Ed Warnicke | cb9cada | 2015-12-08 15:45:58 -0700 | [diff] [blame] | 1 | /* |
| 2 | * Copyright (c) 2015 Cisco and/or its affiliates. |
| 3 | * Licensed under the Apache License, Version 2.0 (the "License"); |
| 4 | * you may not use this file except in compliance with the License. |
| 5 | * You may obtain a copy of the License at: |
| 6 | * |
| 7 | * http://www.apache.org/licenses/LICENSE-2.0 |
| 8 | * |
| 9 | * Unless required by applicable law or agreed to in writing, software |
| 10 | * distributed under the License is distributed on an "AS IS" BASIS, |
| 11 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 12 | * See the License for the specific language governing permissions and |
| 13 | * limitations under the License. |
| 14 | */ |
| 15 | /* |
| 16 | * Defines used for testing various optimisation schemes |
| 17 | */ |
| 18 | #define MAP_ENCAP_DUAL 0 |
| 19 | |
| 20 | #include "map.h" |
| 21 | #include "../ip/ip_frag.h" |
| 22 | |
| 23 | vlib_node_registration_t ip4_map_reass_node; |
| 24 | |
| 25 | enum ip4_map_next_e { |
| 26 | IP4_MAP_NEXT_IP6_LOOKUP, |
| 27 | #ifdef MAP_SKIP_IP6_LOOKUP |
| 28 | IP4_MAP_NEXT_IP6_REWRITE, |
| 29 | #endif |
| 30 | IP4_MAP_NEXT_FRAGMENT, |
| 31 | IP4_MAP_NEXT_REASS, |
| 32 | IP4_MAP_NEXT_DROP, |
| 33 | IP4_MAP_N_NEXT, |
| 34 | }; |
| 35 | |
| 36 | enum ip4_map_reass_next_t { |
| 37 | IP4_MAP_REASS_NEXT_IP6_LOOKUP, |
| 38 | IP4_MAP_REASS_NEXT_IP4_FRAGMENT, |
| 39 | IP4_MAP_REASS_NEXT_DROP, |
| 40 | IP4_MAP_REASS_N_NEXT, |
| 41 | }; |
| 42 | |
| 43 | typedef struct { |
| 44 | u32 map_domain_index; |
| 45 | u16 port; |
| 46 | u8 cached; |
| 47 | } map_ip4_map_reass_trace_t; |
| 48 | |
| 49 | u8 * |
| 50 | format_ip4_map_reass_trace (u8 *s, va_list *args) |
| 51 | { |
| 52 | CLIB_UNUSED(vlib_main_t *vm) = va_arg (*args, vlib_main_t *); |
| 53 | CLIB_UNUSED(vlib_node_t *node) = va_arg (*args, vlib_node_t *); |
| 54 | map_ip4_map_reass_trace_t *t = va_arg (*args, map_ip4_map_reass_trace_t *); |
| 55 | return format(s, "MAP domain index: %d L4 port: %u Status: %s", t->map_domain_index, |
| 56 | t->port, t->cached?"cached":"forwarded"); |
| 57 | } |
| 58 | |
| 59 | /* |
| 60 | * ip4_map_get_port |
| 61 | */ |
| 62 | u16 |
| 63 | ip4_map_get_port (ip4_header_t *ip, map_dir_e dir) |
| 64 | { |
| 65 | /* Find port information */ |
| 66 | if (PREDICT_TRUE((ip->protocol == IP_PROTOCOL_TCP) || |
| 67 | (ip->protocol == IP_PROTOCOL_UDP))) { |
| 68 | udp_header_t *udp = (void *)(ip + 1); |
| 69 | return (dir == MAP_SENDER ? udp->src_port : udp->dst_port); |
| 70 | } else if (ip->protocol == IP_PROTOCOL_ICMP) { |
| 71 | /* |
| 72 | * 1) ICMP Echo request or Echo reply |
| 73 | * 2) ICMP Error with inner packet being UDP or TCP |
| 74 | * 3) ICMP Error with inner packet being ICMP Echo request or Echo reply |
| 75 | */ |
| 76 | icmp46_header_t *icmp = (void *)(ip + 1); |
| 77 | if (icmp->type == ICMP4_echo_request || icmp->type == ICMP4_echo_reply) { |
| 78 | return *((u16 *)(icmp + 1)); |
| 79 | } else if (clib_net_to_host_u16(ip->length) >= 64) { // IP + ICMP + IP + L4 header |
| 80 | ip4_header_t *icmp_ip = (ip4_header_t *)(icmp + 2); |
| 81 | if (PREDICT_TRUE((icmp_ip->protocol == IP_PROTOCOL_TCP) || |
| 82 | (icmp_ip->protocol == IP_PROTOCOL_UDP))) { |
| 83 | udp_header_t *udp = (void *)(icmp_ip + 1); |
| 84 | return (dir == MAP_SENDER ? udp->dst_port : udp->src_port); |
| 85 | } else if (icmp_ip->protocol == IP_PROTOCOL_ICMP) { |
| 86 | icmp46_header_t *inner_icmp = (void *)(icmp_ip + 1); |
| 87 | if (inner_icmp->type == ICMP4_echo_request || inner_icmp->type == ICMP4_echo_reply) |
| 88 | return (*((u16 *)(inner_icmp + 1))); |
| 89 | } |
| 90 | } |
| 91 | } |
| 92 | return (0); |
| 93 | } |
| 94 | |
| 95 | static_always_inline u16 |
| 96 | ip4_map_port_and_security_check (map_domain_t *d, ip4_header_t *ip, u32 *next, u8 *error) |
| 97 | { |
| 98 | u16 port = 0; |
| 99 | |
| 100 | if (d->psid_length > 0) { |
| 101 | if (!ip4_is_fragment(ip)) { |
| 102 | if (PREDICT_FALSE((ip->ip_version_and_header_length != 0x45) || clib_host_to_net_u16(ip->length) < 28)) { |
| 103 | return 0; |
| 104 | } |
| 105 | port = ip4_map_get_port(ip, MAP_RECEIVER); |
| 106 | if (port) { |
| 107 | /* Verify that port is not among the well-known ports */ |
| 108 | if ((d->psid_offset > 0) && (clib_net_to_host_u16(port) < (0x1 << (16 - d->psid_offset)))) { |
| 109 | *error = MAP_ERROR_ENCAP_SEC_CHECK; |
| 110 | } else { |
| 111 | return (port); |
| 112 | } |
| 113 | } else { |
| 114 | *error = MAP_ERROR_BAD_PROTOCOL; |
| 115 | } |
| 116 | } else { |
| 117 | *next = IP4_MAP_NEXT_REASS; |
| 118 | } |
| 119 | } |
| 120 | return (0); |
| 121 | } |
| 122 | |
| 123 | /* |
| 124 | * ip4_map_vtcfl |
| 125 | */ |
| 126 | static_always_inline u32 |
| 127 | ip4_map_vtcfl (ip4_header_t *ip4, vlib_buffer_t *p) |
| 128 | { |
| 129 | map_main_t *mm = &map_main; |
| 130 | u8 tc = mm->tc_copy ? ip4->tos : mm->tc; |
| 131 | u32