); - csum = ip_csum_with_carry(csum, clib_host_to_net_u16(IP_PROTOCOL_UDP)); - csum = ip_csum_with_carry(csum, *((u64 *)(&ip41->src_address))); - *checksum1 = ~ip_csum_fold(csum); - } - - csum0 = ip_csum_sub_even(*checksum0, ip40->src_address.as_u32); - csum1 = ip_csum_sub_even(*checksum1, ip41->src_address.as_u32); - csum0 = ip_csum_sub_even(csum0, ip40->dst_address.as_u32); - csum1 = ip_csum_sub_even(csum1, ip41->dst_address.as_u32); - - // Deal with fragmented packets - if (PREDICT_FALSE(ip40->flags_and_fragment_offset & - clib_host_to_net_u16(IP4_HEADER_FLAG_MORE_FRAGMENTS))) { - ip60 = (ip6_header_t *) u8_ptr_add(ip40, sizeof(*ip40) - sizeof(*ip60) - sizeof(*frag0)); - frag0 = (ip6_frag_hdr_t *) u8_ptr_add(ip40, sizeof(*ip40) - sizeof(*frag0)); - frag_id0 = frag_id_4to6(ip40->fragment_id); - vlib_buffer_advance(p0, sizeof(*ip40) - sizeof(*ip60) - sizeof(*frag0)); - } else { - ip60 = (ip6_header_t *) (((u8 *)ip40) + sizeof(*ip40) - sizeof(*ip60)); - vlib_buffer_advance(p0, sizeof(*ip40) - sizeof(*ip60)); - frag0 = NULL; - } - - if (PREDICT_FALSE(ip41->flags_and_fragment_offset & - clib_host_to_net_u16(IP4_HEADER_FLAG_MORE_FRAGMENTS))) { - ip61 = (ip6_header_t *) u8_ptr_add(ip41, sizeof(*ip40) - sizeof(*ip60) - sizeof(*frag0)); - frag1 = (ip6_frag_hdr_t *) u8_ptr_add(ip41, sizeof(*ip40) - sizeof(*frag0)); - frag_id1 = frag_id_4to6(ip41->fragment_id); - vlib_buffer_advance(p1, sizeof(*ip40) - sizeof(*ip60) - sizeof(*frag0)); - } else { - ip61 = (ip6_header_t *) (((u8 *)ip41) + sizeof(*ip40) - sizeof(*ip60)); - vlib_buffer_advance(p1, sizeof(*ip40) - sizeof(*ip60)); - frag1 = NULL; - } - - ip60->ip_version_traffic_class_and_flow_label = clib_host_to_net_u32((6 << 28) + (ip40->tos << 20)); - ip61->ip_version_traffic_class_and_flow_label = clib_host_to_net_u32((6 << 28) + (ip41->tos << 20)); - ip60->payload_length = u16_net_add(ip40->length, - sizeof(*ip40)); - ip61->payload_length = u16_net_add(ip41->length, - sizeof(*ip40)); - ip60->hop_limit = ip40->ttl; - ip61->hop_limit = ip41->ttl; - ip60->protocol = ip40->protocol; - ip61->protocol = ip41->protocol; - - if (PREDICT_FALSE(frag0 != NULL)) { - frag0->next_hdr = ip60->protocol; - frag0->identification = frag_id0; - frag0->rsv = 0; - frag0->fragment_offset_and_more = ip6_frag_hdr_offset_and_more(0, 1); - ip60->protocol = IP_PROTOCOL_IPV6_FRAGMENTATION; - ip60->payload_length = u16_net_add(ip60->payload_length, sizeof(*frag0)); - } - - if (PREDICT_FALSE(frag1 != NULL)) { - frag1->next_hdr = ip61->protocol; - frag1->identification = frag_id1; - frag1->rsv = 0; - frag1->fragment_offset_and_more = ip6_frag_hdr_offset_and_more(0, 1); - ip61->protocol = IP_PROTOCOL_IPV6_FRAGMENTATION; - ip61->payload_length = u16_net_add(ip61->payload_length, sizeof(*frag0)); - } - - //Finally copying the address - ip60->dst_address.as_u64[0] = pheader0->daddr.as_u64[0]; - ip61->dst_address.as_u64[0] = pheader1->daddr.as_u64[0]; - ip60->dst_address.as_u64[1] = pheader0->daddr.as_u64[1]; - ip61->dst_address.as_u64[1] = pheader1->daddr.as_u64[1]; - ip60->src_address.as_u64[0] = pheader0->saddr.as_u64[0]; - ip61->src_address.as_u64[0] = pheader1->saddr.as_u64[0]; - ip60->src_address.as_u64[1] = pheader0->saddr.as_u64[1]; - ip61->src_address.as_u64[1] = pheader1->saddr.as_u64[1]; - - csum0 = ip_csum_add_even(csum0, ip60->src_address.as_u64[0]); - csum1 = ip_csum_add_even(csum1, ip61->src_address.as_u64[0]); - csum0 = ip_csum_add_even(csum0, ip60->src_address.as_u64[1]); - csum1 = ip_csum_add_even(csum1, ip61->src_address.as_u64[1]); - csum0 = ip_csum_add_even(csum0, ip60->dst_address.as_u64[0]); - csum1 = ip_csum_add_even(csum1, ip61->dst_address.as_u64[0]); - csum0 = ip_csum_add_even(csum0, ip60->dst_address.as_u64[1]); - csum1 = ip_csum_add_even(csum1, ip61->dst_address.as_u64[1]); - *checksum0 = ip_csum_fold(csum0); - *checksum1 = ip_csum_fold(csum1); - - if(vnet_buffer(p0)->map_t.mtu < p0->current_length) { - vnet_buffer(p0)->ip_frag.header_offset = 0; - vnet_buffer(p0)->ip_frag.mtu = vnet_buffer(p0)->map_t.mtu; - vnet_buffer(p0)->ip_frag.next_index = IP6_FRAG_NEXT_IP6_LOOKUP; - next0 = IP4_MAPT_TCP_UDP_NEXT_IP6_FRAG; - } - - if(vnet_buffer(p1)->map_t.mtu < p1->current_length) { - vnet_buffer(p1)->ip_frag.header_offset = 0; - vnet_buffer(p1)->ip_frag.mtu = vnet_buffer(p1)->map_t.mtu; - vnet_buffer(p1)->ip_frag.next_index = IP6_FRAG_NEXT_IP6_LOOKUP; - next1 = IP4_MAPT_TCP_UDP_NEXT_IP6_FRAG; - } - - vlib_validate_buffer_enqueue_x2(vm, node, next_index, - to_next, n_left_to_next, pi0, pi1, - next0, next1); - } + while (n_left_from >= 4 && n_left_to_next >= 2) + { + u32 pi0, pi1; + vlib_buffer_t *p0, *p1; + ip4_header_t *ip40, *ip41; + ip6_header_t *ip60, *ip61; + ip_csum_t csum0, csum1; + u16 *checksum0, *checksum1; + ip6_frag_hdr_t *frag0, *frag1; + u32 frag_id0, frag_id1; + ip4_mapt_pseudo_header_t *pheader0, *pheader1; + ip4_mapt_tcp_udp_next_t next0, next1; + + pi0 = to_next[0] = from[0]; + pi1 = to_next[1] = from[1]; + from += 2; + n_left_from -= 2; + to_next += 2; + n_left_to_next -= 2; + + next0 = IP4_MAPT_TCP_UDP_NEXT_IP6_LOOKUP; + next1 = IP4_MAPT_TCP_UDP_NEXT_IP6_LOOKUP; + p0 = vlib_get_buffer (vm, pi0); + p1 = vlib_get_buffer (vm, pi1); + + //Accessing pseudo header + pheader0 = vlib_buffer_get_current (p0); + pheader1 = vlib_buffer_get_current (p1); + vlib_buffer_advance (p0, sizeof (*pheader0)); + vlib_buffer_advance (p1, sizeof (*pheader1)); + + //Accessing ip4 header + ip40 = vlib_buffer_get_current (p0); + ip41 = vlib_buffer_get_current (p1); + checksum0 = + (u16 *) u8_ptr_add (ip40, + vnet_buffer (p0)->map_t.checksum_offset); + checksum1 = + (u16 *) u8_ptr_add (ip41, + vnet_buffer (p1)->map_t.checksum_offset); + + //UDP checksum is optional over IPv4 but mandatory for IPv6 + //We do not check udp->length sanity but use our safe computed value instead + if (PREDICT_FALSE + (!*checksum0 && ip40->protocol == IP_PROTOCOL_UDP)) + { + u16 udp_len = + clib_host_to_net_u16 (ip40->length) - sizeof (*ip40); + udp_header_t *udp = + (udp_header_t *) u8_ptr_add (ip40, sizeof (*ip40)); + ip_csum_t csum; + csum = ip_incremental_checksum (0, udp, udp_len); + csum = + ip_csum_with_carry (csum, clib_host_to_net_u16 (udp_len)); + csum = + ip_csum_with_carry (csum, + clib_host_to_net_u16 (IP_PROTOCOL_UDP)); + csum = + ip_csum_with_carry (csum, *((u64 *) (&ip40->src_address))); + *checksum0 = ~ip_csum_fold (csum); + } + if (PREDICT_FALSE + (!*checksum1 && ip41->protocol == IP_PROTOCOL_UDP)) + { + u16 udp_len = + clib_host_to_net_u16 (ip41->length) - sizeof (*ip40); + udp_header_t *udp = + (udp_header_t *) u8_ptr_add (ip41, sizeof (*ip40)); + ip_csum_t csum; + csum = ip_incremental_checksum (0, udp, udp_len); + csum = + ip_csum_with_carry (csum, clib_host_to_net_u16 (udp_len)); + csum = + ip_csum_with_carry (csum, + clib_host_to_net_u16 (IP_PROTOCOL_UDP)); + csum = + ip_csum_with_carry (csum, *((u64 *) (&ip41->src_address))); + *checksum1 = ~ip_csum_fold (csum); + } + + csum0 = ip_csum_sub_even (*checksum0, ip40->src_address.as_u32); + csum1 = ip_csum_sub_even (*checksum1, ip41->src_address.as_u32); + csum0 = ip_csum_sub_even (csum0, ip40->dst_address.as_u32); + csum1 = ip_csum_sub_even (csum1, ip41->dst_address.as_u32); + + // Deal with fragmented packets + if (PREDICT_FALSE (ip40->flags_and_fragment_offset & + clib_host_to_net_u16 + (IP4_HEADER_FLAG_MORE_FRAGMENTS))) + { + ip60 = + (ip6_header_t *) u8_ptr_add (ip40, + sizeof (*ip40) - sizeof (*ip60) - + sizeof (*frag0)); + frag0 = + (ip6_frag_hdr_t *) u8_ptr_add (ip40, + sizeof (*ip40) - + sizeof (*frag0)); + frag_id0 = frag_id_4to6 (ip40->fragment_id); + vlib_buffer_advance (p0, + sizeof (*ip40) - sizeof (*ip60) - + sizeof (*frag0)); + } + else + { + ip60 = + (ip6_header_t *) (((u8 *) ip40) + sizeof (*ip40) - + sizeof (*ip60)); + vlib_buffer_advance (p0, sizeof (*ip40) - sizeof (*ip60)); + frag0 = NULL; + } + + if (PREDICT_FALSE (ip41->flags_and_fragment_offset & + clib_host_to_net_u16 + (IP4_HEADER_FLAG_MORE_FRAGMENTS))) + { + ip61 = + (ip6_header_t *) u8_ptr_add (ip41, + sizeof (*ip40) - sizeof (*ip60) - + sizeof (*frag0)); + frag1 = + (ip6_frag_hdr_t *) u8_ptr_add (ip41, + sizeof (*ip40) - + sizeof (*frag0)); + frag_id1 = frag_id_4to6 (ip41->fragment_id); + vlib_buffer_advance (p1, + sizeof (*ip40) - sizeof (*ip60) - + sizeof (*frag0)); + } + else + { + ip61 = + (ip6_header_t *) (((u8 *) ip41) + sizeof (*ip40) - + sizeof (*ip60)); + vlib_buffer_advance (p1, sizeof (*ip40) - sizeof (*ip60)); + frag1 = NULL; + } + + ip60->ip_version_traffic_class_and_flow_label = + clib_host_to_net_u32 ((6 << 28) + (ip40->tos << 20)); + ip61->ip_version_traffic_class_and_flow_label = + clib_host_to_net_u32 ((6 << 28) + (ip41->tos << 20)); + ip60->payload_length = u16_net_add (ip40->length, -sizeof (*ip40)); + ip61->payload_length = u16_net_add (ip41->length, -sizeof (*ip40)); + ip60->hop_limit = ip40->ttl; + ip61->hop_limit = ip41->ttl; + ip60->protocol = ip40->protocol; + ip61->protocol = ip41->protocol; + + if (PREDICT_FALSE (frag0 != NULL)) + { + frag0->next_hdr = ip60->protocol; + frag0->identification = frag_id0; + frag0->rsv = 0; + frag0->fragment_offset_and_more = + ip6_frag_hdr_offset_and_more (0, 1); + ip60->protocol = IP_PROTOCOL_IPV6_FRAGMENTATION; + ip60->payload_length = + u16_net_add (ip60->payload_length, sizeof (*frag0)); + } + + if (PREDICT_FALSE (frag1 != NULL)) + { + frag1->next_hdr = ip61->protocol; + frag1->identification = frag_id1; + frag1->rsv = 0; + frag1->fragment_offset_and_more = + ip6_frag_hdr_offset_and_more (0, 1); + ip61->protocol = IP_PROTOCOL_IPV6_FRAGMENTATION; + ip61->payload_length = + u16_net_add (ip61->payload_length, sizeof (*frag0)); + } + + //Finally copying the address + ip60->dst_address.as_u64[0] = pheader0->daddr.as_u64[0]; + ip61->dst_address.as_u64[0] = pheader1->daddr.as_u64[0]; + ip60->dst_address.as_u64[1] = pheader0->daddr.as_u64[1]; + ip61->dst_address.as_u64[1] = pheader1->daddr.as_u64[1]; + ip60->src_address.as_u64[0] = pheader0->saddr.as_u64[0]; + ip61->src_address.as_u64[0] = pheader1->saddr.as_u64[0]; + ip60->src_address.as_u64[1] = pheader0->saddr.as_u64[1]; + ip61->src_address.as_u64[1] = pheader1->saddr.as_u64[1]; + + csum0 = ip_csum_add_even (csum0, ip60->src_address.as_u64[0]); + csum1 = ip_csum_add_even (csum1, ip61->src_address.as_u64[0]); + csum0 = ip_csum_add_even (csum0, ip60->src_address.as_u64[1]); + csum1 = ip_csum_add_even (csum1, ip61->src_address.as_u64[1]); + csum0 = ip_csum_add_even (csum0, ip60->dst_address.as_u64[0]); + csum1 = ip_csum_add_even (csum1, ip61->dst_address.as_u64[0]); + csum0 = ip_csum_add_even (csum0, ip60->dst_address.as_u64[1]); + csum1 = ip_csum_add_even (csum1, ip61->dst_address.as_u64[1]); + *checksum0 = ip_csum_fold (csum0); + *checksum1 = ip_csum_fold (csum1); + + if (vnet_buffer (p0)->map_t.mtu < p0->current_length) + { + vnet_buffer (p0)->ip_frag.header_offset = 0; + vnet_buffer (p0)->ip_frag.mtu = vnet_buffer (p0)->map_t.mtu; + vnet_buffer (p0)->ip_frag.next_index = IP6_FRAG_NEXT_IP6_LOOKUP; + next0 = IP4_MAPT_TCP_UDP_NEXT_IP6_FRAG; + } + + if (vnet_buffer (p1)->map_t.mtu < p1->current_length) + { + vnet_buffer (p1)->ip_frag.header_offset = 0; + vnet_buffer (p1)->ip_frag.mtu = vnet_buffer (p1)->map_t.mtu; + vnet_buffer (p1)->ip_frag.next_index = IP6_FRAG_NEXT_IP6_LOOKUP; + next1 = IP4_MAPT_TCP_UDP_NEXT_IP6_FRAG; + } + + vlib_validate_buffer_enqueue_x2 (vm, node, next_index, + to_next, n_left_to_next, pi0, pi1, + next0, next1); + } #endif - while (n_left_from > 0 && n_left_to_next > 0) { - u32 pi0; - vlib_buffer_t *p0; - ip4_header_t *ip40; - ip6_header_t *ip60; - ip_csum_t csum0; - u16 *checksum0; - ip6_frag_hdr_t *frag0; - u32 frag_id0; - ip4_mapt_pseudo_header_t *pheader0; - ip4_mapt_tcp_udp_next_t next0; - - pi0 = to_next[0] = from[0]; - from += 1; - n_left_from -= 1; - to_next +=1; - n_left_to_next -= 1; - - next0 = IP4_MAPT_TCP_UDP_NEXT_IP6_LOOKUP; - p0 = vlib_get_buffer(vm, pi0); - - //Accessing pseudo header - pheader0 = vlib_buffer_get_current(p0); - vlib_buffer_advance(p0, sizeof(*pheader0)); - - //Accessing ip4 header - ip40 = vlib_buffer_get_current(p0); - checksum0 = (u16 *) u8_ptr_add(ip40, vnet_buffer(p0)->map_t.checksum_offset); - - //UDP checksum is optional over IPv4 but mandatory for IPv6 - //We do not check udp->length sanity but use our safe computed value instead - if (PREDICT_FALSE(!*checksum0 && ip40->protocol == IP_PROTOCOL_UDP)) { - u16 udp_len = clib_host_to_net_u16(ip40->length) - sizeof(*ip40); - udp_header_t *udp = (udp_header_t *) u8_ptr_add(ip40, sizeof(*ip40)); - ip_csum_t csum; - csum = ip_incremental_checksum(0, udp, udp_len); - csum = ip_csum_with_carry(csum, clib_host_to_net_u16(udp_len)); - csum = ip_csum_with_carry(csum, clib_host_to_net_u16(IP_PROTOCOL_UDP)); - csum = ip_csum_with_carry(csum, *((u64 *)(&ip40->src_address))); - *checksum0 = ~ip_csum_fold(csum); - } - - csum0 = ip_csum_sub_even(*checksum0, ip40->src_address.as_u32); - csum0 = ip_csum_sub_even(csum0, ip40->dst_address.as_u32); - - // Deal with fragmented packets - if (PREDICT_FALSE(ip40->flags_and_fragment_offset & - clib_host_to_net_u16(IP4_HEADER_FLAG_MORE_FRAGMENTS))) { - ip60 = (ip6_header_t *) u8_ptr_add(ip40, sizeof(*ip40) - sizeof(*ip60) - sizeof(*frag0)); - frag0 = (ip6_frag_hdr_t *) u8_ptr_add(ip40, sizeof(*ip40) - sizeof(*frag0)); - frag_id0 = frag_id_4to6(ip40->fragment_id); - vlib_buffer_advance(p0, sizeof(*ip40) - sizeof(*ip60) - sizeof(*frag0)); - } else { - ip60 = (ip6_header_t *) (((u8 *)ip40) + sizeof(*ip40) - sizeof(*ip60)); - vlib_buffer_advance(p0, sizeof(*ip40) - sizeof(*ip60)); - frag0 = NULL; - } - - ip60->ip_version_traffic_class_and_flow_label = clib_host_to_net_u32((6 << 28) + (ip40->tos << 20)); - ip60->payload_length = u16_net_add(ip40->length, - sizeof(*ip40)); - ip60->hop_limit = ip40->ttl; - ip60->protocol = ip40->protocol; - - if (PREDICT_FALSE(frag0 != NULL)) { - frag0->next_hdr = ip60->protocol; - frag0->identification = frag_id0; - frag0->rsv = 0; - frag0->fragment_offset_and_more = ip6_frag_hdr_offset_and_more(0, 1); - ip60->protocol = IP_PROTOCOL_IPV6_FRAGMENTATION; - ip60->payload_length = u16_net_add(ip60->payload_length, sizeof(*frag0)); - } - - //Finally copying the address - ip60->dst_address.as_u64[0] = pheader0->daddr.as_u64[0]; - ip60->dst_address.as_u64[1] = pheader0->daddr.as_u64[1]; - ip60->src_address.as_u64[0] = pheader0->saddr.as_u64[0]; - ip60->src_address.as_u64[1] = pheader0->saddr.as_u64[1]; - - csum0 = ip_csum_add_even(csum0, ip60->src_address.as_u64[0]); - csum0 = ip_csum_add_even(csum0, ip60->src_address.as_u64[1]); - csum0 = ip_csum_add_even(csum0, ip60->dst_address.as_u64[0]); - csum0 = ip_csum_add_even(csum0, ip60->dst_address.as_u64[1]); - *checksum0 = ip_csum_fold(csum0); - - if(vnet_buffer(p0)->map_t.mtu < p0->current_length) { - //Send to fragmentation node if necessary - vnet_buffer(p0)->ip_frag.header_offset = 0; - vnet_buffer(p0)->ip_frag.mtu = vnet_buffer(p0)->map_t.mtu; - vnet_buffer(p0)->ip_frag.next_index = IP6_FRAG_NEXT_IP6_LOOKUP; - next0 = IP4_MAPT_TCP_UDP_NEXT_IP6_FRAG; - } - - vlib_validate_buffer_enqueue_x1(vm, node, next_index, - to_next, n_left_to_next, pi0, - next0); + while (n_left_from > 0 && n_left_to_next > 0) + { + u32 pi0; + vlib_buffer_t *p0; + ip4_header_t *ip40; + ip6_header_t *ip60; + ip_csum_t csum0; + u16 *checksum0; + ip6_frag_hdr_t *frag0; + u32 frag_id0; + ip4_mapt_pseudo_header_t *pheader0; + ip4_mapt_tcp_udp_next_t next0; + + pi0 = to_next[0] = from[0]; + from += 1; + n_left_from -= 1; + to_next += 1; + n_left_to_next -= 1; + + next0 = IP4_MAPT_TCP_UDP_NEXT_IP6_LOOKUP; + p0 = vlib_get_buffer (vm, pi0); + + //Accessing pseudo header + pheader0 = vlib_buffer_get_current (p0); + vlib_buffer_advance (p0, sizeof (*pheader0)); + + //Accessing ip4 header + ip40 = vlib_buffer_get_current (p0); + checksum0 = + (u16 *) u8_ptr_add (ip40, + vnet_buffer (p0)->map_t.checksum_offset); + + //UDP checksum is optional over IPv4 but mandatory for IPv6 + //We do not check udp->length sanity but use our safe computed value instead + if (PREDICT_FALSE + (!*checksum0 && ip40->protocol == IP_PROTOCOL_UDP)) + { + u16 udp_len = + clib_host_to_net_u16 (ip40->length) - sizeof (*ip40); + udp_header_t *udp = + (udp_header_t *) u8_ptr_add (ip40, sizeof (*ip40)); + ip_csum_t csum; + csum = ip_incremental_checksum (0, udp, udp_len); + csum = + ip_csum_with_carry (csum, clib_host_to_net_u16 (udp_len)); + csum = + ip_csum_with_carry (csum, + clib_host_to_net_u16 (IP_PROTOCOL_UDP)); + csum = + ip_csum_with_carry (csum, *((u64 *) (&ip40->src_address))); + *checksum0 = ~ip_csum_fold (csum); + } + + csum0 = ip_csum_sub_even (*checksum0, ip40->src_address.as_u32); + csum0 = ip_csum_sub_even (csum0, ip40->dst_address.as_u32); + + // Deal with fragmented packets + if (PREDICT_FALSE (ip40->flags_and_fragment_offset & + clib_host_to_net_u16 + (IP4_HEADER_FLAG_MORE_FRAGMENTS))) + { + ip60 = + (ip6_header_t *) u8_ptr_add (ip40, + sizeof (*ip40) - sizeof (*ip60) - + sizeof (*frag0)); + frag0 = + (ip6_frag_hdr_t *) u8_ptr_add (ip40, + sizeof (*ip40) - + sizeof (*frag0)); + frag_id0 = frag_id_4to6 (ip40->fragment_id); + vlib_buffer_advance (p0, + sizeof (*ip40) - sizeof (*ip60) - + sizeof (*frag0)); + } + else + { + ip60 = + (ip6_header_t *) (((u8 *) ip40) + sizeof (*ip40) - + sizeof (*ip60)); + vlib_buffer_advance (p0, sizeof (*ip40) - sizeof (*ip60)); + frag0 = NULL; + } + + ip60->ip_version_traffic_class_and_flow_label = + clib_host_to_net_u32 ((6 << 28) + (ip40->tos << 20)); + ip60->payload_length = u16_net_add (ip40->length, -sizeof (*ip40)); + ip60->hop_limit = ip40->ttl; + ip60->protocol = ip40->protocol; + + if (PREDICT_FALSE (frag0 != NULL)) + { + frag0->next_hdr = ip60->protocol; + frag0->identification = frag_id0; + frag0->rsv = 0; + frag0->fragment_offset_and_more = + ip6_frag_hdr_offset_and_more (0, 1); + ip60->protocol = IP_PROTOCOL_IPV6_FRAGMENTATION; + ip60->payload_length = + u16_net_add (ip60->payload_length, sizeof (*frag0)); + } + + //Finally copying the address + ip60->dst_address.as_u64[0] = pheader0->daddr.as_u64[0]; + ip60->dst_address.as_u64[1] = pheader0->daddr.as_u64[1]; + ip60->src_address.as_u64[0] = pheader0->saddr.as_u64[0]; + ip60->src_address.as_u64[1] = pheader0->saddr.as_u64[1]; + + csum0 = ip_csum_add_even (csum0, ip60->src_address.as_u64[0]); + csum0 = ip_csum_add_even (csum0, ip60->src_address.as_u64[1]); + csum0 = ip_csum_add_even (csum0, ip60->dst_address.as_u64[0]); + csum0 = ip_csum_add_even (csum0, ip60->dst_address.as_u64[1]); + *checksum0 = ip_csum_fold (csum0); + + if (vnet_buffer (p0)->map_t.mtu < p0->current_length) + { + //Send to fragmentation node if necessary + vnet_buffer (p0)->ip_frag.header_offset = 0; + vnet_buffer (p0)->ip_frag.mtu = vnet_buffer (p0)->map_t.mtu; + vnet_buffer (p0)->ip_frag.next_index = IP6_FRAG_NEXT_IP6_LOOKUP; + next0 = IP4_MAPT_TCP_UDP_NEXT_IP6_FRAG; + } + + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, + to_next, n_left_to_next, pi0, + next0); + } + vlib_put_next_frame (vm, node, next_index, n_left_to_next); } - vlib_put_next_frame(vm, node, next_index, n_left_to_next); - } return frame->n_vectors; } static_always_inline void -ip4_map_t_classify(vlib_buffer_t *p0, map_domain_t *d0, ip4_header_t *ip40, u16 ip4_len0, - i32 *dst_port0, u8 *error0, ip4_mapt_next_t *next0) +ip4_map_t_classify (vlib_buffer_t * p0, map_domain_t * d0, + ip4_header_t * ip40, u16 ip4_len0, i32 * dst_port0, + u8 * error0, ip4_mapt_next_t * next0) { - if (PREDICT_FALSE(ip4_get_fragment_offset(ip40))) { - *next0 = IP4_MAPT_NEXT_MAPT_FRAGMENTED; - if(d0->ea_bits_len == 0 && d0->rules) { - *dst_port0 = 0; - } else { - *dst_port0 = ip4_map_fragment_get_port(ip40); - *error0 = (*dst_port0 == -1) ? MAP_ERROR_FRAGMENT_MEMORY : *error0; + if (PREDICT_FALSE (ip4_get_fragment_offset (ip40))) + { + *next0 = IP4_MAPT_NEXT_MAPT_FRAGMENTED; + if (d0->ea_bits_len == 0 && d0->rules) + { + *dst_port0 = 0; + } + else + { + *dst_port0 = ip4_map_fragment_get_port (ip40); + *error0 = (*dst_port0 == -1) ? MAP_ERROR_FRAGMENT_MEMORY : *error0; + } + } + else if (PREDICT_TRUE (ip40->protocol == IP_PROTOCOL_TCP)) + { + vnet_buffer (p0)->map_t.checksum_offset = 36; + *next0 = IP4_MAPT_NEXT_MAPT_TCP_UDP; + *error0 = ip4_len0 < 40 ? MAP_ERROR_MALFORMED : *error0; + *dst_port0 = (i32) * ((u16 *) u8_ptr_add (ip40, sizeof (*ip40) + 2)); + } + else if (PREDICT_TRUE (ip40->protocol == IP_PROTOCOL_UDP)) + { + vnet_buffer (p0)->map_t.checksum_offset = 26; + *next0 = IP4_MAPT_NEXT_MAPT_TCP_UDP; + *error0 = ip4_len0 < 28 ? MAP_ERROR_MALFORMED : *error0; + *dst_port0 = (i32) * ((u16 *) u8_ptr_add (ip40, sizeof (*ip40) + 2)); + } + else if (ip40->protocol == IP_PROTOCOL_ICMP) + { + *next0 = IP4_MAPT_NEXT_MAPT_ICMP; + if (d0->ea_bits_len == 0 && d0->rules) + *dst_port0 = 0; + else if (((icmp46_header_t *) u8_ptr_add (ip40, sizeof (*ip40)))->code + == ICMP4_echo_reply + || ((icmp46_header_t *) + u8_ptr_add (ip40, + sizeof (*ip40)))->code == ICMP4_echo_request) + *dst_port0 = (i32) * ((u16 *) u8_ptr_add (ip40, sizeof (*ip40) + 6)); + } + else + { + *error0 = MAP_ERROR_BAD_PROTOCOL; } - } else if (PREDICT_TRUE(ip40->protocol == IP_PROTOCOL_TCP)) { - vnet_buffer(p0)->map_t.checksum_offset = 36; - *next0 = IP4_MAPT_NEXT_MAPT_TCP_UDP; - *error0 = ip4_len0 < 40 ? MAP_ERROR_MALFORMED : *error0; - *dst_port0 = (i32) *((u16 *)u8_ptr_add(ip40, sizeof(*ip40) + 2)); - } else if (PREDICT_TRUE(ip40->protocol == IP_PROTOCOL_UDP)) { - vnet_buffer(p0)->map_t.checksum_offset = 26; - *next0 = IP4_MAPT_NEXT_MAPT_TCP_UDP; - *error0 = ip4_len0 < 28 ? MAP_ERROR_MALFORMED : *error0; - *dst_port0 = (i32) *((u16 *)u8_ptr_add(ip40, sizeof(*ip40) + 2)); - } else if (ip40->protocol == IP_PROTOCOL_ICMP) { - *next0 = IP4_MAPT_NEXT_MAPT_ICMP; - if(d0->ea_bits_len == 0 && d0->rules) - *dst_port0 = 0; - else if (((icmp46_header_t *) u8_ptr_add(ip40, sizeof(*ip40)))->code == ICMP4_echo_reply || - ((icmp46_header_t *) u8_ptr_add(ip40, sizeof(*ip40)))->code == ICMP4_echo_request) - *dst_port0 = (i32) *((u16 *)u8_ptr_add(ip40, sizeof(*ip40) + 6)); - } else { - *error0 = MAP_ERROR_BAD_PROTOCOL; - } } static uword -ip4_map_t (vlib_main_t *vm, - vlib_node_runtime_t *node, - vlib_frame_t *frame) +ip4_map_t (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) { u32 n_left_from, *from, next_index, *to_next, n_left_to_next; - vlib_node_runtime_t *error_node = vlib_node_get_runtime(vm, ip4_map_t_node.index); - from = vlib_frame_vector_args(frame); + vlib_node_runtime_t *error_node = + vlib_node_get_runtime (vm, ip4_map_t_node.index); + from = vlib_frame_vector_args (frame); n_left_from = frame->n_vectors; next_index = node->cached_next_index; vlib_combined_counter_main_t *cm = map_main.domain_counters; - u32 cpu_index = os_get_cpu_number(); + u32 cpu_index = os_get_cpu_number (); - while (n_left_from > 0) { - vlib_get_next_frame(vm, node, next_index, to_next, n_left_to_next); + while (n_left_from > 0) + { + vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); #ifdef IP4_MAP_T_DUAL_LOOP - while (n_left_from >= 4 && n_left_to_next >= 2) { - u32 pi0, pi1; - vlib_buffer_t *p0, *p1; - ip4_header_t *ip40, *ip41; - map_domain_t *d0, *d1; - ip4_mapt_next_t next0 = 0, next1 = 0; - u16 ip4_len0, ip4_len1; - u8 error0, error1; - i32 dst_port0, dst_port1; - ip4_mapt_pseudo_header_t *pheader0, *pheader1; - - pi0 = to_next[0] = from[0]; - pi1 = to_next[1] = from[1]; - from += 2; - n_left_from -= 2; - to_next +=2; - n_left_to_next -= 2; - error0 = MAP_ERROR_NONE; - error1 = MAP_ERROR_NONE; - - p0 = vlib_get_buffer(vm, pi0); - p1 = vlib_get_buffer(vm, pi1); - ip40 = vlib_buffer_get_current(p0); - ip41 = vlib_buffer_get_current(p1); - ip4_len0 = clib_host_to_net_u16(ip40->length); - ip4_len1 = clib_host_to_net_u16(ip41->length); - - if (PREDICT_FALSE(p0->current_length < ip4_len0 || - ip40->ip_version_and_header_length != 0x45)) { - error0 = MAP_ERROR_UNKNOWN; - next0 = IP4_MAPT_NEXT_DROP; - } - - if (PREDICT_FALSE(p1->current_length < ip4_len1 || - ip41->ip_version_and_header_length != 0x45)) { - error1 = MAP_ERROR_UNKNOWN; - next1 = IP4_MAPT_NEXT_DROP; - } - - d0 = ip4_map_get_domain(vnet_buffer(p0)->ip.adj_index[VLIB_TX], - &vnet_buffer(p0)->map_t.map_domain_index); - d1 = ip4_map_get_domain(vnet_buffer(p1)->ip.adj_index[VLIB_TX], - &vnet_buffer(p1)->map_t.map_domain_index); - - vnet_buffer(p0)->map_t.mtu = d0->mtu ? d0->mtu : ~0; - vnet_buffer(p1)->map_t.mtu = d1->mtu ? d1->mtu : ~0; - - dst_port0 = -1; - dst_port1 = -1; - - ip4_map_t_classify(p0, d0, ip40, ip4_len0, &dst_port0, &error0, &next0); - ip4_map_t_classify(p1, d1, ip41, ip4_len1, &dst_port1, &error1, &next1); - - //Add MAP-T pseudo header in front of the packet - vlib_buffer_advance(p0, - sizeof(*pheader0)); - vlib_buffer_advance(p1, - sizeof(*pheader1)); - pheader0 = vlib_buffer_get_current(p0); - pheader1 = vlib_buffer_get_current(p1); - - //Save addresses within the packet - ip4_map_t_embedded_address(d0, &pheader0->saddr, &ip40->src_address); - ip4_map_t_embedded_address(d1, &pheader1->saddr, &ip41->src_address); - pheader0->daddr.as_u64[0] = map_get_pfx_net(d0, ip40->dst_address.as_u32, (u16)dst_port0); - pheader0->daddr.as_u64[1] = map_get_sfx_net(d0, ip40->dst_address.as_u32, (u16)dst_port0); - pheader1->daddr.as_u64[0] = map_get_pfx_net(d1, ip41->dst_address.as_u32, (u16)dst_port1); - pheader1->daddr.as_u64[1] = map_get_sfx_net(d1, ip41->dst_address.as_u32, (u16)dst_port1); - - if (PREDICT_FALSE(ip4_is_first_fragment(ip40) && (dst_port0 != -1) && - (d0->ea_bits_len != 0 || !d0->rules) && - ip4_map_fragment_cache(ip40, dst_port0))) { - error0 = MAP_ERROR_FRAGMENT_MEMORY; - } - - if (PREDICT_FALSE(ip4_is_first_fragment(ip41) && (dst_port1 != -1) && - (d1->ea_bits_len != 0 || !d1->rules) && - ip4_map_fragment_cache(ip41, dst_port1))) { - error1 = MAP_ERROR_FRAGMENT_MEMORY; - } - - if (PREDICT_TRUE(error0 == MAP_ERROR_NONE && next0 != IP4_MAPT_NEXT_MAPT_ICMP)) { - vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_TX, cpu_index, - vnet_buffer(p0)->map_t.map_domain_index, 1, - clib_net_to_host_u16(ip40->length)); - } - - if (PREDICT_TRUE(error1 == MAP_ERROR_NONE && next1 != IP4_MAPT_NEXT_MAPT_ICMP)) { - vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_TX, cpu_index, - vnet_buffer(p1)->map_t.map_domain_index, 1, - clib_net_to_host_u16(ip41->length)); - } - - next0 = (error0 != MAP_ERROR_NONE) ? IP4_MAPT_NEXT_DROP : next0; - next1 = (error1 != MAP_ERROR_NONE) ? IP4_MAPT_NEXT_DROP : next1; - p0->error = error_node->errors[error0]; - p1->error = error_node->errors[error1]; - vlib_validate_buffer_enqueue_x2(vm, node, next_index, to_next, - n_left_to_next, pi0, pi1, next0, next1); - } + while (n_left_from >= 4 && n_left_to_next >= 2) + { + u32 pi0, pi1; + vlib_buffer_t *p0, *p1; + ip4_header_t *ip40, *ip41; + map_domain_t *d0, *d1; + ip4_mapt_next_t next0 = 0, next1 = 0; + u16 ip4_len0, ip4_len1; + u8 error0, error1; + i32 dst_port0, dst_port1; + ip4_mapt_pseudo_header_t *pheader0, *pheader1; + + pi0 = to_next[0] = from[0]; + pi1 = to_next[1] = from[1]; + from += 2; + n_left_from -= 2; + to_next += 2; + n_left_to_next -= 2; + error0 = MAP_ERROR_NONE; + error1 = MAP_ERROR_NONE; + + p0 = vlib_get_buffer (vm, pi0); + p1 = vlib_get_buffer (vm, pi1); + ip40 = vlib_buffer_get_current (p0); + ip41 = vlib_buffer_get_current (p1); + ip4_len0 = clib_host_to_net_u16 (ip40->length); + ip4_len1 = clib_host_to_net_u16 (ip41->length); + + if (PREDICT_FALSE (p0->current_length < ip4_len0 || + ip40->ip_version_and_header_length != 0x45)) + { + error0 = MAP_ERROR_UNKNOWN; + next0 = IP4_MAPT_NEXT_DROP; + } + + if (PREDICT_FALSE (p1->current_length < ip4_len1 || + ip41->ip_version_and_header_length != 0x45)) + { + error1 = MAP_ERROR_UNKNOWN; + next1 = IP4_MAPT_NEXT_DROP; + } + + d0 = ip4_map_get_domain (vnet_buffer (p0)->ip.adj_index[VLIB_TX], + &vnet_buffer (p0)->map_t.map_domain_index); + d1 = ip4_map_get_domain (vnet_buffer (p1)->ip.adj_index[VLIB_TX], + &vnet_buffer (p1)->map_t.map_domain_index); + + vnet_buffer (p0)->map_t.mtu = d0->mtu ? d0->mtu : ~0; + vnet_buffer (p1)->map_t.mtu = d1->mtu ? d1->mtu : ~0; + + dst_port0 = -1; + dst_port1 = -1; + + ip4_map_t_classify (p0, d0, ip40, ip4_len0, &dst_port0, &error0, + &next0); + ip4_map_t_classify (p1, d1, ip41, ip4_len1, &dst_port1, &error1, + &next1); + + //Add MAP-T pseudo header in front of the packet + vlib_buffer_advance (p0, -sizeof (*pheader0)); + vlib_buffer_advance (p1, -sizeof (*pheader1)); + pheader0 = vlib_buffer_get_current (p0); + pheader1 = vlib_buffer_get_current (p1); + + //Save addresses within the packet + ip4_map_t_embedded_address (d0, &pheader0->saddr, + &ip40->src_address); + ip4_map_t_embedded_address (d1, &pheader1->saddr, + &ip41->src_address); + pheader0->daddr.as_u64[0] = + map_get_pfx_net (d0, ip40->dst_address.as_u32, (u16) dst_port0); + pheader0->daddr.as_u64[1] = + map_get_sfx_net (d0, ip40->dst_address.as_u32, (u16) dst_port0); + pheader1->daddr.as_u64[0] = + map_get_pfx_net (d1, ip41->dst_address.as_u32, (u16) dst_port1); + pheader1->daddr.as_u64[1] = + map_get_sfx_net (d1, ip41->dst_address.as_u32, (u16) dst_port1); + + if (PREDICT_FALSE + (ip4_is_first_fragment (ip40) && (dst_port0 != -1) + && (d0->ea_bits_len != 0 || !d0->rules) + && ip4_map_fragment_cache (ip40, dst_port0))) + { + error0 = MAP_ERROR_FRAGMENT_MEMORY; + } + + if (PREDICT_FALSE + (ip4_is_first_fragment (ip41) && (dst_port1 != -1) + && (d1->ea_bits_len != 0 || !d1->rules) + && ip4_map_fragment_cache (ip41, dst_port1))) + { + error1 = MAP_ERROR_FRAGMENT_MEMORY; + } + + if (PREDICT_TRUE + (error0 == MAP_ERROR_NONE && next0 != IP4_MAPT_NEXT_MAPT_ICMP)) + { + vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_TX, + cpu_index, + vnet_buffer (p0)->map_t. + map_domain_index, 1, + clib_net_to_host_u16 (ip40-> + length)); + } + + if (PREDICT_TRUE + (error1 == MAP_ERROR_NONE && next1 != IP4_MAPT_NEXT_MAPT_ICMP)) + { + vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_TX, + cpu_index, + vnet_buffer (p1)->map_t. + map_domain_index, 1, + clib_net_to_host_u16 (ip41-> + length)); + } + + next0 = (error0 != MAP_ERROR_NONE) ? IP4_MAPT_NEXT_DROP : next0; + next1 = (error1 != MAP_ERROR_NONE) ? IP4_MAPT_NEXT_DROP : next1; + p0->error = error_node->errors[error0]; + p1->error = error_node->errors[error1]; + vlib_validate_buffer_enqueue_x2 (vm, node, next_index, to_next, + n_left_to_next, pi0, pi1, next0, + next1); + } #endif - while (n_left_from > 0 && n_left_to_next > 0) { - u32 pi0; - vlib_buffer_t *p0; - ip4_header_t *ip40; - map_domain_t *d0; - ip4_mapt_next_t next0; - u16 ip4_len0; - u8 error0; - i32 dst_port0; - ip4_mapt_pseudo_header_t *pheader0; - - pi0 = to_next[0] = from[0]; - from += 1; - n_left_from -= 1; - to_next +=1; - n_left_to_next -= 1; - error0 = MAP_ERROR_NONE; - - p0 = vlib_get_buffer(vm, pi0); - ip40 = vlib_buffer_get_current(p0); - ip4_len0 = clib_host_to_net_u16(ip40->length); - if (PREDICT_FALSE(p0->current_length < ip4_len0 || - ip40->ip_version_and_header_length != 0x45)) { - error0 = MAP_ERROR_UNKNOWN; - next0 = IP4_MAPT_NEXT_DROP; - } - - d0 = ip4_map_get_domain(vnet_buffer(p0)->ip.adj_index[VLIB_TX], - &vnet_buffer(p0)->map_t.map_domain_index); - - vnet_buffer(p0)->map_t.mtu = d0->mtu ? d0->mtu : ~0; - - dst_port0 = -1; - ip4_map_t_classify(p0, d0, ip40, ip4_len0, &dst_port0, &error0, &next0); - - //Add MAP-T pseudo header in front of the packet - vlib_buffer_advance(p0, - sizeof(*pheader0)); - pheader0 = vlib_buffer_get_current(p0); - - //Save addresses within the packet - ip4_map_t_embedded_address(d0, &pheader0->saddr, &ip40->src_address); - pheader0->daddr.as_u64[0] = map_get_pfx_net(d0, ip40->dst_address.as_u32, (u16)dst_port0); - pheader0->daddr.as_u64[1] = map_get_sfx_net(d0, ip40->dst_address.as_u32, (u16)dst_port0); - - //It is important to cache at this stage because the result might be necessary - //for packets within the same vector. - //Actually, this approach even provides some limited out-of-order fragments support - if (PREDICT_FALSE(ip4_is_first_fragment(ip40) && (dst_port0 != -1) && - (d0->ea_bits_len != 0 || !d0->rules) && - ip4_map_fragment_cache(ip40, dst_port0))) { - error0 = MAP_ERROR_UNKNOWN; - } - - if (PREDICT_TRUE(error0 == MAP_ERROR_NONE && next0 != IP4_MAPT_NEXT_MAPT_ICMP)) { - vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_TX, cpu_index, - vnet_buffer(p0)->map_t.map_domain_index, 1, - clib_net_to_host_u16(ip40->length)); - } - - next0 = (error0 != MAP_ERROR_NONE) ? IP4_MAPT_NEXT_DROP : next0; - p0->error = error_node->errors[error0]; - vlib_validate_buffer_enqueue_x1 (vm, node, next_index, - to_next, n_left_to_next, pi0, - next0); + while (n_left_from > 0 && n_left_to_next > 0) + { + u32 pi0; + vlib_buffer_t *p0; + ip4_header_t *ip40; + map_domain_t *d0; + ip4_mapt_next_t next0; + u16 ip4_len0; + u8 error0; + i32 dst_port0; + ip4_mapt_pseudo_header_t *pheader0; + + pi0 = to_next[0] = from[0]; + from += 1; + n_left_from -= 1; + to_next += 1; + n_left_to_next -= 1; + error0 = MAP_ERROR_NONE; + + p0 = vlib_get_buffer (vm, pi0); + ip40 = vlib_buffer_get_current (p0); + ip4_len0 = clib_host_to_net_u16 (ip40->length); + if (PREDICT_FALSE (p0->current_length < ip4_len0 || + ip40->ip_version_and_header_length != 0x45)) + { + error0 = MAP_ERROR_UNKNOWN; + next0 = IP4_MAPT_NEXT_DROP; + } + + d0 = ip4_map_get_domain (vnet_buffer (p0)->ip.adj_index[VLIB_TX], + &vnet_buffer (p0)->map_t.map_domain_index); + + vnet_buffer (p0)->map_t.mtu = d0->mtu ? d0->mtu : ~0; + + dst_port0 = -1; + ip4_map_t_classify (p0, d0, ip40, ip4_len0, &dst_port0, &error0, + &next0); + + //Add MAP-T pseudo header in front of the packet + vlib_buffer_advance (p0, -sizeof (*pheader0)); + pheader0 = vlib_buffer_get_current (p0); + + //Save addresses within the packet + ip4_map_t_embedded_address (d0, &pheader0->saddr, + &ip40->src_address); + pheader0->daddr.as_u64[0] = + map_get_pfx_net (d0, ip40->dst_address.as_u32, (u16) dst_port0); + pheader0->daddr.as_u64[1] = + map_get_sfx_net (d0, ip40->dst_address.as_u32, (u16) dst_port0); + + //It is important to cache at this stage because the result might be necessary + //for packets within the same vector. + //Actually, this approach even provides some limited out-of-order fragments support + if (PREDICT_FALSE + (ip4_is_first_fragment (ip40) && (dst_port0 != -1) + && (d0->ea_bits_len != 0 || !d0->rules) + && ip4_map_fragment_cache (ip40, dst_port0))) + { + error0 = MAP_ERROR_UNKNOWN; + } + + if (PREDICT_TRUE + (error0 == MAP_ERROR_NONE && next0 != IP4_MAPT_NEXT_MAPT_ICMP)) + { + vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_TX, + cpu_index, + vnet_buffer (p0)->map_t. + map_domain_index, 1, + clib_net_to_host_u16 (ip40-> + length)); + } + + next0 = (error0 != MAP_ERROR_NONE) ? IP4_MAPT_NEXT_DROP : next0; + p0->error = error_node->errors[error0]; + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, + to_next, n_left_to_next, pi0, + next0); + } + vlib_put_next_frame (vm, node, next_index, n_left_to_next); } - vlib_put_next_frame(vm, node, next_index, n_left_to_next); - } return frame->n_vectors; } @@ -1018,6 +1272,7 @@ static char *map_t_error_strings[] = { #undef _ }; +/* *INDENT-OFF* */ VLIB_REGISTER_NODE(ip4_map_t_fragmented_node) = { .function = ip4_map_t_fragmented, .name = "ip4-map-t-fragmented", @@ -1035,7 +1290,9 @@ VLIB_REGISTER_NODE(ip4_map_t_fragmented_node) = { [IP4_MAPT_FRAGMENTED_NEXT_DROP] = "error-drop", }, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_REGISTER_NODE(ip4_map_t_icmp_node) = { .function = ip4_map_t_icmp, .name = "ip4-map-t-icmp", @@ -1053,7 +1310,9 @@ VLIB_REGISTER_NODE(ip4_map_t_icmp_node) = { [IP4_MAPT_ICMP_NEXT_DROP] = "error-drop", }, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_REGISTER_NODE(ip4_map_t_tcp_udp_node) = { .function = ip4_map_t_tcp_udp, .name = "ip4-map-t-tcp-udp", @@ -1071,7 +1330,9 @@ VLIB_REGISTER_NODE(ip4_map_t_tcp_udp_node) = { [IP4_MAPT_TCP_UDP_NEXT_DROP] = "error-drop", }, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_REGISTER_NODE(ip4_map_t_node) = { .function = ip4_map_t, .name = "ip4-map-t", @@ -1090,3 +1351,12 @@ VLIB_REGISTER_NODE(ip4_map_t_node) = { [IP4_MAPT_NEXT_DROP] = "error-drop", }, }; +/* *INDENT-ON* */ + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/vnet/vnet/map/ip6_map.c b/vnet/vnet/map/ip6_map.c index 178b7c3fb9e..b51e8358d41 100644 --- a/vnet/vnet/map/ip6_map.c +++ b/vnet/vnet/map/ip6_map.c @@ -16,7 +16,8 @@ #include "../ip/ip_frag.h" -enum ip6_map_next_e { +enum ip6_map_next_e +{ IP6_MAP_NEXT_IP4_LOOKUP, #ifdef MAP_SKIP_IP6_LOOKUP IP6_MAP_NEXT_IP4_REWRITE, @@ -31,20 +32,23 @@ enum ip6_map_next_e { IP6_MAP_N_NEXT, }; -enum ip6_map_ip6_reass_next_e { +enum ip6_map_ip6_reass_next_e +{ IP6_MAP_IP6_REASS_NEXT_IP6_MAP, IP6_MAP_IP6_REASS_NEXT_DROP, IP6_MAP_IP6_REASS_N_NEXT, }; -enum ip6_map_ip4_reass_next_e { +enum ip6_map_ip4_reass_next_e +{ IP6_MAP_IP4_REASS_NEXT_IP4_LOOKUP, IP6_MAP_IP4_REASS_NEXT_IP4_FRAGMENT, IP6_MAP_IP4_REASS_NEXT_DROP, IP6_MAP_IP4_REASS_N_NEXT, }; -enum ip6_icmp_relay_next_e { +enum ip6_icmp_relay_next_e +{ IP6_ICMP_RELAY_NEXT_IP4_LOOKUP, IP6_ICMP_RELAY_NEXT_DROP, IP6_ICMP_RELAY_N_NEXT, @@ -54,91 +58,113 @@ vlib_node_registration_t ip6_map_ip4_reass_node; vlib_node_registration_t ip6_map_ip6_reass_node; static vlib_node_registration_t ip6_map_icmp_relay_node; -typedef struct { +typedef struct +{ u32 map_domain_index; u16 port; u8 cached; } map_ip6_map_ip4_reass_trace_t; u8 * -format_ip6_map_ip4_reass_trace (u8 *s, va_list *args) +format_ip6_map_ip4_reass_trace (u8 * s, va_list * args) { - CLIB_UNUSED(vlib_main_t *vm) = va_arg (*args, vlib_main_t *); - CLIB_UNUSED(vlib_node_t *node) = va_arg (*args, vlib_node_t *); - map_ip6_map_ip4_reass_trace_t *t = va_arg (*args, map_ip6_map_ip4_reass_trace_t *); - return format(s, "MAP domain index: %d L4 port: %u Status: %s", t->map_domain_index, - t->port, t->cached?"cached":"forwarded"); + CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *); + CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *); + map_ip6_map_ip4_reass_trace_t *t = + va_arg (*args, map_ip6_map_ip4_reass_trace_t *); + return format (s, "MAP domain index: %d L4 port: %u Status: %s", + t->map_domain_index, t->port, + t->cached ? "cached" : "forwarded"); } -typedef struct { +typedef struct +{ u16 offset; u16 frag_len; u8 out; } map_ip6_map_ip6_reass_trace_t; u8 * -format_ip6_map_ip6_reass_trace (u8 *s, va_list *args) +format_ip6_map_ip6_reass_trace (u8 * s, va_list * args) { - CLIB_UNUSED(vlib_main_t *vm) = va_arg (*args, vlib_main_t *); - CLIB_UNUSED(vlib_node_t *node) = va_arg (*args, vlib_node_t *); - map_ip6_map_ip6_reass_trace_t *t = va_arg (*args, map_ip6_map_ip6_reass_trace_t *); - return format(s, "Offset: %d Fragment length: %d Status: %s", t->offset, t->frag_len, t->out?"out":"in"); + CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *); + CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *); + map_ip6_map_ip6_reass_trace_t *t = + va_arg (*args, map_ip6_map_ip6_reass_trace_t *); + return format (s, "Offset: %d Fragment length: %d Status: %s", t->offset, + t->frag_len, t->out ? "out" : "in"); } /* * ip6_map_sec_check */ static_always_inline bool -ip6_map_sec_check (map_domain_t *d, u16 port, ip4_header_t *ip4, ip6_header_t *ip6) +ip6_map_sec_check (map_domain_t * d, u16 port, ip4_header_t * ip4, + ip6_header_t * ip6) { - u16 sp4 = clib_net_to_host_u16(port); - u32 sa4 = clib_net_to_host_u32(ip4->src_address.as_u32); - u64 sal6 = map_get_pfx(d, sa4, sp4); - u64 sar6 = map_get_sfx(d, sa4, sp4); - - if (PREDICT_FALSE(sal6 != clib_net_to_host_u64(ip6->src_address.as_u64[0]) || - sar6 != clib_net_to_host_u64(ip6->src_address.as_u64[1]))) + u16 sp4 = clib_net_to_host_u16 (port); + u32 sa4 = clib_net_to_host_u32 (ip4->src_address.as_u32); + u64 sal6 = map_get_pfx (d, sa4, sp4); + u64 sar6 = map_get_sfx (d, sa4, sp4); + + if (PREDICT_FALSE + (sal6 != clib_net_to_host_u64 (ip6->src_address.as_u64[0]) + || sar6 != clib_net_to_host_u64 (ip6->src_address.as_u64[1]))) return (false); return (true); } static_always_inline void -ip6_map_security_check (map_domain_t *d, ip4_header_t *ip4, ip6_header_t *ip6, u32 *next, u8 *error) +ip6_map_security_check (map_domain_t * d, ip4_header_t * ip4, + ip6_header_t * ip6, u32 * next, u8 * error) { map_main_t *mm = &map_main; - if (d->ea_bits_len || d->rules) { - if (d->psid_length > 0) { - if (!ip4_is_fragment(ip4)) { - u16 port = ip4_map_get_port(ip4, MAP_SENDER); - if (port) { - if (mm->sec_check) - *error = ip6_map_sec_check(d, port, ip4, ip6) ? MAP_ERROR_NONE : MAP_ERROR_DECAP_SEC_CHECK; - } else { - *error = MAP_ERROR_BAD_PROTOCOL; + if (d->ea_bits_len || d->rules) + { + if (d->psid_length > 0) + { + if (!ip4_is_fragment (ip4)) + { + u16 port = ip4_map_get_port (ip4, MAP_SENDER); + if (port) + { + if (mm->sec_check) + *error = + ip6_map_sec_check (d, port, ip4, + ip6) ? MAP_ERROR_NONE : + MAP_ERROR_DECAP_SEC_CHECK; + } + else + { + *error = MAP_ERROR_BAD_PROTOCOL; + } + } + else + { + *next = mm->sec_check_frag ? IP6_MAP_NEXT_IP4_REASS : *next; + } } - } else { - *next = mm->sec_check_frag ? IP6_MAP_NEXT_IP4_REASS : *next; - } } - } } static_always_inline bool -ip6_map_ip4_lookup_bypass (vlib_buffer_t *p0, ip4_header_t *ip) +ip6_map_ip4_lookup_bypass (vlib_buffer_t * p0, ip4_header_t * ip) { #ifdef MAP_SKIP_IP6_LOOKUP map_main_t *mm = &map_main; u32 adj_index0 = mm->adj4_index; - if (adj_index0 > 0) { - ip_lookup_main_t *lm4 = &ip4_main.lookup_main; - ip_adjacency_t *adj = ip_get_adjacency(lm4, mm->adj4_index); - if (adj->n_adj > 1) { - u32 hash_c0 = ip4_compute_flow_hash(ip, IP_FLOW_HASH_DEFAULT); - adj_index0 += (hash_c0 & (adj->n_adj - 1)); + if (adj_index0 > 0) + { + ip_lookup_main_t *lm4 = &ip4_main.lookup_main; + ip_adjacency_t *adj = ip_get_adjacency (lm4, mm->adj4_index); + if (adj->n_adj > 1) + { + u32 hash_c0 = ip4_compute_flow_hash (ip, IP_FLOW_HASH_DEFAULT); + adj_index0 += (hash_c0 & (adj->n_adj - 1)); + } + vnet_buffer (p0)->ip.adj_index[VLIB_TX] = adj_index0; + return (true); } - vnet_buffer(p0)->ip.adj_index[VLIB_TX] = adj_index0; - return (true); - } #endif return (false); } @@ -147,376 +173,503 @@ ip6_map_ip4_lookup_bypass (vlib_buffer_t *p0, ip4_header_t *ip) * ip6_map */ static uword -ip6_map (vlib_main_t *vm, - vlib_node_runtime_t *node, - vlib_frame_t *frame) +ip6_map (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) { u32 n_left_from, *from, next_index, *to_next, n_left_to_next; - vlib_node_runtime_t *error_node = vlib_node_get_runtime(vm, ip6_map_node.index); + vlib_node_runtime_t *error_node = + vlib_node_get_runtime (vm, ip6_map_node.index); map_main_t *mm = &map_main; vlib_combined_counter_main_t *cm = mm->domain_counters; - u32 cpu_index = os_get_cpu_number(); + u32 cpu_index = os_get_cpu_number (); - from = vlib_frame_vector_args(frame); + from = vlib_frame_vector_args (frame); n_left_from = frame->n_vectors; next_index = node->cached_next_index; - while (n_left_from > 0) { - vlib_get_next_frame(vm, node, next_index, to_next, n_left_to_next); - - /* Dual loop */ - while (n_left_from >= 4 && n_left_to_next >= 2) { - u32 pi0, pi1; - vlib_buffer_t *p0, *p1; - u8 error0 = MAP_ERROR_NONE; - u8 error1 = MAP_ERROR_NONE; - map_domain_t *d0 = 0, *d1 = 0; - ip4_header_t *ip40, *ip41; - ip6_header_t *ip60, *ip61; - u16 port0 = 0, port1 = 0; - u32 map_domain_index0 = ~0, map_domain_index1 = ~0; - u32 next0 = IP6_MAP_NEXT_IP4_LOOKUP; - u32 next1 = IP6_MAP_NEXT_IP4_LOOKUP; - - /* Prefetch next iteration. */ - { - vlib_buffer_t *p2, *p3; - - p2 = vlib_get_buffer(vm, from[2]); - p3 = vlib_get_buffer(vm, from[3]); - - vlib_prefetch_buffer_header(p2, LOAD); - vlib_prefetch_buffer_header(p3, LOAD); - - /* IPv6 + IPv4 header + 8 bytes of ULP */ - CLIB_PREFETCH(p2->data, 68, LOAD); - CLIB_PREFETCH(p3->data, 68, LOAD); - } - - pi0 = to_next[0] = from[0]; - pi1 = to_next[1] = from[1]; - from += 2; - n_left_from -= 2; - to_next +=2; - n_left_to_next -= 2; - - p0 = vlib_get_buffer(vm, pi0); - p1 = vlib_get_buffer(vm, pi1); - ip60 = vlib_buffer_get_current(p0); - ip61 = vlib_buffer_get_current(p1); - vlib_buffer_advance(p0, sizeof(ip6_header_t)); - vlib_buffer_advance(p1, sizeof(ip6_header_t)); - ip40 = vlib_buffer_get_current(p0); - ip41 = vlib_buffer_get_current(p1); - - /* - * Encapsulated IPv4 packet - * - IPv4 fragmented -> Pass to virtual reassembly unless security check disabled - * - Lookup/Rewrite or Fragment node in case of packet > MTU - * Fragmented IPv6 packet - * ICMP IPv6 packet - * - Error -> Pass to ICMPv6/ICMPv4 relay - * - Info -> Pass to IPv6 local - * Anything else -> drop - */ - if (PREDICT_TRUE(ip60->protocol == IP_PROTOCOL_IP_IN_IP && clib_net_to_host_u16(ip60->payload_length) > 20)) { - d0 = ip6_map_get_domain(vnet_buffer(p0)->ip.adj_index[VLIB_TX], (ip4_address_t *)&ip40->src_address.as_u32, - &map_domain_index0, &error0); - } else if (ip60->protocol == IP_PROTOCOL_ICMP6 && - clib_net_to_host_u16(ip60->payload_length) > sizeof(icmp46_header_t)) { - icmp46_header_t *icmp = (void *)(ip60 + 1); - next0 = (icmp->type == ICMP6_echo_request || icmp->type == ICMP6_echo_reply) ? - IP6_MAP_NEXT_IP6_LOCAL : IP6_MAP_NEXT_IP6_ICMP_RELAY; - } else if (ip60->protocol == IP_PROTOCOL_IPV6_FRAGMENTATION) { - next0 = IP6_MAP_NEXT_IP6_REASS; - } else { - error0 = MAP_ERROR_BAD_PROTOCOL; - } - if (PREDICT_TRUE(ip61->protocol == IP_PROTOCOL_IP_IN_IP && clib_net_to_host_u16(ip61->payload_length) > 20)) { - d1 = ip6_map_get_domain(vnet_buffer(p1)->ip.adj_index[VLIB_TX], (ip4_address_t *)&ip41->src_address.as_u32, - &map_domain_index1, &error1); - } else if (ip61->protocol == IP_PROTOCOL_ICMP6 && - clib_net_to_host_u16(ip61->payload_length) > sizeof(icmp46_header_t)) { - icmp46_header_t *icmp = (void *)(ip61 + 1); - next1 = (icmp->type == ICMP6_echo_request || icmp->type == ICMP6_echo_reply) ? - IP6_MAP_NEXT_IP6_LOCAL : IP6_MAP_NEXT_IP6_ICMP_RELAY; - } else if (ip61->protocol == IP_PROTOCOL_IPV6_FRAGMENTATION) { - next1 = IP6_MAP_NEXT_IP6_REASS; - } else { - error1 = MAP_ERROR_BAD_PROTOCOL; - } - - if (d0) { - /* MAP inbound security check */ - ip6_map_security_check(d0, ip40, ip60, &next0, &error0); - - if (PREDICT_TRUE(error0 == MAP_ERROR_NONE && - next0 == IP6_MAP_NEXT_IP4_LOOKUP)) { - if (PREDICT_FALSE(d0->mtu && (clib_host_to_net_u16(ip40->length) > d0->mtu))) { - vnet_buffer(p0)->ip_frag.header_offset = 0; - vnet_buffer(p0)->ip_frag.flags = 0; - vnet_buffer(p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP4_LOOKUP; - vnet_buffer(p0)->ip_frag.mtu = d0->mtu; - next0 = IP6_MAP_NEXT_IP4_FRAGMENT; - } else { - next0 = ip6_map_ip4_lookup_bypass(p0, ip40) ? IP6_MAP_NEXT_IP4_REWRITE : next0; - } - vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_RX, cpu_index, map_domain_index0, 1, - clib_net_to_host_u16(ip40->length)); - } - } - if (d1) { - /* MAP inbound security check */ - ip6_map_security_check(d1, ip41, ip61, &next1, &error1); - - if (PREDICT_TRUE(error1 == MAP_ERROR_NONE && - next1 == IP6_MAP_NEXT_IP4_LOOKUP)) { - if (PREDICT_FALSE(d1->mtu && (clib_host_to_net_u16(ip41->length) > d1->mtu))) { - vnet_buffer(p1)->ip_frag.header_offset = 0; - vnet_buffer(p1)->ip_frag.flags = 0; - vnet_buffer(p1)->ip_frag.next_index = IP4_FRAG_NEXT_IP4_LOOKUP; - vnet_buffer(p1)->ip_frag.mtu = d1->mtu; - next1 = IP6_MAP_NEXT_IP4_FRAGMENT; - } else { - next1 = ip6_map_ip4_lookup_bypass(p1, ip41) ? IP6_MAP_NEXT_IP4_REWRITE : next1; + while (n_left_from > 0) + { + vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); + + /* Dual loop */ + while (n_left_from >= 4 && n_left_to_next >= 2) + { + u32 pi0, pi1; + vlib_buffer_t *p0, *p1; + u8 error0 = MAP_ERROR_NONE; + u8 error1 = MAP_ERROR_NONE; + map_domain_t *d0 = 0, *d1 = 0; + ip4_header_t *ip40, *ip41; + ip6_header_t *ip60, *ip61; + u16 port0 = 0, port1 = 0; + u32 map_domain_index0 = ~0, map_domain_index1 = ~0; + u32 next0 = IP6_MAP_NEXT_IP4_LOOKUP; + u32 next1 = IP6_MAP_NEXT_IP4_LOOKUP; + + /* Prefetch next iteration. */ + { + vlib_buffer_t *p2, *p3; + + p2 = vlib_get_buffer (vm, from[2]); + p3 = vlib_get_buffer (vm, from[3]); + + vlib_prefetch_buffer_header (p2, LOAD); + vlib_prefetch_buffer_header (p3, LOAD); + + /* IPv6 + IPv4 header + 8 bytes of ULP */ + CLIB_PREFETCH (p2->data, 68, LOAD); + CLIB_PREFETCH (p3->data, 68, LOAD); } - vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_RX, cpu_index, map_domain_index1, 1, - clib_net_to_host_u16(ip41->length)); + + pi0 = to_next[0] = from[0]; + pi1 = to_next[1] = from[1]; + from += 2; + n_left_from -= 2; + to_next += 2; + n_left_to_next -= 2; + + p0 = vlib_get_buffer (vm, pi0); + p1 = vlib_get_buffer (vm, pi1); + ip60 = vlib_buffer_get_current (p0); + ip61 = vlib_buffer_get_current (p1); + vlib_buffer_advance (p0, sizeof (ip6_header_t)); + vlib_buffer_advance (p1, sizeof (ip6_header_t)); + ip40 = vlib_buffer_get_current (p0); + ip41 = vlib_buffer_get_current (p1); + + /* + * Encapsulated IPv4 packet + * - IPv4 fragmented -> Pass to virtual reassembly unless security check disabled + * - Lookup/Rewrite or Fragment node in case of packet > MTU + * Fragmented IPv6 packet + * ICMP IPv6 packet + * - Error -> Pass to ICMPv6/ICMPv4 relay + * - Info -> Pass to IPv6 local + * Anything else -> drop + */ + if (PREDICT_TRUE + (ip60->protocol == IP_PROTOCOL_IP_IN_IP + && clib_net_to_host_u16 (ip60->payload_length) > 20)) + { + d0 = + ip6_map_get_domain (vnet_buffer (p0)->ip.adj_index[VLIB_TX], + (ip4_address_t *) & ip40->src_address. + as_u32, &map_domain_index0, &error0); + } + else if (ip60->protocol == IP_PROTOCOL_ICMP6 && + clib_net_to_host_u16 (ip60->payload_length) > + sizeof (icmp46_header_t)) + { + icmp46_header_t *icmp = (void *) (ip60 + 1); + next0 = (icmp->type == ICMP6_echo_request + || icmp->type == + ICMP6_echo_reply) ? IP6_MAP_NEXT_IP6_LOCAL : + IP6_MAP_NEXT_IP6_ICMP_RELAY; + } + else if (ip60->protocol == IP_PROTOCOL_IPV6_FRAGMENTATION) + { + next0 = IP6_MAP_NEXT_IP6_REASS; + } + else + { + error0 = MAP_ERROR_BAD_PROTOCOL; + } + if (PREDICT_TRUE + (ip61->protocol == IP_PROTOCOL_IP_IN_IP + && clib_net_to_host_u16 (ip61->payload_length) > 20)) + { + d1 = + ip6_map_get_domain (vnet_buffer (p1)->ip.adj_index[VLIB_TX], + (ip4_address_t *) & ip41->src_address. + as_u32, &map_domain_index1, &error1); + } + else if (ip61->protocol == IP_PROTOCOL_ICMP6 && + clib_net_to_host_u16 (ip61->payload_length) > + sizeof (icmp46_header_t)) + { + icmp46_header_t *icmp = (void *) (ip61 + 1); + next1 = (icmp->type == ICMP6_echo_request + || icmp->type == + ICMP6_echo_reply) ? IP6_MAP_NEXT_IP6_LOCAL : + IP6_MAP_NEXT_IP6_ICMP_RELAY; + } + else if (ip61->protocol == IP_PROTOCOL_IPV6_FRAGMENTATION) + { + next1 = IP6_MAP_NEXT_IP6_REASS; + } + else + { + error1 = MAP_ERROR_BAD_PROTOCOL; + } + + if (d0) + { + /* MAP inbound security check */ + ip6_map_security_check (d0, ip40, ip60, &next0, &error0); + + if (PREDICT_TRUE (error0 == MAP_ERROR_NONE && + next0 == IP6_MAP_NEXT_IP4_LOOKUP)) + { + if (PREDICT_FALSE + (d0->mtu + && (clib_host_to_net_u16 (ip40->length) > d0->mtu))) + { + vnet_buffer (p0)->ip_frag.header_offset = 0; + vnet_buffer (p0)->ip_frag.flags = 0; + vnet_buffer (p0)->ip_frag.next_index = + IP4_FRAG_NEXT_IP4_LOOKUP; + vnet_buffer (p0)->ip_frag.mtu = d0->mtu; + next0 = IP6_MAP_NEXT_IP4_FRAGMENT; + } + else + { + next0 = + ip6_map_ip4_lookup_bypass (p0, + ip40) ? + IP6_MAP_NEXT_IP4_REWRITE : next0; + } + vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_RX, + cpu_index, + map_domain_index0, 1, + clib_net_to_host_u16 + (ip40->length)); + } + } + if (d1) + { + /* MAP inbound security check */ + ip6_map_security_check (d1, ip41, ip61, &next1, &error1); + + if (PREDICT_TRUE (error1 == MAP_ERROR_NONE && + next1 == IP6_MAP_NEXT_IP4_LOOKUP)) + { + if (PREDICT_FALSE + (d1->mtu + && (clib_host_to_net_u16 (ip41->length) > d1->mtu))) + { + vnet_buffer (p1)->ip_frag.header_offset = 0; + vnet_buffer (p1)->ip_frag.flags = 0; + vnet_buffer (p1)->ip_frag.next_index = + IP4_FRAG_NEXT_IP4_LOOKUP; + vnet_buffer (p1)->ip_frag.mtu = d1->mtu; + next1 = IP6_MAP_NEXT_IP4_FRAGMENT; + } + else + { + next1 = + ip6_map_ip4_lookup_bypass (p1, + ip41) ? + IP6_MAP_NEXT_IP4_REWRITE : next1; + } + vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_RX, + cpu_index, + map_domain_index1, 1, + clib_net_to_host_u16 + (ip41->length)); + } + } + + if (PREDICT_FALSE (p0->flags & VLIB_BUFFER_IS_TRACED)) + { + map_trace_t *tr = vlib_add_trace (vm, node, p0, sizeof (*tr)); + tr->map_domain_index = map_domain_index0; + tr->port = port0; + } + + if (PREDICT_FALSE (p1->flags & VLIB_BUFFER_IS_TRACED)) + { + map_trace_t *tr = vlib_add_trace (vm, node, p1, sizeof (*tr)); + tr->map_domain_index = map_domain_index1; + tr->port = port1; + } + + if (error0 == MAP_ERROR_DECAP_SEC_CHECK && mm->icmp6_enabled) + { + /* Set ICMP parameters */ + vlib_buffer_advance (p0, -sizeof (ip6_header_t)); + icmp6_error_set_vnet_buffer (p0, ICMP6_destination_unreachable, + ICMP6_destination_unreachable_source_address_failed_policy, + 0); + next0 = IP6_MAP_NEXT_ICMP; + } + else + { + next0 = (error0 == MAP_ERROR_NONE) ? next0 : IP6_MAP_NEXT_DROP; + } + + if (error1 == MAP_ERROR_DECAP_SEC_CHECK && mm->icmp6_enabled) + { + /* Set ICMP parameters */ + vlib_buffer_advance (p1, -sizeof (ip6_header_t)); + icmp6_error_set_vnet_buffer (p1, ICMP6_destination_unreachable, + ICMP6_destination_unreachable_source_address_failed_policy, + 0); + next1 = IP6_MAP_NEXT_ICMP; + } + else + { + next1 = (error1 == MAP_ERROR_NONE) ? next1 : IP6_MAP_NEXT_DROP; + } + + /* Reset packet */ + if (next0 == IP6_MAP_NEXT_IP6_LOCAL) + vlib_buffer_advance (p0, -sizeof (ip6_header_t)); + if (next1 == IP6_MAP_NEXT_IP6_LOCAL) + vlib_buffer_advance (p1, -sizeof (ip6_header_t)); + + p0->error = error_node->errors[error0]; + p1->error = error_node->errors[error1]; + vlib_validate_buffer_enqueue_x2 (vm, node, next_index, to_next, + n_left_to_next, pi0, pi1, next0, + next1); } - } - - if (PREDICT_FALSE(p0->flags & VLIB_BUFFER_IS_TRACED)) { - map_trace_t *tr = vlib_add_trace(vm, node, p0, sizeof(*tr)); - tr->map_domain_index = map_domain_index0; - tr->port = port0; - } - - if (PREDICT_FALSE(p1->flags & VLIB_BUFFER_IS_TRACED)) { - map_trace_t *tr = vlib_add_trace(vm, node, p1, sizeof(*tr)); - tr->map_domain_index = map_domain_index1; - tr->port = port1; - } - - if (error0 == MAP_ERROR_DECAP_SEC_CHECK && mm->icmp6_enabled) { - /* Set ICMP parameters */ - vlib_buffer_advance(p0, -sizeof(ip6_header_t)); - icmp6_error_set_vnet_buffer(p0, ICMP6_destination_unreachable, - ICMP6_destination_unreachable_source_address_failed_policy, 0); - next0 = IP6_MAP_NEXT_ICMP; - } else { - next0 = (error0 == MAP_ERROR_NONE) ? next0 : IP6_MAP_NEXT_DROP; - } - - if (error1 == MAP_ERROR_DECAP_SEC_CHECK && mm->icmp6_enabled) { - /* Set ICMP parameters */ - vlib_buffer_advance(p1, -sizeof(ip6_header_t)); - icmp6_error_set_vnet_buffer(p1, ICMP6_destination_unreachable, - ICMP6_destination_unreachable_source_address_failed_policy, 0); - next1 = IP6_MAP_NEXT_ICMP; - } else { - next1 = (error1 == MAP_ERROR_NONE) ? next1 : IP6_MAP_NEXT_DROP; - } - - /* Reset packet */ - if (next0 == IP6_MAP_NEXT_IP6_LOCAL) - vlib_buffer_advance(p0, -sizeof(ip6_header_t)); - if (next1 == IP6_MAP_NEXT_IP6_LOCAL) - vlib_buffer_advance(p1, -sizeof(ip6_header_t)); - - p0->error = error_node->errors[error0]; - p1->error = error_node->errors[error1]; - vlib_validate_buffer_enqueue_x2(vm, node, next_index, to_next, n_left_to_next, pi0, pi1, next0, next1); - } - /* Single loop */ - while (n_left_from > 0 && n_left_to_next > 0) { - u32 pi0; - vlib_buffer_t *p0; - u8 error0 = MAP_ERROR_NONE; - map_domain_t *d0 = 0; - ip4_header_t *ip40; - ip6_header_t *ip60; - i32 port0 = 0; - u32 map_domain_index0 = ~0; - u32 next0 = IP6_MAP_NEXT_IP4_LOOKUP; - - pi0 = to_next[0] = from[0]; - from += 1; - n_left_from -= 1; - to_next +=1; - n_left_to_next -= 1; - - p0 = vlib_get_buffer(vm, pi0); - ip60 = vlib_buffer_get_current(p0); - vlib_buffer_advance(p0, sizeof(ip6_header_t)); - ip40 = vlib_buffer_get_current(p0); - - /* - * Encapsulated IPv4 packet - * - IPv4 fragmented -> Pass to virtual reassembly unless security check disabled - * - Lookup/Rewrite or Fragment node in case of packet > MTU - * Fragmented IPv6 packet - * ICMP IPv6 packet - * - Error -> Pass to ICMPv6/ICMPv4 relay - * - Info -> Pass to IPv6 local - * Anything else -> drop - */ - if (PREDICT_TRUE(ip60->protocol == IP_PROTOCOL_IP_IN_IP && clib_net_to_host_u16(ip60->payload_length) > 20)) { - d0 = ip6_map_get_domain(vnet_buffer(p0)->ip.adj_index[VLIB_TX], (ip4_address_t *)&ip40->src_address.as_u32, - &map_domain_index0, &error0); - } else if (ip60->protocol == IP_PROTOCOL_ICMP6 && - clib_net_to_host_u16(ip60->payload_length) > sizeof(icmp46_header_t)) { - icmp46_header_t *icmp = (void *)(ip60 + 1); - next0 = (icmp->type == ICMP6_echo_request || icmp->type == ICMP6_echo_reply) ? - IP6_MAP_NEXT_IP6_LOCAL : IP6_MAP_NEXT_IP6_ICMP_RELAY; - } else if (ip60->protocol == IP_PROTOCOL_IPV6_FRAGMENTATION && - (((ip6_frag_hdr_t *)(ip60+1))->next_hdr == IP_PROTOCOL_IP_IN_IP)) { - next0 = IP6_MAP_NEXT_IP6_REASS; - } else { - error0 = MAP_ERROR_BAD_PROTOCOL; - } - - if (d0) { - /* MAP inbound security check */ - ip6_map_security_check(d0, ip40, ip60, &next0, &error0); - - if (PREDICT_TRUE(error0 == MAP_ERROR_NONE && - next0 == IP6_MAP_NEXT_IP4_LOOKUP)) { - if (PREDICT_FALSE(d0->mtu && (clib_host_to_net_u16(ip40->length) > d0->mtu))) { - vnet_buffer(p0)->ip_frag.header_offset = 0; - vnet_buffer(p0)->ip_frag.flags = 0; - vnet_buffer(p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP4_LOOKUP; - vnet_buffer(p0)->ip_frag.mtu = d0->mtu; - next0 = IP6_MAP_NEXT_IP4_FRAGMENT; - } else { - next0 = ip6_map_ip4_lookup_bypass(p0, ip40) ? IP6_MAP_NEXT_IP4_REWRITE : next0; - } - vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_RX, cpu_index, map_domain_index0, 1, - clib_net_to_host_u16(ip40->length)); + /* Single loop */ + while (n_left_from > 0 && n_left_to_next > 0) + { + u32 pi0; + vlib_buffer_t *p0; + u8 error0 = MAP_ERROR_NONE; + map_domain_t *d0 = 0; + ip4_header_t *ip40; + ip6_header_t *ip60; + i32 port0 = 0; + u32 map_domain_index0 = ~0; + u32 next0 = IP6_MAP_NEXT_IP4_LOOKUP; + + pi0 = to_next[0] = from[0]; + from += 1; + n_left_from -= 1; + to_next += 1; + n_left_to_next -= 1; + + p0 = vlib_get_buffer (vm, pi0); + ip60 = vlib_buffer_get_current (p0); + vlib_buffer_advance (p0, sizeof (ip6_header_t)); + ip40 = vlib_buffer_get_current (p0); + + /* + * Encapsulated IPv4 packet + * - IPv4 fragmented -> Pass to virtual reassembly unless security check disabled + * - Lookup/Rewrite or Fragment node in case of packet > MTU + * Fragmented IPv6 packet + * ICMP IPv6 packet + * - Error -> Pass to ICMPv6/ICMPv4 relay + * - Info -> Pass to IPv6 local + * Anything else -> drop + */ + if (PREDICT_TRUE + (ip60->protocol == IP_PROTOCOL_IP_IN_IP + && clib_net_to_host_u16 (ip60->payload_length) > 20)) + { + d0 = + ip6_map_get_domain (vnet_buffer (p0)->ip.adj_index[VLIB_TX], + (ip4_address_t *) & ip40->src_address. + as_u32, &map_domain_index0, &error0); + } + else if (ip60->protocol == IP_PROTOCOL_ICMP6 && + clib_net_to_host_u16 (ip60->payload_length) > + sizeof (icmp46_header_t)) + { + icmp46_header_t *icmp = (void *) (ip60 + 1); + next0 = (icmp->type == ICMP6_echo_request + || icmp->type == + ICMP6_echo_reply) ? IP6_MAP_NEXT_IP6_LOCAL : + IP6_MAP_NEXT_IP6_ICMP_RELAY; + } + else if (ip60->protocol == IP_PROTOCOL_IPV6_FRAGMENTATION && + (((ip6_frag_hdr_t *) (ip60 + 1))->next_hdr == + IP_PROTOCOL_IP_IN_IP)) + { + next0 = IP6_MAP_NEXT_IP6_REASS; + } + else + { + error0 = MAP_ERROR_BAD_PROTOCOL; + } + + if (d0) + { + /* MAP inbound security check */ + ip6_map_security_check (d0, ip40, ip60, &next0, &error0); + + if (PREDICT_TRUE (error0 == MAP_ERROR_NONE && + next0 == IP6_MAP_NEXT_IP4_LOOKUP)) + { + if (PREDICT_FALSE + (d0->mtu + && (clib_host_to_net_u16 (ip40->length) > d0->mtu))) + { + vnet_buffer (p0)->ip_frag.header_offset = 0; + vnet_buffer (p0)->ip_frag.flags = 0; + vnet_buffer (p0)->ip_frag.next_index = + IP4_FRAG_NEXT_IP4_LOOKUP; + vnet_buffer (p0)->ip_frag.mtu = d0->mtu; + next0 = IP6_MAP_NEXT_IP4_FRAGMENT; + } + else + { + next0 = + ip6_map_ip4_lookup_bypass (p0, + ip40) ? + IP6_MAP_NEXT_IP4_REWRITE : next0; + } + vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_RX, + cpu_index, + map_domain_index0, 1, + clib_net_to_host_u16 + (ip40->length)); + } + } + + if (PREDICT_FALSE (p0->flags & VLIB_BUFFER_IS_TRACED)) + { + map_trace_t *tr = vlib_add_trace (vm, node, p0, sizeof (*tr)); + tr->map_domain_index = map_domain_index0; + tr->port = (u16) port0; + } + + if (mm->icmp6_enabled && + (error0 == MAP_ERROR_DECAP_SEC_CHECK + || error0 == MAP_ERROR_NO_DOMAIN)) + { + /* Set ICMP parameters */ + vlib_buffer_advance (p0, -sizeof (ip6_header_t)); + icmp6_error_set_vnet_buffer (p0, ICMP6_destination_unreachable, + ICMP6_destination_unreachable_source_address_failed_policy, + 0); + next0 = IP6_MAP_NEXT_ICMP; + } + else + { + next0 = (error0 == MAP_ERROR_NONE) ? next0 : IP6_MAP_NEXT_DROP; + } + + /* Reset packet */ + if (next0 == IP6_MAP_NEXT_IP6_LOCAL) + vlib_buffer_advance (p0, -sizeof (ip6_header_t)); + + p0->error = error_node->errors[error0]; + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, to_next, + n_left_to_next, pi0, next0); } - } - - if (PREDICT_FALSE(p0->flags & VLIB_BUFFER_IS_TRACED)) { - map_trace_t *tr = vlib_add_trace(vm, node, p0, sizeof(*tr)); - tr->map_domain_index = map_domain_index0; - tr->port = (u16)port0; - } - - if (mm->icmp6_enabled && - (error0 == MAP_ERROR_DECAP_SEC_CHECK || error0 == MAP_ERROR_NO_DOMAIN)) { - /* Set ICMP parameters */ - vlib_buffer_advance(p0, -sizeof(ip6_header_t)); - icmp6_error_set_vnet_buffer(p0, ICMP6_destination_unreachable, - ICMP6_destination_unreachable_source_address_failed_policy, 0); - next0 = IP6_MAP_NEXT_ICMP; - } else { - next0 = (error0 == MAP_ERROR_NONE) ? next0 : IP6_MAP_NEXT_DROP; - } - - /* Reset packet */ - if (next0 == IP6_MAP_NEXT_IP6_LOCAL) - vlib_buffer_advance(p0, -sizeof(ip6_header_t)); - - p0->error = error_node->errors[error0]; - vlib_validate_buffer_enqueue_x1(vm, node, next_index, to_next, n_left_to_next, pi0, next0); + vlib_put_next_frame (vm, node, next_index, n_left_to_next); } - vlib_put_next_frame(vm, node, next_index, n_left_to_next); - } return frame->n_vectors; } static_always_inline void -ip6_map_ip6_reass_prepare(vlib_main_t *vm, vlib_node_runtime_t *node, map_ip6_reass_t *r, - u32 **fragments_ready, u32 **fragments_to_drop) +ip6_map_ip6_reass_prepare (vlib_main_t * vm, vlib_node_runtime_t * node, + map_ip6_reass_t * r, u32 ** fragments_ready, + u32 ** fragments_to_drop) { ip4_header_t *ip40; ip6_header_t *ip60; ip6_frag_hdr_t *frag0; vlib_buffer_t *p0; - if(!r->ip4_header.ip_version_and_header_length) + if (!r->ip4_header.ip_version_and_header_length) return; //The IP header is here, we need to check for packets //that can be forwarded int i; - for (i=0; i<MAP_IP6_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++) { - if (r->fragments[i].pi == ~0 || - ((!r->fragments[i].next_data_len) && (r->fragments[i].next_data_offset != (0xffff)))) - continue; - - p0 = vlib_get_buffer(vm, r->fragments[i].pi); - ip60 = vlib_buffer_get_current(p0); - frag0 = (ip6_frag_hdr_t *)(ip60 + 1); - ip40 = (ip4_header_t *)(frag0 + 1); - - if (ip6_frag_hdr_offset(frag0)) { - //Not first fragment, add the IPv4 header - clib_memcpy(ip40, &r->ip4_header, 20); - } + for (i = 0; i < MAP_IP6_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++) + { + if (r->fragments[i].pi == ~0 || + ((!r->fragments[i].next_data_len) + && (r->fragments[i].next_data_offset != (0xffff)))) + continue; + + p0 = vlib_get_buffer (vm, r->fragments[i].pi); + ip60 = vlib_buffer_get_current (p0); + frag0 = (ip6_frag_hdr_t *) (ip60 + 1); + ip40 = (ip4_header_t *) (frag0 + 1); + + if (ip6_frag_hdr_offset (frag0)) + { + //Not first fragment, add the IPv4 header + clib_memcpy (ip40, &r->ip4_header, 20); + } #ifdef MAP_IP6_REASS_COUNT_BYTES - r->forwarded += clib_net_to_host_u16(ip60->payload_length) - sizeof(*frag0); + r->forwarded += + clib_net_to_host_u16 (ip60->payload_length) - sizeof (*frag0); #endif - if (ip6_frag_hdr_more(frag0)) { - //Not last fragment, we copy end of next - clib_memcpy(u8_ptr_add(ip60, p0->current_length), r->fragments[i].next_data, 20); - p0->current_length += 20; - ip60->payload_length = u16_net_add(ip60->payload_length, 20); - } - - if (!ip4_is_fragment(ip40)) { - ip40->fragment_id = frag_id_6to4(frag0->identification); - ip40->flags_and_fragment_offset = clib_host_to_net_u16(ip6_frag_hdr_offset(frag0)); - } else { - ip40->flags_and_fragment_offset = clib_host_to_net_u16(ip4_get_fragment_offset(ip40) + ip6_frag_hdr_offset(frag0)); - } - - if (ip6_frag_hdr_more(frag0)) - ip40->flags_and_fragment_offset |= clib_host_to_net_u16(IP4_HEADER_FLAG_MORE_FRAGMENTS); + if (ip6_frag_hdr_more (frag0)) + { + //Not last fragment, we copy end of next + clib_memcpy (u8_ptr_add (ip60, p0->current_length), + r->fragments[i].next_data, 20); + p0->current_length += 20; + ip60->payload_length = u16_net_add (ip60->payload_length, 20); + } - ip40->length = clib_host_to_net_u16(p0->current_length - sizeof(*ip60) - sizeof(*frag0)); - ip40->checksum = ip4_header_checksum(ip40); + if (!ip4_is_fragment (ip40)) + { + ip40->fragment_id = frag_id_6to4 (frag0->identification); + ip40->flags_and_fragment_offset = + clib_host_to_net_u16 (ip6_frag_hdr_offset (frag0)); + } + else + { + ip40->flags_and_fragment_offset = + clib_host_to_net_u16 (ip4_get_fragment_offset (ip40) + + ip6_frag_hdr_offset (frag0)); + } - if (PREDICT_FALSE(p0->flags & VLIB_BUFFER_IS_TRACED)) { - map_ip6_map_ip6_reass_trace_t *tr = vlib_add_trace(vm, node, p0, sizeof(*tr)); - tr->offset = ip4_get_fragment_offset(ip40); - tr->frag_len = clib_net_to_host_u16(ip40->length) - sizeof(*ip40); - tr->out = 1; - } + if (ip6_frag_hdr_more (frag0)) + ip40->flags_and_fragment_offset |= + clib_host_to_net_u16 (IP4_HEADER_FLAG_MORE_FRAGMENTS); + + ip40->length = + clib_host_to_net_u16 (p0->current_length - sizeof (*ip60) - + sizeof (*frag0)); + ip40->checksum = ip4_header_checksum (ip40); + + if (PREDICT_FALSE (p0->flags & VLIB_BUFFER_IS_TRACED)) + { + map_ip6_map_ip6_reass_trace_t *tr = + vlib_add_trace (vm, node, p0, sizeof (*tr)); + tr->offset = ip4_get_fragment_offset (ip40); + tr->frag_len = clib_net_to_host_u16 (ip40->length) - sizeof (*ip40); + tr->out = 1; + } - vec_add1(*fragments_ready, r->fragments[i].pi); - r->fragments[i].pi = ~0; - r->fragments[i].next_data_len = 0; - r->fragments[i].next_data_offset = 0; - map_main.ip6_reass_buffered_counter--; + vec_add1 (*fragments_ready, r->fragments[i].pi); + r->fragments[i].pi = ~0; + r->fragments[i].next_data_len = 0; + r->fragments[i].next_data_offset = 0; + map_main.ip6_reass_buffered_counter--; - //TODO: Best solution would be that ip6_map handles extension headers - // and ignores atomic fragment. But in the meantime, let's just copy the header. + //TODO: Best solution would be that ip6_map handles extension headers + // and ignores atomic fragment. But in the meantime, let's just copy the header. - u8 protocol = frag0->next_hdr; - memmove(u8_ptr_add(ip40, - sizeof(*ip60)), ip60, sizeof(*ip60)); - ((ip6_header_t *)u8_ptr_add(ip40, - sizeof(*ip60)))->protocol = protocol; - vlib_buffer_advance(p0, sizeof(*frag0)); - } + u8 protocol = frag0->next_hdr; + memmove (u8_ptr_add (ip40, -sizeof (*ip60)), ip60, sizeof (*ip60)); + ((ip6_header_t *) u8_ptr_add (ip40, -sizeof (*ip60)))->protocol = + protocol; + vlib_buffer_advance (p0, sizeof (*frag0)); + } } void -map_ip6_drop_pi(u32 pi) +map_ip6_drop_pi (u32 pi) { - vlib_main_t *vm = vlib_get_main(); - vlib_node_runtime_t *n = vlib_node_get_runtime(vm, ip6_map_ip6_reass_node.index); - vlib_set_next_frame_buffer(vm, n, IP6_MAP_IP6_REASS_NEXT_DROP, pi); + vlib_main_t *vm = vlib_get_main (); + vlib_node_runtime_t *n = + vlib_node_get_runtime (vm, ip6_map_ip6_reass_node.index); + vlib_set_next_frame_buffer (vm, n, IP6_MAP_IP6_REASS_NEXT_DROP, pi); } void -map_ip4_drop_pi(u32 pi) +map_ip4_drop_pi (u32 pi) { - vlib_main_t *vm = vlib_get_main(); - vlib_node_runtime_t *n = vlib_node_get_runtime(vm, ip6_map_ip4_reass_node.index); - vlib_set_next_frame_buffer(vm, n, IP6_MAP_IP4_REASS_NEXT_DROP, pi); + vlib_main_t *vm = vlib_get_main (); + vlib_node_runtime_t *n = + vlib_node_get_runtime (vm, ip6_map_ip4_reass_node.index); + vlib_set_next_frame_buffer (vm, n, IP6_MAP_IP4_REASS_NEXT_DROP, pi); } /* @@ -527,109 +680,139 @@ map_ip4_drop_pi(u32 pi) * have been forwarded. */ static uword -ip6_map_ip6_reass (vlib_main_t *vm, - vlib_node_runtime_t *node, - vlib_frame_t *frame) +ip6_map_ip6_reass (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) { u32 n_left_from, *from, next_index, *to_next, n_left_to_next; - vlib_node_runtime_t *error_node = vlib_node_get_runtime(vm, ip6_map_ip6_reass_node.index); + vlib_node_runtime_t *error_node = + vlib_node_get_runtime (vm, ip6_map_ip6_reass_node.index); u32 *fragments_to_drop = NULL; u32 *fragments_ready = NULL; - from = vlib_frame_vector_args(frame); + from = vlib_frame_vector_args (frame); n_left_from = frame->n_vectors; next_index = node->cached_next_index; - while (n_left_from > 0) { - vlib_get_next_frame(vm, node, next_index, to_next, n_left_to_next); - - /* Single loop */ - while (n_left_from > 0 && n_left_to_next > 0) { - u32 pi0; - vlib_buffer_t *p0; - u8 error0 = MAP_ERROR_NONE; - ip6_header_t *ip60; - ip6_frag_hdr_t *frag0; - u16 offset; - u16 next_offset; - u16 frag_len; - - pi0 = to_next[0] = from[0]; - from += 1; - n_left_from -= 1; - to_next +=1; - n_left_to_next -= 1; - - p0 = vlib_get_buffer(vm, pi0); - ip60 = vlib_buffer_get_current(p0); - frag0 = (ip6_frag_hdr_t *)(ip60 + 1); - offset = clib_host_to_net_u16(frag0->fragment_offset_and_more) & (~7); - frag_len = clib_net_to_host_u16(ip60->payload_length) - sizeof(*frag0); - next_offset = ip6_frag_hdr_more(frag0) ? (offset + frag_len) : (0xffff); - - //FIXME: Support other extension headers, maybe - - if (PREDICT_FALSE(p0->flags & VLIB_BUFFER_IS_TRACED)) { - map_ip6_map_ip6_reass_trace_t *tr = vlib_add_trace(vm, node, p0, sizeof(*tr)); - tr->offset = offset; - tr->frag_len = frag_len; - tr->out = 0; - } - - map_ip6_reass_lock(); - map_ip6_reass_t *r = map_ip6_reass_get(&ip60->src_address, &ip60->dst_address, - frag0->identification, frag0->next_hdr, &fragments_to_drop); - //FIXME: Use better error codes - if (PREDICT_FALSE(!r)) { - // Could not create a caching entry - error0 = MAP_ERROR_FRAGMENT_MEMORY; - } else if (PREDICT_FALSE((frag_len <= 20 && - (ip6_frag_hdr_more(frag0) || (!offset))))) { - //Very small fragment are restricted to the last one and - //can't be the first one - error0 = MAP_ERROR_FRAGMENT_MALFORMED; - } else if (map_ip6_reass_add_fragment(r, pi0, offset, next_offset, (u8 *)(frag0 + 1), frag_len)) { - map_ip6_reass_free(r, &fragments_to_drop); - error0 = MAP_ERROR_FRAGMENT_MEMORY; - } else { + while (n_left_from > 0) + { + vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); + + /* Single loop */ + while (n_left_from > 0 && n_left_to_next > 0) + { + u32 pi0; + vlib_buffer_t *p0; + u8 error0 = MAP_ERROR_NONE; + ip6_header_t *ip60; + ip6_frag_hdr_t *frag0; + u16 offset; + u16 next_offset; + u16 frag_len; + + pi0 = to_next[0] = from[0]; + from += 1; + n_left_from -= 1; + to_next += 1; + n_left_to_next -= 1; + + p0 = vlib_get_buffer (vm, pi0); + ip60 = vlib_buffer_get_current (p0); + frag0 = (ip6_frag_hdr_t *) (ip60 + 1); + offset = + clib_host_to_net_u16 (frag0->fragment_offset_and_more) & (~7); + frag_len = + clib_net_to_host_u16 (ip60->payload_length) - sizeof (*frag0); + next_offset = + ip6_frag_hdr_more (frag0) ? (offset + frag_len) : (0xffff); + + //FIXME: Support other extension headers, maybe + + if (PREDICT_FALSE (p0->flags & VLIB_BUFFER_IS_TRACED)) + { + map_ip6_map_ip6_reass_trace_t *tr = + vlib_add_trace (vm, node, p0, sizeof (*tr)); + tr->offset = offset; + tr->frag_len = frag_len; + tr->out = 0; + } + + map_ip6_reass_lock (); + map_ip6_reass_t *r = + map_ip6_reass_get (&ip60->src_address, &ip60->dst_address, + frag0->identification, frag0->next_hdr, + &fragments_to_drop); + //FIXME: Use better error codes + if (PREDICT_FALSE (!r)) + { + // Could not create a caching entry + error0 = MAP_ERROR_FRAGMENT_MEMORY; + } + else if (PREDICT_FALSE ((frag_len <= 20 && + (ip6_frag_hdr_more (frag0) || (!offset))))) + { + //Very small fragment are restricted to the last one and + //can't be the first one + error0 = MAP_ERROR_FRAGMENT_MALFORMED; + } + else + if (map_ip6_reass_add_fragment + (r, pi0, offset, next_offset, (u8 *) (frag0 + 1), frag_len)) + { + map_ip6_reass_free (r, &fragments_to_drop); + error0 = MAP_ERROR_FRAGMENT_MEMORY; + } + else + { #ifdef MAP_IP6_REASS_COUNT_BYTES - if (!ip6_frag_hdr_more(frag0)) - r->expected_total = offset + frag_len; + if (!ip6_frag_hdr_more (frag0)) + r->expected_total = offset + frag_len; #endif - ip6_map_ip6_reass_prepare(vm, node, r, &fragments_ready, &fragments_to_drop); + ip6_map_ip6_reass_prepare (vm, node, r, &fragments_ready, + &fragments_to_drop); #ifdef MAP_IP6_REASS_COUNT_BYTES - if(r->forwarded >= r->expected_total) - map_ip6_reass_free(r, &fragments_to_drop); + if (r->forwarded >= r->expected_total) + map_ip6_reass_free (r, &fragments_to_drop); #endif - } - map_ip6_reass_unlock(); - - if (error0 == MAP_ERROR_NONE) { - if (frag_len > 20) { - //Dequeue the packet - n_left_to_next++; - to_next--; - } else { - //All data from that packet was copied no need to keep it, but this is not an error - p0->error = error_node->errors[MAP_ERROR_NONE]; - vlib_validate_buffer_enqueue_x1(vm, node, next_index, to_next, n_left_to_next, pi0, IP6_MAP_IP6_REASS_NEXT_DROP); - } - } else { - p0->error = error_node->errors[error0]; - vlib_validate_buffer_enqueue_x1(vm, node, next_index, to_next, n_left_to_next, pi0, IP6_MAP_IP6_REASS_NEXT_DROP); - } + } + map_ip6_reass_unlock (); + + if (error0 == MAP_ERROR_NONE) + { + if (frag_len > 20) + { + //Dequeue the packet + n_left_to_next++; + to_next--; + } + else + { + //All data from that packet was copied no need to keep it, but this is not an error + p0->error = error_node->errors[MAP_ERROR_NONE]; + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, + to_next, n_left_to_next, + pi0, + IP6_MAP_IP6_REASS_NEXT_DROP); + } + } + else + { + p0->error = error_node->errors[error0]; + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, to_next, + n_left_to_next, pi0, + IP6_MAP_IP6_REASS_NEXT_DROP); + } + } + vlib_put_next_frame (vm, node, next_index, n_left_to_next); } - vlib_put_next_frame(vm, node, next_index, n_left_to_next); - } - - map_send_all_to_node(vm, fragments_ready, node, - &error_node->errors[MAP_ERROR_NONE], - IP6_MAP_IP6_REASS_NEXT_IP6_MAP); - map_send_all_to_node(vm, fragments_to_drop, node, - &error_node->errors[MAP_ERROR_FRAGMENT_DROPPED], - IP6_MAP_IP6_REASS_NEXT_DROP); - - vec_free(fragments_to_drop); - vec_free(fragments_ready); + + map_send_all_to_node (vm, fragments_ready, node, + &error_node->errors[MAP_ERROR_NONE], + IP6_MAP_IP6_REASS_NEXT_IP6_MAP); + map_send_all_to_node (vm, fragments_to_drop, node, + &error_node->errors[MAP_ERROR_FRAGMENT_DROPPED], + IP6_MAP_IP6_REASS_NEXT_DROP); + + vec_free (fragments_to_drop); + vec_free (fragments_ready); return frame->n_vectors; } @@ -637,147 +820,195 @@ ip6_map_ip6_reass (vlib_main_t *vm, * ip6_ip4_virt_reass */ static uword -ip6_map_ip4_reass (vlib_main_t *vm, - vlib_node_runtime_t *node, - vlib_frame_t *frame) +ip6_map_ip4_reass (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) { u32 n_left_from, *from, next_index, *to_next, n_left_to_next; - vlib_node_runtime_t *error_node = vlib_node_get_runtime(vm, ip6_map_ip4_reass_node.index); + vlib_node_runtime_t *error_node = + vlib_node_get_runtime (vm, ip6_map_ip4_reass_node.index); map_main_t *mm = &map_main; vlib_combined_counter_main_t *cm = mm->domain_counters; - u32 cpu_index = os_get_cpu_number(); + u32 cpu_index = os_get_cpu_number (); u32 *fragments_to_drop = NULL; u32 *fragments_to_loopback = NULL; - from = vlib_frame_vector_args(frame); + from = vlib_frame_vector_args (frame); n_left_from = frame->n_vectors; next_index = node->cached_next_index; - while (n_left_from > 0) { - vlib_get_next_frame(vm, node, next_index, to_next, n_left_to_next); - - /* Single loop */ - while (n_left_from > 0 && n_left_to_next > 0) { - u32 pi0; - vlib_buffer_t *p0; - u8 error0 = MAP_ERROR_NONE; - map_domain_t *d0; - ip4_header_t *ip40; - ip6_header_t *ip60; - i32 port0 = 0; - u32 map_domain_index0 = ~0; - u32 next0 = IP6_MAP_IP4_REASS_NEXT_IP4_LOOKUP; - u8 cached = 0; - - pi0 = to_next[0] = from[0]; - from += 1; - n_left_from -= 1; - to_next +=1; - n_left_to_next -= 1; - - p0 = vlib_get_buffer(vm, pi0); - ip40 = vlib_buffer_get_current(p0); - ip60 = ((ip6_header_t *)ip40) - 1; - - d0 = ip6_map_get_domain(vnet_buffer(p0)->ip.adj_index[VLIB_TX], (ip4_address_t *)&ip40->src_address.as_u32, - &map_domain_index0, &error0); - - map_ip4_reass_lock(); - //This node only deals with fragmented ip4 - map_ip4_reass_t *r = map_ip4_reass_get(ip40->src_address.as_u32, ip40->dst_address.as_u32, - ip40->fragment_id, ip40->protocol, &fragments_to_drop); - if (PREDICT_FALSE(!r)) { - // Could not create a caching entry - error0 = MAP_ERROR_FRAGMENT_MEMORY; - } else if (PREDICT_TRUE(ip4_get_fragment_offset(ip40))) { - // This is a fragment - if (r->port >= 0) { - // We know the port already - port0 = r->port; - } else if (map_ip4_reass_add_fragment(r, pi0)) { - // Not enough space for caching - error0 = MAP_ERROR_FRAGMENT_MEMORY; - map_ip4_reass_free(r, &fragments_to_drop); - } else { - cached = 1; - } - } else if ((port0 = ip4_get_port(ip40, MAP_SENDER, p0->current_length)) < 0) { - // Could not find port from first fragment. Stop reassembling. - error0 = MAP_ERROR_BAD_PROTOCOL; - port0 = 0; - map_ip4_reass_free(r, &fragments_to_drop); - } else { - // Found port. Remember it and loopback saved fragments - r->port = port0; - map_ip4_reass_get_fragments(r, &fragments_to_loopback); - } + while (n_left_from > 0) + { + vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); + + /* Single loop */ + while (n_left_from > 0 && n_left_to_next > 0) + { + u32 pi0; + vlib_buffer_t *p0; + u8 error0 = MAP_ERROR_NONE; + map_domain_t *d0; + ip4_header_t *ip40; + ip6_header_t *ip60; + i32 port0 = 0; + u32 map_domain_index0 = ~0; + u32 next0 = IP6_MAP_IP4_REASS_NEXT_IP4_LOOKUP; + u8 cached = 0; + + pi0 = to_next[0] = from[0]; + from += 1; + n_left_from -= 1; + to_next += 1; + n_left_to_next -= 1; + + p0 = vlib_get_buffer (vm, pi0); + ip40 = vlib_buffer_get_current (p0); + ip60 = ((ip6_header_t *) ip40) - 1; + + d0 = + ip6_map_get_domain (vnet_buffer (p0)->ip.adj_index[VLIB_TX], + (ip4_address_t *) & ip40->src_address.as_u32, + &map_domain_index0, &error0); + + map_ip4_reass_lock (); + //This node only deals with fragmented ip4 + map_ip4_reass_t *r = + map_ip4_reass_get (ip40->src_address.as_u32, + ip40->dst_address.as_u32, + ip40->fragment_id, ip40->protocol, + &fragments_to_drop); + if (PREDICT_FALSE (!r)) + { + // Could not create a caching entry + error0 = MAP_ERROR_FRAGMENT_MEMORY; + } + else if (PREDICT_TRUE (ip4_get_fragment_offset (ip40))) + { + // This is a fragment + if (r->port >= 0) + { + // We know the port already + port0 = r->port; + } + else if (map_ip4_reass_add_fragment (r, pi0)) + { + // Not enough space for caching + error0 = MAP_ERROR_FRAGMENT_MEMORY; + map_ip4_reass_free (r, &fragments_to_drop); + } + else + { + cached = 1; + } + } + else + if ((port0 = + ip4_get_port (ip40, MAP_SENDER, p0->current_length)) < 0) + { + // Could not find port from first fragment. Stop reassembling. + error0 = MAP_ERROR_BAD_PROTOCOL; + port0 = 0; + map_ip4_reass_free (r, &fragments_to_drop); + } + else + { + // Found port. Remember it and loopback saved fragments + r->port = port0; + map_ip4_reass_get_fragments (r, &fragments_to_loopback); + } #ifdef MAP_IP4_REASS_COUNT_BYTES - if (!cached && r) { - r->forwarded += clib_host_to_net_u16(ip40->length) - 20; - if (!ip4_get_fragment_more(ip40)) - r->expected_total = ip4_get_fragment_offset(ip40) * 8 + clib_host_to_net_u16(ip40->length) - 20; - if(r->forwarded >= r->expected_total) - map_ip4_reass_free(r, &fragments_to_drop); - } + if (!cached && r) + { + r->forwarded += clib_host_to_net_u16 (ip40->length) - 20; + if (!ip4_get_fragment_more (ip40)) + r->expected_total = + ip4_get_fragment_offset (ip40) * 8 + + clib_host_to_net_u16 (ip40->length) - 20; + if (r->forwarded >= r->expected_total) + map_ip4_reass_free (r, &fragments_to_drop); + } #endif - map_ip4_reass_unlock(); - - if(PREDICT_TRUE(error0 == MAP_ERROR_NONE)) - error0 = ip6_map_sec_check(d0, port0, ip40, ip60) ? MAP_ERROR_NONE : MAP_ERROR_DECAP_SEC_CHECK; - - if (PREDICT_FALSE(d0->mtu && (clib_host_to_net_u16(ip40->length) > d0->mtu) && - error0 == MAP_ERROR_NONE && !cached)) { - vnet_buffer(p0)->ip_frag.header_offset = 0; - vnet_buffer(p0)->ip_frag.flags = 0; - vnet_buffer(p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP4_LOOKUP; - vnet_buffer(p0)->ip_frag.mtu = d0->mtu; - next0 = IP6_MAP_IP4_REASS_NEXT_IP4_FRAGMENT; - } - - if (PREDICT_FALSE(p0->flags & VLIB_BUFFER_IS_TRACED)) { - map_ip6_map_ip4_reass_trace_t *tr = vlib_add_trace(vm, node, p0, sizeof(*tr)); - tr->map_domain_index = map_domain_index0; - tr->port = port0; - tr->cached = cached; - } - - if (cached) { - //Dequeue the packet - n_left_to_next++; - to_next--; - } else { - if (error0 == MAP_ERROR_NONE) - vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_RX, cpu_index, map_domain_index0, 1, - clib_net_to_host_u16(ip40->length)); - next0 = (error0 == MAP_ERROR_NONE) ? next0 : IP6_MAP_IP4_REASS_NEXT_DROP; - p0->error = error_node->errors[error0]; - vlib_validate_buffer_enqueue_x1(vm, node, next_index, to_next, n_left_to_next, pi0, next0); - } - - //Loopback when we reach the end of the inpu vector - if(n_left_from == 0 && vec_len(fragments_to_loopback)) { - from = vlib_frame_vector_args(frame); - u32 len = vec_len(fragments_to_loopback); - if(len <= VLIB_FRAME_SIZE) { - clib_memcpy(from, fragments_to_loopback, sizeof(u32)*len); - n_left_from = len; - vec_reset_length(fragments_to_loopback); - } else { - clib_memcpy(from, fragments_to_loopback + (len - VLIB_FRAME_SIZE), sizeof(u32)*VLIB_FRAME_SIZE); - n_left_from = VLIB_FRAME_SIZE; - _vec_len(fragments_to_loopback) = len - VLIB_FRAME_SIZE; - } - } + map_ip4_reass_unlock (); + + if (PREDICT_TRUE (error0 == MAP_ERROR_NONE)) + error0 = + ip6_map_sec_check (d0, port0, ip40, + ip60) ? MAP_ERROR_NONE : + MAP_ERROR_DECAP_SEC_CHECK; + + if (PREDICT_FALSE + (d0->mtu && (clib_host_to_net_u16 (ip40->length) > d0->mtu) + && error0 == MAP_ERROR_NONE && !cached)) + { + vnet_buffer (p0)->ip_frag.header_offset = 0; + vnet_buffer (p0)->ip_frag.flags = 0; + vnet_buffer (p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP4_LOOKUP; + vnet_buffer (p0)->ip_frag.mtu = d0->mtu; + next0 = IP6_MAP_IP4_REASS_NEXT_IP4_FRAGMENT; + } + + if (PREDICT_FALSE (p0->flags & VLIB_BUFFER_IS_TRACED)) + { + map_ip6_map_ip4_reass_trace_t *tr = + vlib_add_trace (vm, node, p0, sizeof (*tr)); + tr->map_domain_index = map_domain_index0; + tr->port = port0; + tr->cached = cached; + } + + if (cached) + { + //Dequeue the packet + n_left_to_next++; + to_next--; + } + else + { + if (error0 == MAP_ERROR_NONE) + vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_RX, + cpu_index, map_domain_index0, + 1, + clib_net_to_host_u16 (ip40-> + length)); + next0 = + (error0 == + MAP_ERROR_NONE) ? next0 : IP6_MAP_IP4_REASS_NEXT_DROP; + p0->error = error_node->errors[error0]; + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, to_next, + n_left_to_next, pi0, next0); + } + + //Loopback when we reach the end of the inpu vector + if (n_left_from == 0 && vec_len (fragments_to_loopback)) + { + from = vlib_frame_vector_args (frame); + u32 len = vec_len (fragments_to_loopback); + if (len <= VLIB_FRAME_SIZE) + { + clib_memcpy (from, fragments_to_loopback, + sizeof (u32) * len); + n_left_from = len; + vec_reset_length (fragments_to_loopback); + } + else + { + clib_memcpy (from, + fragments_to_loopback + (len - + VLIB_FRAME_SIZE), + sizeof (u32) * VLIB_FRAME_SIZE); + n_left_from = VLIB_FRAME_SIZE; + _vec_len (fragments_to_loopback) = len - VLIB_FRAME_SIZE; + } + } + } + vlib_put_next_frame (vm, node, next_index, n_left_to_next); } - vlib_put_next_frame(vm, node, next_index, n_left_to_next); - } - map_send_all_to_node(vm, fragments_to_drop, node, - &error_node->errors[MAP_ERROR_FRAGMENT_DROPPED], - IP6_MAP_IP4_REASS_NEXT_DROP); - - vec_free(fragments_to_drop); - vec_free(fragments_to_loopback); + map_send_all_to_node (vm, fragments_to_drop, node, + &error_node->errors[MAP_ERROR_FRAGMENT_DROPPED], + IP6_MAP_IP4_REASS_NEXT_DROP); + + vec_free (fragments_to_drop); + vec_free (fragments_to_loopback); return frame->n_vectors; } @@ -785,142 +1016,160 @@ ip6_map_ip4_reass (vlib_main_t *vm, * ip6_icmp_relay */ static uword -ip6_map_icmp_relay (vlib_main_t *vm, - vlib_node_runtime_t *node, - vlib_frame_t *frame) +ip6_map_icmp_relay (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) { u32 n_left_from, *from, next_index, *to_next, n_left_to_next; - vlib_node_runtime_t *error_node = vlib_node_get_runtime(vm, ip6_map_icmp_relay_node.index); + vlib_node_runtime_t *error_node = + vlib_node_get_runtime (vm, ip6_map_icmp_relay_node.index); map_main_t *mm = &map_main; - u32 cpu_index = os_get_cpu_number(); + u32 cpu_index = os_get_cpu_number (); u16 *fragment_ids, *fid; - from = vlib_frame_vector_args(frame); + from = vlib_frame_vector_args (frame); n_left_from = frame->n_vectors; next_index = node->cached_next_index; /* Get random fragment IDs for replies. */ - fid = fragment_ids = clib_random_buffer_get_data (&vm->random_buffer, n_left_from * sizeof (fragment_ids[0])); - - while (n_left_from > 0) { - vlib_get_next_frame(vm, node, next_index, to_next, n_left_to_next); - - /* Single loop */ - while (n_left_from > 0 && n_left_to_next > 0) { - u32 pi0; - vlib_buffer_t *p0; - u8 error0 = MAP_ERROR_NONE; - ip6_header_t *ip60; - u32 next0 = IP6_ICMP_RELAY_NEXT_IP4_LOOKUP; - u32 mtu; - - pi0 = to_next[0] = from[0]; - from += 1; - n_left_from -= 1; - to_next +=1; - n_left_to_next -= 1; - - p0 = vlib_get_buffer(vm, pi0); - ip60 = vlib_buffer_get_current(p0); - u16 tlen = clib_net_to_host_u16(ip60->payload_length); - - /* - * In: - * IPv6 header (40) - * ICMPv6 header (8) - * IPv6 header (40) - * Original IPv4 header / packet - * Out: - * New IPv4 header - * New ICMP header - * Original IPv4 header / packet - */ - - /* Need at least ICMP(8) + IPv6(40) + IPv4(20) + L4 header(8) */ - if (tlen < 76) { - error0 = MAP_ERROR_ICMP_RELAY; - goto error; - } - - icmp46_header_t *icmp60 = (icmp46_header_t *)(ip60 + 1); - ip6_header_t *inner_ip60 = (ip6_header_t *)(icmp60 + 2); - - if (inner_ip60->protocol != IP_PROTOCOL_IP_IN_IP) { - error0 = MAP_ERROR_ICMP_RELAY; - goto error; - } - - ip4_header_t *inner_ip40 = (ip4_header_t *)(inner_ip60 + 1); - vlib_buffer_advance(p0, 60); /* sizeof ( IPv6 + ICMP + IPv6 - IPv4 - ICMP ) */ - ip4_header_t *new_ip40 = vlib_buffer_get_current(p0); - icmp46_header_t *new_icmp40 = (icmp46_header_t *)(new_ip40 + 1); - - /* - * Relay according to RFC2473, section 8.3 - */ - switch (icmp60->type) { - case ICMP6_destination_unreachable: - case ICMP6_time_exceeded: - case ICMP6_parameter_problem: - /* Type 3 - destination unreachable, Code 1 - host unreachable */ - new_icmp40->type = ICMP4_destination_unreachable; - new_icmp40->code = ICMP4_destination_unreachable_destination_unreachable_host; - break; - - case ICMP6_packet_too_big: - /* Type 3 - destination unreachable, Code 4 - packet too big */ - /* Potential TODO: Adjust domain tunnel MTU based on the value received here */ - mtu = clib_net_to_host_u32(*((u32 *)(icmp60 + 1))); - - /* Check DF flag */ - if (!(inner_ip40->flags_and_fragment_offset & clib_host_to_net_u16(IP4_HEADER_FLAG_DONT_FRAGMENT))) { - error0 = MAP_ERROR_ICMP_RELAY; - goto error; + fid = fragment_ids = + clib_random_buffer_get_data (&vm->random_buffer, + n_left_from * sizeof (fragment_ids[0])); + + while (n_left_from > 0) + { + vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); + + /* Single loop */ + while (n_left_from > 0 && n_left_to_next > 0) + { + u32 pi0; + vlib_buffer_t *p0; + u8 error0 = MAP_ERROR_NONE; + ip6_header_t *ip60; + u32 next0 = IP6_ICMP_RELAY_NEXT_IP4_LOOKUP; + u32 mtu; + + pi0 = to_next[0] = from[0]; + from += 1; + n_left_from -= 1; + to_next += 1; + n_left_to_next -= 1; + + p0 = vlib_get_buffer (vm, pi0); + ip60 = vlib_buffer_get_current (p0); + u16 tlen = clib_net_to_host_u16 (ip60->payload_length); + + /* + * In: + * IPv6 header (40) + * ICMPv6 header (8) + * IPv6 header (40) + * Original IPv4 header / packet + * Out: + * New IPv4 header + * New ICMP header + * Original IPv4 header / packet + */ + + /* Need at least ICMP(8) + IPv6(40) + IPv4(20) + L4 header(8) */ + if (tlen < 76) + { + error0 = MAP_ERROR_ICMP_RELAY; + goto error; + } + + icmp46_header_t *icmp60 = (icmp46_header_t *) (ip60 + 1); + ip6_header_t *inner_ip60 = (ip6_header_t *) (icmp60 + 2); + + if (inner_ip60->protocol != IP_PROTOCOL_IP_IN_IP) + { + error0 = MAP_ERROR_ICMP_RELAY; + goto error; + } + + ip4_header_t *inner_ip40 = (ip4_header_t *) (inner_ip60 + 1); + vlib_buffer_advance (p0, 60); /* sizeof ( IPv6 + ICMP + IPv6 - IPv4 - ICMP ) */ + ip4_header_t *new_ip40 = vlib_buffer_get_current (p0); + icmp46_header_t *new_icmp40 = (icmp46_header_t *) (new_ip40 + 1); + + /* + * Relay according to RFC2473, section 8.3 + */ + switch (icmp60->type) + { + case ICMP6_destination_unreachable: + case ICMP6_time_exceeded: + case ICMP6_parameter_problem: + /* Type 3 - destination unreachable, Code 1 - host unreachable */ + new_icmp40->type = ICMP4_destination_unreachable; + new_icmp40->code = + ICMP4_destination_unreachable_destination_unreachable_host; + break; + + case ICMP6_packet_too_big: + /* Type 3 - destination unreachable, Code 4 - packet too big */ + /* Potential TODO: Adjust domain tunnel MTU based on the value received here */ + mtu = clib_net_to_host_u32 (*((u32 *) (icmp60 + 1))); + + /* Check DF flag */ + if (! + (inner_ip40-> + flags_and_fragment_offset & + clib_host_to_net_u16 (IP4_HEADER_FLAG_DONT_FRAGMENT))) + { + error0 = MAP_ERROR_ICMP_RELAY; + goto error; + } + + new_icmp40->type = ICMP4_destination_unreachable; + new_icmp40->code = + ICMP4_destination_unreachable_fragmentation_needed_and_dont_fragment_set; + *((u32 *) (new_icmp40 + 1)) = + clib_host_to_net_u32 (mtu < 1280 ? 1280 : mtu); + break; + + default: + error0 = MAP_ERROR_ICMP_RELAY; + break; + } + + /* + * Ensure the total ICMP packet is no longer than 576 bytes (RFC1812) + */ + new_ip40->ip_version_and_header_length = 0x45; + new_ip40->tos = 0; + u16 nlen = (tlen - 20) > 576 ? 576 : tlen - 20; + new_ip40->length = clib_host_to_net_u16 (nlen); + new_ip40->fragment_id = fid[0]; + fid++; + new_ip40->ttl = 64; + new_ip40->protocol = IP_PROTOCOL_ICMP; + new_ip40->src_address = mm->icmp4_src_address; + new_ip40->dst_address = inner_ip40->src_address; + new_ip40->checksum = ip4_header_checksum (new_ip40); + + new_icmp40->checksum = 0; + ip_csum_t sum = ip_incremental_checksum (0, new_icmp40, nlen - 20); + new_icmp40->checksum = ~ip_csum_fold (sum); + + vlib_increment_simple_counter (&mm->icmp_relayed, cpu_index, 0, 1); + + error: + if (PREDICT_FALSE (p0->flags & VLIB_BUFFER_IS_TRACED)) + { + map_trace_t *tr = vlib_add_trace (vm, node, p0, sizeof (*tr)); + tr->map_domain_index = 0; + tr->port = 0; + } + + next0 = + (error0 == MAP_ERROR_NONE) ? next0 : IP6_ICMP_RELAY_NEXT_DROP; + p0->error = error_node->errors[error0]; + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, to_next, + n_left_to_next, pi0, next0); } - - new_icmp40->type = ICMP4_destination_unreachable; - new_icmp40->code = ICMP4_destination_unreachable_fragmentation_needed_and_dont_fragment_set; - *((u32 *)(new_icmp40 + 1)) = clib_host_to_net_u32(mtu < 1280 ? 1280 : mtu); - break; - - default: - error0 = MAP_ERROR_ICMP_RELAY; - break; - } - - /* - * Ensure the total ICMP packet is no longer than 576 bytes (RFC1812) - */ - new_ip40->ip_version_and_header_length = 0x45; - new_ip40->tos = 0; - u16 nlen = (tlen - 20) > 576 ? 576 : tlen - 20; - new_ip40->length = clib_host_to_net_u16(nlen); - new_ip40->fragment_id = fid[0]; fid++; - new_ip40->ttl = 64; - new_ip40->protocol = IP_PROTOCOL_ICMP; - new_ip40->src_address = mm->icmp4_src_address; - new_ip40->dst_address = inner_ip40->src_address; - new_ip40->checksum = ip4_header_checksum(new_ip40); - - new_icmp40->checksum = 0; - ip_csum_t sum = ip_incremental_checksum(0, new_icmp40, nlen - 20); - new_icmp40->checksum = ~ip_csum_fold(sum); - - vlib_increment_simple_counter(&mm->icmp_relayed, cpu_index, 0, 1); - - error: - if (PREDICT_FALSE(p0->flags & VLIB_BUFFER_IS_TRACED)) { - map_trace_t *tr = vlib_add_trace(vm, node, p0, sizeof(*tr)); - tr->map_domain_index = 0; - tr->port = 0; - } - - next0 = (error0 == MAP_ERROR_NONE) ? next0 : IP6_ICMP_RELAY_NEXT_DROP; - p0->error = error_node->errors[error0]; - vlib_validate_buffer_enqueue_x1(vm, node, next_index, to_next, n_left_to_next, pi0, next0); + vlib_put_next_frame (vm, node, next_index, n_left_to_next); } - vlib_put_next_frame(vm, node, next_index, n_left_to_next); - } return frame->n_vectors; @@ -932,6 +1181,7 @@ static char *map_error_strings[] = { #undef _ }; +/* *INDENT-OFF* */ VLIB_REGISTER_NODE(ip6_map_node) = { .function = ip6_map, .name = "ip6-map", @@ -957,7 +1207,9 @@ VLIB_REGISTER_NODE(ip6_map_node) = { [IP6_MAP_NEXT_ICMP] = "ip6-icmp-error", }, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_REGISTER_NODE(ip6_map_ip6_reass_node) = { .function = ip6_map_ip6_reass, .name = "ip6-map-ip6-reass", @@ -972,7 +1224,9 @@ VLIB_REGISTER_NODE(ip6_map_ip6_reass_node) = { [IP6_MAP_IP6_REASS_NEXT_DROP] = "error-drop", }, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_REGISTER_NODE(ip6_map_ip4_reass_node) = { .function = ip6_map_ip4_reass, .name = "ip6-map-ip4-reass", @@ -988,7 +1242,9 @@ VLIB_REGISTER_NODE(ip6_map_ip4_reass_node) = { [IP6_MAP_IP4_REASS_NEXT_DROP] = "error-drop", }, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_REGISTER_NODE(ip6_map_icmp_relay_node, static) = { .function = ip6_map_icmp_relay, .name = "ip6-map-icmp-relay", @@ -1003,3 +1259,12 @@ VLIB_REGISTER_NODE(ip6_map_icmp_relay_node, static) = { [IP6_ICMP_RELAY_NEXT_DROP] = "error-drop", }, }; +/* *INDENT-ON* */ + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/vnet/vnet/map/ip6_map_t.c b/vnet/vnet/map/ip6_map_t.c index 39e8e0abb78..eb3996c2467 100644 --- a/vnet/vnet/map/ip6_map_t.c +++ b/vnet/vnet/map/ip6_map_t.c @@ -18,7 +18,8 @@ #define IP6_MAP_T_DUAL_LOOP -typedef enum { +typedef enum +{ IP6_MAPT_NEXT_MAPT_TCP_UDP, IP6_MAPT_NEXT_MAPT_ICMP, IP6_MAPT_NEXT_MAPT_FRAGMENTED, @@ -26,21 +27,24 @@ typedef enum { IP6_MAPT_N_NEXT } ip6_mapt_next_t; -typedef enum { +typedef enum +{ IP6_MAPT_ICMP_NEXT_IP4_LOOKUP, IP6_MAPT_ICMP_NEXT_IP4_FRAG, IP6_MAPT_ICMP_NEXT_DROP, IP6_MAPT_ICMP_N_NEXT } ip6_mapt_icmp_next_t; -typedef enum { +typedef enum +{ IP6_MAPT_TCP_UDP_NEXT_IP4_LOOKUP, IP6_MAPT_TCP_UDP_NEXT_IP4_FRAG, IP6_MAPT_TCP_UDP_NEXT_DROP, IP6_MAPT_TCP_UDP_N_NEXT } ip6_mapt_tcp_udp_next_t; -typedef enum { +typedef enum +{ IP6_MAPT_FRAGMENTED_NEXT_IP4_LOOKUP, IP6_MAPT_FRAGMENTED_NEXT_IP4_FRAG, IP6_MAPT_FRAGMENTED_NEXT_DROP, @@ -48,131 +52,153 @@ typedef enum { } ip6_mapt_fragmented_next_t; static_always_inline int -ip6_map_fragment_cache (ip6_header_t *ip6, ip6_frag_hdr_t *frag, map_domain_t *d, u16 port) +ip6_map_fragment_cache (ip6_header_t * ip6, ip6_frag_hdr_t * frag, + map_domain_t * d, u16 port) { u32 *ignore = NULL; - map_ip4_reass_lock(); - map_ip4_reass_t *r = map_ip4_reass_get(map_get_ip4(&ip6->src_address), ip6_map_t_embedded_address(d, &ip6->dst_address), - frag_id_6to4(frag->identification), - (ip6->protocol == IP_PROTOCOL_ICMP6) ? IP_PROTOCOL_ICMP : ip6->protocol, - &ignore); + map_ip4_reass_lock (); + map_ip4_reass_t *r = map_ip4_reass_get (map_get_ip4 (&ip6->src_address), + ip6_map_t_embedded_address (d, + &ip6-> + dst_address), + frag_id_6to4 (frag->identification), + (ip6->protocol == + IP_PROTOCOL_ICMP6) ? + IP_PROTOCOL_ICMP : ip6->protocol, + &ignore); if (r) r->port = port; - map_ip4_reass_unlock(); + map_ip4_reass_unlock (); return !r; } /* Returns the associated port or -1 */ static_always_inline i32 -ip6_map_fragment_get(ip6_header_t *ip6, ip6_frag_hdr_t *frag, map_domain_t *d) +ip6_map_fragment_get (ip6_header_t * ip6, ip6_frag_hdr_t * frag, + map_domain_t * d) { u32 *ignore = NULL; - map_ip4_reass_lock(); - map_ip4_reass_t *r = map_ip4_reass_get(map_get_ip4(&ip6->src_address), ip6_map_t_embedded_address(d, &ip6->dst_address), - frag_id_6to4(frag->identification), - (ip6->protocol == IP_PROTOCOL_ICMP6) ? IP_PROTOCOL_ICMP : ip6->protocol, - &ignore); - i32 ret = r?r->port:-1; - map_ip4_reass_unlock(); + map_ip4_reass_lock (); + map_ip4_reass_t *r = map_ip4_reass_get (map_get_ip4 (&ip6->src_address), + ip6_map_t_embedded_address (d, + &ip6-> + dst_address), + frag_id_6to4 (frag->identification), + (ip6->protocol == + IP_PROTOCOL_ICMP6) ? + IP_PROTOCOL_ICMP : ip6->protocol, + &ignore); + i32 ret = r ? r->port : -1; + map_ip4_reass_unlock (); return ret; } static_always_inline u8 -ip6_translate_tos(const ip6_header_t *ip6) +ip6_translate_tos (const ip6_header_t * ip6) { #ifdef IP6_MAP_T_OVERRIDE_TOS return IP6_MAP_T_OVERRIDE_TOS; #else - return (clib_net_to_host_u32(ip6->ip_version_traffic_class_and_flow_label) & 0x0ff00000) >> 20; + return (clib_net_to_host_u32 (ip6->ip_version_traffic_class_and_flow_label) + & 0x0ff00000) >> 20; #endif } //TODO: Find right place in memory for that +/* *INDENT-OFF* */ static u8 icmp6_to_icmp_updater_pointer_table[] = - { 0, 1,~0,~0, - 2, 2, 9, 8, - 12,12,12,12, - 12,12,12,12, - 12,12,12,12, - 12,12,12,12, - 24,24,24,24, - 24,24,24,24, - 24,24,24,24, - 24,24,24,24 - }; + { 0, 1, ~0, ~0, + 2, 2, 9, 8, + 12, 12, 12, 12, + 12, 12, 12, 12, + 12, 12, 12, 12, + 12, 12, 12, 12, + 24, 24, 24, 24, + 24, 24, 24, 24, + 24, 24, 24, 24, + 24, 24, 24, 24 + }; +/* *INDENT-ON* */ static_always_inline int -ip6_icmp_to_icmp6_in_place (icmp46_header_t *icmp, u32 icmp_len, - i32 *sender_port, ip6_header_t **inner_ip6) +ip6_icmp_to_icmp6_in_place (icmp46_header_t * icmp, u32 icmp_len, + i32 * sender_port, ip6_header_t ** inner_ip6) { *inner_ip6 = NULL; - switch (icmp->type) { + switch (icmp->type) + { case ICMP6_echo_request: - *sender_port = ((u16 *)icmp)[2]; + *sender_port = ((u16 *) icmp)[2]; icmp->type = ICMP4_echo_request; break; case ICMP6_echo_reply: - *sender_port = ((u16 *)icmp)[2]; + *sender_port = ((u16 *) icmp)[2]; icmp->type = ICMP4_echo_reply; break; case ICMP6_destination_unreachable: - *inner_ip6 = (ip6_header_t *) u8_ptr_add(icmp, 8); - *sender_port = ip6_get_port(*inner_ip6, MAP_RECEIVER, icmp_len); - - switch (icmp->code) { - case ICMP6_destination_unreachable_no_route_to_destination: //0 - case ICMP6_destination_unreachable_beyond_scope_of_source_address: //2 - case ICMP6_destination_unreachable_address_unreachable: //3 - icmp->type = ICMP4_destination_unreachable; - icmp->code = ICMP4_destination_unreachable_destination_unreachable_host; - break; - case ICMP6_destination_unreachable_destination_administratively_prohibited: //1 - icmp->type = ICMP4_destination_unreachable; - icmp->code = ICMP4_destination_unreachable_communication_administratively_prohibited; - break; - case ICMP6_destination_unreachable_port_unreachable: - icmp->type = ICMP4_destination_unreachable; - icmp->code = ICMP4_destination_unreachable_port_unreachable; - break; - default: - return -1; - } - break; + *inner_ip6 = (ip6_header_t *) u8_ptr_add (icmp, 8); + *sender_port = ip6_get_port (*inner_ip6, MAP_RECEIVER, icmp_len); + + switch (icmp->code) + { + case ICMP6_destination_unreachable_no_route_to_destination: //0 + case ICMP6_destination_unreachable_beyond_scope_of_source_address: //2 + case ICMP6_destination_unreachable_address_unreachable: //3 + icmp->type = ICMP4_destination_unreachable; + icmp->code = + ICMP4_destination_unreachable_destination_unreachable_host; + break; + case ICMP6_destination_unreachable_destination_administratively_prohibited: //1 + icmp->type = + ICMP4_destination_unreachable; + icmp->code = + ICMP4_destination_unreachable_communication_administratively_prohibited; + break; + case ICMP6_destination_unreachable_port_unreachable: + icmp->type = ICMP4_destination_unreachable; + icmp->code = ICMP4_destination_unreachable_port_unreachable; + break; + default: + return -1; + } + break; case ICMP6_packet_too_big: - *inner_ip6 = (ip6_header_t *) u8_ptr_add(icmp, 8); - *sender_port = ip6_get_port(*inner_ip6, MAP_RECEIVER, icmp_len); + *inner_ip6 = (ip6_header_t *) u8_ptr_add (icmp, 8); + *sender_port = ip6_get_port (*inner_ip6, MAP_RECEIVER, icmp_len); icmp->type = ICMP4_destination_unreachable; icmp->code = 4; { - u32 advertised_mtu = clib_net_to_host_u32(*((u32 *)(icmp + 1))); + u32 advertised_mtu = clib_net_to_host_u32 (*((u32 *) (icmp + 1))); advertised_mtu -= 20; - //FIXME: = minimum(advertised MTU-20, MTU_of_IPv4_nexthop, (MTU_of_IPv6_nexthop)-20) - ((u16 *)(icmp))[3] = clib_host_to_net_u16(advertised_mtu); + //FIXME: = minimum(advertised MTU-20, MTU_of_IPv4_nexthop, (MTU_of_IPv6_nexthop)-20) + ((u16 *) (icmp))[3] = clib_host_to_net_u16 (advertised_mtu); } break; case ICMP6_time_exceeded: - *inner_ip6 = (ip6_header_t *) u8_ptr_add(icmp, 8); - *sender_port = ip6_get_port(*inner_ip6, MAP_RECEIVER, icmp_len); + *inner_ip6 = (ip6_header_t *) u8_ptr_add (icmp, 8); + *sender_port = ip6_get_port (*inner_ip6, MAP_RECEIVER, icmp_len); icmp->type = ICMP4_time_exceeded; break; case ICMP6_parameter_problem: - *inner_ip6 = (ip6_header_t *) u8_ptr_add(icmp, 8); - *sender_port = ip6_get_port(*inner_ip6, MAP_RECEIVER, icmp_len); + *inner_ip6 = (ip6_header_t *) u8_ptr_add (icmp, 8); + *sender_port = ip6_get_port (*inner_ip6, MAP_RECEIVER, icmp_len); - switch (icmp->code) { + switch (icmp->code) + { case ICMP6_parameter_problem_erroneous_header_field: icmp->type = ICMP4_parameter_problem; icmp->code = ICMP4_parameter_problem_pointer_indicates_error; - u32 pointer = clib_net_to_host_u32(*((u32*)(icmp + 1))); + u32 pointer = clib_net_to_host_u32 (*((u32 *) (icmp + 1))); if (pointer >= 40) return -1; - ((u8*)(icmp + 1))[0] = icmp6_to_icmp_updater_pointer_table[pointer]; + ((u8 *) (icmp + 1))[0] = + icmp6_to_icmp_updater_pointer_table[pointer]; break; case ICMP6_parameter_problem_unrecognized_next_header: icmp->type = ICMP4_destination_unreachable; @@ -181,17 +207,17 @@ ip6_icmp_to_icmp6_in_place (icmp46_header_t *icmp, u32 icmp_len, case ICMP6_parameter_problem_unrecognized_option: default: return -1; - } + } break; default: return -1; break; - } + } return 0; } static_always_inline void -_ip6_map_t_icmp (map_domain_t *d, vlib_buffer_t *p, u8 *error) +_ip6_map_t_icmp (map_domain_t * d, vlib_buffer_t * p, u8 * error) { ip6_header_t *ip6, *inner_ip6; ip4_header_t *ip4, *inner_ip4; @@ -201,872 +227,1207 @@ _ip6_map_t_icmp (map_domain_t *d, vlib_buffer_t *p, u8 *error) ip_csum_t csum; u32 ip4_sadr, inner_ip4_dadr; - ip6 = vlib_buffer_get_current(p); - ip6_pay_len = clib_net_to_host_u16(ip6->payload_length); - icmp = (icmp46_header_t *)(ip6 + 1); - ASSERT(ip6_pay_len + sizeof(*ip6) <= p->current_length); - - if (ip6->protocol != IP_PROTOCOL_ICMP6) { - //No extensions headers allowed here - //TODO: SR header - *error = MAP_ERROR_MALFORMED; - return; - } - - //There are no fragmented ICMP messages, so no extension header for now - - if (ip6_icmp_to_icmp6_in_place(icmp, ip6_pay_len, &sender_port, &inner_ip6)) { - //TODO: In case of 1:1 mapping it is not necessary to have the sender port - *error = MAP_ERROR_ICMP; - return; - } - - if (sender_port < 0) { - // In case of 1:1 mapping, we don't care about the port - if(d->ea_bits_len == 0 && d->rules) { - sender_port = 0; - } else { - *error = MAP_ERROR_ICMP; - return; - } - } + ip6 = vlib_buffer_get_current (p); + ip6_pay_len = clib_net_to_host_u16 (ip6->payload_length); + icmp = (icmp46_header_t *) (ip6 + 1); + ASSERT (ip6_pay_len + sizeof (*ip6) <= p->current_length); - //Security check - //Note that this prevents an intermediate IPv6 router from answering the request - ip4_sadr = map_get_ip4(&ip6->src_address); - if (ip6->src_address.as_u64[0] != map_get_pfx_net(d, ip4_sadr, sender_port) || - ip6->src_address.as_u64[1] != map_get_sfx_net(d, ip4_sadr, sender_port)) { - *error = MAP_ERROR_SEC_CHECK; - return; - } - - if (inner_ip6) { - u16 *inner_L4_checksum, inner_l4_offset, inner_frag_offset, inner_frag_id; - u8 *inner_l4, inner_protocol; - - //We have two headers to translate - // FROM - // [ IPv6 ]<- ext ->[IC][ IPv6 ]<- ext ->[L4 header ... - // Handled cases: - // [ IPv6 ][IC][ IPv6 ][L4 header ... - // [ IPv6 ][IC][ IPv6 ][Fr][L4 header ... - // TO - // [ IPv4][IC][ IPv4][L4 header ... - - //TODO: This was already done deep in ip6_icmp_to_icmp6_in_place - //We shouldn't have to do it again - if (ip6_parse(inner_ip6, ip6_pay_len - 8, - &inner_protocol, &inner_l4_offset, &inner_frag_offset)) { + if (ip6->protocol != IP_PROTOCOL_ICMP6) + { + //No extensions headers allowed here + //TODO: SR header *error = MAP_ERROR_MALFORMED; return; } - inner_l4 = u8_ptr_add(inner_ip6, inner_l4_offset); - inner_ip4 = (ip4_header_t *) u8_ptr_add(inner_l4, - sizeof(*inner_ip4)); - if (inner_frag_offset) { - ip6_frag_hdr_t *inner_frag = (ip6_frag_hdr_t *) u8_ptr_add(inner_ip6, inner_frag_offset); - inner_frag_id = frag_id_6to4(inner_frag->identification); - } else { - inner_frag_id = 0; - } + //There are no fragmented ICMP messages, so no extension header for now - //Do the translation of the inner packet - if (inner_protocol == IP_PROTOCOL_TCP) { - inner_L4_checksum = (u16 *) u8_ptr_add(inner_l4, 16); - } else if (inner_protocol == IP_PROTOCOL_UDP) { - inner_L4_checksum = (u16 *) u8_ptr_add(inner_l4, 6); - } else if (inner_protocol == IP_PROTOCOL_ICMP6) { - icmp46_header_t *inner_icmp = (icmp46_header_t *) inner_l4; - csum = inner_icmp->checksum; - csum = ip_csum_sub_even(csum, *((u16 *)inner_icmp)); - //It cannot be of a different type as ip6_icmp_to_icmp6_in_place succeeded - inner_icmp->type = (inner_icmp->type == ICMP6_echo_request) ? - ICMP4_echo_request : ICMP4_echo_reply; - csum = ip_csum_add_even(csum, *((u16 *)inner_icmp)); - inner_icmp->checksum = ip_csum_fold(csum); - inner_protocol = IP_PROTOCOL_ICMP; //Will be copied to ip6 later - inner_L4_checksum = &inner_icmp->checksum; - } else { - *error = MAP_ERROR_BAD_PROTOCOL; + if (ip6_icmp_to_icmp6_in_place + (icmp, ip6_pay_len, &sender_port, &inner_ip6)) + { + //TODO: In case of 1:1 mapping it is not necessary to have the sender port + *error = MAP_ERROR_ICMP; return; } - csum = *inner_L4_checksum; - csum = ip_csum_sub_even(csum, inner_ip6->src_address.as_u64[0]); - csum = ip_csum_sub_even(csum, inner_ip6->src_address.as_u64[1]); - csum = ip_csum_sub_even(csum, inner_ip6->dst_address.as_u64[0]); - csum = ip_csum_sub_even(csum, inner_ip6->dst_address.as_u64[1]); - - //Sanity check of the outer destination address - if (ip6->dst_address.as_u64[0] != inner_ip6->src_address.as_u64[0] && - ip6->dst_address.as_u64[1] != inner_ip6->src_address.as_u64[1]) { - *error = MAP_ERROR_SEC_CHECK; - return; + if (sender_port < 0) + { + // In case of 1:1 mapping, we don't care about the port + if (d->ea_bits_len == 0 && d->rules) + { + sender_port = 0; + } + else + { + *error = MAP_ERROR_ICMP; + return; + } } - //Security check of inner packet - inner_ip4_dadr = map_get_ip4(&inner_ip6->dst_address); - if (inner_ip6->dst_address.as_u64[0] != map_get_pfx_net(d, inner_ip4_dadr, sender_port) || - inner_ip6->dst_address.as_u64[1] != map_get_sfx_net(d, inner_ip4_dadr, sender_port)) { + //Security check + //Note that this prevents an intermediate IPv6 router from answering the request + ip4_sadr = map_get_ip4 (&ip6->src_address); + if (ip6->src_address.as_u64[0] != map_get_pfx_net (d, ip4_sadr, sender_port) + || ip6->src_address.as_u64[1] != map_get_sfx_net (d, ip4_sadr, + sender_port)) + { *error = MAP_ERROR_SEC_CHECK; return; } - inner_ip4->dst_address.as_u32 = inner_ip4_dadr; - inner_ip4->src_address.as_u32 = ip6_map_t_embedded_address(d, &inner_ip6->src_address); - inner_ip4->ip_version_and_header_length = IP4_VERSION_AND_HEADER_LENGTH_NO_OPTIONS; - inner_ip4->tos = ip6_translate_tos(inner_ip6); - inner_ip4->length = u16_net_add(inner_ip6->payload_length, sizeof(*ip4) + sizeof(*ip6) - - inner_l4_offset); - inner_ip4->fragment_id = inner_frag_id; - inner_ip4->flags_and_fragment_offset = clib_host_to_net_u16(IP4_HEADER_FLAG_MORE_FRAGMENTS); - inner_ip4->ttl = inner_ip6->hop_limit; - inner_ip4->protocol = inner_protocol; - inner_ip4->checksum = ip4_header_checksum(inner_ip4); - - if (inner_ip4->protocol == IP_PROTOCOL_ICMP) { - //Remove remainings of the pseudo-header in the csum - csum = ip_csum_sub_even(csum, clib_host_to_net_u16(IP_PROTOCOL_ICMP6)); - csum = ip_csum_sub_even(csum, inner_ip4->length - sizeof(*inner_ip4)); - } else { - //Update to new pseudo-header - csum = ip_csum_add_even(csum, inner_ip4->src_address.as_u32); - csum = ip_csum_add_even(csum, inner_ip4->dst_address.as_u32); + if (inner_ip6) + { + u16 *inner_L4_checksum, inner_l4_offset, inner_frag_offset, + inner_frag_id; + u8 *inner_l4, inner_protocol; + + //We have two headers to translate + // FROM + // [ IPv6 ]<- ext ->[IC][ IPv6 ]<- ext ->[L4 header ... + // Handled cases: + // [ IPv6 ][IC][ IPv6 ][L4 header ... + // [ IPv6 ][IC][ IPv6 ][Fr][L4 header ... + // TO + // [ IPv4][IC][ IPv4][L4 header ... + + //TODO: This was already done deep in ip6_icmp_to_icmp6_in_place + //We shouldn't have to do it again + if (ip6_parse (inner_ip6, ip6_pay_len - 8, + &inner_protocol, &inner_l4_offset, &inner_frag_offset)) + { + *error = MAP_ERROR_MALFORMED; + return; + } + + inner_l4 = u8_ptr_add (inner_ip6, inner_l4_offset); + inner_ip4 = + (ip4_header_t *) u8_ptr_add (inner_l4, -sizeof (*inner_ip4)); + if (inner_frag_offset) + { + ip6_frag_hdr_t *inner_frag = + (ip6_frag_hdr_t *) u8_ptr_add (inner_ip6, inner_frag_offset); + inner_frag_id = frag_id_6to4 (inner_frag->identification); + } + else + { + inner_frag_id = 0; + } + + //Do the translation of the inner packet + if (inner_protocol == IP_PROTOCOL_TCP) + { + inner_L4_checksum = (u16 *) u8_ptr_add (inner_l4, 16); + } + else if (inner_protocol == IP_PROTOCOL_UDP) + { + inner_L4_checksum = (u16 *) u8_ptr_add (inner_l4, 6); + } + else if (inner_protocol == IP_PROTOCOL_ICMP6) + { + icmp46_header_t *inner_icmp = (icmp46_header_t *) inner_l4; + csum = inner_icmp->checksum; + csum = ip_csum_sub_even (csum, *((u16 *) inner_icmp)); + //It cannot be of a different type as ip6_icmp_to_icmp6_in_place succeeded + inner_icmp->type = (inner_icmp->type == ICMP6_echo_request) ? + ICMP4_echo_request : ICMP4_echo_reply; + csum = ip_csum_add_even (csum, *((u16 *) inner_icmp)); + inner_icmp->checksum = ip_csum_fold (csum); + inner_protocol = IP_PROTOCOL_ICMP; //Will be copied to ip6 later + inner_L4_checksum = &inner_icmp->checksum; + } + else + { + *error = MAP_ERROR_BAD_PROTOCOL; + return; + } + + csum = *inner_L4_checksum; + csum = ip_csum_sub_even (csum, inner_ip6->src_address.as_u64[0]); + csum = ip_csum_sub_even (csum, inner_ip6->src_address.as_u64[1]); + csum = ip_csum_sub_even (csum, inner_ip6->dst_address.as_u64[0]); + csum = ip_csum_sub_even (csum, inner_ip6->dst_address.as_u64[1]); + + //Sanity check of the outer destination address + if (ip6->dst_address.as_u64[0] != inner_ip6->src_address.as_u64[0] && + ip6->dst_address.as_u64[1] != inner_ip6->src_address.as_u64[1]) + { + *error = MAP_ERROR_SEC_CHECK; + return; + } + + //Security check of inner packet + inner_ip4_dadr = map_get_ip4 (&inner_ip6->dst_address); + if (inner_ip6->dst_address.as_u64[0] != + map_get_pfx_net (d, inner_ip4_dadr, sender_port) + || inner_ip6->dst_address.as_u64[1] != map_get_sfx_net (d, + inner_ip4_dadr, + sender_port)) + { + *error = MAP_ERROR_SEC_CHECK; + return; + } + + inner_ip4->dst_address.as_u32 = inner_ip4_dadr; + inner_ip4->src_address.as_u32 = + ip6_map_t_embedded_address (d, &inner_ip6->src_address); + inner_ip4->ip_version_and_header_length = + IP4_VERSION_AND_HEADER_LENGTH_NO_OPTIONS; + inner_ip4->tos = ip6_translate_tos (inner_ip6); + inner_ip4->length = + u16_net_add (inner_ip6->payload_length, + sizeof (*ip4) + sizeof (*ip6) - inner_l4_offset); + inner_ip4->fragment_id = inner_frag_id; + inner_ip4->flags_and_fragment_offset = + clib_host_to_net_u16 (IP4_HEADER_FLAG_MORE_FRAGMENTS); + inner_ip4->ttl = inner_ip6->hop_limit; + inner_ip4->protocol = inner_protocol; + inner_ip4->checksum = ip4_header_checksum (inner_ip4); + + if (inner_ip4->protocol == IP_PROTOCOL_ICMP) + { + //Remove remainings of the pseudo-header in the csum + csum = + ip_csum_sub_even (csum, clib_host_to_net_u16 (IP_PROTOCOL_ICMP6)); + csum = + ip_csum_sub_even (csum, inner_ip4->length - sizeof (*inner_ip4)); + } + else + { + //Update to new pseudo-header + csum = ip_csum_add_even (csum, inner_ip4->src_address.as_u32); + csum = ip_csum_add_even (csum, inner_ip4->dst_address.as_u32); + } + *inner_L4_checksum = ip_csum_fold (csum); + + //Move up icmp header + ip4 = (ip4_header_t *) u8_ptr_add (inner_l4, -2 * sizeof (*ip4) - 8); + clib_memcpy (u8_ptr_add (inner_l4, -sizeof (*ip4) - 8), icmp, 8); + icmp = (icmp46_header_t *) u8_ptr_add (inner_l4, -sizeof (*ip4) - 8); } - *inner_L4_checksum = ip_csum_fold(csum); - - //Move up icmp header - ip4 = (ip4_header_t *) u8_ptr_add(inner_l4, - 2 * sizeof(*ip4) - 8); - clib_memcpy(u8_ptr_add(inner_l4, - sizeof(*ip4) - 8), icmp, 8); - icmp = (icmp46_header_t *) u8_ptr_add(inner_l4, - sizeof(*ip4) - 8); - } else { - //Only one header to translate - ip4 = (ip4_header_t *) u8_ptr_add(ip6, sizeof(*ip6) - sizeof(*ip4)); - } - vlib_buffer_advance(p, (u32) (((u8 *)ip4) - ((u8 *)ip6))); - - ip4->dst_address.as_u32 = ip6_map_t_embedded_address(d, &ip6->dst_address); + else + { + //Only one header to translate + ip4 = (ip4_header_t *) u8_ptr_add (ip6, sizeof (*ip6) - sizeof (*ip4)); + } + vlib_buffer_advance (p, (u32) (((u8 *) ip4) - ((u8 *) ip6))); + + ip4->dst_address.as_u32 = ip6_map_t_embedded_address (d, &ip6->dst_address); ip4->src_address.as_u32 = ip4_sadr; - ip4->ip_version_and_header_length = IP4_VERSION_AND_HEADER_LENGTH_NO_OPTIONS; - ip4->tos = ip6_translate_tos(ip6); + ip4->ip_version_and_header_length = + IP4_VERSION_AND_HEADER_LENGTH_NO_OPTIONS; + ip4->tos = ip6_translate_tos (ip6); ip4->fragment_id = 0; ip4->flags_and_fragment_offset = 0; ip4->ttl = ip6->hop_limit; ip4->protocol = IP_PROTOCOL_ICMP; //TODO fix the length depending on offset length - ip4->length = u16_net_add(ip6->payload_length, - (inner_ip6 == NULL)?sizeof(*ip4):(2*sizeof(*ip4) - sizeof(*ip6))); - ip4->checksum = ip4_header_checksum(ip4); + ip4->length = u16_net_add (ip6->payload_length, + (inner_ip6 == + NULL) ? sizeof (*ip4) : (2 * sizeof (*ip4) - + sizeof (*ip6))); + ip4->checksum = ip4_header_checksum (ip4); //TODO: We could do an easy diff-checksum for echo requests/replies //Recompute ICMP checksum icmp->checksum = 0; - csum = ip_incremental_checksum(0, icmp, clib_net_to_host_u16(ip4->length) - sizeof(*ip4)); + csum = + ip_incremental_checksum (0, icmp, + clib_net_to_host_u16 (ip4->length) - + sizeof (*ip4)); icmp->checksum = ~ip_csum_fold (csum); } static uword -ip6_map_t_icmp (vlib_main_t *vm, - vlib_node_runtime_t *node, - vlib_frame_t *frame) +ip6_map_t_icmp (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) { u32 n_left_from, *from, next_index, *to_next, n_left_to_next; - vlib_node_runtime_t *error_node = vlib_node_get_runtime (vm, ip6_map_t_icmp_node.index); + vlib_node_runtime_t *error_node = + vlib_node_get_runtime (vm, ip6_map_t_icmp_node.index); from = vlib_frame_vector_args (frame); n_left_from = frame->n_vectors; next_index = node->cached_next_index; vlib_combined_counter_main_t *cm = map_main.domain_counters; - u32 cpu_index = os_get_cpu_number(); - - while (n_left_from > 0) { - vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); - - while (n_left_from > 0 && n_left_to_next > 0) { - u32 pi0; - vlib_buffer_t *p0; - u8 error0; - ip6_mapt_icmp_next_t next0; - map_domain_t *d0; - u16 len0; - - pi0 = to_next[0] = from[0]; - from += 1; - n_left_from -= 1; - to_next +=1; - n_left_to_next -= 1; - error0 = MAP_ERROR_NONE; - next0 = IP6_MAPT_ICMP_NEXT_IP4_LOOKUP; - - p0 = vlib_get_buffer(vm, pi0); - len0 = clib_net_to_host_u16(((ip6_header_t *)vlib_buffer_get_current(p0))->payload_length); - d0 = pool_elt_at_index(map_main.domains, vnet_buffer(p0)->map_t.map_domain_index); - _ip6_map_t_icmp(d0, p0, &error0); - - if(vnet_buffer(p0)->map_t.mtu < p0->current_length) { - //Send to fragmentation node if necessary - vnet_buffer(p0)->ip_frag.mtu = vnet_buffer(p0)->map_t.mtu; - vnet_buffer(p0)->ip_frag.header_offset = 0; - vnet_buffer(p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP4_LOOKUP; - next0 = IP6_MAPT_ICMP_NEXT_IP4_FRAG; - } - - if (PREDICT_TRUE(error0 == MAP_ERROR_NONE)) { - vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_RX, cpu_index, - vnet_buffer(p0)->map_t.map_domain_index, 1, - len0); - } else { - next0 = IP6_MAPT_ICMP_NEXT_DROP; - } - - p0->error = error_node->errors[error0]; - vlib_validate_buffer_enqueue_x1 (vm, node, next_index, - to_next, n_left_to_next, pi0, - next0); + u32 cpu_index = os_get_cpu_number (); + + while (n_left_from > 0) + { + vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); + + while (n_left_from > 0 && n_left_to_next > 0) + { + u32 pi0; + vlib_buffer_t *p0; + u8 error0; + ip6_mapt_icmp_next_t next0; + map_domain_t *d0; + u16 len0; + + pi0 = to_next[0] = from[0]; + from += 1; + n_left_from -= 1; + to_next += 1; + n_left_to_next -= 1; + error0 = MAP_ERROR_NONE; + next0 = IP6_MAPT_ICMP_NEXT_IP4_LOOKUP; + + p0 = vlib_get_buffer (vm, pi0); + len0 = + clib_net_to_host_u16 (((ip6_header_t *) + vlib_buffer_get_current + (p0))->payload_length); + d0 = + pool_elt_at_index (map_main.domains, + vnet_buffer (p0)->map_t.map_domain_index); + _ip6_map_t_icmp (d0, p0, &error0); + + if (vnet_buffer (p0)->map_t.mtu < p0->current_length) + { + //Send to fragmentation node if necessary + vnet_buffer (p0)->ip_frag.mtu = vnet_buffer (p0)->map_t.mtu; + vnet_buffer (p0)->ip_frag.header_offset = 0; + vnet_buffer (p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP4_LOOKUP; + next0 = IP6_MAPT_ICMP_NEXT_IP4_FRAG; + } + + if (PREDICT_TRUE (error0 == MAP_ERROR_NONE)) + { + vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_RX, + cpu_index, + vnet_buffer (p0)-> + map_t.map_domain_index, 1, + len0); + } + else + { + next0 = IP6_MAPT_ICMP_NEXT_DROP; + } + + p0->error = error_node->errors[error0]; + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, + to_next, n_left_to_next, pi0, + next0); + } + vlib_put_next_frame (vm, node, next_index, n_left_to_next); } - vlib_put_next_frame (vm, node, next_index, n_left_to_next); - } return frame->n_vectors; } static uword -ip6_map_t_fragmented (vlib_main_t *vm, - vlib_node_runtime_t *node, - vlib_frame_t *frame) +ip6_map_t_fragmented (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) { u32 n_left_from, *from, next_index, *to_next, n_left_to_next; - from = vlib_frame_vector_args(frame); + from = vlib_frame_vector_args (frame); n_left_from = frame->n_vectors; next_index = node->cached_next_index; - while (n_left_from > 0) { - vlib_get_next_frame(vm, node, next_index, to_next, n_left_to_next); + while (n_left_from > 0) + { + vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); #ifdef IP6_MAP_T_DUAL_LOOP - while(n_left_from >= 4 && n_left_to_next >= 2) { - u32 pi0, pi1; - vlib_buffer_t *p0, *p1; - ip6_header_t *ip60, *ip61; - ip6_frag_hdr_t *frag0, *frag1; - ip4_header_t *ip40, *ip41; - u16 frag_id0, frag_offset0, - frag_id1, frag_offset1; - u8 frag_more0, frag_more1; - u32 next0, next1; - - pi0 = to_next[0] = from[0]; - pi1 = to_next[1] = from[1]; - from += 2; - n_left_from -= 2; - to_next += 2; - n_left_to_next -= 2; - - next0 = IP6_MAPT_TCP_UDP_NEXT_IP4_LOOKUP; - next1 = IP6_MAPT_TCP_UDP_NEXT_IP4_LOOKUP; - p0 = vlib_get_buffer(vm, pi0); - p1 = vlib_get_buffer(vm, pi1); - ip60 = vlib_buffer_get_current(p0); - ip61 = vlib_buffer_get_current(p1); - frag0 = (ip6_frag_hdr_t *)u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.frag_offset); - frag1 = (ip6_frag_hdr_t *)u8_ptr_add(ip61, vnet_buffer(p1)->map_t.v6.frag_offset); - ip40 = (ip4_header_t *)u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.l4_offset - sizeof(*ip40)); - ip41 = (ip4_header_t *)u8_ptr_add(ip61, vnet_buffer(p1)->map_t.v6.l4_offset - sizeof(*ip40)); - vlib_buffer_advance(p0, vnet_buffer(p0)->map_t.v6.l4_offset - sizeof(*ip40)); - vlib_buffer_advance(p1, vnet_buffer(p1)->map_t.v6.l4_offset - sizeof(*ip40)); - - frag_id0 = frag_id_6to4(frag0->identification); - frag_id1 = frag_id_6to4(frag1->identification); - frag_more0 = ip6_frag_hdr_more(frag0); - frag_more1 = ip6_frag_hdr_more(frag1); - frag_offset0 = ip6_frag_hdr_offset(frag0); - frag_offset1 = ip6_frag_hdr_offset(frag1); - - ip40->dst_address.as_u32 = vnet_buffer(p0)->map_t.v6.daddr; - ip41->dst_address.as_u32 = vnet_buffer(p1)->map_t.v6.daddr; - ip40->src_address.as_u32 = vnet_buffer(p0)->map_t.v6.saddr; - ip41->src_address.as_u32 = vnet_buffer(p1)->map_t.v6.saddr; - ip40->ip_version_and_header_length = IP4_VERSION_AND_HEADER_LENGTH_NO_OPTIONS; - ip41->ip_version_and_header_length = IP4_VERSION_AND_HEADER_LENGTH_NO_OPTIONS; - ip40->tos = ip6_translate_tos(ip60); - ip41->tos = ip6_translate_tos(ip61); - ip40->length = u16_net_add(ip60->payload_length, - sizeof(*ip40) - vnet_buffer(p0)->map_t.v6.l4_offset + sizeof(*ip60)); - ip41->length = u16_net_add(ip61->payload_length, - sizeof(*ip40) - vnet_buffer(p1)->map_t.v6.l4_offset + sizeof(*ip60)); - ip40->fragment_id = frag_id0; - ip41->fragment_id = frag_id1; - ip40->flags_and_fragment_offset = - clib_host_to_net_u16(frag_offset0 | (frag_more0?IP4_HEADER_FLAG_MORE_FRAGMENTS:0)); - ip41->flags_and_fragment_offset = - clib_host_to_net_u16(frag_offset1 | (frag_more1?IP4_HEADER_FLAG_MORE_FRAGMENTS:0)); - ip40->ttl = ip60->hop_limit; - ip41->ttl = ip61->hop_limit; - ip40->protocol = (vnet_buffer(p0)->map_t.v6.l4_protocol == IP_PROTOCOL_ICMP6)? - IP_PROTOCOL_ICMP:vnet_buffer(p0)->map_t.v6.l4_protocol; - ip41->protocol = (vnet_buffer(p1)->map_t.v6.l4_protocol == IP_PROTOCOL_ICMP6)? - IP_PROTOCOL_ICMP:vnet_buffer(p1)->map_t.v6.l4_protocol; - ip40->checksum = ip4_header_checksum(ip40); - ip41->checksum = ip4_header_checksum(ip41); - - if(vnet_buffer(p0)->map_t.mtu < p0->current_length) { - vnet_buffer(p0)->ip_frag.mtu = vnet_buffer(p0)->map_t.mtu; - vnet_buffer(p0)->ip_frag.header_offset = 0; - vnet_buffer(p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP4_LOOKUP; - next0 = IP6_MAPT_FRAGMENTED_NEXT_IP4_FRAG; - } - - if(vnet_buffer(p1)->map_t.mtu < p1->current_length) { - vnet_buffer(p1)->ip_frag.mtu = vnet_buffer(p1)->map_t.mtu; - vnet_buffer(p1)->ip_frag.header_offset = 0; - vnet_buffer(p1)->ip_frag.next_index = IP4_FRAG_NEXT_IP4_LOOKUP; - next1 = IP6_MAPT_FRAGMENTED_NEXT_IP4_FRAG; - } - - vlib_validate_buffer_enqueue_x2(vm, node, next_index, - to_next, n_left_to_next, pi0, pi1, - next0, next1); - } + while (n_left_from >= 4 && n_left_to_next >= 2) + { + u32 pi0, pi1; + vlib_buffer_t *p0, *p1; + ip6_header_t *ip60, *ip61; + ip6_frag_hdr_t *frag0, *frag1; + ip4_header_t *ip40, *ip41; + u16 frag_id0, frag_offset0, frag_id1, frag_offset1; + u8 frag_more0, frag_more1; + u32 next0, next1; + + pi0 = to_next[0] = from[0]; + pi1 = to_next[1] = from[1]; + from += 2; + n_left_from -= 2; + to_next += 2; + n_left_to_next -= 2; + + next0 = IP6_MAPT_TCP_UDP_NEXT_IP4_LOOKUP; + next1 = IP6_MAPT_TCP_UDP_NEXT_IP4_LOOKUP; + p0 = vlib_get_buffer (vm, pi0); + p1 = vlib_get_buffer (vm, pi1); + ip60 = vlib_buffer_get_current (p0); + ip61 = vlib_buffer_get_current (p1); + frag0 = + (ip6_frag_hdr_t *) u8_ptr_add (ip60, + vnet_buffer (p0)->map_t. + v6.frag_offset); + frag1 = + (ip6_frag_hdr_t *) u8_ptr_add (ip61, + vnet_buffer (p1)->map_t. + v6.frag_offset); + ip40 = + (ip4_header_t *) u8_ptr_add (ip60, + vnet_buffer (p0)->map_t. + v6.l4_offset - sizeof (*ip40)); + ip41 = + (ip4_header_t *) u8_ptr_add (ip61, + vnet_buffer (p1)->map_t. + v6.l4_offset - sizeof (*ip40)); + vlib_buffer_advance (p0, + vnet_buffer (p0)->map_t.v6.l4_offset - + sizeof (*ip40)); + vlib_buffer_advance (p1, + vnet_buffer (p1)->map_t.v6.l4_offset - + sizeof (*ip40)); + + frag_id0 = frag_id_6to4 (frag0->identification); + frag_id1 = frag_id_6to4 (frag1->identification); + frag_more0 = ip6_frag_hdr_more (frag0); + frag_more1 = ip6_frag_hdr_more (frag1); + frag_offset0 = ip6_frag_hdr_offset (frag0); + frag_offset1 = ip6_frag_hdr_offset (frag1); + + ip40->dst_address.as_u32 = vnet_buffer (p0)->map_t.v6.daddr; + ip41->dst_address.as_u32 = vnet_buffer (p1)->map_t.v6.daddr; + ip40->src_address.as_u32 = vnet_buffer (p0)->map_t.v6.saddr; + ip41->src_address.as_u32 = vnet_buffer (p1)->map_t.v6.saddr; + ip40->ip_version_and_header_length = + IP4_VERSION_AND_HEADER_LENGTH_NO_OPTIONS; + ip41->ip_version_and_header_length = + IP4_VERSION_AND_HEADER_LENGTH_NO_OPTIONS; + ip40->tos = ip6_translate_tos (ip60); + ip41->tos = ip6_translate_tos (ip61); + ip40->length = u16_net_add (ip60->payload_length, + sizeof (*ip40) - + vnet_buffer (p0)->map_t.v6.l4_offset + + sizeof (*ip60)); + ip41->length = + u16_net_add (ip61->payload_length, + sizeof (*ip40) - + vnet_buffer (p1)->map_t.v6.l4_offset + + sizeof (*ip60)); + ip40->fragment_id = frag_id0; + ip41->fragment_id = frag_id1; + ip40->flags_and_fragment_offset = + clib_host_to_net_u16 (frag_offset0 | + (frag_more0 ? IP4_HEADER_FLAG_MORE_FRAGMENTS + : 0)); + ip41->flags_and_fragment_offset = + clib_host_to_net_u16 (frag_offset1 | + (frag_more1 ? IP4_HEADER_FLAG_MORE_FRAGMENTS + : 0)); + ip40->ttl = ip60->hop_limit; + ip41->ttl = ip61->hop_limit; + ip40->protocol = + (vnet_buffer (p0)->map_t.v6.l4_protocol == + IP_PROTOCOL_ICMP6) ? IP_PROTOCOL_ICMP : vnet_buffer (p0)-> + map_t.v6.l4_protocol; + ip41->protocol = + (vnet_buffer (p1)->map_t.v6.l4_protocol == + IP_PROTOCOL_ICMP6) ? IP_PROTOCOL_ICMP : vnet_buffer (p1)-> + map_t.v6.l4_protocol; + ip40->checksum = ip4_header_checksum (ip40); + ip41->checksum = ip4_header_checksum (ip41); + + if (vnet_buffer (p0)->map_t.mtu < p0->current_length) + { + vnet_buffer (p0)->ip_frag.mtu = vnet_buffer (p0)->map_t.mtu; + vnet_buffer (p0)->ip_frag.header_offset = 0; + vnet_buffer (p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP4_LOOKUP; + next0 = IP6_MAPT_FRAGMENTED_NEXT_IP4_FRAG; + } + + if (vnet_buffer (p1)->map_t.mtu < p1->current_length) + { + vnet_buffer (p1)->ip_frag.mtu = vnet_buffer (p1)->map_t.mtu; + vnet_buffer (p1)->ip_frag.header_offset = 0; + vnet_buffer (p1)->ip_frag.next_index = IP4_FRAG_NEXT_IP4_LOOKUP; + next1 = IP6_MAPT_FRAGMENTED_NEXT_IP4_FRAG; + } + + vlib_validate_buffer_enqueue_x2 (vm, node, next_index, + to_next, n_left_to_next, pi0, pi1, + next0, next1); + } #endif - while (n_left_from > 0 && n_left_to_next > 0) { - u32 pi0; - vlib_buffer_t *p0; - ip6_header_t *ip60; - ip6_frag_hdr_t *frag0; - ip4_header_t *ip40; - u16 frag_id0; - u8 frag_more0; - u16 frag_offset0; - u32 next0; - - pi0 = to_next[0] = from[0]; - from += 1; - n_left_from -= 1; - to_next +=1; - n_left_to_next -= 1; - - next0 = IP6_MAPT_TCP_UDP_NEXT_IP4_LOOKUP; - p0 = vlib_get_buffer(vm, pi0); - ip60 = vlib_buffer_get_current(p0); - frag0 = (ip6_frag_hdr_t *)u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.frag_offset); - ip40 = (ip4_header_t *)u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.l4_offset - sizeof(*ip40)); - vlib_buffer_advance(p0, vnet_buffer(p0)->map_t.v6.l4_offset - sizeof(*ip40)); - - frag_id0 = frag_id_6to4(frag0->identification); - frag_more0 = ip6_frag_hdr_more(frag0); - frag_offset0 = ip6_frag_hdr_offset(frag0); - - ip40->dst_address.as_u32 = vnet_buffer(p0)->map_t.v6.daddr; - ip40->src_address.as_u32 = vnet_buffer(p0)->map_t.v6.saddr; - ip40->ip_version_and_header_length = IP4_VERSION_AND_HEADER_LENGTH_NO_OPTIONS; - ip40->tos = ip6_translate_tos(ip60); - ip40->length = u16_net_add(ip60->payload_length, - sizeof(*ip40) - vnet_buffer(p0)->map_t.v6.l4_offset + sizeof(*ip60)); - ip40->fragment_id = frag_id0; - ip40->flags_and_fragment_offset = - clib_host_to_net_u16(frag_offset0 | (frag_more0?IP4_HEADER_FLAG_MORE_FRAGMENTS:0)); - ip40->ttl = ip60->hop_limit; - ip40->protocol = (vnet_buffer(p0)->map_t.v6.l4_protocol == IP_PROTOCOL_ICMP6)? - IP_PROTOCOL_ICMP:vnet_buffer(p0)->map_t.v6.l4_protocol; - ip40->checksum = ip4_header_checksum(ip40); - - if(vnet_buffer(p0)->map_t.mtu < p0->current_length) { - //Send to fragmentation node if necessary - vnet_buffer(p0)->ip_frag.mtu = vnet_buffer(p0)->map_t.mtu; - vnet_buffer(p0)->ip_frag.header_offset = 0; - vnet_buffer(p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP4_LOOKUP; - next0 = IP6_MAPT_FRAGMENTED_NEXT_IP4_FRAG; - } - - vlib_validate_buffer_enqueue_x1(vm, node, next_index, - to_next, n_left_to_next, pi0, - next0); + while (n_left_from > 0 && n_left_to_next > 0) + { + u32 pi0; + vlib_buffer_t *p0; + ip6_header_t *ip60; + ip6_frag_hdr_t *frag0; + ip4_header_t *ip40; + u16 frag_id0; + u8 frag_more0; + u16 frag_offset0; + u32 next0; + + pi0 = to_next[0] = from[0]; + from += 1; + n_left_from -= 1; + to_next += 1; + n_left_to_next -= 1; + + next0 = IP6_MAPT_TCP_UDP_NEXT_IP4_LOOKUP; + p0 = vlib_get_buffer (vm, pi0); + ip60 = vlib_buffer_get_current (p0); + frag0 = + (ip6_frag_hdr_t *) u8_ptr_add (ip60, + vnet_buffer (p0)->map_t. + v6.frag_offset); + ip40 = + (ip4_header_t *) u8_ptr_add (ip60, + vnet_buffer (p0)->map_t. + v6.l4_offset - sizeof (*ip40)); + vlib_buffer_advance (p0, + vnet_buffer (p0)->map_t.v6.l4_offset - + sizeof (*ip40)); + + frag_id0 = frag_id_6to4 (frag0->identification); + frag_more0 = ip6_frag_hdr_more (frag0); + frag_offset0 = ip6_frag_hdr_offset (frag0); + + ip40->dst_address.as_u32 = vnet_buffer (p0)->map_t.v6.daddr; + ip40->src_address.as_u32 = vnet_buffer (p0)->map_t.v6.saddr; + ip40->ip_version_and_header_length = + IP4_VERSION_AND_HEADER_LENGTH_NO_OPTIONS; + ip40->tos = ip6_translate_tos (ip60); + ip40->length = u16_net_add (ip60->payload_length, + sizeof (*ip40) - + vnet_buffer (p0)->map_t.v6.l4_offset + + sizeof (*ip60)); + ip40->fragment_id = frag_id0; + ip40->flags_and_fragment_offset = + clib_host_to_net_u16 (frag_offset0 | + (frag_more0 ? IP4_HEADER_FLAG_MORE_FRAGMENTS + : 0)); + ip40->ttl = ip60->hop_limit; + ip40->protocol = + (vnet_buffer (p0)->map_t.v6.l4_protocol == + IP_PROTOCOL_ICMP6) ? IP_PROTOCOL_ICMP : vnet_buffer (p0)-> + map_t.v6.l4_protocol; + ip40->checksum = ip4_header_checksum (ip40); + + if (vnet_buffer (p0)->map_t.mtu < p0->current_length) + { + //Send to fragmentation node if necessary + vnet_buffer (p0)->ip_frag.mtu = vnet_buffer (p0)->map_t.mtu; + vnet_buffer (p0)->ip_frag.header_offset = 0; + vnet_buffer (p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP4_LOOKUP; + next0 = IP6_MAPT_FRAGMENTED_NEXT_IP4_FRAG; + } + + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, + to_next, n_left_to_next, pi0, + next0); + } + vlib_put_next_frame (vm, node, next_index, n_left_to_next); } - vlib_put_next_frame(vm, node, next_index, n_left_to_next); - } return frame->n_vectors; } static uword -ip6_map_t_tcp_udp (vlib_main_t *vm, - vlib_node_runtime_t *node, - vlib_frame_t *frame) +ip6_map_t_tcp_udp (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) { u32 n_left_from, *from, next_index, *to_next, n_left_to_next; - from = vlib_frame_vector_args(frame); + from = vlib_frame_vector_args (frame); n_left_from = frame->n_vectors; next_index = node->cached_next_index; - while (n_left_from > 0) { - vlib_get_next_frame(vm, node, next_index, to_next, n_left_to_next); + while (n_left_from > 0) + { + vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); #ifdef IP6_MAP_T_DUAL_LOOP - while(n_left_from >= 4 && n_left_to_next >= 2) { - u32 pi0, pi1; - vlib_buffer_t *p0, *p1; - ip6_header_t *ip60, *ip61; - ip_csum_t csum0, csum1; - ip4_header_t *ip40, *ip41; - u16 fragment_id0, flags0, *checksum0, - fragment_id1, flags1, *checksum1; - ip6_mapt_tcp_udp_next_t next0, next1; - - pi0 = to_next[0] = from[0]; - pi1 = to_next[1] = from[1]; - from += 2; - n_left_from -= 2; - to_next += 2; - n_left_to_next -= 2; - next0 = IP6_MAPT_TCP_UDP_NEXT_IP4_LOOKUP; - next1 = IP6_MAPT_TCP_UDP_NEXT_IP4_LOOKUP; - - p0 = vlib_get_buffer(vm, pi0); - p1 = vlib_get_buffer(vm, pi1); - ip60 = vlib_buffer_get_current(p0); - ip61 = vlib_buffer_get_current(p1); - ip40 = (ip4_header_t *) u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.l4_offset - sizeof(*ip40)); - ip41 = (ip4_header_t *) u8_ptr_add(ip61, vnet_buffer(p1)->map_t.v6.l4_offset - sizeof(*ip40)); - vlib_buffer_advance(p0, vnet_buffer(p0)->map_t.v6.l4_offset - sizeof(*ip40)); - vlib_buffer_advance(p1, vnet_buffer(p1)->map_t.v6.l4_offset - sizeof(*ip40)); - checksum0 = (u16 *) u8_ptr_add(ip60, vnet_buffer(p0)->map_t.checksum_offset); - checksum1 = (u16 *) u8_ptr_add(ip61, vnet_buffer(p1)->map_t.checksum_offset); - - csum0 = ip_csum_sub_even(*checksum0, ip60->src_address.as_u64[0]); - csum1 = ip_csum_sub_even(*checksum1, ip61->src_address.as_u64[0]); - csum0 = ip_csum_sub_even(csum0, ip60->src_address.as_u64[1]); - csum1 = ip_csum_sub_even(csum1, ip61->src_address.as_u64[1]); - csum0 = ip_csum_sub_even(csum0, ip60->dst_address.as_u64[0]); - csum1 = ip_csum_sub_even(csum0, ip61->dst_address.as_u64[0]); - csum0 = ip_csum_sub_even(csum0, ip60->dst_address.as_u64[1]); - csum1 = ip_csum_sub_even(csum1, ip61->dst_address.as_u64[1]); - csum0 = ip_csum_add_even(csum0, vnet_buffer(p0)->map_t.v6.daddr); - csum1 = ip_csum_add_even(csum1, vnet_buffer(p1)->map_t.v6.daddr); - csum0 = ip_csum_add_even(csum0, vnet_buffer(p0)->map_t.v6.saddr); - csum1 = ip_csum_add_even(csum1, vnet_buffer(p1)->map_t.v6.saddr); - *checksum0 = ip_csum_fold(csum0); - *checksum1 = ip_csum_fold(csum1); - - if (PREDICT_FALSE(vnet_buffer(p0)->map_t.v6.frag_offset)) { - ip6_frag_hdr_t *hdr = (ip6_frag_hdr_t *) u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.frag_offset); - fragment_id0 = frag_id_6to4(hdr->identification); - flags0 = clib_host_to_net_u16(IP4_HEADER_FLAG_MORE_FRAGMENTS); - } else { - fragment_id0 = 0; - flags0 = 0; - } - - if (PREDICT_FALSE(vnet_buffer(p1)->map_t.v6.frag_offset)) { - ip6_frag_hdr_t *hdr = (ip6_frag_hdr_t *) u8_ptr_add(ip61, vnet_buffer(p1)->map_t.v6.frag_offset); - fragment_id1 = frag_id_6to4(hdr->identification); - flags1 = clib_host_to_net_u16(IP4_HEADER_FLAG_MORE_FRAGMENTS); - } else { - fragment_id1 = 0; - flags1 = 0; - } - - ip40->dst_address.as_u32 = vnet_buffer(p0)->map_t.v6.daddr; - ip41->dst_address.as_u32 = vnet_buffer(p1)->map_t.v6.daddr; - ip40->src_address.as_u32 = vnet_buffer(p0)->map_t.v6.saddr; - ip41->src_address.as_u32 = vnet_buffer(p1)->map_t.v6.saddr; - ip40->ip_version_and_header_length = IP4_VERSION_AND_HEADER_LENGTH_NO_OPTIONS; - ip41->ip_version_and_header_length = IP4_VERSION_AND_HEADER_LENGTH_NO_OPTIONS; - ip40->tos = ip6_translate_tos(ip60); - ip41->tos = ip6_translate_tos(ip61); - ip40->length = u16_net_add(ip60->payload_length, - sizeof(*ip40) + sizeof(*ip60) - vnet_buffer(p0)->map_t.v6.l4_offset); - ip41->length = u16_net_add(ip61->payload_length, - sizeof(*ip40) + sizeof(*ip60) - vnet_buffer(p1)->map_t.v6.l4_offset); - ip40->fragment_id = fragment_id0; - ip41->fragment_id = fragment_id1; - ip40->flags_and_fragment_offset = flags0; - ip41->flags_and_fragment_offset = flags1; - ip40->ttl = ip60->hop_limit; - ip41->ttl = ip61->hop_limit; - ip40->protocol = vnet_buffer(p0)->map_t.v6.l4_protocol; - ip41->protocol = vnet_buffer(p1)->map_t.v6.l4_protocol; - ip40->checksum = ip4_header_checksum(ip40); - ip41->checksum = ip4_header_checksum(ip41); - - if(vnet_buffer(p0)->map_t.mtu < p0->current_length) { - vnet_buffer(p0)->ip_frag.mtu = vnet_buffer(p0)->map_t.mtu; - vnet_buffer(p0)->ip_frag.header_offset = 0; - vnet_buffer(p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP4_LOOKUP; - next0 = IP6_MAPT_TCP_UDP_NEXT_IP4_FRAG; - } - - if(vnet_buffer(p1)->map_t.mtu < p1->current_length) { - vnet_buffer(p1)->ip_frag.mtu = vnet_buffer(p1)->map_t.mtu; - vnet_buffer(p1)->ip_frag.header_offset = 0; - vnet_buffer(p1)->ip_frag.next_index = IP4_FRAG_NEXT_IP4_LOOKUP; - next1 = IP6_MAPT_TCP_UDP_NEXT_IP4_FRAG; - } - - vlib_validate_buffer_enqueue_x2(vm, node, next_index, to_next, - n_left_to_next, pi0, pi1, next0, next1); - } + while (n_left_from >= 4 && n_left_to_next >= 2) + { + u32 pi0, pi1; + vlib_buffer_t *p0, *p1; + ip6_header_t *ip60, *ip61; + ip_csum_t csum0, csum1; + ip4_header_t *ip40, *ip41; + u16 fragment_id0, flags0, *checksum0, + fragment_id1, flags1, *checksum1; + ip6_mapt_tcp_udp_next_t next0, next1; + + pi0 = to_next[0] = from[0]; + pi1 = to_next[1] = from[1]; + from += 2; + n_left_from -= 2; + to_next += 2; + n_left_to_next -= 2; + next0 = IP6_MAPT_TCP_UDP_NEXT_IP4_LOOKUP; + next1 = IP6_MAPT_TCP_UDP_NEXT_IP4_LOOKUP; + + p0 = vlib_get_buffer (vm, pi0); + p1 = vlib_get_buffer (vm, pi1); + ip60 = vlib_buffer_get_current (p0); + ip61 = vlib_buffer_get_current (p1); + ip40 = + (ip4_header_t *) u8_ptr_add (ip60, + vnet_buffer (p0)->map_t. + v6.l4_offset - sizeof (*ip40)); + ip41 = + (ip4_header_t *) u8_ptr_add (ip61, + vnet_buffer (p1)->map_t. + v6.l4_offset - sizeof (*ip40)); + vlib_buffer_advance (p0, + vnet_buffer (p0)->map_t.v6.l4_offset - + sizeof (*ip40)); + vlib_buffer_advance (p1, + vnet_buffer (p1)->map_t.v6.l4_offset - + sizeof (*ip40)); + checksum0 = + (u16 *) u8_ptr_add (ip60, + vnet_buffer (p0)->map_t.checksum_offset); + checksum1 = + (u16 *) u8_ptr_add (ip61, + vnet_buffer (p1)->map_t.checksum_offset); + + csum0 = ip_csum_sub_even (*checksum0, ip60->src_address.as_u64[0]); + csum1 = ip_csum_sub_even (*checksum1, ip61->src_address.as_u64[0]); + csum0 = ip_csum_sub_even (csum0, ip60->src_address.as_u64[1]); + csum1 = ip_csum_sub_even (csum1, ip61->src_address.as_u64[1]); + csum0 = ip_csum_sub_even (csum0, ip60->dst_address.as_u64[0]); + csum1 = ip_csum_sub_even (csum0, ip61->dst_address.as_u64[0]); + csum0 = ip_csum_sub_even (csum0, ip60->dst_address.as_u64[1]); + csum1 = ip_csum_sub_even (csum1, ip61->dst_address.as_u64[1]); + csum0 = ip_csum_add_even (csum0, vnet_buffer (p0)->map_t.v6.daddr); + csum1 = ip_csum_add_even (csum1, vnet_buffer (p1)->map_t.v6.daddr); + csum0 = ip_csum_add_even (csum0, vnet_buffer (p0)->map_t.v6.saddr); + csum1 = ip_csum_add_even (csum1, vnet_buffer (p1)->map_t.v6.saddr); + *checksum0 = ip_csum_fold (csum0); + *checksum1 = ip_csum_fold (csum1); + + if (PREDICT_FALSE (vnet_buffer (p0)->map_t.v6.frag_offset)) + { + ip6_frag_hdr_t *hdr = (ip6_frag_hdr_t *) u8_ptr_add (ip60, + vnet_buffer + (p0)-> + map_t. + v6.frag_offset); + fragment_id0 = frag_id_6to4 (hdr->identification); + flags0 = clib_host_to_net_u16 (IP4_HEADER_FLAG_MORE_FRAGMENTS); + } + else + { + fragment_id0 = 0; + flags0 = 0; + } + + if (PREDICT_FALSE (vnet_buffer (p1)->map_t.v6.frag_offset)) + { + ip6_frag_hdr_t *hdr = (ip6_frag_hdr_t *) u8_ptr_add (ip61, + vnet_buffer + (p1)-> + map_t. + v6.frag_offset); + fragment_id1 = frag_id_6to4 (hdr->identification); + flags1 = clib_host_to_net_u16 (IP4_HEADER_FLAG_MORE_FRAGMENTS); + } + else + { + fragment_id1 = 0; + flags1 = 0; + } + + ip40->dst_address.as_u32 = vnet_buffer (p0)->map_t.v6.daddr; + ip41->dst_address.as_u32 = vnet_buffer (p1)->map_t.v6.daddr; + ip40->src_address.as_u32 = vnet_buffer (p0)->map_t.v6.saddr; + ip41->src_address.as_u32 = vnet_buffer (p1)->map_t.v6.saddr; + ip40->ip_version_and_header_length = + IP4_VERSION_AND_HEADER_LENGTH_NO_OPTIONS; + ip41->ip_version_and_header_length = + IP4_VERSION_AND_HEADER_LENGTH_NO_OPTIONS; + ip40->tos = ip6_translate_tos (ip60); + ip41->tos = ip6_translate_tos (ip61); + ip40->length = u16_net_add (ip60->payload_length, + sizeof (*ip40) + sizeof (*ip60) - + vnet_buffer (p0)->map_t.v6.l4_offset); + ip41->length = + u16_net_add (ip61->payload_length, + sizeof (*ip40) + sizeof (*ip60) - + vnet_buffer (p1)->map_t.v6.l4_offset); + ip40->fragment_id = fragment_id0; + ip41->fragment_id = fragment_id1; + ip40->flags_and_fragment_offset = flags0; + ip41->flags_and_fragment_offset = flags1; + ip40->ttl = ip60->hop_limit; + ip41->ttl = ip61->hop_limit; + ip40->protocol = vnet_buffer (p0)->map_t.v6.l4_protocol; + ip41->protocol = vnet_buffer (p1)->map_t.v6.l4_protocol; + ip40->checksum = ip4_header_checksum (ip40); + ip41->checksum = ip4_header_checksum (ip41); + + if (vnet_buffer (p0)->map_t.mtu < p0->current_length) + { + vnet_buffer (p0)->ip_frag.mtu = vnet_buffer (p0)->map_t.mtu; + vnet_buffer (p0)->ip_frag.header_offset = 0; + vnet_buffer (p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP4_LOOKUP; + next0 = IP6_MAPT_TCP_UDP_NEXT_IP4_FRAG; + } + + if (vnet_buffer (p1)->map_t.mtu < p1->current_length) + { + vnet_buffer (p1)->ip_frag.mtu = vnet_buffer (p1)->map_t.mtu; + vnet_buffer (p1)->ip_frag.header_offset = 0; + vnet_buffer (p1)->ip_frag.next_index = IP4_FRAG_NEXT_IP4_LOOKUP; + next1 = IP6_MAPT_TCP_UDP_NEXT_IP4_FRAG; + } + + vlib_validate_buffer_enqueue_x2 (vm, node, next_index, to_next, + n_left_to_next, pi0, pi1, next0, + next1); + } #endif - while (n_left_from > 0 && n_left_to_next > 0) { - u32 pi0; - vlib_buffer_t *p0; - ip6_header_t *ip60; - u16 *checksum0; - ip_csum_t csum0; - ip4_header_t *ip40; - u16 fragment_id0; - u16 flags0; - ip6_mapt_tcp_udp_next_t next0; - - pi0 = to_next[0] = from[0]; - from += 1; - n_left_from -= 1; - to_next +=1; - n_left_to_next -= 1; - next0 = IP6_MAPT_TCP_UDP_NEXT_IP4_LOOKUP; - - p0 = vlib_get_buffer(vm, pi0); - ip60 = vlib_buffer_get_current(p0); - ip40 = (ip4_header_t *) u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.l4_offset - sizeof(*ip40)); - vlib_buffer_advance(p0, vnet_buffer(p0)->map_t.v6.l4_offset - sizeof(*ip40)); - checksum0 = (u16 *) u8_ptr_add(ip60, vnet_buffer(p0)->map_t.checksum_offset); - - //TODO: This can probably be optimized - csum0 = ip_csum_sub_even(*checksum0, ip60->src_address.as_u64[0]); - csum0 = ip_csum_sub_even(csum0, ip60->src_address.as_u64[1]); - csum0 = ip_csum_sub_even(csum0, ip60->dst_address.as_u64[0]); - csum0 = ip_csum_sub_even(csum0, ip60->dst_address.as_u64[1]); - csum0 = ip_csum_add_even(csum0, vnet_buffer(p0)->map_t.v6.daddr); - csum0 = ip_csum_add_even(csum0, vnet_buffer(p0)->map_t.v6.saddr); - *checksum0 = ip_csum_fold(csum0); - - if (PREDICT_FALSE(vnet_buffer(p0)->map_t.v6.frag_offset)) { - //Only the first fragment - ip6_frag_hdr_t *hdr = (ip6_frag_hdr_t *) u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.frag_offset); - fragment_id0 = frag_id_6to4(hdr->identification); - flags0 = clib_host_to_net_u16(IP4_HEADER_FLAG_MORE_FRAGMENTS); - } else { - fragment_id0 = 0; - flags0 = 0; - } - - ip40->dst_address.as_u32 = vnet_buffer(p0)->map_t.v6.daddr; - ip40->src_address.as_u32 = vnet_buffer(p0)->map_t.v6.saddr; - ip40->ip_version_and_header_length = IP4_VERSION_AND_HEADER_LENGTH_NO_OPTIONS; - ip40->tos = ip6_translate_tos(ip60); - ip40->length = u16_net_add(ip60->payload_length, - sizeof(*ip40) + sizeof(*ip60) - vnet_buffer(p0)->map_t.v6.l4_offset); - ip40->fragment_id = fragment_id0; - ip40->flags_and_fragment_offset = flags0; - ip40->ttl = ip60->hop_limit; - ip40->protocol = vnet_buffer(p0)->map_t.v6.l4_protocol; - ip40->checksum = ip4_header_checksum(ip40); - - if(vnet_buffer(p0)->map_t.mtu < p0->current_length) { - //Send to fragmentation node if necessary - vnet_buffer(p0)->ip_frag.mtu = vnet_buffer(p0)->map_t.mtu; - vnet_buffer(p0)->ip_frag.header_offset = 0; - vnet_buffer(p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP4_LOOKUP; - next0 = IP6_MAPT_TCP_UDP_NEXT_IP4_FRAG; - } - - vlib_validate_buffer_enqueue_x1(vm, node, next_index, - to_next, n_left_to_next, pi0, - next0); + while (n_left_from > 0 && n_left_to_next > 0) + { + u32 pi0; + vlib_buffer_t *p0; + ip6_header_t *ip60; + u16 *checksum0; + ip_csum_t csum0; + ip4_header_t *ip40; + u16 fragment_id0; + u16 flags0; + ip6_mapt_tcp_udp_next_t next0; + + pi0 = to_next[0] = from[0]; + from += 1; + n_left_from -= 1; + to_next += 1; + n_left_to_next -= 1; + next0 = IP6_MAPT_TCP_UDP_NEXT_IP4_LOOKUP; + + p0 = vlib_get_buffer (vm, pi0); + ip60 = vlib_buffer_get_current (p0); + ip40 = + (ip4_header_t *) u8_ptr_add (ip60, + vnet_buffer (p0)->map_t. + v6.l4_offset - sizeof (*ip40)); + vlib_buffer_advance (p0, + vnet_buffer (p0)->map_t.v6.l4_offset - + sizeof (*ip40)); + checksum0 = + (u16 *) u8_ptr_add (ip60, + vnet_buffer (p0)->map_t.checksum_offset); + + //TODO: This can probably be optimized + csum0 = ip_csum_sub_even (*checksum0, ip60->src_address.as_u64[0]); + csum0 = ip_csum_sub_even (csum0, ip60->src_address.as_u64[1]); + csum0 = ip_csum_sub_even (csum0, ip60->dst_address.as_u64[0]); + csum0 = ip_csum_sub_even (csum0, ip60->dst_address.as_u64[1]); + csum0 = ip_csum_add_even (csum0, vnet_buffer (p0)->map_t.v6.daddr); + csum0 = ip_csum_add_even (csum0, vnet_buffer (p0)->map_t.v6.saddr); + *checksum0 = ip_csum_fold (csum0); + + if (PREDICT_FALSE (vnet_buffer (p0)->map_t.v6.frag_offset)) + { + //Only the first fragment + ip6_frag_hdr_t *hdr = (ip6_frag_hdr_t *) u8_ptr_add (ip60, + vnet_buffer + (p0)-> + map_t. + v6.frag_offset); + fragment_id0 = frag_id_6to4 (hdr->identification); + flags0 = clib_host_to_net_u16 (IP4_HEADER_FLAG_MORE_FRAGMENTS); + } + else + { + fragment_id0 = 0; + flags0 = 0; + } + + ip40->dst_address.as_u32 = vnet_buffer (p0)->map_t.v6.daddr; + ip40->src_address.as_u32 = vnet_buffer (p0)->map_t.v6.saddr; + ip40->ip_version_and_header_length = + IP4_VERSION_AND_HEADER_LENGTH_NO_OPTIONS; + ip40->tos = ip6_translate_tos (ip60); + ip40->length = u16_net_add (ip60->payload_length, + sizeof (*ip40) + sizeof (*ip60) - + vnet_buffer (p0)->map_t.v6.l4_offset); + ip40->fragment_id = fragment_id0; + ip40->flags_and_fragment_offset = flags0; + ip40->ttl = ip60->hop_limit; + ip40->protocol = vnet_buffer (p0)->map_t.v6.l4_protocol; + ip40->checksum = ip4_header_checksum (ip40); + + if (vnet_buffer (p0)->map_t.mtu < p0->current_length) + { + //Send to fragmentation node if necessary + vnet_buffer (p0)->ip_frag.mtu = vnet_buffer (p0)->map_t.mtu; + vnet_buffer (p0)->ip_frag.header_offset = 0; + vnet_buffer (p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP4_LOOKUP; + next0 = IP6_MAPT_TCP_UDP_NEXT_IP4_FRAG; + } + + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, + to_next, n_left_to_next, pi0, + next0); + } + vlib_put_next_frame (vm, node, next_index, n_left_to_next); } - vlib_put_next_frame(vm, node, next_index, n_left_to_next); - } return frame->n_vectors; } static_always_inline void -ip6_map_t_classify(vlib_buffer_t *p0, ip6_header_t *ip60, - map_domain_t *d0, i32 *src_port0, - u8 *error0, ip6_mapt_next_t *next0, - u32 l4_len0, ip6_frag_hdr_t *frag0) +ip6_map_t_classify (vlib_buffer_t * p0, ip6_header_t * ip60, + map_domain_t * d0, i32 * src_port0, + u8 * error0, ip6_mapt_next_t * next0, + u32 l4_len0, ip6_frag_hdr_t * frag0) { - if (PREDICT_FALSE(vnet_buffer(p0)->map_t.v6.frag_offset && - ip6_frag_hdr_offset(frag0))) { - *next0 = IP6_MAPT_NEXT_MAPT_FRAGMENTED; - if(d0->ea_bits_len == 0 && d0->rules) { - *src_port0 = 0; - } else { - *src_port0 = ip6_map_fragment_get(ip60, frag0, d0); - *error0 = (*src_port0 != -1) ? *error0 : MAP_ERROR_FRAGMENT_DROPPED; + if (PREDICT_FALSE (vnet_buffer (p0)->map_t.v6.frag_offset && + ip6_frag_hdr_offset (frag0))) + { + *next0 = IP6_MAPT_NEXT_MAPT_FRAGMENTED; + if (d0->ea_bits_len == 0 && d0->rules) + { + *src_port0 = 0; + } + else + { + *src_port0 = ip6_map_fragment_get (ip60, frag0, d0); + *error0 = (*src_port0 != -1) ? *error0 : MAP_ERROR_FRAGMENT_DROPPED; + } + } + else + if (PREDICT_TRUE + (vnet_buffer (p0)->map_t.v6.l4_protocol == IP_PROTOCOL_TCP)) + { + *error0 = + l4_len0 < sizeof (tcp_header_t) ? MAP_ERROR_MALFORMED : *error0; + vnet_buffer (p0)->map_t.checksum_offset = + vnet_buffer (p0)->map_t.v6.l4_offset + 16; + *next0 = IP6_MAPT_NEXT_MAPT_TCP_UDP; + *src_port0 = + (i32) * + ((u16 *) u8_ptr_add (ip60, vnet_buffer (p0)->map_t.v6.l4_offset)); + } + else + if (PREDICT_TRUE + (vnet_buffer (p0)->map_t.v6.l4_protocol == IP_PROTOCOL_UDP)) + { + *error0 = + l4_len0 < sizeof (udp_header_t) ? MAP_ERROR_MALFORMED : *error0; + vnet_buffer (p0)->map_t.checksum_offset = + vnet_buffer (p0)->map_t.v6.l4_offset + 6; + *next0 = IP6_MAPT_NEXT_MAPT_TCP_UDP; + *src_port0 = + (i32) * + ((u16 *) u8_ptr_add (ip60, vnet_buffer (p0)->map_t.v6.l4_offset)); } - } else if (PREDICT_TRUE(vnet_buffer(p0)->map_t.v6.l4_protocol == IP_PROTOCOL_TCP)) { - *error0 = l4_len0 < sizeof(tcp_header_t) ? MAP_ERROR_MALFORMED : *error0; - vnet_buffer(p0)->map_t.checksum_offset = vnet_buffer(p0)->map_t.v6.l4_offset + 16; - *next0 = IP6_MAPT_NEXT_MAPT_TCP_UDP; - *src_port0 = (i32) *((u16*)u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.l4_offset)); - } else if (PREDICT_TRUE(vnet_buffer(p0)->map_t.v6.l4_protocol == IP_PROTOCOL_UDP)) { - *error0 = l4_len0 < sizeof(udp_header_t) ? MAP_ERROR_MALFORMED : *error0; - vnet_buffer(p0)->map_t.checksum_offset = vnet_buffer(p0)->map_t.v6.l4_offset + 6; - *next0 = IP6_MAPT_NEXT_MAPT_TCP_UDP; - *src_port0 = (i32) *((u16*)u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.l4_offset)); - } else if (vnet_buffer(p0)->map_t.v6.l4_protocol == IP_PROTOCOL_ICMP6) { - *error0 = l4_len0 < sizeof(icmp46_header_t) ? MAP_ERROR_MALFORMED : *error0; - *next0 = IP6_MAPT_NEXT_MAPT_ICMP; - if(d0->ea_bits_len == 0 && d0->rules) { - *src_port0 = 0; - } else if (((icmp46_header_t *) u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.l4_offset))->code == ICMP6_echo_reply || - ((icmp46_header_t *) u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.l4_offset))->code == ICMP6_echo_request) { - *src_port0 = (i32) *((u16 *)u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.l4_offset + 6)); + else if (vnet_buffer (p0)->map_t.v6.l4_protocol == IP_PROTOCOL_ICMP6) + { + *error0 = + l4_len0 < sizeof (icmp46_header_t) ? MAP_ERROR_MALFORMED : *error0; + *next0 = IP6_MAPT_NEXT_MAPT_ICMP; + if (d0->ea_bits_len == 0 && d0->rules) + { + *src_port0 = 0; + } + else + if (((icmp46_header_t *) + u8_ptr_add (ip60, + vnet_buffer (p0)->map_t.v6.l4_offset))->code == + ICMP6_echo_reply + || ((icmp46_header_t *) + u8_ptr_add (ip60, + vnet_buffer (p0)->map_t.v6.l4_offset))->code == + ICMP6_echo_request) + { + *src_port0 = + (i32) * + ((u16 *) + u8_ptr_add (ip60, vnet_buffer (p0)->map_t.v6.l4_offset + 6)); + } + } + else + { + //TODO: In case of 1:1 mapping, it might be possible to do something with those packets. + *error0 = MAP_ERROR_BAD_PROTOCOL; } - } else { - //TODO: In case of 1:1 mapping, it might be possible to do something with those packets. - *error0 = MAP_ERROR_BAD_PROTOCOL; - } } static uword -ip6_map_t (vlib_main_t *vm, - vlib_node_runtime_t *node, - vlib_frame_t *frame) +ip6_map_t (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) { u32 n_left_from, *from, next_index, *to_next, n_left_to_next; - vlib_node_runtime_t *error_node = vlib_node_get_runtime(vm, ip6_map_t_node.index); + vlib_node_runtime_t *error_node = + vlib_node_get_runtime (vm, ip6_map_t_node.index); vlib_combined_counter_main_t *cm = map_main.domain_counters; - u32 cpu_index = os_get_cpu_number(); + u32 cpu_index = os_get_cpu_number (); - from = vlib_frame_vector_args(frame); + from = vlib_frame_vector_args (frame); n_left_from = frame->n_vectors; next_index = node->cached_next_index; - while (n_left_from > 0) { - vlib_get_next_frame(vm, node, next_index, to_next, n_left_to_next); + while (n_left_from > 0) + { + vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); #ifdef IP6_MAP_T_DUAL_LOOP - while (n_left_from >= 4 && n_left_to_next >=2) { - u32 pi0, pi1; - vlib_buffer_t *p0, *p1; - ip6_header_t *ip60, *ip61; - u8 error0, error1; - ip6_mapt_next_t next0, next1; - u32 l4_len0, l4_len1; - i32 src_port0, src_port1; - map_domain_t *d0, *d1; - ip6_frag_hdr_t *frag0, *frag1; - u32 saddr0, saddr1; - next0 = next1 = 0; //Because compiler whines - - pi0 = to_next[0] = from[0]; - pi1 = to_next[1] = from[1]; - from += 2; - n_left_from -= 2; - to_next += 2; - n_left_to_next -= 2; - - error0 = MAP_ERROR_NONE; - error1 = MAP_ERROR_NONE; - - p0 = vlib_get_buffer(vm, pi0); - p1 = vlib_get_buffer(vm, pi1); - ip60 = vlib_buffer_get_current(p0); - ip61 = vlib_buffer_get_current(p1); - - saddr0 = map_get_ip4(&ip60->src_address); - saddr1 = map_get_ip4(&ip61->src_address); - d0 = ip6_map_get_domain(vnet_buffer(p0)->ip.adj_index[VLIB_TX], - (ip4_address_t *)&saddr0, - &vnet_buffer(p0)->map_t.map_domain_index, &error0); - d1 = ip6_map_get_domain(vnet_buffer(p1)->ip.adj_index[VLIB_TX], - (ip4_address_t *)&saddr1, - &vnet_buffer(p1)->map_t.map_domain_index, &error1); - - vnet_buffer(p0)->map_t.v6.saddr = saddr0; - vnet_buffer(p1)->map_t.v6.saddr = saddr1; - vnet_buffer(p0)->map_t.v6.daddr = ip6_map_t_embedded_address(d0, &ip60->dst_address); - vnet_buffer(p1)->map_t.v6.daddr = ip6_map_t_embedded_address(d1, &ip61->dst_address); - vnet_buffer(p0)->map_t.mtu = d0->mtu ? d0->mtu : ~0; - vnet_buffer(p1)->map_t.mtu = d1->mtu ? d1->mtu : ~0; - - if (PREDICT_FALSE(ip6_parse(ip60, p0->current_length, - &(vnet_buffer(p0)->map_t.v6.l4_protocol), - &(vnet_buffer(p0)->map_t.v6.l4_offset), - &(vnet_buffer(p0)->map_t.v6.frag_offset)))) { - error0 = MAP_ERROR_MALFORMED; - next0 = IP6_MAPT_NEXT_DROP; - } - - if (PREDICT_FALSE(ip6_parse(ip61, p1->current_length, - &(vnet_buffer(p1)->map_t.v6.l4_protocol), - &(vnet_buffer(p1)->map_t.v6.l4_offset), - &(vnet_buffer(p1)->map_t.v6.frag_offset)))) { - error1 = MAP_ERROR_MALFORMED; - next1 = IP6_MAPT_NEXT_DROP; - } - - src_port0 = src_port1 = -1; - l4_len0 = (u32)clib_net_to_host_u16(ip60->payload_length) + - sizeof(*ip60) - vnet_buffer(p0)->map_t.v6.l4_offset; - l4_len1 = (u32)clib_net_to_host_u16(ip61->payload_length) + - sizeof(*ip60) - vnet_buffer(p1)->map_t.v6.l4_offset; - frag0 = (ip6_frag_hdr_t *) u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.frag_offset); - frag1 = (ip6_frag_hdr_t *) u8_ptr_add(ip61, vnet_buffer(p1)->map_t.v6.frag_offset); - - ip6_map_t_classify(p0, ip60, d0, &src_port0, &error0, &next0, l4_len0, frag0); - ip6_map_t_classify(p1, ip61, d1, &src_port1, &error1, &next1, l4_len1, frag1); - - if (PREDICT_FALSE((src_port0 != -1) && ( - ip60->src_address.as_u64[0] != map_get_pfx_net(d0, vnet_buffer(p0)->map_t.v6.saddr, src_port0) || - ip60->src_address.as_u64[1] != map_get_sfx_net(d0, vnet_buffer(p0)->map_t.v6.saddr, src_port0)))) { - error0 = MAP_ERROR_SEC_CHECK; - } - - if (PREDICT_FALSE((src_port1 != -1) && ( - ip61->src_address.as_u64[0] != map_get_pfx_net(d1, vnet_buffer(p1)->map_t.v6.saddr, src_port1) || - ip61->src_address.as_u64[1] != map_get_sfx_net(d1, vnet_buffer(p1)->map_t.v6.saddr, src_port1)))) { - error1 = MAP_ERROR_SEC_CHECK; - } - - if (PREDICT_FALSE(vnet_buffer(p0)->map_t.v6.frag_offset && - !ip6_frag_hdr_offset((ip6_frag_hdr_t *) - u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.frag_offset))) && - (src_port0 != -1) && (d0->ea_bits_len != 0 || !d0->rules) && (error0 == MAP_ERROR_NONE)) { - ip6_map_fragment_cache(ip60, - (ip6_frag_hdr_t *)u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.frag_offset), - d0, src_port0); - } - - if (PREDICT_FALSE(vnet_buffer(p1)->map_t.v6.frag_offset && - !ip6_frag_hdr_offset((ip6_frag_hdr_t *) - u8_ptr_add(ip61, vnet_buffer(p1)->map_t.v6.frag_offset))) && - (src_port1 != -1) && (d1->ea_bits_len != 0 || !d1->rules) && (error1 == MAP_ERROR_NONE)) { - ip6_map_fragment_cache(ip61, - (ip6_frag_hdr_t *)u8_ptr_add(ip61, vnet_buffer(p1)->map_t.v6.frag_offset), - d1, src_port1); - } - - if (PREDICT_TRUE(error0 == MAP_ERROR_NONE && next0 != IP6_MAPT_NEXT_MAPT_ICMP)) { - vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_RX, cpu_index, - vnet_buffer(p0)->map_t.map_domain_index, 1, - clib_net_to_host_u16(ip60->payload_length)); - } - - if (PREDICT_TRUE(error1 == MAP_ERROR_NONE && next1 != IP6_MAPT_NEXT_MAPT_ICMP)) { - vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_RX, cpu_index, - vnet_buffer(p1)->map_t.map_domain_index, 1, - clib_net_to_host_u16(ip61->payload_length)); - } - - next0 = (error0 != MAP_ERROR_NONE) ? IP6_MAPT_NEXT_DROP : next0; - next1 = (error1 != MAP_ERROR_NONE) ? IP6_MAPT_NEXT_DROP : next1; - p0->error = error_node->errors[error0]; - p1->error = error_node->errors[error1]; - vlib_validate_buffer_enqueue_x2(vm, node, next_index, to_next, n_left_to_next, pi0, pi1, next0, next1); - } + while (n_left_from >= 4 && n_left_to_next >= 2) + { + u32 pi0, pi1; + vlib_buffer_t *p0, *p1; + ip6_header_t *ip60, *ip61; + u8 error0, error1; + ip6_mapt_next_t next0, next1; + u32 l4_len0, l4_len1; + i32 src_port0, src_port1; + map_domain_t *d0, *d1; + ip6_frag_hdr_t *frag0, *frag1; + u32 saddr0, saddr1; + next0 = next1 = 0; //Because compiler whines + + pi0 = to_next[0] = from[0]; + pi1 = to_next[1] = from[1]; + from += 2; + n_left_from -= 2; + to_next += 2; + n_left_to_next -= 2; + + error0 = MAP_ERROR_NONE; + error1 = MAP_ERROR_NONE; + + p0 = vlib_get_buffer (vm, pi0); + p1 = vlib_get_buffer (vm, pi1); + ip60 = vlib_buffer_get_current (p0); + ip61 = vlib_buffer_get_current (p1); + + saddr0 = map_get_ip4 (&ip60->src_address); + saddr1 = map_get_ip4 (&ip61->src_address); + d0 = ip6_map_get_domain (vnet_buffer (p0)->ip.adj_index[VLIB_TX], + (ip4_address_t *) & saddr0, + &vnet_buffer (p0)->map_t.map_domain_index, + &error0); + d1 = + ip6_map_get_domain (vnet_buffer (p1)->ip.adj_index[VLIB_TX], + (ip4_address_t *) & saddr1, + &vnet_buffer (p1)->map_t.map_domain_index, + &error1); + + vnet_buffer (p0)->map_t.v6.saddr = saddr0; + vnet_buffer (p1)->map_t.v6.saddr = saddr1; + vnet_buffer (p0)->map_t.v6.daddr = + ip6_map_t_embedded_address (d0, &ip60->dst_address); + vnet_buffer (p1)->map_t.v6.daddr = + ip6_map_t_embedded_address (d1, &ip61->dst_address); + vnet_buffer (p0)->map_t.mtu = d0->mtu ? d0->mtu : ~0; + vnet_buffer (p1)->map_t.mtu = d1->mtu ? d1->mtu : ~0; + + if (PREDICT_FALSE (ip6_parse (ip60, p0->current_length, + &(vnet_buffer (p0)->map_t. + v6.l4_protocol), + &(vnet_buffer (p0)->map_t. + v6.l4_offset), + &(vnet_buffer (p0)->map_t. + v6.frag_offset)))) + { + error0 = MAP_ERROR_MALFORMED; + next0 = IP6_MAPT_NEXT_DROP; + } + + if (PREDICT_FALSE (ip6_parse (ip61, p1->current_length, + &(vnet_buffer (p1)->map_t. + v6.l4_protocol), + &(vnet_buffer (p1)->map_t. + v6.l4_offset), + &(vnet_buffer (p1)->map_t. + v6.frag_offset)))) + { + error1 = MAP_ERROR_MALFORMED; + next1 = IP6_MAPT_NEXT_DROP; + } + + src_port0 = src_port1 = -1; + l4_len0 = (u32) clib_net_to_host_u16 (ip60->payload_length) + + sizeof (*ip60) - vnet_buffer (p0)->map_t.v6.l4_offset; + l4_len1 = (u32) clib_net_to_host_u16 (ip61->payload_length) + + sizeof (*ip60) - vnet_buffer (p1)->map_t.v6.l4_offset; + frag0 = + (ip6_frag_hdr_t *) u8_ptr_add (ip60, + vnet_buffer (p0)->map_t. + v6.frag_offset); + frag1 = + (ip6_frag_hdr_t *) u8_ptr_add (ip61, + vnet_buffer (p1)->map_t. + v6.frag_offset); + + ip6_map_t_classify (p0, ip60, d0, &src_port0, &error0, &next0, + l4_len0, frag0); + ip6_map_t_classify (p1, ip61, d1, &src_port1, &error1, &next1, + l4_len1, frag1); + + if (PREDICT_FALSE + ((src_port0 != -1) + && (ip60->src_address.as_u64[0] != + map_get_pfx_net (d0, vnet_buffer (p0)->map_t.v6.saddr, + src_port0) + || ip60->src_address.as_u64[1] != map_get_sfx_net (d0, + vnet_buffer + (p0)->map_t.v6.saddr, + src_port0)))) + { + error0 = MAP_ERROR_SEC_CHECK; + } + + if (PREDICT_FALSE + ((src_port1 != -1) + && (ip61->src_address.as_u64[0] != + map_get_pfx_net (d1, vnet_buffer (p1)->map_t.v6.saddr, + src_port1) + || ip61->src_address.as_u64[1] != map_get_sfx_net (d1, + vnet_buffer + (p1)->map_t.v6.saddr, + src_port1)))) + { + error1 = MAP_ERROR_SEC_CHECK; + } + + if (PREDICT_FALSE (vnet_buffer (p0)->map_t.v6.frag_offset && + !ip6_frag_hdr_offset ((ip6_frag_hdr_t *) + u8_ptr_add (ip60, + vnet_buffer + (p0)->map_t. + v6.frag_offset))) + && (src_port0 != -1) && (d0->ea_bits_len != 0 || !d0->rules) + && (error0 == MAP_ERROR_NONE)) + { + ip6_map_fragment_cache (ip60, + (ip6_frag_hdr_t *) u8_ptr_add (ip60, + vnet_buffer + (p0)->map_t. + v6.frag_offset), + d0, src_port0); + } + + if (PREDICT_FALSE (vnet_buffer (p1)->map_t.v6.frag_offset && + !ip6_frag_hdr_offset ((ip6_frag_hdr_t *) + u8_ptr_add (ip61, + vnet_buffer + (p1)->map_t. + v6.frag_offset))) + && (src_port1 != -1) && (d1->ea_bits_len != 0 || !d1->rules) + && (error1 == MAP_ERROR_NONE)) + { + ip6_map_fragment_cache (ip61, + (ip6_frag_hdr_t *) u8_ptr_add (ip61, + vnet_buffer + (p1)->map_t. + v6.frag_offset), + d1, src_port1); + } + + if (PREDICT_TRUE + (error0 == MAP_ERROR_NONE && next0 != IP6_MAPT_NEXT_MAPT_ICMP)) + { + vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_RX, + cpu_index, + vnet_buffer (p0)-> + map_t.map_domain_index, 1, + clib_net_to_host_u16 + (ip60->payload_length)); + } + + if (PREDICT_TRUE + (error1 == MAP_ERROR_NONE && next1 != IP6_MAPT_NEXT_MAPT_ICMP)) + { + vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_RX, + cpu_index, + vnet_buffer (p1)-> + map_t.map_domain_index, 1, + clib_net_to_host_u16 + (ip61->payload_length)); + } + + next0 = (error0 != MAP_ERROR_NONE) ? IP6_MAPT_NEXT_DROP : next0; + next1 = (error1 != MAP_ERROR_NONE) ? IP6_MAPT_NEXT_DROP : next1; + p0->error = error_node->errors[error0]; + p1->error = error_node->errors[error1]; + vlib_validate_buffer_enqueue_x2 (vm, node, next_index, to_next, + n_left_to_next, pi0, pi1, next0, + next1); + } #endif - while (n_left_from > 0 && n_left_to_next > 0) { - u32 pi0; - vlib_buffer_t *p0; - ip6_header_t *ip60; - u8 error0; - u32 l4_len0; - i32 src_port0; - map_domain_t *d0; - ip6_frag_hdr_t *frag0; - ip6_mapt_next_t next0 = 0; - u32 saddr; - - pi0 = to_next[0] = from[0]; - from += 1; - n_left_from -= 1; - to_next +=1; - n_left_to_next -= 1; - error0 = MAP_ERROR_NONE; - - p0 = vlib_get_buffer(vm, pi0); - ip60 = vlib_buffer_get_current(p0); - //Save saddr in a different variable to not overwrite ip.adj_index - saddr = map_get_ip4(&ip60->src_address); - d0 = ip6_map_get_domain(vnet_buffer(p0)->ip.adj_index[VLIB_TX], - (ip4_address_t *)&saddr, - &vnet_buffer(p0)->map_t.map_domain_index, &error0); - - //FIXME: What if d0 is null - vnet_buffer(p0)->map_t.v6.saddr = saddr; - vnet_buffer(p0)->map_t.v6.daddr = ip6_map_t_embedded_address(d0, &ip60->dst_address); - vnet_buffer(p0)->map_t.mtu = d0->mtu ? d0->mtu : ~0; - - if (PREDICT_FALSE(ip6_parse(ip60, p0->current_length, - &(vnet_buffer(p0)->map_t.v6.l4_protocol), - &(vnet_buffer(p0)->map_t.v6.l4_offset), - &(vnet_buffer(p0)->map_t.v6.frag_offset)))) { - error0 = MAP_ERROR_MALFORMED; - next0 = IP6_MAPT_NEXT_DROP; - } - - src_port0 = -1; - l4_len0 = (u32)clib_net_to_host_u16(ip60->payload_length) + - sizeof(*ip60) - vnet_buffer(p0)->map_t.v6.l4_offset; - frag0 = (ip6_frag_hdr_t *) u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.frag_offset); - - - if (PREDICT_FALSE(vnet_buffer(p0)->map_t.v6.frag_offset && - ip6_frag_hdr_offset(frag0))) { - src_port0 = ip6_map_fragment_get(ip60, frag0, d0); - error0 = (src_port0 != -1) ? error0 : MAP_ERROR_FRAGMENT_MEMORY; - next0 = IP6_MAPT_NEXT_MAPT_FRAGMENTED; - } else if (PREDICT_TRUE(vnet_buffer(p0)->map_t.v6.l4_protocol == IP_PROTOCOL_TCP)) { - error0 = l4_len0 < sizeof(tcp_header_t) ? MAP_ERROR_MALFORMED : error0; - vnet_buffer(p0)->map_t.checksum_offset = vnet_buffer(p0)->map_t.v6.l4_offset + 16; - next0 = IP6_MAPT_NEXT_MAPT_TCP_UDP; - src_port0 = (i32) *((u16*)u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.l4_offset)); - } else if (PREDICT_TRUE(vnet_buffer(p0)->map_t.v6.l4_protocol == IP_PROTOCOL_UDP)) { - error0 = l4_len0 < sizeof(udp_header_t) ? MAP_ERROR_MALFORMED : error0; - vnet_buffer(p0)->map_t.checksum_offset = vnet_buffer(p0)->map_t.v6.l4_offset + 6; - next0 = IP6_MAPT_NEXT_MAPT_TCP_UDP; - src_port0 = (i32) *((u16*)u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.l4_offset)); - } else if (vnet_buffer(p0)->map_t.v6.l4_protocol == IP_PROTOCOL_ICMP6) { - error0 = l4_len0 < sizeof(icmp46_header_t) ? MAP_ERROR_MALFORMED : error0; - next0 = IP6_MAPT_NEXT_MAPT_ICMP; - if (((icmp46_header_t *) u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.l4_offset))->code == ICMP6_echo_reply || - ((icmp46_header_t *) u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.l4_offset))->code == ICMP6_echo_request) - src_port0 = (i32) *((u16 *)u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.l4_offset + 6)); - } else { - //TODO: In case of 1:1 mapping, it might be possible to do something with those packets. - error0 = MAP_ERROR_BAD_PROTOCOL; - } - - //Security check - if (PREDICT_FALSE((src_port0 != -1) && ( - ip60->src_address.as_u64[0] != map_get_pfx_net(d0, vnet_buffer(p0)->map_t.v6.saddr, src_port0) || - ip60->src_address.as_u64[1] != map_get_sfx_net(d0, vnet_buffer(p0)->map_t.v6.saddr, src_port0)))) { - //Security check when src_port0 is not zero (non-first fragment, UDP or TCP) - error0 = MAP_ERROR_SEC_CHECK; - } - - //Fragmented first packet needs to be cached for following packets - if (PREDICT_FALSE(vnet_buffer(p0)->map_t.v6.frag_offset && - !ip6_frag_hdr_offset((ip6_frag_hdr_t *) - u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.frag_offset))) && - (src_port0 != -1) && (d0->ea_bits_len != 0 || !d0->rules) && (error0 == MAP_ERROR_NONE)) { - ip6_map_fragment_cache(ip60, - (ip6_frag_hdr_t *)u8_ptr_add(ip60, vnet_buffer(p0)->map_t.v6.frag_offset), - d0, src_port0); - } - - if (PREDICT_TRUE(error0 == MAP_ERROR_NONE && next0 != IP6_MAPT_NEXT_MAPT_ICMP)) { - vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_RX, cpu_index, - vnet_buffer(p0)->map_t.map_domain_index, 1, - clib_net_to_host_u16(ip60->payload_length)); - } - - next0 = (error0 != MAP_ERROR_NONE) ? IP6_MAPT_NEXT_DROP : next0; - p0->error = error_node->errors[error0]; - vlib_validate_buffer_enqueue_x1(vm, node, next_index, - to_next, n_left_to_next, pi0, - next0); + while (n_left_from > 0 && n_left_to_next > 0) + { + u32 pi0; + vlib_buffer_t *p0; + ip6_header_t *ip60; + u8 error0; + u32 l4_len0; + i32 src_port0; + map_domain_t *d0; + ip6_frag_hdr_t *frag0; + ip6_mapt_next_t next0 = 0; + u32 saddr; + + pi0 = to_next[0] = from[0]; + from += 1; + n_left_from -= 1; + to_next += 1; + n_left_to_next -= 1; + error0 = MAP_ERROR_NONE; + + p0 = vlib_get_buffer (vm, pi0); + ip60 = vlib_buffer_get_current (p0); + //Save saddr in a different variable to not overwrite ip.adj_index + saddr = map_get_ip4 (&ip60->src_address); + d0 = ip6_map_get_domain (vnet_buffer (p0)->ip.adj_index[VLIB_TX], + (ip4_address_t *) & saddr, + &vnet_buffer (p0)->map_t.map_domain_index, + &error0); + + //FIXME: What if d0 is null + vnet_buffer (p0)->map_t.v6.saddr = saddr; + vnet_buffer (p0)->map_t.v6.daddr = + ip6_map_t_embedded_address (d0, &ip60->dst_address); + vnet_buffer (p0)->map_t.mtu = d0->mtu ? d0->mtu : ~0; + + if (PREDICT_FALSE (ip6_parse (ip60, p0->current_length, + &(vnet_buffer (p0)->map_t. + v6.l4_protocol), + &(vnet_buffer (p0)->map_t. + v6.l4_offset), + &(vnet_buffer (p0)->map_t. + v6.frag_offset)))) + { + error0 = MAP_ERROR_MALFORMED; + next0 = IP6_MAPT_NEXT_DROP; + } + + src_port0 = -1; + l4_len0 = (u32) clib_net_to_host_u16 (ip60->payload_length) + + sizeof (*ip60) - vnet_buffer (p0)->map_t.v6.l4_offset; + frag0 = + (ip6_frag_hdr_t *) u8_ptr_add (ip60, + vnet_buffer (p0)->map_t. + v6.frag_offset); + + + if (PREDICT_FALSE (vnet_buffer (p0)->map_t.v6.frag_offset && + ip6_frag_hdr_offset (frag0))) + { + src_port0 = ip6_map_fragment_get (ip60, frag0, d0); + error0 = (src_port0 != -1) ? error0 : MAP_ERROR_FRAGMENT_MEMORY; + next0 = IP6_MAPT_NEXT_MAPT_FRAGMENTED; + } + else + if (PREDICT_TRUE + (vnet_buffer (p0)->map_t.v6.l4_protocol == IP_PROTOCOL_TCP)) + { + error0 = + l4_len0 < + sizeof (tcp_header_t) ? MAP_ERROR_MALFORMED : error0; + vnet_buffer (p0)->map_t.checksum_offset = + vnet_buffer (p0)->map_t.v6.l4_offset + 16; + next0 = IP6_MAPT_NEXT_MAPT_TCP_UDP; + src_port0 = + (i32) * + ((u16 *) + u8_ptr_add (ip60, vnet_buffer (p0)->map_t.v6.l4_offset)); + } + else + if (PREDICT_TRUE + (vnet_buffer (p0)->map_t.v6.l4_protocol == IP_PROTOCOL_UDP)) + { + error0 = + l4_len0 < + sizeof (udp_header_t) ? MAP_ERROR_MALFORMED : error0; + vnet_buffer (p0)->map_t.checksum_offset = + vnet_buffer (p0)->map_t.v6.l4_offset + 6; + next0 = IP6_MAPT_NEXT_MAPT_TCP_UDP; + src_port0 = + (i32) * + ((u16 *) + u8_ptr_add (ip60, vnet_buffer (p0)->map_t.v6.l4_offset)); + } + else if (vnet_buffer (p0)->map_t.v6.l4_protocol == + IP_PROTOCOL_ICMP6) + { + error0 = + l4_len0 < + sizeof (icmp46_header_t) ? MAP_ERROR_MALFORMED : error0; + next0 = IP6_MAPT_NEXT_MAPT_ICMP; + if (((icmp46_header_t *) + u8_ptr_add (ip60, + vnet_buffer (p0)->map_t.v6.l4_offset))->code == + ICMP6_echo_reply + || ((icmp46_header_t *) + u8_ptr_add (ip60, + vnet_buffer (p0)->map_t.v6. + l4_offset))->code == ICMP6_echo_request) + src_port0 = + (i32) * + ((u16 *) + u8_ptr_add (ip60, + vnet_buffer (p0)->map_t.v6.l4_offset + 6)); + } + else + { + //TODO: In case of 1:1 mapping, it might be possible to do something with those packets. + error0 = MAP_ERROR_BAD_PROTOCOL; + } + + //Security check + if (PREDICT_FALSE + ((src_port0 != -1) + && (ip60->src_address.as_u64[0] != + map_get_pfx_net (d0, vnet_buffer (p0)->map_t.v6.saddr, + src_port0) + || ip60->src_address.as_u64[1] != map_get_sfx_net (d0, + vnet_buffer + (p0)->map_t.v6.saddr, + src_port0)))) + { + //Security check when src_port0 is not zero (non-first fragment, UDP or TCP) + error0 = MAP_ERROR_SEC_CHECK; + } + + //Fragmented first packet needs to be cached for following packets + if (PREDICT_FALSE (vnet_buffer (p0)->map_t.v6.frag_offset && + !ip6_frag_hdr_offset ((ip6_frag_hdr_t *) + u8_ptr_add (ip60, + vnet_buffer + (p0)->map_t. + v6.frag_offset))) + && (src_port0 != -1) && (d0->ea_bits_len != 0 || !d0->rules) + && (error0 == MAP_ERROR_NONE)) + { + ip6_map_fragment_cache (ip60, + (ip6_frag_hdr_t *) u8_ptr_add (ip60, + vnet_buffer + (p0)->map_t. + v6.frag_offset), + d0, src_port0); + } + + if (PREDICT_TRUE + (error0 == MAP_ERROR_NONE && next0 != IP6_MAPT_NEXT_MAPT_ICMP)) + { + vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_RX, + cpu_index, + vnet_buffer (p0)-> + map_t.map_domain_index, 1, + clib_net_to_host_u16 + (ip60->payload_length)); + } + + next0 = (error0 != MAP_ERROR_NONE) ? IP6_MAPT_NEXT_DROP : next0; + p0->error = error_node->errors[error0]; + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, + to_next, n_left_to_next, pi0, + next0); + } + vlib_put_next_frame (vm, node, next_index, n_left_to_next); } - vlib_put_next_frame(vm, node, next_index, n_left_to_next); - } return frame->n_vectors; } static char *map_t_error_strings[] = { #define _(sym,string) string, - foreach_map_error + foreach_map_error #undef _ }; +/* *INDENT-OFF* */ VLIB_REGISTER_NODE(ip6_map_t_fragmented_node) = { .function = ip6_map_t_fragmented, .name = "ip6-map-t-fragmented", @@ -1084,7 +1445,9 @@ VLIB_REGISTER_NODE(ip6_map_t_fragmented_node) = { [IP6_MAPT_FRAGMENTED_NEXT_DROP] = "error-drop", }, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_REGISTER_NODE(ip6_map_t_icmp_node) = { .function = ip6_map_t_icmp, .name = "ip6-map-t-icmp", @@ -1102,7 +1465,9 @@ VLIB_REGISTER_NODE(ip6_map_t_icmp_node) = { [IP6_MAPT_ICMP_NEXT_DROP] = "error-drop", }, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_REGISTER_NODE(ip6_map_t_tcp_udp_node) = { .function = ip6_map_t_tcp_udp, .name = "ip6-map-t-tcp-udp", @@ -1120,7 +1485,9 @@ VLIB_REGISTER_NODE(ip6_map_t_tcp_udp_node) = { [IP6_MAPT_TCP_UDP_NEXT_DROP] = "error-drop", }, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_REGISTER_NODE(ip6_map_t_node) = { .function = ip6_map_t, .name = "ip6-map-t", @@ -1139,3 +1506,12 @@ VLIB_REGISTER_NODE(ip6_map_t_node) = { [IP6_MAPT_NEXT_DROP] = "error-drop", }, }; +/* *INDENT-ON* */ + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/vnet/vnet/map/map.c b/vnet/vnet/map/map.c index f91a8718aff..cd327777404 100644 --- a/vnet/vnet/map/map.c +++ b/vnet/vnet/map/map.c @@ -21,10 +21,10 @@ #include <vppinfra/xxhash.h> static inline u32 -crc_u32(u32 data, u32 value) +crc_u32 (u32 data, u32 value) { - u64 tmp = ((u64)data<<32) | (u64) value; - return (u32) clib_xxhash(tmp); + u64 tmp = ((u64) data << 32) | (u64) value; + return (u32) clib_xxhash (tmp); } #endif @@ -63,86 +63,100 @@ crc_u32(u32 data, u32 value) i32 -ip4_get_port (ip4_header_t *ip, map_dir_e dir, u16 buffer_len) +ip4_get_port (ip4_header_t * ip, map_dir_e dir, u16 buffer_len) { //TODO: use buffer length if (ip->ip_version_and_header_length != 0x45 || - ip4_get_fragment_offset(ip)) - return -1; + ip4_get_fragment_offset (ip)) + return -1; - if (PREDICT_TRUE((ip->protocol == IP_PROTOCOL_TCP) || - (ip->protocol == IP_PROTOCOL_UDP))) { - udp_header_t *udp = (void *)(ip + 1); - return (dir == MAP_SENDER) ? udp->src_port : udp->dst_port; - } else if (ip->protocol == IP_PROTOCOL_ICMP) { - icmp46_header_t *icmp = (void *)(ip + 1); - if (icmp->type == ICMP4_echo_request || - icmp->type == ICMP4_echo_reply) { - return *((u16 *)(icmp + 1)); - } else if (clib_net_to_host_u16(ip->length) >= 64) { - ip = (ip4_header_t *)(icmp + 2); - if (PREDICT_TRUE((ip->protocol == IP_PROTOCOL_TCP) || - (ip->protocol == IP_PROTOCOL_UDP))) { - udp_header_t *udp = (void *)(ip + 1); - return (dir == MAP_SENDER) ? udp->dst_port : udp->src_port; - } else if (ip->protocol == IP_PROTOCOL_ICMP) { - icmp46_header_t *icmp = (void *)(ip + 1); - if (icmp->type == ICMP4_echo_request || - icmp->type == ICMP4_echo_reply) { - return *((u16 *)(icmp + 1)); - } - } + if (PREDICT_TRUE ((ip->protocol == IP_PROTOCOL_TCP) || + (ip->protocol == IP_PROTOCOL_UDP))) + { + udp_header_t *udp = (void *) (ip + 1); + return (dir == MAP_SENDER) ? udp->src_port : udp->dst_port; + } + else if (ip->protocol == IP_PROTOCOL_ICMP) + { + icmp46_header_t *icmp = (void *) (ip + 1); + if (icmp->type == ICMP4_echo_request || icmp->type == ICMP4_echo_reply) + { + return *((u16 *) (icmp + 1)); + } + else if (clib_net_to_host_u16 (ip->length) >= 64) + { + ip = (ip4_header_t *) (icmp + 2); + if (PREDICT_TRUE ((ip->protocol == IP_PROTOCOL_TCP) || + (ip->protocol == IP_PROTOCOL_UDP))) + { + udp_header_t *udp = (void *) (ip + 1); + return (dir == MAP_SENDER) ? udp->dst_port : udp->src_port; + } + else if (ip->protocol == IP_PROTOCOL_ICMP) + { + icmp46_header_t *icmp = (void *) (ip + 1); + if (icmp->type == ICMP4_echo_request || + icmp->type == ICMP4_echo_reply) + { + return *((u16 *) (icmp + 1)); + } + } + } } - } return -1; } i32 -ip6_get_port (ip6_header_t *ip6, map_dir_e dir, u16 buffer_len) +ip6_get_port (ip6_header_t * ip6, map_dir_e dir, u16 buffer_len) { u8 l4_protocol; u16 l4_offset; u16 frag_offset; u8 *l4; - if (ip6_parse(ip6, buffer_len, &l4_protocol, &l4_offset, &frag_offset)) + if (ip6_parse (ip6, buffer_len, &l4_protocol, &l4_offset, &frag_offset)) return -1; //TODO: Use buffer length if (frag_offset && - ip6_frag_hdr_offset(((ip6_frag_hdr_t *)u8_ptr_add(ip6, frag_offset)))) - return -1; //Can't deal with non-first fragment for now - - l4 = u8_ptr_add(ip6, l4_offset); - if (l4_protocol == IP_PROTOCOL_TCP || - l4_protocol == IP_PROTOCOL_UDP) { - return (dir == MAP_SENDER) ? ((udp_header_t *)(l4))->src_port : ((udp_header_t *)(l4))->dst_port; - } else if (l4_protocol == IP_PROTOCOL_ICMP6) { - icmp46_header_t *icmp = (icmp46_header_t *)(l4); - if (icmp->type == ICMP6_echo_request) { - return (dir == MAP_SENDER) ? ((u16*)(icmp))[2] : -1; - } else if (icmp->type == ICMP6_echo_reply) { - return (dir == MAP_SENDER) ? -1 : ((u16*)(icmp))[2]; + ip6_frag_hdr_offset (((ip6_frag_hdr_t *) + u8_ptr_add (ip6, frag_offset)))) + return -1; //Can't deal with non-first fragment for now + + l4 = u8_ptr_add (ip6, l4_offset); + if (l4_protocol == IP_PROTOCOL_TCP || l4_protocol == IP_PROTOCOL_UDP) + { + return (dir == + MAP_SENDER) ? ((udp_header_t *) (l4))-> + src_port : ((udp_header_t *) (l4))->dst_port; + } + else if (l4_protocol == IP_PROTOCOL_ICMP6) + { + icmp46_header_t *icmp = (icmp46_header_t *) (l4); + if (icmp->type == ICMP6_echo_request) + { + return (dir == MAP_SENDER) ? ((u16 *) (icmp))[2] : -1; + } + else if (icmp->type == ICMP6_echo_reply) + { + return (dir == MAP_SENDER) ? -1 : ((u16 *) (icmp))[2]; + } } - } return -1; } int -map_create_domain (ip4_address_t *ip4_prefix, - u8 ip4_prefix_len, - ip6_address_t *ip6_prefix, - u8 ip6_prefix_len, - ip6_address_t *ip6_src, - u8 ip6_src_len, - u8 ea_bits_len, - u8 psid_offset, - u8 psid_length, - u32 *map_domain_index, - u16 mtu, - u8 flags) +map_create_domain (ip4_address_t * ip4_prefix, + u8 ip4_prefix_len, + ip6_address_t * ip6_prefix, + u8 ip6_prefix_len, + ip6_address_t * ip6_src, + u8 ip6_src_len, + u8 ea_bits_len, + u8 psid_offset, + u8 psid_length, u32 * map_domain_index, u16 mtu, u8 flags) { map_main_t *mm = &map_main; ip4_main_t *im4 = &ip4_main; @@ -159,21 +173,27 @@ map_create_domain (ip4_address_t *ip4_prefix, return -1; /* Sanity check on the src prefix length */ - if (flags & MAP_DOMAIN_TRANSLATION) { - if (ip6_src_len != 96) { - clib_warning("MAP-T only supports ip6_src_len = 96 for now."); + if (flags & MAP_DOMAIN_TRANSLATION) + { + if (ip6_src_len != 96) + { + clib_warning ("MAP-T only supports ip6_src_len = 96 for now."); return -1; - } - } else { - if (ip6_src_len != 128) { - clib_warning("MAP-E requires a BR address, not a prefix (ip6_src_len should be 128)."); + } + } + else + { + if (ip6_src_len != 128) + { + clib_warning + ("MAP-E requires a BR address, not a prefix (ip6_src_len should be 128)."); return -1; - } - } + } + } /* Get domain index */ - pool_get_aligned(mm->domains, d, CLIB_CACHE_LINE_BYTES); - memset(d, 0, sizeof (*d)); + pool_get_aligned (mm->domains, d, CLIB_CACHE_LINE_BYTES); + memset (d, 0, sizeof (*d)); *map_domain_index = d - mm->domains; /* Init domain struct */ @@ -190,34 +210,42 @@ map_create_domain (ip4_address_t *ip4_prefix, d->flags = flags; /* How many, and which bits to grab from the IPv4 DA */ - if (ip4_prefix_len + ea_bits_len < 32) { - d->flags |= MAP_DOMAIN_PREFIX; - suffix_len = d->suffix_shift = 32 - ip4_prefix_len - ea_bits_len; - } else { - d->suffix_shift = 0; - suffix_len = 32 - ip4_prefix_len; - } - d->suffix_mask = (1<<suffix_len) - 1; + if (ip4_prefix_len + ea_bits_len < 32) + { + d->flags |= MAP_DOMAIN_PREFIX; + suffix_len = d->suffix_shift = 32 - ip4_prefix_len - ea_bits_len; + } + else + { + d->suffix_shift = 0; + suffix_len = 32 - ip4_prefix_len; + } + d->suffix_mask = (1 << suffix_len) - 1; d->psid_shift = 16 - psid_length - psid_offset; d->psid_mask = (1 << d->psid_length) - 1; d->ea_shift = 64 - ip6_prefix_len - suffix_len - d->psid_length; /* Init IP adjacency */ - memset(&adj, 0, sizeof(adj)); + memset (&adj, 0, sizeof (adj)); adj.explicit_fib_index = ~0; - adj.lookup_next_index = (d->flags & MAP_DOMAIN_TRANSLATION) ? IP_LOOKUP_NEXT_MAP_T : IP_LOOKUP_NEXT_MAP; - p = (uword *)&adj.rewrite_data[0]; + adj.lookup_next_index = + (d-> + flags & MAP_DOMAIN_TRANSLATION) ? IP_LOOKUP_NEXT_MAP_T : + IP_LOOKUP_NEXT_MAP; + p = (uword *) & adj.rewrite_data[0]; *p = (uword) (*map_domain_index); - if (ip4_get_route(im4, 0, 0, (u8 *)ip4_prefix, ip4_prefix_len)) { - clib_warning("IPv4 route already defined: %U/%d", format_ip4_address, ip4_prefix, ip4_prefix_len); - pool_put(mm->domains, d); - return -1; - } - + if (ip4_get_route (im4, 0, 0, (u8 *) ip4_prefix, ip4_prefix_len)) + { + clib_warning ("IPv4 route already defined: %U/%d", format_ip4_address, + ip4_prefix, ip4_prefix_len); + pool_put (mm->domains, d); + return -1; + } + /* Create ip4 adjacency */ - memset(&args4, 0, sizeof(args4)); + memset (&args4, 0, sizeof (args4)); args4.table_index_or_table_id = 0; args4.flags = IP4_ROUTE_FLAG_ADD; args4.dst_address.as_u32 = ip4_prefix->as_u32; @@ -226,51 +254,61 @@ map_create_domain (ip4_address_t *ip4_prefix, args4.adj_index = ~0; args4.add_adj = &adj; args4.n_add_adj = 1; - ip4_add_del_route(im4, &args4); + ip4_add_del_route (im4, &args4); /* Multiple MAP domains may share same source IPv6 TEP */ - u32 ai = ip6_get_route(im6, 0, 0, ip6_src, ip6_src_len); - if (ai > 0) { - ip_lookup_main_t *lm6 = &ip6_main.lookup_main; - ip_adjacency_t *adj6 = ip_get_adjacency(lm6, ai); - if (adj6->lookup_next_index != IP_LOOKUP_NEXT_MAP && - adj6->lookup_next_index != IP_LOOKUP_NEXT_MAP_T) { - clib_warning("BR source address already assigned: %U", format_ip6_address, ip6_src); - pool_put(mm->domains, d); - return -1; + u32 ai = ip6_get_route (im6, 0, 0, ip6_src, ip6_src_len); + if (ai > 0) + { + ip_lookup_main_t *lm6 = &ip6_main.lookup_main; + ip_adjacency_t *adj6 = ip_get_adjacency (lm6, ai); + if (adj6->lookup_next_index != IP_LOOKUP_NEXT_MAP && + adj6->lookup_next_index != IP_LOOKUP_NEXT_MAP_T) + { + clib_warning ("BR source address already assigned: %U", + format_ip6_address, ip6_src); + pool_put (mm->domains, d); + return -1; + } + /* Shared source */ + p = (uword *) & adj6->rewrite_data[0]; + p[0] = ~0; + + /* Add refcount, so we don't accidentially delete the route underneath someone */ + p[1]++; + } + else + { + /* Create ip6 adjacency. */ + memset (&args6, 0, sizeof (args6)); + args6.table_index_or_table_id = 0; + args6.flags = IP6_ROUTE_FLAG_ADD; + args6.dst_address.as_u64[0] = ip6_src->as_u64[0]; + args6.dst_address.as_u64[1] = ip6_src->as_u64[1]; + args6.dst_address_length = ip6_src_len; + args6.adj_index = ~0; + args6.add_adj = &adj; + args6.n_add_adj = 1; + ip6_add_del_route (im6, &args6); } - /* Shared source */ - p = (uword *)&adj6->rewrite_data[0]; - p[0] = ~0; - - /* Add refcount, so we don't accidentially delete the route underneath someone */ - p[1]++; - } else { - /* Create ip6 adjacency. */ - memset(&args6, 0, sizeof(args6)); - args6.table_index_or_table_id = 0; - args6.flags = IP6_ROUTE_FLAG_ADD; - args6.dst_address.as_u64[0] = ip6_src->as_u64[0]; - args6.dst_address.as_u64[1] = ip6_src->as_u64[1]; - args6.dst_address_length = ip6_src_len; - args6.adj_index = ~0; - args6.add_adj = &adj; - args6.n_add_adj = 1; - ip6_add_del_route(im6, &args6); - } /* Validate packet/byte counters */ - map_domain_counter_lock(mm); + map_domain_counter_lock (mm); int i; - for (i = 0; i < vec_len(mm->simple_domain_counters); i++) { - vlib_validate_simple_counter(&mm->simple_domain_counters[i], *map_domain_index); - vlib_zero_simple_counter(&mm->simple_domain_counters[i], *map_domain_index); - } - for (i = 0; i < vec_len(mm->domain_counters); i++) { - vlib_validate_combined_counter(&mm->domain_counters[i], *map_domain_index); - vlib_zero_combined_counter(&mm->domain_counters[i], *map_domain_index); - } - map_domain_counter_unlock(mm); + for (i = 0; i < vec_len (mm->simple_domain_counters); i++) + { + vlib_validate_simple_counter (&mm->simple_domain_counters[i], + *map_domain_index); + vlib_zero_simple_counter (&mm->simple_domain_counters[i], + *map_domain_index); + } + for (i = 0; i < vec_len (mm->domain_counters); i++) + { + vlib_validate_combined_counter (&mm->domain_counters[i], + *map_domain_index); + vlib_zero_combined_counter (&mm->domain_counters[i], *map_domain_index); + } + map_domain_counter_unlock (mm); return 0; } @@ -289,19 +327,24 @@ map_delete_domain (u32 map_domain_index) ip4_add_del_route_args_t args4; ip6_add_del_route_args_t args6; - if (pool_is_free_index(mm->domains, map_domain_index)) { - clib_warning("MAP domain delete: domain does not exist: %d", map_domain_index); - return -1; - } + if (pool_is_free_index (mm->domains, map_domain_index)) + { + clib_warning ("MAP domain delete: domain does not exist: %d", + map_domain_index); + return -1; + } - d = pool_elt_at_index(mm->domains, map_domain_index); + d = pool_elt_at_index (mm->domains, map_domain_index); - memset(&adj, 0, sizeof(adj)); + memset (&adj, 0, sizeof (adj)); adj.explicit_fib_index = ~0; - adj.lookup_next_index = (d->flags & MAP_DOMAIN_TRANSLATION) ? IP_LOOKUP_NEXT_MAP_T : IP_LOOKUP_NEXT_MAP; + adj.lookup_next_index = + (d-> + flags & MAP_DOMAIN_TRANSLATION) ? IP_LOOKUP_NEXT_MAP_T : + IP_LOOKUP_NEXT_MAP; /* Delete ip4 adjacency */ - memset(&args4, 0, sizeof(args4)); + memset (&args4, 0, sizeof (args4)); args4.table_index_or_table_id = 0; args4.flags = IP4_ROUTE_FLAG_DEL; args4.dst_address.as_u32 = d->ip4_prefix.as_u32; @@ -309,147 +352,160 @@ map_delete_domain (u32 map_domain_index) args4.adj_index = 0; args4.add_adj = &adj; args4.n_add_adj = 0; - ip4_add_del_route(im4, &args4); + ip4_add_del_route (im4, &args4); /* Delete ip6 adjacency */ - u32 ai = ip6_get_route(im6, 0, 0, &d->ip6_src, d->ip6_src_len); - if (ai > 0) { - ip_lookup_main_t *lm6 = &ip6_main.lookup_main; - ip_adjacency_t *adj6 = ip_get_adjacency(lm6, ai); - - uword *p = (uword *)&adj6->rewrite_data[0]; - /* Delete route when no other domains use this source */ - if (p[1] == 0) { - memset(&args6, 0, sizeof (args6)); - args6.table_index_or_table_id = 0; - args6.flags = IP6_ROUTE_FLAG_DEL; - args6.dst_address.as_u64[0] = d->ip6_src.as_u64[0]; - args6.dst_address.as_u64[1] = d->ip6_src.as_u64[1]; - args6.dst_address_length = d->ip6_src_len; - args6.adj_index = 0; - args6.add_adj = &adj; - args6.n_add_adj = 0; - ip6_add_del_route(im6, &args6); + u32 ai = ip6_get_route (im6, 0, 0, &d->ip6_src, d->ip6_src_len); + if (ai > 0) + { + ip_lookup_main_t *lm6 = &ip6_main.lookup_main; + ip_adjacency_t *adj6 = ip_get_adjacency (lm6, ai); + + uword *p = (uword *) & adj6->rewrite_data[0]; + /* Delete route when no other domains use this source */ + if (p[1] == 0) + { + memset (&args6, 0, sizeof (args6)); + args6.table_index_or_table_id = 0; + args6.flags = IP6_ROUTE_FLAG_DEL; + args6.dst_address.as_u64[0] = d->ip6_src.as_u64[0]; + args6.dst_address.as_u64[1] = d->ip6_src.as_u64[1]; + args6.dst_address_length = d->ip6_src_len; + args6.adj_index = 0; + args6.add_adj = &adj; + args6.n_add_adj = 0; + ip6_add_del_route (im6, &args6); + } + p[1]--; } - p[1]--; - } /* Deleting rules */ if (d->rules) - clib_mem_free(d->rules); + clib_mem_free (d->rules); - pool_put(mm->domains, d); + pool_put (mm->domains, d); return 0; } int -map_add_del_psid (u32 map_domain_index, u16 psid, ip6_address_t *tep, +map_add_del_psid (u32 map_domain_index, u16 psid, ip6_address_t * tep, u8 is_add) { map_domain_t *d; map_main_t *mm = &map_main; - if (pool_is_free_index(mm->domains, map_domain_index)) { - clib_warning("MAP rule: domain does not exist: %d", map_domain_index); - return -1; - } - d = pool_elt_at_index(mm->domains, map_domain_index); + if (pool_is_free_index (mm->domains, map_domain_index)) + { + clib_warning ("MAP rule: domain does not exist: %d", map_domain_index); + return -1; + } + d = pool_elt_at_index (mm->domains, map_domain_index); /* Rules are only used in 1:1 independent case */ if (d->ea_bits_len > 0) return (-1); - if (!d->rules) { - u32 l = (0x1 << d->psid_length) * sizeof(ip6_address_t); - d->rules = clib_mem_alloc_aligned(l, CLIB_CACHE_LINE_BYTES); - if (!d->rules) return -1; - memset(d->rules, 0, l); - } + if (!d->rules) + { + u32 l = (0x1 << d->psid_length) * sizeof (ip6_address_t); + d->rules = clib_mem_alloc_aligned (l, CLIB_CACHE_LINE_BYTES); + if (!d->rules) + return -1; + memset (d->rules, 0, l); + } - if (psid >= (0x1 << d->psid_length)) { - clib_warning("MAP rule: PSID outside bounds: %d [%d]", psid, 0x1 << d->psid_length); - return -1; - } + if (psid >= (0x1 << d->psid_length)) + { + clib_warning ("MAP rule: PSID outside bounds: %d [%d]", psid, + 0x1 << d->psid_length); + return -1; + } - if (is_add) { - d->rules[psid] = *tep; - } else { - memset(&d->rules[psid], 0, sizeof(ip6_address_t)); - } + if (is_add) + { + d->rules[psid] = *tep; + } + else + { + memset (&d->rules[psid], 0, sizeof (ip6_address_t)); + } return 0; } #ifdef MAP_SKIP_IP6_LOOKUP static void -map_pre_resolve (ip4_address_t *ip4, ip6_address_t *ip6) +map_pre_resolve (ip4_address_t * ip4, ip6_address_t * ip6) { map_main_t *mm = &map_main; ip4_main_t *im4 = &ip4_main; ip6_main_t *im6 = &ip6_main; - if (ip6->as_u64[0] != 0 || ip6->as_u64[1] != 0) { - mm->adj6_index = ip6_fib_lookup_with_table(im6, 0, ip6); - clib_warning("FIB lookup results in: %u", mm->adj6_index); - } - if (ip4->as_u32 != 0) { - mm->adj4_index = ip4_fib_lookup_with_table(im4, 0, ip4, 0); - clib_warning("FIB lookup results in: %u", mm->adj4_index); - } + if (ip6->as_u64[0] != 0 || ip6->as_u64[1] != 0) + { + mm->adj6_index = ip6_fib_lookup_with_table (im6, 0, ip6); + clib_warning ("FIB lookup results in: %u", mm->adj6_index); + } + if (ip4->as_u32 != 0) + { + mm->adj4_index = ip4_fib_lookup_with_table (im4, 0, ip4, 0); + clib_warning ("FIB lookup results in: %u", mm->adj4_index); + } } #endif static clib_error_t * -map_security_check_command_fn (vlib_main_t *vm, - unformat_input_t *input, - vlib_cli_command_t *cmd) +map_security_check_command_fn (vlib_main_t * vm, + unformat_input_t * input, + vlib_cli_command_t * cmd) { unformat_input_t _line_input, *line_input = &_line_input; map_main_t *mm = &map_main; /* Get a line of input. */ - if (!unformat_user(input, unformat_line_input, line_input)) + if (!unformat_user (input, unformat_line_input, line_input)) return 0; - - while (unformat_check_input(line_input) != UNFORMAT_END_OF_INPUT) { - if (unformat(line_input, "off")) - mm->sec_check = false; - else if (unformat(line_input, "on")) - mm->sec_check = true; - else - return clib_error_return(0, "unknown input `%U'", - format_unformat_error, input); - } - unformat_free(line_input); + + while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) + { + if (unformat (line_input, "off")) + mm->sec_check = false; + else if (unformat (line_input, "on")) + mm->sec_check = true; + else + return clib_error_return (0, "unknown input `%U'", + format_unformat_error, input); + } + unformat_free (line_input); return 0; } static clib_error_t * -map_security_check_frag_command_fn (vlib_main_t *vm, - unformat_input_t *input, - vlib_cli_command_t *cmd) +map_security_check_frag_command_fn (vlib_main_t * vm, + unformat_input_t * input, + vlib_cli_command_t * cmd) { unformat_input_t _line_input, *line_input = &_line_input; map_main_t *mm = &map_main; /* Get a line of input. */ - if (!unformat_user(input, unformat_line_input, line_input)) + if (!unformat_user (input, unformat_line_input, line_input)) return 0; - - while (unformat_check_input(line_input) != UNFORMAT_END_OF_INPUT) { - if (unformat(line_input, "off")) - mm->sec_check_frag = false; - else if (unformat(line_input, "on")) - mm->sec_check_frag = true; - else - return clib_error_return(0, "unknown input `%U'", - format_unformat_error, input); - } - unformat_free(line_input); + + while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) + { + if (unformat (line_input, "off")) + mm->sec_check_frag = false; + else if (unformat (line_input, "on")) + mm->sec_check_frag = true; + else + return clib_error_return (0, "unknown input `%U'", + format_unformat_error, input); + } + unformat_free (line_input); return 0; } static clib_error_t * -map_add_domain_command_fn (vlib_main_t *vm, - unformat_input_t *input, - vlib_cli_command_t *cmd) +map_add_domain_command_fn (vlib_main_t * vm, + unformat_input_t * input, vlib_cli_command_t * cmd) { unformat_input_t _line_input, *line_input = &_line_input; ip4_address_t ip4_prefix; @@ -464,150 +520,165 @@ map_add_domain_command_fn (vlib_main_t *vm, ip6_src_len = 128; /* Get a line of input. */ - if (!unformat_user(input, unformat_line_input, line_input)) + if (!unformat_user (input, unformat_line_input, line_input)) return 0; - - while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) { - if (unformat(line_input, "ip4-pfx %U/%d", unformat_ip4_address, &ip4_prefix, &ip4_prefix_len)) - num_m_args++; - else if (unformat(line_input, "ip6-pfx %U/%d", unformat_ip6_address, &ip6_prefix, &ip6_prefix_len)) - num_m_args++; - else if (unformat(line_input, "ip6-src %U/%d", unformat_ip6_address, &ip6_src, &ip6_src_len)) - num_m_args++; - else if (unformat(line_input, "ip6-src %U", unformat_ip6_address, &ip6_src)) - num_m_args++; - else if (unformat(line_input, "ea-bits-len %d", &ea_bits_len)) - num_m_args++; - else if (unformat(line_input, "psid-offset %d", &psid_offset)) - num_m_args++; - else if (unformat(line_input, "psid-len %d", &psid_length)) - num_m_args++; - else if (unformat(line_input, "mtu %d", &mtu)) - num_m_args++; - else if (unformat(line_input, "map-t")) - flags |= MAP_DOMAIN_TRANSLATION; - else - return clib_error_return(0, "unknown input `%U'", - format_unformat_error, input); - } - unformat_free(line_input); + + while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) + { + if (unformat + (line_input, "ip4-pfx %U/%d", unformat_ip4_address, &ip4_prefix, + &ip4_prefix_len)) + num_m_args++; + else + if (unformat + (line_input, "ip6-pfx %U/%d", unformat_ip6_address, &ip6_prefix, + &ip6_prefix_len)) + num_m_args++; + else + if (unformat + (line_input, "ip6-src %U/%d", unformat_ip6_address, &ip6_src, + &ip6_src_len)) + num_m_args++; + else + if (unformat + (line_input, "ip6-src %U", unformat_ip6_address, &ip6_src)) + num_m_args++; + else if (unformat (line_input, "ea-bits-len %d", &ea_bits_len)) + num_m_args++; + else if (unformat (line_input, "psid-offset %d", &psid_offset)) + num_m_args++; + else if (unformat (line_input, "psid-len %d", &psid_length)) + num_m_args++; + else if (unformat (line_input, "mtu %d", &mtu)) + num_m_args++; + else if (unformat (line_input, "map-t")) + flags |= MAP_DOMAIN_TRANSLATION; + else + return clib_error_return (0, "unknown input `%U'", + format_unformat_error, input); + } + unformat_free (line_input); if (num_m_args < 3) - return clib_error_return(0, "mandatory argument(s) missing"); + return clib_error_return (0, "mandatory argument(s) missing"); - map_create_domain(&ip4_prefix, ip4_prefix_len, - &ip6_prefix, ip6_prefix_len, &ip6_src, ip6_src_len, - ea_bits_len, psid_offset, psid_length, &map_domain_index, - mtu, flags); + map_create_domain (&ip4_prefix, ip4_prefix_len, + &ip6_prefix, ip6_prefix_len, &ip6_src, ip6_src_len, + ea_bits_len, psid_offset, psid_length, &map_domain_index, + mtu, flags); return 0; } static clib_error_t * -map_del_domain_command_fn (vlib_main_t *vm, - unformat_input_t *input, - vlib_cli_command_t *cmd) +map_del_domain_command_fn (vlib_main_t * vm, + unformat_input_t * input, vlib_cli_command_t * cmd) { unformat_input_t _line_input, *line_input = &_line_input; u32 num_m_args = 0; u32 map_domain_index; /* Get a line of input. */ - if (! unformat_user(input, unformat_line_input, line_input)) + if (!unformat_user (input, unformat_line_input, line_input)) return 0; - - while (unformat_check_input(line_input) != UNFORMAT_END_OF_INPUT) { - if (unformat(line_input, "index %d", &map_domain_index)) - num_m_args++; - else - return clib_error_return(0, "unknown input `%U'", - format_unformat_error, input); - } - unformat_free(line_input); + + while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) + { + if (unformat (line_input, "index %d", &map_domain_index)) + num_m_args++; + else + return clib_error_return (0, "unknown input `%U'", + format_unformat_error, input); + } + unformat_free (line_input); if (num_m_args != 1) - return clib_error_return(0, "mandatory argument(s) missing"); + return clib_error_return (0, "mandatory argument(s) missing"); - map_delete_domain(map_domain_index); + map_delete_domain (map_domain_index); return 0; } static clib_error_t * -map_add_rule_command_fn (vlib_main_t *vm, - unformat_input_t *input, - vlib_cli_command_t *cmd) +map_add_rule_command_fn (vlib_main_t * vm, + unformat_input_t * input, vlib_cli_command_t * cmd) { unformat_input_t _line_input, *line_input = &_line_input; ip6_address_t tep; u32 num_m_args = 0; u32 psid, map_domain_index; - + /* Get a line of input. */ - if (! unformat_user(input, unformat_line_input, line_input)) + if (!unformat_user (input, unformat_line_input, line_input)) return 0; - while (unformat_check_input(line_input) != UNFORMAT_END_OF_INPUT) { - if (unformat(line_input, "index %d", &map_domain_index)) - num_m_args++; - else if (unformat(line_input, "psid %d", &psid)) - num_m_args++; - else if (unformat(line_input, "ip6-dst %U", unformat_ip6_address, &tep)) - num_m_args++; - else - return clib_error_return(0, "unknown input `%U'", - format_unformat_error, input); - } - unformat_free(line_input); + while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) + { + if (unformat (line_input, "index %d", &map_domain_index)) + num_m_args++; + else if (unformat (line_input, "psid %d", &psid)) + num_m_args++; + else + if (unformat (line_input, "ip6-dst %U", unformat_ip6_address, &tep)) + num_m_args++; + else + return clib_error_return (0, "unknown input `%U'", + format_unformat_error, input); + } + unformat_free (line_input); if (num_m_args != 3) - return clib_error_return(0, "mandatory argument(s) missing"); + return clib_error_return (0, "mandatory argument(s) missing"); - if (map_add_del_psid(map_domain_index, psid, &tep, 1) != 0) { - return clib_error_return(0, "Failing to add Mapping Rule"); - } + if (map_add_del_psid (map_domain_index, psid, &tep, 1) != 0) + { + return clib_error_return (0, "Failing to add Mapping Rule"); + } return 0; } #if MAP_SKIP_IP6_LOOKUP static clib_error_t * -map_pre_resolve_command_fn (vlib_main_t *vm, - unformat_input_t *input, - vlib_cli_command_t *cmd) +map_pre_resolve_command_fn (vlib_main_t * vm, + unformat_input_t * input, + vlib_cli_command_t * cmd) { unformat_input_t _line_input, *line_input = &_line_input; ip4_address_t ip4nh; ip6_address_t ip6nh; map_main_t *mm = &map_main; - memset(&ip4nh, 0, sizeof(ip4nh)); - memset(&ip6nh, 0, sizeof(ip6nh)); + memset (&ip4nh, 0, sizeof (ip4nh)); + memset (&ip6nh, 0, sizeof (ip6nh)); /* Get a line of input. */ - if (!unformat_user(input, unformat_line_input, line_input)) + if (!unformat_user (input, unformat_line_input, line_input)) return 0; - - while (unformat_check_input(line_input) != UNFORMAT_END_OF_INPUT) { - if (unformat(line_input, "ip4-nh %U", unformat_ip4_address, &ip4nh)) - mm->preresolve_ip4 = ip4nh; - else if (unformat(line_input, "ip6-nh %U", unformat_ip6_address, &ip6nh)) - mm->preresolve_ip6 = ip6nh; - else - return clib_error_return(0, "unknown input `%U'", - format_unformat_error, input); - } - unformat_free(line_input); - map_pre_resolve(&ip4nh, &ip6nh); + while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) + { + if (unformat (line_input, "ip4-nh %U", unformat_ip4_address, &ip4nh)) + mm->preresolve_ip4 = ip4nh; + else + if (unformat (line_input, "ip6-nh %U", unformat_ip6_address, &ip6nh)) + mm->preresolve_ip6 = ip6nh; + else + return clib_error_return (0, "unknown input `%U'", + format_unformat_error, input); + } + unformat_free (line_input); + + map_pre_resolve (&ip4nh, &ip6nh); return 0; } #endif static clib_error_t * -map_icmp_relay_source_address_command_fn (vlib_main_t *vm, - unformat_input_t *input, - vlib_cli_command_t *cmd) +map_icmp_relay_source_address_command_fn (vlib_main_t * vm, + unformat_input_t * input, + vlib_cli_command_t * cmd) { unformat_input_t _line_input, *line_input = &_line_input; ip4_address_t icmp_src_address; @@ -616,109 +687,113 @@ map_icmp_relay_source_address_command_fn (vlib_main_t *vm, mm->icmp4_src_address.as_u32 = 0; /* Get a line of input. */ - if (!unformat_user(input, unformat_line_input, line_input)) + if (!unformat_user (input, unformat_line_input, line_input)) return 0; - - while (unformat_check_input(line_input) != UNFORMAT_END_OF_INPUT) { - if (unformat(line_input, "%U", unformat_ip4_address, &icmp_src_address)) - mm->icmp4_src_address = icmp_src_address; - else - return clib_error_return(0, "unknown input `%U'", - format_unformat_error, input); - } - unformat_free(line_input); + + while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) + { + if (unformat + (line_input, "%U", unformat_ip4_address, &icmp_src_address)) + mm->icmp4_src_address = icmp_src_address; + else + return clib_error_return (0, "unknown input `%U'", + format_unformat_error, input); + } + unformat_free (line_input); return 0; } static clib_error_t * -map_icmp_unreachables_command_fn (vlib_main_t *vm, - unformat_input_t *input, - vlib_cli_command_t *cmd) +map_icmp_unreachables_command_fn (vlib_main_t * vm, + unformat_input_t * input, + vlib_cli_command_t * cmd) { unformat_input_t _line_input, *line_input = &_line_input; map_main_t *mm = &map_main; int num_m_args = 0; /* Get a line of input. */ - if (!unformat_user(input, unformat_line_input, line_input)) + if (!unformat_user (input, unformat_line_input, line_input)) return 0; - - while (unformat_check_input(line_input) != UNFORMAT_END_OF_INPUT) { - num_m_args++; - if (unformat(line_input, "on")) - mm->icmp6_enabled = true; - else if (unformat(line_input, "off")) - mm->icmp6_enabled = false; - else - return clib_error_return(0, "unknown input `%U'", - format_unformat_error, input); - } - unformat_free(line_input); + + while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) + { + num_m_args++; + if (unformat (line_input, "on")) + mm->icmp6_enabled = true; + else if (unformat (line_input, "off")) + mm->icmp6_enabled = false; + else + return clib_error_return (0, "unknown input `%U'", + format_unformat_error, input); + } + unformat_free (line_input); if (num_m_args != 1) - return clib_error_return(0, "mandatory argument(s) missing"); + return clib_error_return (0, "mandatory argument(s) missing"); return 0; } static clib_error_t * -map_fragment_command_fn (vlib_main_t *vm, - unformat_input_t *input, - vlib_cli_command_t *cmd) +map_fragment_command_fn (vlib_main_t * vm, + unformat_input_t * input, vlib_cli_command_t * cmd) { unformat_input_t _line_input, *line_input = &_line_input; map_main_t *mm = &map_main; /* Get a line of input. */ - if (!unformat_user(input, unformat_line_input, line_input)) + if (!unformat_user (input, unformat_line_input, line_input)) return 0; - - while (unformat_check_input(line_input) != UNFORMAT_END_OF_INPUT) { - if (unformat(line_input, "inner")) - mm->frag_inner = true; - else if (unformat(line_input, "outer")) - mm->frag_inner = false; - else - return clib_error_return(0, "unknown input `%U'", - format_unformat_error, input); - } - unformat_free(line_input); + + while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) + { + if (unformat (line_input, "inner")) + mm->frag_inner = true; + else if (unformat (line_input, "outer")) + mm->frag_inner = false; + else + return clib_error_return (0, "unknown input `%U'", + format_unformat_error, input); + } + unformat_free (line_input); return 0; } static clib_error_t * -map_fragment_df_command_fn (vlib_main_t *vm, - unformat_input_t *input, - vlib_cli_command_t *cmd) +map_fragment_df_command_fn (vlib_main_t * vm, + unformat_input_t * input, + vlib_cli_command_t * cmd) { unformat_input_t _line_input, *line_input = &_line_input; map_main_t *mm = &map_main; /* Get a line of input. */ - if (!unformat_user(input, unformat_line_input, line_input)) + if (!unformat_user (input, unformat_line_input, line_input)) return 0; - - while (unformat_check_input(line_input) != UNFORMAT_END_OF_INPUT) { - if (unformat(line_input, "on")) - mm->frag_ignore_df = true; - else if (unformat(line_input, "off")) - mm->frag_ignore_df = false; - else - return clib_error_return(0, "unknown input `%U'", - format_unformat_error, input); - } - unformat_free(line_input); + + while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) + { + if (unformat (line_input, "on")) + mm->frag_ignore_df = true; + else if (unformat (line_input, "off")) + mm->frag_ignore_df = false; + else + return clib_error_return (0, "unknown input `%U'", + format_unformat_error, input); + } + unformat_free (line_input); return 0; } static clib_error_t * -map_traffic_class_command_fn (vlib_main_t *vm, - unformat_input_t *input, - vlib_cli_command_t *cmd) +map_traffic_class_command_fn (vlib_main_t * vm, + unformat_input_t * input, + vlib_cli_command_t * cmd) { unformat_input_t _line_input, *line_input = &_line_input; map_main_t *mm = &map_main; @@ -727,104 +802,115 @@ map_traffic_class_command_fn (vlib_main_t *vm, mm->tc_copy = false; /* Get a line of input. */ - if (!unformat_user(input, unformat_line_input, line_input)) + if (!unformat_user (input, unformat_line_input, line_input)) return 0; - - while (unformat_check_input(line_input) != UNFORMAT_END_OF_INPUT) { - if (unformat(line_input, "copy")) - mm->tc_copy = true; - else if (unformat(line_input, "%x", &tc)) - mm->tc = tc & 0xff; - else - return clib_error_return(0, "unknown input `%U'", - format_unformat_error, input); - } - unformat_free(line_input); + + while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) + { + if (unformat (line_input, "copy")) + mm->tc_copy = true; + else if (unformat (line_input, "%x", &tc)) + mm->tc = tc & 0xff; + else + return clib_error_return (0, "unknown input `%U'", + format_unformat_error, input); + } + unformat_free (line_input); return 0; } static u8 * -format_map_domain (u8 *s, va_list *args) +format_map_domain (u8 * s, va_list * args) { - map_domain_t *d = va_arg(*args, map_domain_t *); - bool counters = va_arg(*args, int); + map_domain_t *d = va_arg (*args, map_domain_t *); + bool counters = va_arg (*args, int); map_main_t *mm = &map_main; ip6_address_t ip6_prefix; if (d->rules) - memset(&ip6_prefix, 0, sizeof(ip6_prefix)); + memset (&ip6_prefix, 0, sizeof (ip6_prefix)); else ip6_prefix = d->ip6_prefix; - - s = format(s, - "[%d] ip4-pfx %U/%d ip6-pfx %U/%d ip6-src %U/%d ea_bits_len %d psid-offset %d psid-len %d mtu %d %s", - d - mm->domains, - format_ip4_address, &d->ip4_prefix, d->ip4_prefix_len, - format_ip6_address, &ip6_prefix, d->ip6_prefix_len, - format_ip6_address, &d->ip6_src, d->ip6_src_len, - d->ea_bits_len, d->psid_offset, d->psid_length, d->mtu, - (d->flags & MAP_DOMAIN_TRANSLATION) ? "map-t" : ""); - - if (counters) { - map_domain_counter_lock(mm); - vlib_counter_t v; - vlib_get_combined_counter(&mm->domain_counters[MAP_DOMAIN_COUNTER_TX], d - mm->domains, &v); - s = format(s, " TX: %lld/%lld", v.packets, v.bytes); - vlib_get_combined_counter(&mm->domain_counters[MAP_DOMAIN_COUNTER_RX], d - mm->domains, &v); - s = format(s, " RX: %lld/%lld", v.packets, v.bytes); - map_domain_counter_unlock(mm); - } - s = format(s, "\n"); - - if (d->rules) { - int i; - ip6_address_t dst; - for (i = 0; i < (0x1 << d->psid_length); i++) { - dst = d->rules[i]; - if (dst.as_u64[0] == 0 && dst.as_u64[1] == 0 ) - continue; - s = format(s, - " rule psid: %d ip6-dst %U\n", i, format_ip6_address, &dst); + + s = format (s, + "[%d] ip4-pfx %U/%d ip6-pfx %U/%d ip6-src %U/%d ea_bits_len %d psid-offset %d psid-len %d mtu %d %s", + d - mm->domains, + format_ip4_address, &d->ip4_prefix, d->ip4_prefix_len, + format_ip6_address, &ip6_prefix, d->ip6_prefix_len, + format_ip6_address, &d->ip6_src, d->ip6_src_len, + d->ea_bits_len, d->psid_offset, d->psid_length, d->mtu, + (d->flags & MAP_DOMAIN_TRANSLATION) ? "map-t" : ""); + + if (counters) + { + map_domain_counter_lock (mm); + vlib_counter_t v; + vlib_get_combined_counter (&mm->domain_counters[MAP_DOMAIN_COUNTER_TX], + d - mm->domains, &v); + s = format (s, " TX: %lld/%lld", v.packets, v.bytes); + vlib_get_combined_counter (&mm->domain_counters[MAP_DOMAIN_COUNTER_RX], + d - mm->domains, &v); + s = format (s, " RX: %lld/%lld", v.packets, v.bytes); + map_domain_counter_unlock (mm); + } + s = format (s, "\n"); + + if (d->rules) + { + int i; + ip6_address_t dst; + for (i = 0; i < (0x1 << d->psid_length); i++) + { + dst = d->rules[i]; + if (dst.as_u64[0] == 0 && dst.as_u64[1] == 0) + continue; + s = format (s, + " rule psid: %d ip6-dst %U\n", i, format_ip6_address, + &dst); + } } - } return s; } static u8 * -format_map_ip4_reass (u8 *s, va_list *args) +format_map_ip4_reass (u8 * s, va_list * args) { map_main_t *mm = &map_main; - map_ip4_reass_t *r = va_arg(*args, map_ip4_reass_t *); + map_ip4_reass_t *r = va_arg (*args, map_ip4_reass_t *); map_ip4_reass_key_t *k = &r->key; - f64 now = vlib_time_now(mm->vlib_main); - f64 lifetime = (((f64)mm->ip4_reass_conf_lifetime_ms) / 1000); + f64 now = vlib_time_now (mm->vlib_main); + f64 lifetime = (((f64) mm->ip4_reass_conf_lifetime_ms) / 1000); f64 dt = (r->ts + lifetime > now) ? (r->ts + lifetime - now) : -1; - s = format(s, - "ip4-reass src=%U dst=%U protocol=%d identifier=%d port=%d lifetime=%.3lf\n", - format_ip4_address, &k->src.as_u8, format_ip4_address, &k->dst.as_u8, - k->protocol, clib_net_to_host_u16(k->fragment_id), (r->port >= 0)?clib_net_to_host_u16(r->port):-1, dt); + s = format (s, + "ip4-reass src=%U dst=%U protocol=%d identifier=%d port=%d lifetime=%.3lf\n", + format_ip4_address, &k->src.as_u8, format_ip4_address, + &k->dst.as_u8, k->protocol, + clib_net_to_host_u16 (k->fragment_id), + (r->port >= 0) ? clib_net_to_host_u16 (r->port) : -1, dt); return s; } static u8 * -format_map_ip6_reass (u8 *s, va_list *args) +format_map_ip6_reass (u8 * s, va_list * args) { map_main_t *mm = &map_main; - map_ip6_reass_t *r = va_arg(*args, map_ip6_reass_t *); + map_ip6_reass_t *r = va_arg (*args, map_ip6_reass_t *); map_ip6_reass_key_t *k = &r->key; - f64 now = vlib_time_now(mm->vlib_main); - f64 lifetime = (((f64)mm->ip6_reass_conf_lifetime_ms) / 1000); + f64 now = vlib_time_now (mm->vlib_main); + f64 lifetime = (((f64) mm->ip6_reass_conf_lifetime_ms) / 1000); f64 dt = (r->ts + lifetime > now) ? (r->ts + lifetime - now) : -1; - s = format(s, - "ip6-reass src=%U dst=%U protocol=%d identifier=%d lifetime=%.3lf\n", - format_ip6_address, &k->src.as_u8, format_ip6_address, &k->dst.as_u8, - k->protocol, clib_net_to_host_u32(k->fragment_id), dt); + s = format (s, + "ip6-reass src=%U dst=%U protocol=%d identifier=%d lifetime=%.3lf\n", + format_ip6_address, &k->src.as_u8, format_ip6_address, + &k->dst.as_u8, k->protocol, + clib_net_to_host_u32 (k->fragment_id), dt); return s; } static clib_error_t * -show_map_domain_command_fn (vlib_main_t *vm, unformat_input_t *input, vlib_cli_command_t *cmd) +show_map_domain_command_fn (vlib_main_t * vm, unformat_input_t * input, + vlib_cli_command_t * cmd) { unformat_input_t _line_input, *line_input = &_line_input; map_main_t *mm = &map_main; @@ -833,60 +919,73 @@ show_map_domain_command_fn (vlib_main_t *vm, unformat_input_t *input, vlib_cli_c u32 map_domain_index = ~0; /* Get a line of input. */ - if (!unformat_user(input, unformat_line_input, line_input)) + if (!unformat_user (input, unformat_line_input, line_input)) return 0; - - while (unformat_check_input(line_input) != UNFORMAT_END_OF_INPUT) { - if (unformat(line_input, "counters")) - counters = true; - else if (unformat(line_input, "index %d", &map_domain_index)) - ; - else - return clib_error_return(0, "unknown input `%U'", - format_unformat_error, input); - } - unformat_free(line_input); - if (pool_elts(mm->domains) == 0) - vlib_cli_output(vm, "No MAP domains are configured..."); + while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) + { + if (unformat (line_input, "counters")) + counters = true; + else if (unformat (line_input, "index %d", &map_domain_index)) + ; + else + return clib_error_return (0, "unknown input `%U'", + format_unformat_error, input); + } + unformat_free (line_input); + + if (pool_elts (mm->domains) == 0) + vlib_cli_output (vm, "No MAP domains are configured..."); - if (map_domain_index == ~0) { + if (map_domain_index == ~0) + { + /* *INDENT-OFF* */ pool_foreach(d, mm->domains, ({vlib_cli_output(vm, "%U", format_map_domain, d, counters);})); - } else { - if (pool_is_free_index(mm->domains, map_domain_index)) { - return clib_error_return(0, "MAP domain does not exists %d", map_domain_index); + /* *INDENT-ON* */ + } + else + { + if (pool_is_free_index (mm->domains, map_domain_index)) + { + return clib_error_return (0, "MAP domain does not exists %d", + map_domain_index); + } + + d = pool_elt_at_index (mm->domains, map_domain_index); + vlib_cli_output (vm, "%U", format_map_domain, d, counters); } - - d = pool_elt_at_index(mm->domains, map_domain_index); - vlib_cli_output(vm, "%U", format_map_domain, d, counters); - } return 0; } static clib_error_t * -show_map_fragments_command_fn (vlib_main_t *vm, unformat_input_t *input, vlib_cli_command_t *cmd) +show_map_fragments_command_fn (vlib_main_t * vm, unformat_input_t * input, + vlib_cli_command_t * cmd) { map_main_t *mm = &map_main; map_ip4_reass_t *f4; map_ip6_reass_t *f6; + /* *INDENT-OFF* */ pool_foreach(f4, mm->ip4_reass_pool, ({vlib_cli_output (vm, "%U", format_map_ip4_reass, f4);})); + /* *INDENT-ON* */ + /* *INDENT-OFF* */ pool_foreach(f6, mm->ip6_reass_pool, ({vlib_cli_output (vm, "%U", format_map_ip6_reass, f6);})); + /* *INDENT-ON* */ return (0); } u64 map_error_counter_get (u32 node_index, map_error_t map_error) { - vlib_main_t *vm = vlib_get_main(); - vlib_node_runtime_t *error_node = vlib_node_get_runtime(vm, node_index); + vlib_main_t *vm = vlib_get_main (); + vlib_node_runtime_t *error_node = vlib_node_get_runtime (vm, node_index); vlib_error_main_t *em = &vm->error_main; vlib_error_t e = error_node->errors[map_error]; - vlib_node_t *n = vlib_get_node(vm, node_index); + vlib_node_t *n = vlib_get_node (vm, node_index); u32 ci; - ci = vlib_error_get_code(e); + ci = vlib_error_get_code (e); ASSERT (ci < n->n_errors); ci += n->error_heap_index; @@ -894,14 +993,16 @@ map_error_counter_get (u32 node_index, map_error_t map_error) } static clib_error_t * -show_map_stats_command_fn (vlib_main_t *vm, unformat_input_t *input, vlib_cli_command_t *cmd) +show_map_stats_command_fn (vlib_main_t * vm, unformat_input_t * input, + vlib_cli_command_t * cmd) { map_main_t *mm = &map_main; map_domain_t *d; int domains = 0, rules = 0, domaincount = 0, rulecount = 0; if (pool_elts (mm->domains) == 0) - vlib_cli_output(vm, "No MAP domains are configured..."); + vlib_cli_output (vm, "No MAP domains are configured..."); + /* *INDENT-OFF* */ pool_foreach(d, mm->domains, ({ if (d->rules) { rulecount+= 0x1 << d->psid_length; @@ -910,30 +1011,38 @@ show_map_stats_command_fn (vlib_main_t *vm, unformat_input_t *input, vlib_cli_co domains += sizeof(*d); domaincount++; })); + /* *INDENT-ON* */ - vlib_cli_output(vm, "MAP domains structure: %d\n", sizeof (map_domain_t)); - vlib_cli_output(vm, "MAP domains: %d (%d bytes)\n", domaincount, domains); - vlib_cli_output(vm, "MAP rules: %d (%d bytes)\n", rulecount, rules); - vlib_cli_output(vm, "Total: %d bytes)\n", rules + domains); + vlib_cli_output (vm, "MAP domains structure: %d\n", sizeof (map_domain_t)); + vlib_cli_output (vm, "MAP domains: %d (%d bytes)\n", domaincount, domains); + vlib_cli_output (vm, "MAP rules: %d (%d bytes)\n", rulecount, rules); + vlib_cli_output (vm, "Total: %d bytes)\n", rules + domains); #if MAP_SKIP_IP6_LOOKUP - vlib_cli_output(vm, "MAP pre-resolve: IP6 next-hop: %U (%u), IP4 next-hop: %U (%u)\n", - format_ip6_address, &mm->preresolve_ip6, mm->adj6_index, - format_ip4_address, &mm->preresolve_ip4, mm->adj4_index); + vlib_cli_output (vm, + "MAP pre-resolve: IP6 next-hop: %U (%u), IP4 next-hop: %U (%u)\n", + format_ip6_address, &mm->preresolve_ip6, mm->adj6_index, + format_ip4_address, &mm->preresolve_ip4, mm->adj4_index); #endif if (mm->tc_copy) - vlib_cli_output(vm, "MAP traffic-class: copy"); + vlib_cli_output (vm, "MAP traffic-class: copy"); else - vlib_cli_output(vm, "MAP traffic-class: %x", mm->tc); - - vlib_cli_output(vm, "MAP IPv6 inbound security check: %s, fragmented packet security check: %s", mm->sec_check ? "enabled" : "disabled", - mm->sec_check_frag ? "enabled" : "disabled"); - - vlib_cli_output(vm, "ICMP-relay IPv4 source address: %U\n", format_ip4_address, &mm->icmp4_src_address); - vlib_cli_output(vm, "ICMP6 unreachables sent for unmatched packets: %s\n", mm->icmp6_enabled ? "enabled" : "disabled"); - vlib_cli_output(vm, "Inner fragmentation: %s\n", mm->frag_inner ? "enabled" : "disabled"); - vlib_cli_output(vm, "Fragment packets regardless of DF flag: %s\n", mm->frag_ignore_df ? "enabled" : "disabled"); + vlib_cli_output (vm, "MAP traffic-class: %x", mm->tc); + + vlib_cli_output (vm, + "MAP IPv6 inbound security check: %s, fragmented packet security check: %s", + mm->sec_check ? "enabled" : "disabled", + mm->sec_check_frag ? "enabled" : "disabled"); + + vlib_cli_output (vm, "ICMP-relay IPv4 source address: %U\n", + format_ip4_address, &mm->icmp4_src_address); + vlib_cli_output (vm, "ICMP6 unreachables sent for unmatched packets: %s\n", + mm->icmp6_enabled ? "enabled" : "disabled"); + vlib_cli_output (vm, "Inner fragmentation: %s\n", + mm->frag_inner ? "enabled" : "disabled"); + vlib_cli_output (vm, "Fragment packets regardless of DF flag: %s\n", + mm->frag_ignore_df ? "enabled" : "disabled"); /* * Counters @@ -948,147 +1057,203 @@ show_map_stats_command_fn (vlib_main_t *vm, unformat_input_t *input, vlib_cli_co memset (total_bytes, 0, sizeof (total_bytes)); map_domain_counter_lock (mm); - vec_foreach (cm, mm->domain_counters) { + vec_foreach (cm, mm->domain_counters) + { which = cm - mm->domain_counters; - for (i = 0; i < vec_len (cm->maxi); i++) { - vlib_get_combined_counter (cm, i, &v); - total_pkts[which] += v.packets; - total_bytes[which] += v.bytes; - } + for (i = 0; i < vec_len (cm->maxi); i++) + { + vlib_get_combined_counter (cm, i, &v); + total_pkts[which] += v.packets; + total_bytes[which] += v.bytes; + } } map_domain_counter_unlock (mm); - vlib_cli_output(vm, "Encapsulated packets: %lld bytes: %lld\n", total_pkts[MAP_DOMAIN_COUNTER_TX], - total_bytes[MAP_DOMAIN_COUNTER_TX]); - vlib_cli_output(vm, "Decapsulated packets: %lld bytes: %lld\n", total_pkts[MAP_DOMAIN_COUNTER_RX], - total_bytes[MAP_DOMAIN_COUNTER_RX]); + vlib_cli_output (vm, "Encapsulated packets: %lld bytes: %lld\n", + total_pkts[MAP_DOMAIN_COUNTER_TX], + total_bytes[MAP_DOMAIN_COUNTER_TX]); + vlib_cli_output (vm, "Decapsulated packets: %lld bytes: %lld\n", + total_pkts[MAP_DOMAIN_COUNTER_RX], + total_bytes[MAP_DOMAIN_COUNTER_RX]); - vlib_cli_output(vm, "ICMP relayed packets: %d\n", vlib_get_simple_counter(&mm->icmp_relayed, 0)); + vlib_cli_output (vm, "ICMP relayed packets: %d\n", + vlib_get_simple_counter (&mm->icmp_relayed, 0)); return 0; } static clib_error_t * -map_params_reass_command_fn (vlib_main_t *vm, unformat_input_t *input, vlib_cli_command_t *cmd) +map_params_reass_command_fn (vlib_main_t * vm, unformat_input_t * input, + vlib_cli_command_t * cmd) { unformat_input_t _line_input, *line_input = &_line_input; u32 lifetime = ~0; - f64 ht_ratio = (MAP_IP4_REASS_CONF_HT_RATIO_MAX+1); + f64 ht_ratio = (MAP_IP4_REASS_CONF_HT_RATIO_MAX + 1); u32 pool_size = ~0; u64 buffers = ~(0ull); u8 ip4 = 0, ip6 = 0; - if (!unformat_user(input, unformat_line_input, line_input)) - return 0; + if (!unformat_user (input, unformat_line_input, line_input)) + return 0; - while (unformat_check_input(line_input) != UNFORMAT_END_OF_INPUT) { - if (!unformat(line_input, "lifetime %u", &lifetime) && - !unformat(line_input, "ht-ratio %lf", &ht_ratio) && - !unformat(line_input, "pool-size %u", &pool_size) && - !unformat(line_input, "buffers %llu", &buffers) && - !((unformat(line_input, "ip4")) && (ip4 = 1)) && - !((unformat(line_input, "ip6")) && (ip6 = 1))) { - unformat_free(line_input); - return clib_error_return(0, "invalid input"); + while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) + { + if (!unformat (line_input, "lifetime %u", &lifetime) && + !unformat (line_input, "ht-ratio %lf", &ht_ratio) && + !unformat (line_input, "pool-size %u", &pool_size) && + !unformat (line_input, "buffers %llu", &buffers) && + !((unformat (line_input, "ip4")) && (ip4 = 1)) && + !((unformat (line_input, "ip6")) && (ip6 = 1))) + { + unformat_free (line_input); + return clib_error_return (0, "invalid input"); + } } - } - unformat_free(line_input); + unformat_free (line_input); if (!ip4 && !ip6) - return clib_error_return(0, "must specify ip4 and/or ip6"); - - if (ip4) { - if (pool_size != ~0 && pool_size > MAP_IP4_REASS_CONF_POOL_SIZE_MAX) - return clib_error_return(0, "invalid ip4-reass pool-size ( > %d)", MAP_IP4_REASS_CONF_POOL_SIZE_MAX); - if (ht_ratio != (MAP_IP4_REASS_CONF_HT_RATIO_MAX+1) && ht_ratio > MAP_IP4_REASS_CONF_HT_RATIO_MAX) - return clib_error_return(0, "invalid ip4-reass ht-ratio ( > %d)", MAP_IP4_REASS_CONF_HT_RATIO_MAX); - if (lifetime != ~0 && lifetime > MAP_IP4_REASS_CONF_LIFETIME_MAX) - return clib_error_return(0, "invalid ip4-reass lifetime ( > %d)", MAP_IP4_REASS_CONF_LIFETIME_MAX); - if (buffers != ~(0ull) && buffers > MAP_IP4_REASS_CONF_BUFFERS_MAX) - return clib_error_return(0, "invalid ip4-reass buffers ( > %ld)", MAP_IP4_REASS_CONF_BUFFERS_MAX); - } - - if (ip6) { - if (pool_size != ~0 && pool_size > MAP_IP6_REASS_CONF_POOL_SIZE_MAX) - return clib_error_return(0, "invalid ip6-reass pool-size ( > %d)", MAP_IP6_REASS_CONF_POOL_SIZE_MAX); - if (ht_ratio != (MAP_IP4_REASS_CONF_HT_RATIO_MAX+1) && ht_ratio > MAP_IP6_REASS_CONF_HT_RATIO_MAX) - return clib_error_return(0, "invalid ip6-reass ht-log2len ( > %d)", MAP_IP6_REASS_CONF_HT_RATIO_MAX); - if (lifetime != ~0 && lifetime > MAP_IP6_REASS_CONF_LIFETIME_MAX) - return clib_error_return(0, "invalid ip6-reass lifetime ( > %d)", MAP_IP6_REASS_CONF_LIFETIME_MAX); - if (buffers != ~(0ull) && buffers > MAP_IP6_REASS_CONF_BUFFERS_MAX) - return clib_error_return(0, "invalid ip6-reass buffers ( > %ld)", MAP_IP6_REASS_CONF_BUFFERS_MAX); - } - - if (ip4) { - u32 reass = 0, packets = 0; - if (pool_size != ~0) { - if (map_ip4_reass_conf_pool_size(pool_size, &reass, &packets)) { - vlib_cli_output(vm, "Could not set ip4-reass pool-size"); - } else { - vlib_cli_output(vm, "Setting ip4-reass pool-size (destroyed-reassembly=%u , dropped-fragments=%u)", reass, packets); - } - } - if (ht_ratio != (MAP_IP4_REASS_CONF_HT_RATIO_MAX+1)) { - if (map_ip4_reass_conf_ht_ratio(ht_ratio, &reass, &packets)) { - vlib_cli_output(vm, "Could not set ip4-reass ht-log2len"); - } else { - vlib_cli_output(vm, "Setting ip4-reass ht-log2len (destroyed-reassembly=%u , dropped-fragments=%u)", reass, packets); - } - } - if (lifetime != ~0) { - if (map_ip4_reass_conf_lifetime(lifetime)) - vlib_cli_output(vm, "Could not set ip4-reass lifetime"); - else - vlib_cli_output(vm, "Setting ip4-reass lifetime"); - } - if (buffers != ~(0ull)) { - if (map_ip4_reass_conf_buffers(buffers)) - vlib_cli_output(vm, "Could not set ip4-reass buffers"); - else - vlib_cli_output(vm, "Setting ip4-reass buffers"); + return clib_error_return (0, "must specify ip4 and/or ip6"); + + if (ip4) + { + if (pool_size != ~0 && pool_size > MAP_IP4_REASS_CONF_POOL_SIZE_MAX) + return clib_error_return (0, "invalid ip4-reass pool-size ( > %d)", + MAP_IP4_REASS_CONF_POOL_SIZE_MAX); + if (ht_ratio != (MAP_IP4_REASS_CONF_HT_RATIO_MAX + 1) + && ht_ratio > MAP_IP4_REASS_CONF_HT_RATIO_MAX) + return clib_error_return (0, "invalid ip4-reass ht-ratio ( > %d)", + MAP_IP4_REASS_CONF_HT_RATIO_MAX); + if (lifetime != ~0 && lifetime > MAP_IP4_REASS_CONF_LIFETIME_MAX) + return clib_error_return (0, "invalid ip4-reass lifetime ( > %d)", + MAP_IP4_REASS_CONF_LIFETIME_MAX); + if (buffers != ~(0ull) && buffers > MAP_IP4_REASS_CONF_BUFFERS_MAX) + return clib_error_return (0, "invalid ip4-reass buffers ( > %ld)", + MAP_IP4_REASS_CONF_BUFFERS_MAX); } - if (map_main.ip4_reass_conf_buffers > - map_main.ip4_reass_conf_pool_size * MAP_IP4_REASS_MAX_FRAGMENTS_PER_REASSEMBLY) { - vlib_cli_output(vm, "Note: 'ip4-reass buffers' > pool-size * max-fragments-per-reassembly."); + if (ip6) + { + if (pool_size != ~0 && pool_size > MAP_IP6_REASS_CONF_POOL_SIZE_MAX) + return clib_error_return (0, "invalid ip6-reass pool-size ( > %d)", + MAP_IP6_REASS_CONF_POOL_SIZE_MAX); + if (ht_ratio != (MAP_IP4_REASS_CONF_HT_RATIO_MAX + 1) + && ht_ratio > MAP_IP6_REASS_CONF_HT_RATIO_MAX) + return clib_error_return (0, "invalid ip6-reass ht-log2len ( > %d)", + MAP_IP6_REASS_CONF_HT_RATIO_MAX); + if (lifetime != ~0 && lifetime > MAP_IP6_REASS_CONF_LIFETIME_MAX) + return clib_error_return (0, "invalid ip6-reass lifetime ( > %d)", + MAP_IP6_REASS_CONF_LIFETIME_MAX); + if (buffers != ~(0ull) && buffers > MAP_IP6_REASS_CONF_BUFFERS_MAX) + return clib_error_return (0, "invalid ip6-reass buffers ( > %ld)", + MAP_IP6_REASS_CONF_BUFFERS_MAX); } - } - if (ip6) { - u32 reass = 0, packets = 0; - if (pool_size != ~0) { - if (map_ip6_reass_conf_pool_size(pool_size, &reass, &packets)) { - vlib_cli_output(vm, "Could not set ip6-reass pool-size"); - } else { - vlib_cli_output(vm, "Setting ip6-reass pool-size (destroyed-reassembly=%u , dropped-fragments=%u)", reass, packets); - } - } - if (ht_ratio != (MAP_IP4_REASS_CONF_HT_RATIO_MAX+1)) { - if (map_ip6_reass_conf_ht_ratio(ht_ratio, &reass, &packets)) { - vlib_cli_output(vm, "Could not set ip6-reass ht-log2len"); - } else { - vlib_cli_output(vm, "Setting ip6-reass ht-log2len (destroyed-reassembly=%u , dropped-fragments=%u)", reass, packets); - } - } - if (lifetime != ~0) { - if (map_ip6_reass_conf_lifetime(lifetime)) - vlib_cli_output(vm, "Could not set ip6-reass lifetime"); - else - vlib_cli_output(vm, "Setting ip6-reass lifetime"); - } - if (buffers != ~(0ull)) { - if (map_ip6_reass_conf_buffers(buffers)) - vlib_cli_output(vm, "Could not set ip6-reass buffers"); - else - vlib_cli_output(vm, "Setting ip6-reass buffers"); + if (ip4) + { + u32 reass = 0, packets = 0; + if (pool_size != ~0) + { + if (map_ip4_reass_conf_pool_size (pool_size, &reass, &packets)) + { + vlib_cli_output (vm, "Could not set ip4-reass pool-size"); + } + else + { + vlib_cli_output (vm, + "Setting ip4-reass pool-size (destroyed-reassembly=%u , dropped-fragments=%u)", + reass, packets); + } + } + if (ht_ratio != (MAP_IP4_REASS_CONF_HT_RATIO_MAX + 1)) + { + if (map_ip4_reass_conf_ht_ratio (ht_ratio, &reass, &packets)) + { + vlib_cli_output (vm, "Could not set ip4-reass ht-log2len"); + } + else + { + vlib_cli_output (vm, + "Setting ip4-reass ht-log2len (destroyed-reassembly=%u , dropped-fragments=%u)", + reass, packets); + } + } + if (lifetime != ~0) + { + if (map_ip4_reass_conf_lifetime (lifetime)) + vlib_cli_output (vm, "Could not set ip4-reass lifetime"); + else + vlib_cli_output (vm, "Setting ip4-reass lifetime"); + } + if (buffers != ~(0ull)) + { + if (map_ip4_reass_conf_buffers (buffers)) + vlib_cli_output (vm, "Could not set ip4-reass buffers"); + else + vlib_cli_output (vm, "Setting ip4-reass buffers"); + } + + if (map_main.ip4_reass_conf_buffers > + map_main.ip4_reass_conf_pool_size * + MAP_IP4_REASS_MAX_FRAGMENTS_PER_REASSEMBLY) + { + vlib_cli_output (vm, + "Note: 'ip4-reass buffers' > pool-size * max-fragments-per-reassembly."); + } } - if (map_main.ip6_reass_conf_buffers > - map_main.ip6_reass_conf_pool_size * MAP_IP6_REASS_MAX_FRAGMENTS_PER_REASSEMBLY) { - vlib_cli_output(vm, "Note: 'ip6-reass buffers' > pool-size * max-fragments-per-reassembly."); + if (ip6) + { + u32 reass = 0, packets = 0; + if (pool_size != ~0) + { + if (map_ip6_reass_conf_pool_size (pool_size, &reass, &packets)) + { + vlib_cli_output (vm, "Could not set ip6-reass pool-size"); + } + else + { + vlib_cli_output (vm, + "Setting ip6-reass pool-size (destroyed-reassembly=%u , dropped-fragments=%u)", + reass, packets); + } + } + if (ht_ratio != (MAP_IP4_REASS_CONF_HT_RATIO_MAX + 1)) + { + if (map_ip6_reass_conf_ht_ratio (ht_ratio, &reass, &packets)) + { + vlib_cli_output (vm, "Could not set ip6-reass ht-log2len"); + } + else + { + vlib_cli_output (vm, + "Setting ip6-reass ht-log2len (destroyed-reassembly=%u , dropped-fragments=%u)", + reass, packets); + } + } + if (lifetime != ~0) + { + if (map_ip6_reass_conf_lifetime (lifetime)) + vlib_cli_output (vm, "Could not set ip6-reass lifetime"); + else + vlib_cli_output (vm, "Setting ip6-reass lifetime"); + } + if (buffers != ~(0ull)) + { + if (map_ip6_reass_conf_buffers (buffers)) + vlib_cli_output (vm, "Could not set ip6-reass buffers"); + else + vlib_cli_output (vm, "Setting ip6-reass buffers"); + } + + if (map_main.ip6_reass_conf_buffers > + map_main.ip6_reass_conf_pool_size * + MAP_IP6_REASS_MAX_FRAGMENTS_PER_REASSEMBLY) + { + vlib_cli_output (vm, + "Note: 'ip6-reass buffers' > pool-size * max-fragments-per-reassembly."); + } } - } return 0; } @@ -1098,114 +1263,130 @@ map_params_reass_command_fn (vlib_main_t *vm, unformat_input_t *input, vlib_cli_ * packet trace format function */ u8 * -format_map_trace (u8 *s, va_list *args) +format_map_trace (u8 * s, va_list * args) { - CLIB_UNUSED(vlib_main_t *vm) = va_arg (*args, vlib_main_t *); - CLIB_UNUSED(vlib_node_t *node) = va_arg (*args, vlib_node_t *); + CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *); + CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *); map_trace_t *t = va_arg (*args, map_trace_t *); u32 map_domain_index = t->map_domain_index; u16 port = t->port; - s = format(s, "MAP domain index: %d L4 port: %u", map_domain_index, clib_net_to_host_u16(port)); + s = + format (s, "MAP domain index: %d L4 port: %u", map_domain_index, + clib_net_to_host_u16 (port)); return s; } static_always_inline map_ip4_reass_t * -map_ip4_reass_lookup(map_ip4_reass_key_t *k, u32 bucket, f64 now) +map_ip4_reass_lookup (map_ip4_reass_key_t * k, u32 bucket, f64 now) { map_main_t *mm = &map_main; u32 ri = mm->ip4_reass_hash_table[bucket]; - while(ri != MAP_REASS_INDEX_NONE) { - map_ip4_reass_t * r = pool_elt_at_index(mm->ip4_reass_pool, ri); - if (r->key.as_u64[0] == k->as_u64[0] && - r->key.as_u64[1] == k->as_u64[1] && - now < r->ts + (((f64)mm->ip4_reass_conf_lifetime_ms) / 1000)) { - return r; - } - ri = r->bucket_next; - } + while (ri != MAP_REASS_INDEX_NONE) + { + map_ip4_reass_t *r = pool_elt_at_index (mm->ip4_reass_pool, ri); + if (r->key.as_u64[0] == k->as_u64[0] && + r->key.as_u64[1] == k->as_u64[1] && + now < r->ts + (((f64) mm->ip4_reass_conf_lifetime_ms) / 1000)) + { + return r; + } + ri = r->bucket_next; + } return NULL; } #define map_ip4_reass_pool_index(r) (r - map_main.ip4_reass_pool) void -map_ip4_reass_free(map_ip4_reass_t *r, u32 **pi_to_drop) +map_ip4_reass_free (map_ip4_reass_t * r, u32 ** pi_to_drop) { map_main_t *mm = &map_main; - map_ip4_reass_get_fragments(r, pi_to_drop); + map_ip4_reass_get_fragments (r, pi_to_drop); // Unlink in hash bucket map_ip4_reass_t *r2 = NULL; u32 r2i = mm->ip4_reass_hash_table[r->bucket]; - while (r2i != map_ip4_reass_pool_index(r)) { - ASSERT(r2i != MAP_REASS_INDEX_NONE); - r2 = pool_elt_at_index(mm->ip4_reass_pool, r2i); - r2i = r2->bucket_next; - } - if (r2) { - r2->bucket_next = r->bucket_next; - } else { - mm->ip4_reass_hash_table[r->bucket] = r->bucket_next; - } + while (r2i != map_ip4_reass_pool_index (r)) + { + ASSERT (r2i != MAP_REASS_INDEX_NONE); + r2 = pool_elt_at_index (mm->ip4_reass_pool, r2i); + r2i = r2->bucket_next; + } + if (r2) + { + r2->bucket_next = r->bucket_next; + } + else + { + mm->ip4_reass_hash_table[r->bucket] = r->bucket_next; + } // Unlink in list - if (r->fifo_next == map_ip4_reass_pool_index(r)) { - mm->ip4_reass_fifo_last = MAP_REASS_INDEX_NONE; - } else { - if(mm->ip4_reass_fifo_last == map_ip4_reass_pool_index(r)) - mm->ip4_reass_fifo_last = r->fifo_prev; - pool_elt_at_index(mm->ip4_reass_pool, r->fifo_prev)->fifo_next = r->fifo_next; - pool_elt_at_index(mm->ip4_reass_pool, r->fifo_next)->fifo_prev = r->fifo_prev; - } + if (r->fifo_next == map_ip4_reass_pool_index (r)) + { + mm->ip4_reass_fifo_last = MAP_REASS_INDEX_NONE; + } + else + { + if (mm->ip4_reass_fifo_last == map_ip4_reass_pool_index (r)) + mm->ip4_reass_fifo_last = r->fifo_prev; + pool_elt_at_index (mm->ip4_reass_pool, r->fifo_prev)->fifo_next = + r->fifo_next; + pool_elt_at_index (mm->ip4_reass_pool, r->fifo_next)->fifo_prev = + r->fifo_prev; + } - pool_put(mm->ip4_reass_pool, r); + pool_put (mm->ip4_reass_pool, r); mm->ip4_reass_allocated--; } map_ip4_reass_t * -map_ip4_reass_get(u32 src, u32 dst, u16 fragment_id, - u8 protocol, u32 **pi_to_drop) +map_ip4_reass_get (u32 src, u32 dst, u16 fragment_id, + u8 protocol, u32 ** pi_to_drop) { - map_ip4_reass_t * r; + map_ip4_reass_t *r; map_main_t *mm = &map_main; map_ip4_reass_key_t k = {.src.data_u32 = src, - .dst.data_u32 = dst, - .fragment_id = fragment_id, - .protocol = protocol }; + .dst.data_u32 = dst, + .fragment_id = fragment_id, + .protocol = protocol + }; u32 h = 0; - h = crc_u32(k.as_u32[0], h); - h = crc_u32(k.as_u32[1], h); - h = crc_u32(k.as_u32[2], h); - h = crc_u32(k.as_u32[3], h); + h = crc_u32 (k.as_u32[0], h); + h = crc_u32 (k.as_u32[1], h); + h = crc_u32 (k.as_u32[2], h); + h = crc_u32 (k.as_u32[3], h); h = h >> (32 - mm->ip4_reass_ht_log2len); - f64 now = vlib_time_now(mm->vlib_main); + f64 now = vlib_time_now (mm->vlib_main); //Cache garbage collection - while (mm->ip4_reass_fifo_last != MAP_REASS_INDEX_NONE) { - map_ip4_reass_t *last = pool_elt_at_index(mm->ip4_reass_pool, mm->ip4_reass_fifo_last); - if (last->ts + (((f64)mm->ip4_reass_conf_lifetime_ms) / 1000) < now) - map_ip4_reass_free(last, pi_to_drop); - else - break; - } + while (mm->ip4_reass_fifo_last != MAP_REASS_INDEX_NONE) + { + map_ip4_reass_t *last = + pool_elt_at_index (mm->ip4_reass_pool, mm->ip4_reass_fifo_last); + if (last->ts + (((f64) mm->ip4_reass_conf_lifetime_ms) / 1000) < now) + map_ip4_reass_free (last, pi_to_drop); + else + break; + } - if ((r = map_ip4_reass_lookup(&k, h, now))) + if ((r = map_ip4_reass_lookup (&k, h, now))) return r; if (mm->ip4_reass_allocated >= mm->ip4_reass_conf_pool_size) return NULL; - pool_get(mm->ip4_reass_pool, r); + pool_get (mm->ip4_reass_pool, r); mm->ip4_reass_allocated++; int i; - for (i=0; i<MAP_IP4_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++) + for (i = 0; i < MAP_IP4_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++) r->fragments[i] = ~0; - u32 ri = map_ip4_reass_pool_index(r); + u32 ri = map_ip4_reass_pool_index (r); //Link in new bucket r->bucket = h; @@ -1213,15 +1394,20 @@ map_ip4_reass_get(u32 src, u32 dst, u16 fragment_id, mm->ip4_reass_hash_table[h] = ri; //Link in fifo - if(mm->ip4_reass_fifo_last != MAP_REASS_INDEX_NONE) { - r->fifo_next = pool_elt_at_index(mm->ip4_reass_pool, mm->ip4_reass_fifo_last)->fifo_next; - r->fifo_prev = mm->ip4_reass_fifo_last; - pool_elt_at_index(mm->ip4_reass_pool, r->fifo_prev)->fifo_next = ri; - pool_elt_at_index(mm->ip4_reass_pool, r->fifo_next)->fifo_prev = ri; - } else { - r->fifo_next = r->fifo_prev = ri; - mm->ip4_reass_fifo_last = ri; - } + if (mm->ip4_reass_fifo_last != MAP_REASS_INDEX_NONE) + { + r->fifo_next = + pool_elt_at_index (mm->ip4_reass_pool, + mm->ip4_reass_fifo_last)->fifo_next; + r->fifo_prev = mm->ip4_reass_fifo_last; + pool_elt_at_index (mm->ip4_reass_pool, r->fifo_prev)->fifo_next = ri; + pool_elt_at_index (mm->ip4_reass_pool, r->fifo_next)->fifo_prev = ri; + } + else + { + r->fifo_next = r->fifo_prev = ri; + mm->ip4_reass_fifo_last = ri; + } //Set other fields r->ts = now; @@ -1236,128 +1422,144 @@ map_ip4_reass_get(u32 src, u32 dst, u16 fragment_id, } int -map_ip4_reass_add_fragment(map_ip4_reass_t *r, u32 pi) +map_ip4_reass_add_fragment (map_ip4_reass_t * r, u32 pi) { if (map_main.ip4_reass_buffered_counter >= map_main.ip4_reass_conf_buffers) return -1; int i; - for (i=0; i<MAP_IP4_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++) - if(r->fragments[i] == ~0) { - r->fragments[i] = pi; - map_main.ip4_reass_buffered_counter++; - return 0; - } + for (i = 0; i < MAP_IP4_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++) + if (r->fragments[i] == ~0) + { + r->fragments[i] = pi; + map_main.ip4_reass_buffered_counter++; + return 0; + } return -1; } static_always_inline map_ip6_reass_t * -map_ip6_reass_lookup(map_ip6_reass_key_t *k, u32 bucket, f64 now) +map_ip6_reass_lookup (map_ip6_reass_key_t * k, u32 bucket, f64 now) { map_main_t *mm = &map_main; u32 ri = mm->ip6_reass_hash_table[bucket]; - while(ri != MAP_REASS_INDEX_NONE) { - map_ip6_reass_t * r = pool_elt_at_index(mm->ip6_reass_pool, ri); - if(now < r->ts + (((f64)mm->ip6_reass_conf_lifetime_ms) / 1000) && - r->key.as_u64[0] == k->as_u64[0] && - r->key.as_u64[1] == k->as_u64[1] && - r->key.as_u64[2] == k->as_u64[2] && - r->key.as_u64[3] == k->as_u64[3] && - r->key.as_u64[4] == k->as_u64[4]) - return r; - ri = r->bucket_next; - } + while (ri != MAP_REASS_INDEX_NONE) + { + map_ip6_reass_t *r = pool_elt_at_index (mm->ip6_reass_pool, ri); + if (now < r->ts + (((f64) mm->ip6_reass_conf_lifetime_ms) / 1000) && + r->key.as_u64[0] == k->as_u64[0] && + r->key.as_u64[1] == k->as_u64[1] && + r->key.as_u64[2] == k->as_u64[2] && + r->key.as_u64[3] == k->as_u64[3] && + r->key.as_u64[4] == k->as_u64[4]) + return r; + ri = r->bucket_next; + } return NULL; } #define map_ip6_reass_pool_index(r) (r - map_main.ip6_reass_pool) void -map_ip6_reass_free(map_ip6_reass_t *r, u32 **pi_to_drop) +map_ip6_reass_free (map_ip6_reass_t * r, u32 ** pi_to_drop) { map_main_t *mm = &map_main; int i; - for (i=0; i<MAP_IP6_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++) - if(r->fragments[i].pi != ~0) { - vec_add1(*pi_to_drop, r->fragments[i].pi); - r->fragments[i].pi = ~0; - map_main.ip6_reass_buffered_counter--; - } + for (i = 0; i < MAP_IP6_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++) + if (r->fragments[i].pi != ~0) + { + vec_add1 (*pi_to_drop, r->fragments[i].pi); + r->fragments[i].pi = ~0; + map_main.ip6_reass_buffered_counter--; + } // Unlink in hash bucket map_ip6_reass_t *r2 = NULL; u32 r2i = mm->ip6_reass_hash_table[r->bucket]; - while (r2i != map_ip6_reass_pool_index(r)) { - ASSERT(r2i != MAP_REASS_INDEX_NONE); - r2 = pool_elt_at_index(mm->ip6_reass_pool, r2i); - r2i = r2->bucket_next; - } - if (r2) { - r2->bucket_next = r->bucket_next; - } else { - mm->ip6_reass_hash_table[r->bucket] = r->bucket_next; - } + while (r2i != map_ip6_reass_pool_index (r)) + { + ASSERT (r2i != MAP_REASS_INDEX_NONE); + r2 = pool_elt_at_index (mm->ip6_reass_pool, r2i); + r2i = r2->bucket_next; + } + if (r2) + { + r2->bucket_next = r->bucket_next; + } + else + { + mm->ip6_reass_hash_table[r->bucket] = r->bucket_next; + } // Unlink in list - if (r->fifo_next == map_ip6_reass_pool_index(r)) { - //Single element in the list, list is now empty - mm->ip6_reass_fifo_last = MAP_REASS_INDEX_NONE; - } else { - if (mm->ip6_reass_fifo_last == map_ip6_reass_pool_index(r)) //First element - mm->ip6_reass_fifo_last = r->fifo_prev; - pool_elt_at_index(mm->ip6_reass_pool, r->fifo_prev)->fifo_next = r->fifo_next; - pool_elt_at_index(mm->ip6_reass_pool, r->fifo_next)->fifo_prev = r->fifo_prev; - } + if (r->fifo_next == map_ip6_reass_pool_index (r)) + { + //Single element in the list, list is now empty + mm->ip6_reass_fifo_last = MAP_REASS_INDEX_NONE; + } + else + { + if (mm->ip6_reass_fifo_last == map_ip6_reass_pool_index (r)) //First element + mm->ip6_reass_fifo_last = r->fifo_prev; + pool_elt_at_index (mm->ip6_reass_pool, r->fifo_prev)->fifo_next = + r->fifo_next; + pool_elt_at_index (mm->ip6_reass_pool, r->fifo_next)->fifo_prev = + r->fifo_prev; + } // Free from pool if necessary - pool_put(mm->ip6_reass_pool, r); + pool_put (mm->ip6_reass_pool, r); mm->ip6_reass_allocated--; } map_ip6_reass_t * -map_ip6_reass_get(ip6_address_t *src, ip6_address_t *dst, u32 fragment_id, - u8 protocol, u32 **pi_to_drop) +map_ip6_reass_get (ip6_address_t * src, ip6_address_t * dst, u32 fragment_id, + u8 protocol, u32 ** pi_to_drop) { - map_ip6_reass_t * r; + map_ip6_reass_t *r; map_main_t *mm = &map_main; map_ip6_reass_key_t k = { - .src = *src, - .dst = *dst, - .fragment_id = fragment_id, - .protocol = protocol }; + .src = *src, + .dst = *dst, + .fragment_id = fragment_id, + .protocol = protocol + }; u32 h = 0; int i; - for (i=0; i<10; i++) - h = crc_u32(k.as_u32[i], h); + for (i = 0; i < 10; i++) + h = crc_u32 (k.as_u32[i], h); h = h >> (32 - mm->ip6_reass_ht_log2len); - f64 now = vlib_time_now(mm->vlib_main); + f64 now = vlib_time_now (mm->vlib_main); //Cache garbage collection - while (mm->ip6_reass_fifo_last != MAP_REASS_INDEX_NONE) { - map_ip6_reass_t *last = pool_elt_at_index(mm->ip6_reass_pool, mm->ip6_reass_fifo_last); - if (last->ts + (((f64)mm->ip6_reass_conf_lifetime_ms) / 1000) < now) - map_ip6_reass_free(last, pi_to_drop); - else - break; - } + while (mm->ip6_reass_fifo_last != MAP_REASS_INDEX_NONE) + { + map_ip6_reass_t *last = + pool_elt_at_index (mm->ip6_reass_pool, mm->ip6_reass_fifo_last); + if (last->ts + (((f64) mm->ip6_reass_conf_lifetime_ms) / 1000) < now) + map_ip6_reass_free (last, pi_to_drop); + else + break; + } - if ((r = map_ip6_reass_lookup(&k, h, now))) + if ((r = map_ip6_reass_lookup (&k, h, now))) return r; if (mm->ip6_reass_allocated >= mm->ip6_reass_conf_pool_size) return NULL; - pool_get(mm->ip6_reass_pool, r); + pool_get (mm->ip6_reass_pool, r); mm->ip6_reass_allocated++; - for (i=0; i<MAP_IP6_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++) { - r->fragments[i].pi = ~0; - r->fragments[i].next_data_len = 0; - r->fragments[i].next_data_offset = 0; - } + for (i = 0; i < MAP_IP6_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++) + { + r->fragments[i].pi = ~0; + r->fragments[i].next_data_len = 0; + r->fragments[i].next_data_offset = 0; + } - u32 ri = map_ip6_reass_pool_index(r); + u32 ri = map_ip6_reass_pool_index (r); //Link in new bucket r->bucket = h; @@ -1365,15 +1567,20 @@ map_ip6_reass_get(ip6_address_t *src, ip6_address_t *dst, u32 fragment_id, mm->ip6_reass_hash_table[h] = ri; //Link in fifo - if(mm->ip6_reass_fifo_last != MAP_REASS_INDEX_NONE) { - r->fifo_next = pool_elt_at_index(mm->ip6_reass_pool, mm->ip6_reass_fifo_last)->fifo_next; - r->fifo_prev = mm->ip6_reass_fifo_last; - pool_elt_at_index(mm->ip6_reass_pool, r->fifo_prev)->fifo_next = ri; - pool_elt_at_index(mm->ip6_reass_pool, r->fifo_next)->fifo_prev = ri; - } else { - r->fifo_next = r->fifo_prev = ri; - mm->ip6_reass_fifo_last = ri; - } + if (mm->ip6_reass_fifo_last != MAP_REASS_INDEX_NONE) + { + r->fifo_next = + pool_elt_at_index (mm->ip6_reass_pool, + mm->ip6_reass_fifo_last)->fifo_next; + r->fifo_prev = mm->ip6_reass_fifo_last; + pool_elt_at_index (mm->ip6_reass_pool, r->fifo_prev)->fifo_next = ri; + pool_elt_at_index (mm->ip6_reass_pool, r->fifo_next)->fifo_prev = ri; + } + else + { + r->fifo_next = r->fifo_prev = ri; + mm->ip6_reass_fifo_last = ri; + } //Set other fields r->ts = now; @@ -1387,9 +1594,9 @@ map_ip6_reass_get(ip6_address_t *src, ip6_address_t *dst, u32 fragment_id, } int -map_ip6_reass_add_fragment(map_ip6_reass_t *r, u32 pi, - u16 data_offset, u16 next_data_offset, - u8 *data_start, u16 data_len) +map_ip6_reass_add_fragment (map_ip6_reass_t * r, u32 pi, + u16 data_offset, u16 next_data_offset, + u8 * data_start, u16 data_len) { map_ip6_fragment_t *f = NULL, *prev_f = NULL; u16 copied_len = (data_len > 20) ? 20 : data_len; @@ -1400,258 +1607,310 @@ map_ip6_reass_add_fragment(map_ip6_reass_t *r, u32 pi, //Lookup for fragments for the current buffer //and the one before that int i; - for (i=0; i<MAP_IP6_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++) { - if (data_offset && r->fragments[i].next_data_offset == data_offset) { - prev_f = &r->fragments[i]; // This is buffer for previous packet - } else if (r->fragments[i].next_data_offset == next_data_offset) { - f = &r->fragments[i]; // This is a buffer for the current packet - } else if (r->fragments[i].next_data_offset == 0) { //Available - if (f == NULL) - f = &r->fragments[i]; - else if (prev_f == NULL) - prev_f = &r->fragments[i]; + for (i = 0; i < MAP_IP6_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++) + { + if (data_offset && r->fragments[i].next_data_offset == data_offset) + { + prev_f = &r->fragments[i]; // This is buffer for previous packet + } + else if (r->fragments[i].next_data_offset == next_data_offset) + { + f = &r->fragments[i]; // This is a buffer for the current packet + } + else if (r->fragments[i].next_data_offset == 0) + { //Available + if (f == NULL) + f = &r->fragments[i]; + else if (prev_f == NULL) + prev_f = &r->fragments[i]; + } } - } if (!f || f->pi != ~0) return -1; - if (data_offset) { - if (!prev_f) - return -1; + if (data_offset) + { + if (!prev_f) + return -1; - clib_memcpy(prev_f->next_data, data_start, copied_len); - prev_f->next_data_len = copied_len; - prev_f->next_data_offset = data_offset; - } else { - if (((ip4_header_t *)data_start)->ip_version_and_header_length != 0x45) - return -1; + clib_memcpy (prev_f->next_data, data_start, copied_len); + prev_f->next_data_len = copied_len; + prev_f->next_data_offset = data_offset; + } + else + { + if (((ip4_header_t *) data_start)->ip_version_and_header_length != 0x45) + return -1; - if (r->ip4_header.ip_version_and_header_length == 0) - clib_memcpy(&r->ip4_header, data_start, sizeof(ip4_header_t)); - } + if (r->ip4_header.ip_version_and_header_length == 0) + clib_memcpy (&r->ip4_header, data_start, sizeof (ip4_header_t)); + } - if(data_len > 20) { - f->next_data_offset = next_data_offset; - f->pi = pi; - map_main.ip6_reass_buffered_counter++; - } + if (data_len > 20) + { + f->next_data_offset = next_data_offset; + f->pi = pi; + map_main.ip6_reass_buffered_counter++; + } return 0; } -void map_ip4_reass_reinit(u32 *trashed_reass, u32 *dropped_packets) +void +map_ip4_reass_reinit (u32 * trashed_reass, u32 * dropped_packets) { map_main_t *mm = &map_main; int i; - if(dropped_packets) + if (dropped_packets) *dropped_packets = mm->ip4_reass_buffered_counter; - if(trashed_reass) + if (trashed_reass) *trashed_reass = mm->ip4_reass_allocated; - if (mm->ip4_reass_fifo_last != MAP_REASS_INDEX_NONE) { - u16 ri = mm->ip4_reass_fifo_last; - do { - map_ip4_reass_t *r = pool_elt_at_index(mm->ip4_reass_pool, ri); - for (i=0; i<MAP_IP4_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++) - if (r->fragments[i] != ~0) - map_ip4_drop_pi(r->fragments[i]); - - ri = r->fifo_next; - pool_put(mm->ip4_reass_pool, r); - } while (ri != mm->ip4_reass_fifo_last); - } + if (mm->ip4_reass_fifo_last != MAP_REASS_INDEX_NONE) + { + u16 ri = mm->ip4_reass_fifo_last; + do + { + map_ip4_reass_t *r = pool_elt_at_index (mm->ip4_reass_pool, ri); + for (i = 0; i < MAP_IP4_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++) + if (r->fragments[i] != ~0) + map_ip4_drop_pi (r->fragments[i]); + + ri = r->fifo_next; + pool_put (mm->ip4_reass_pool, r); + } + while (ri != mm->ip4_reass_fifo_last); + } - vec_free(mm->ip4_reass_hash_table); - vec_resize(mm->ip4_reass_hash_table, 1 << mm->ip4_reass_ht_log2len); - for (i=0; i<(1 << mm->ip4_reass_ht_log2len); i++) + vec_free (mm->ip4_reass_hash_table); + vec_resize (mm->ip4_reass_hash_table, 1 << mm->ip4_reass_ht_log2len); + for (i = 0; i < (1 << mm->ip4_reass_ht_log2len); i++) mm->ip4_reass_hash_table[i] = MAP_REASS_INDEX_NONE; - pool_free(mm->ip4_reass_pool); - pool_alloc(mm->ip4_reass_pool, mm->ip4_reass_conf_pool_size); + pool_free (mm->ip4_reass_pool); + pool_alloc (mm->ip4_reass_pool, mm->ip4_reass_conf_pool_size); mm->ip4_reass_allocated = 0; mm->ip4_reass_fifo_last = MAP_REASS_INDEX_NONE; mm->ip4_reass_buffered_counter = 0; } -u8 map_get_ht_log2len(f32 ht_ratio, u16 pool_size) +u8 +map_get_ht_log2len (f32 ht_ratio, u16 pool_size) { - u32 desired_size = (u32)(pool_size * ht_ratio); + u32 desired_size = (u32) (pool_size * ht_ratio); u8 i; - for (i=1; i<31; i++) + for (i = 1; i < 31; i++) if ((1 << i) >= desired_size) return i; return 4; } -int map_ip4_reass_conf_ht_ratio(f32 ht_ratio, u32 *trashed_reass, u32 *dropped_packets) +int +map_ip4_reass_conf_ht_ratio (f32 ht_ratio, u32 * trashed_reass, + u32 * dropped_packets) { map_main_t *mm = &map_main; if (ht_ratio > MAP_IP4_REASS_CONF_HT_RATIO_MAX) return -1; - map_ip4_reass_lock(); + map_ip4_reass_lock (); mm->ip4_reass_conf_ht_ratio = ht_ratio; - mm->ip4_reass_ht_log2len = map_get_ht_log2len(ht_ratio, mm->ip4_reass_conf_pool_size); - map_ip4_reass_reinit(trashed_reass, dropped_packets); - map_ip4_reass_unlock(); + mm->ip4_reass_ht_log2len = + map_get_ht_log2len (ht_ratio, mm->ip4_reass_conf_pool_size); + map_ip4_reass_reinit (trashed_reass, dropped_packets); + map_ip4_reass_unlock (); return 0; } -int map_ip4_reass_conf_pool_size(u16 pool_size, u32 *trashed_reass, u32 *dropped_packets) +int +map_ip4_reass_conf_pool_size (u16 pool_size, u32 * trashed_reass, + u32 * dropped_packets) { map_main_t *mm = &map_main; if (pool_size > MAP_IP4_REASS_CONF_POOL_SIZE_MAX) return -1; - map_ip4_reass_lock(); + map_ip4_reass_lock (); mm->ip4_reass_conf_pool_size = pool_size; - map_ip4_reass_reinit(trashed_reass, dropped_packets); - map_ip4_reass_unlock(); + map_ip4_reass_reinit (trashed_reass, dropped_packets); + map_ip4_reass_unlock (); return 0; } -int map_ip4_reass_conf_lifetime(u16 lifetime_ms) +int +map_ip4_reass_conf_lifetime (u16 lifetime_ms) { map_main.ip4_reass_conf_lifetime_ms = lifetime_ms; return 0; } -int map_ip4_reass_conf_buffers(u32 buffers) +int +map_ip4_reass_conf_buffers (u32 buffers) { map_main.ip4_reass_conf_buffers = buffers; return 0; } -void map_ip6_reass_reinit(u32 *trashed_reass, u32 *dropped_packets) +void +map_ip6_reass_reinit (u32 * trashed_reass, u32 * dropped_packets) { map_main_t *mm = &map_main; - if(dropped_packets) + if (dropped_packets) *dropped_packets = mm->ip6_reass_buffered_counter; - if(trashed_reass) + if (trashed_reass) *trashed_reass = mm->ip6_reass_allocated; int i; - if (mm->ip6_reass_fifo_last != MAP_REASS_INDEX_NONE) { - u16 ri = mm->ip6_reass_fifo_last; - do { - map_ip6_reass_t *r = pool_elt_at_index(mm->ip6_reass_pool, ri); - for (i=0; i<MAP_IP6_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++) - if (r->fragments[i].pi != ~0) - map_ip6_drop_pi(r->fragments[i].pi); - - ri = r->fifo_next; - pool_put(mm->ip6_reass_pool, r); - } while (ri != mm->ip6_reass_fifo_last); - mm->ip6_reass_fifo_last = MAP_REASS_INDEX_NONE; - } + if (mm->ip6_reass_fifo_last != MAP_REASS_INDEX_NONE) + { + u16 ri = mm->ip6_reass_fifo_last; + do + { + map_ip6_reass_t *r = pool_elt_at_index (mm->ip6_reass_pool, ri); + for (i = 0; i < MAP_IP6_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++) + if (r->fragments[i].pi != ~0) + map_ip6_drop_pi (r->fragments[i].pi); + + ri = r->fifo_next; + pool_put (mm->ip6_reass_pool, r); + } + while (ri != mm->ip6_reass_fifo_last); + mm->ip6_reass_fifo_last = MAP_REASS_INDEX_NONE; + } - vec_free(mm->ip6_reass_hash_table); - vec_resize(mm->ip6_reass_hash_table, 1 << mm->ip6_reass_ht_log2len); - for(i=0; i<(1 << mm->ip6_reass_ht_log2len); i++) + vec_free (mm->ip6_reass_hash_table); + vec_resize (mm->ip6_reass_hash_table, 1 << mm->ip6_reass_ht_log2len); + for (i = 0; i < (1 << mm->ip6_reass_ht_log2len); i++) mm->ip6_reass_hash_table[i] = MAP_REASS_INDEX_NONE; - pool_free(mm->ip6_reass_pool); - pool_alloc(mm->ip6_reass_pool, mm->ip4_reass_conf_pool_size); + pool_free (mm->ip6_reass_pool); + pool_alloc (mm->ip6_reass_pool, mm->ip4_reass_conf_pool_size); mm->ip6_reass_allocated = 0; mm->ip6_reass_buffered_counter = 0; } -int map_ip6_reass_conf_ht_ratio(f32 ht_ratio, u32 *trashed_reass, u32 *dropped_packets) +int +map_ip6_reass_conf_ht_ratio (f32 ht_ratio, u32 * trashed_reass, + u32 * dropped_packets) { map_main_t *mm = &map_main; if (ht_ratio > MAP_IP6_REASS_CONF_HT_RATIO_MAX) return -1; - map_ip6_reass_lock(); + map_ip6_reass_lock (); mm->ip6_reass_conf_ht_ratio = ht_ratio; - mm->ip6_reass_ht_log2len = map_get_ht_log2len(ht_ratio, mm->ip6_reass_conf_pool_size); - map_ip6_reass_reinit(trashed_reass, dropped_packets); - map_ip6_reass_unlock(); + mm->ip6_reass_ht_log2len = + map_get_ht_log2len (ht_ratio, mm->ip6_reass_conf_pool_size); + map_ip6_reass_reinit (trashed_reass, dropped_packets); + map_ip6_reass_unlock (); return 0; } -int map_ip6_reass_conf_pool_size(u16 pool_size, u32 *trashed_reass, u32 *dropped_packets) +int +map_ip6_reass_conf_pool_size (u16 pool_size, u32 * trashed_reass, + u32 * dropped_packets) { map_main_t *mm = &map_main; if (pool_size > MAP_IP6_REASS_CONF_POOL_SIZE_MAX) return -1; - map_ip6_reass_lock(); + map_ip6_reass_lock (); mm->ip6_reass_conf_pool_size = pool_size; - map_ip6_reass_reinit(trashed_reass, dropped_packets); - map_ip6_reass_unlock(); + map_ip6_reass_reinit (trashed_reass, dropped_packets); + map_ip6_reass_unlock (); return 0; } -int map_ip6_reass_conf_lifetime(u16 lifetime_ms) +int +map_ip6_reass_conf_lifetime (u16 lifetime_ms) { map_main.ip6_reass_conf_lifetime_ms = lifetime_ms; return 0; } -int map_ip6_reass_conf_buffers(u32 buffers) +int +map_ip6_reass_conf_buffers (u32 buffers) { map_main.ip6_reass_conf_buffers = buffers; return 0; } +/* *INDENT-OFF* */ VLIB_CLI_COMMAND(map_ip4_reass_lifetime_command, static) = { .path = "map params reassembly", .short_help = "[ip4 | ip6] [lifetime <lifetime-ms>] [pool-size <pool-size>] [buffers <buffers>] [ht-ratio <ht-ratio>]", .function = map_params_reass_command_fn, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_CLI_COMMAND(map_traffic_class_command, static) = { .path = "map params traffic-class", .short_help = "traffic-class {0x0-0xff | copy}", .function = map_traffic_class_command_fn, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_CLI_COMMAND(map_pre_resolve_command, static) = { .path = "map params pre-resolve", .short_help = "pre-resolve {ip4-nh <address>} | {ip6-nh <address>}", .function = map_pre_resolve_command_fn, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_CLI_COMMAND(map_security_check_command, static) = { .path = "map params security-check", .short_help = "security-check on|off", .function = map_security_check_command_fn, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_CLI_COMMAND(map_icmp_relay_source_address_command, static) = { .path = "map params icmp source-address", .short_help = "source-address <ip4-address>", .function = map_icmp_relay_source_address_command_fn, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_CLI_COMMAND(map_icmp_unreachables_command, static) = { .path = "map params icmp6 unreachables", .short_help = "unreachables {on|off}", .function = map_icmp_unreachables_command_fn, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_CLI_COMMAND(map_fragment_command, static) = { .path = "map params fragment", .short_help = "[inner|outer] [ignore-df [on|off]]", .function = map_fragment_command_fn, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_CLI_COMMAND(map_fragment_df_command, static) = { .path = "map params fragment ignore-df", .short_help = "on|off", .function = map_fragment_df_command_fn, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_CLI_COMMAND(map_security_check_frag_command, static) = { .path = "map params security-check fragments", .short_help = "fragments on|off", .function = map_security_check_frag_command_fn, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_CLI_COMMAND(map_add_domain_command, static) = { .path = "map add domain", .short_help = @@ -1659,48 +1918,60 @@ VLIB_CLI_COMMAND(map_add_domain_command, static) = { "ea-bits-len <n> psid-offset <n> psid-len <n> [map-t] [mtu <mtu>]", .function = map_add_domain_command_fn, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_CLI_COMMAND(map_add_rule_command, static) = { .path = "map add rule", .short_help = "map add rule index <domain> psid <psid> ip6-dst <ip6-addr>", .function = map_add_rule_command_fn, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_CLI_COMMAND(map_del_command, static) = { .path = "map del domain", .short_help = "map del domain index <domain>", .function = map_del_domain_command_fn, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_CLI_COMMAND(show_map_domain_command, static) = { .path = "show map domain", .function = show_map_domain_command_fn, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_CLI_COMMAND(show_map_stats_command, static) = { .path = "show map stats", .function = show_map_stats_command_fn, }; +/* *INDENT-ON* */ +/* *INDENT-OFF* */ VLIB_CLI_COMMAND(show_map_fragments_command, static) = { .path = "show map fragments", .function = show_map_fragments_command_fn, }; +/* *INDENT-ON* */ /* * map_init */ -clib_error_t *map_init (vlib_main_t *vm) +clib_error_t * +map_init (vlib_main_t * vm) { map_main_t *mm = &map_main; - mm->vnet_main = vnet_get_main(); + mm->vnet_main = vnet_get_main (); mm->vlib_main = vm; -#ifdef MAP_SKIP_IP6_LOOKUP - memset(&mm->preresolve_ip4, 0, sizeof(mm->preresolve_ip4)); - memset(&mm->preresolve_ip6, 0, sizeof(mm->preresolve_ip6)); +#ifdef MAP_SKIP_IP6_LOOKUP + memset (&mm->preresolve_ip4, 0, sizeof (mm->preresolve_ip4)); + memset (&mm->preresolve_ip6, 0, sizeof (mm->preresolve_ip6)); mm->adj4_index = 0; mm->adj6_index = 0; #endif @@ -1720,38 +1991,52 @@ clib_error_t *map_init (vlib_main_t *vm) mm->frag_inner = false; mm->frag_ignore_df = false; - vec_validate(mm->domain_counters, MAP_N_DOMAIN_COUNTER - 1); + vec_validate (mm->domain_counters, MAP_N_DOMAIN_COUNTER - 1); mm->domain_counters[MAP_DOMAIN_COUNTER_RX].name = "rx"; mm->domain_counters[MAP_DOMAIN_COUNTER_TX].name = "tx"; - vlib_validate_simple_counter(&mm->icmp_relayed, 0); - vlib_zero_simple_counter(&mm->icmp_relayed, 0); + vlib_validate_simple_counter (&mm->icmp_relayed, 0); + vlib_zero_simple_counter (&mm->icmp_relayed, 0); /* IP4 virtual reassembly */ mm->ip4_reass_hash_table = 0; mm->ip4_reass_pool = 0; - mm->ip4_reass_lock = clib_mem_alloc_aligned(CLIB_CACHE_LINE_BYTES, CLIB_CACHE_LINE_BYTES); + mm->ip4_reass_lock = + clib_mem_alloc_aligned (CLIB_CACHE_LINE_BYTES, CLIB_CACHE_LINE_BYTES); mm->ip4_reass_conf_ht_ratio = MAP_IP4_REASS_HT_RATIO_DEFAULT; mm->ip4_reass_conf_lifetime_ms = MAP_IP4_REASS_LIFETIME_DEFAULT; mm->ip4_reass_conf_pool_size = MAP_IP4_REASS_POOL_SIZE_DEFAULT; mm->ip4_reass_conf_buffers = MAP_IP4_REASS_BUFFERS_DEFAULT; - mm->ip4_reass_ht_log2len = map_get_ht_log2len(mm->ip4_reass_conf_ht_ratio, mm->ip4_reass_conf_pool_size); + mm->ip4_reass_ht_log2len = + map_get_ht_log2len (mm->ip4_reass_conf_ht_ratio, + mm->ip4_reass_conf_pool_size); mm->ip4_reass_fifo_last = MAP_REASS_INDEX_NONE; - map_ip4_reass_reinit(NULL, NULL); + map_ip4_reass_reinit (NULL, NULL); /* IP6 virtual reassembly */ mm->ip6_reass_hash_table = 0; mm->ip6_reass_pool = 0; - mm->ip6_reass_lock = clib_mem_alloc_aligned(CLIB_CACHE_LINE_BYTES, CLIB_CACHE_LINE_BYTES); + mm->ip6_reass_lock = + clib_mem_alloc_aligned (CLIB_CACHE_LINE_BYTES, CLIB_CACHE_LINE_BYTES); mm->ip6_reass_conf_ht_ratio = MAP_IP6_REASS_HT_RATIO_DEFAULT; mm->ip6_reass_conf_lifetime_ms = MAP_IP6_REASS_LIFETIME_DEFAULT; mm->ip6_reass_conf_pool_size = MAP_IP6_REASS_POOL_SIZE_DEFAULT; mm->ip6_reass_conf_buffers = MAP_IP6_REASS_BUFFERS_DEFAULT; - mm->ip6_reass_ht_log2len = map_get_ht_log2len(mm->ip6_reass_conf_ht_ratio, mm->ip6_reass_conf_pool_size); + mm->ip6_reass_ht_log2len = + map_get_ht_log2len (mm->ip6_reass_conf_ht_ratio, + mm->ip6_reass_conf_pool_size); mm->ip6_reass_fifo_last = MAP_REASS_INDEX_NONE; - map_ip6_reass_reinit(NULL, NULL); + map_ip6_reass_reinit (NULL, NULL); return 0; } -VLIB_INIT_FUNCTION(map_init); +VLIB_INIT_FUNCTION (map_init); + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/vnet/vnet/map/map.h b/vnet/vnet/map/map.h index a79da2cdabe..4b3df5f0842 100644 --- a/vnet/vnet/map/map.h +++ b/vnet/vnet/map/map.h @@ -20,26 +20,28 @@ #define MAP_SKIP_IP6_LOOKUP 1 -typedef enum { +typedef enum +{ MAP_SENDER, MAP_RECEIVER } map_dir_e; -int map_create_domain(ip4_address_t *ip4_prefix, u8 ip4_prefix_len, - ip6_address_t *ip6_prefix, u8 ip6_prefix_len, - ip6_address_t *ip6_src, u8 ip6_src_len, - u8 ea_bits_len, u8 psid_offset, u8 psid_length, - u32 *map_domain_index, u16 mtu, u8 flags); -int map_delete_domain(u32 map_domain_index); -int map_add_del_psid(u32 map_domain_index, u16 psid, ip6_address_t *tep, u8 is_add); -u8 *format_map_trace(u8 *s, va_list *args); -i32 ip4_get_port(ip4_header_t *ip, map_dir_e dir, u16 buffer_len); -i32 ip6_get_port(ip6_header_t *ip6, map_dir_e dir, u16 buffer_len); -u16 ip4_map_get_port (ip4_header_t *ip, map_dir_e dir); - -typedef enum __attribute__ ((__packed__)) { - MAP_DOMAIN_PREFIX = 1 << 0, - MAP_DOMAIN_TRANSLATION = 1 << 1, // The domain uses MAP-T +int map_create_domain (ip4_address_t * ip4_prefix, u8 ip4_prefix_len, + ip6_address_t * ip6_prefix, u8 ip6_prefix_len, + ip6_address_t * ip6_src, u8 ip6_src_len, + u8 ea_bits_len, u8 psid_offset, u8 psid_length, + u32 * map_domain_index, u16 mtu, u8 flags); +int map_delete_domain (u32 map_domain_index); +int map_add_del_psid (u32 map_domain_index, u16 psid, ip6_address_t * tep, + u8 is_add); +u8 *format_map_trace (u8 * s, va_list * args); +i32 ip4_get_port (ip4_header_t * ip, map_dir_e dir, u16 buffer_len); +i32 ip6_get_port (ip6_header_t * ip6, map_dir_e dir, u16 buffer_len); +u16 ip4_map_get_port (ip4_header_t * ip, map_dir_e dir); + +typedef enum __attribute__ ((__packed__)) +{ + MAP_DOMAIN_PREFIX = 1 << 0, MAP_DOMAIN_TRANSLATION = 1 << 1, // The domain uses MAP-T } map_domain_flags_e; /** @@ -55,16 +57,16 @@ typedef enum __attribute__ ((__packed__)) { * In case no structure can be allocated, the fragment is dropped. */ -#define MAP_IP4_REASS_LIFETIME_DEFAULT (100) /* ms */ +#define MAP_IP4_REASS_LIFETIME_DEFAULT (100) /* ms */ #define MAP_IP4_REASS_HT_RATIO_DEFAULT (1.0) -#define MAP_IP4_REASS_POOL_SIZE_DEFAULT 1024 // Number of reassembly structures +#define MAP_IP4_REASS_POOL_SIZE_DEFAULT 1024 // Number of reassembly structures #define MAP_IP4_REASS_BUFFERS_DEFAULT 2048 -#define MAP_IP4_REASS_MAX_FRAGMENTS_PER_REASSEMBLY 5 // Number of fragment per reassembly +#define MAP_IP4_REASS_MAX_FRAGMENTS_PER_REASSEMBLY 5 // Number of fragment per reassembly -#define MAP_IP6_REASS_LIFETIME_DEFAULT (100) /* ms */ +#define MAP_IP6_REASS_LIFETIME_DEFAULT (100) /* ms */ #define MAP_IP6_REASS_HT_RATIO_DEFAULT (1.0) -#define MAP_IP6_REASS_POOL_SIZE_DEFAULT 1024 // Number of reassembly structures +#define MAP_IP6_REASS_POOL_SIZE_DEFAULT 1024 // Number of reassembly structures #define MAP_IP6_REASS_BUFFERS_DEFAULT 2048 #define MAP_IP6_REASS_MAX_FRAGMENTS_PER_REASSEMBLY 5 @@ -78,7 +80,8 @@ typedef enum __attribute__ ((__packed__)) { * This structure _MUST_ be no larger than a single cache line (64 bytes). * If more space is needed make a union of ip6_prefix and *rules, those are mutually exclusive. */ -typedef struct { +typedef struct +{ ip6_address_t ip6_src; ip6_address_t ip6_prefix; ip6_address_t *rules; @@ -107,6 +110,7 @@ typedef struct { /* * Hash key, padded out to 16 bytes for fast compare */ +/* *INDENT-OFF* */ typedef union { CLIB_PACKED (struct { ip4_address_t src; @@ -117,8 +121,10 @@ typedef union { u64 as_u64[2]; u32 as_u32[4]; } map_ip4_reass_key_t; +/* *INDENT-ON* */ -typedef struct { +typedef struct +{ map_ip4_reass_key_t key; f64 ts; #ifdef MAP_IP4_REASS_COUNT_BYTES @@ -136,7 +142,8 @@ typedef struct { /* * MAP domain counters */ -typedef enum { +typedef enum +{ /* Simple counters */ MAP_DOMAIN_IPV4_FRAGMENT = 0, /* Combined counters */ @@ -148,6 +155,7 @@ typedef enum { /* * main_main_t */ +/* *INDENT-OFF* */ typedef union { CLIB_PACKED (struct { ip6_address_t src; @@ -158,6 +166,7 @@ typedef union { u64 as_u64[5]; u32 as_u32[10]; } map_ip6_reass_key_t; +/* *INDENT-OFF* */ typedef struct { u32 pi; //Cached packet or ~0 @@ -565,3 +574,11 @@ map_send_all_to_node(vlib_main_t *vm, u32 *pi_vector, vlib_put_next_frame(vm, node, next_index, n_left_to_next); } } + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ |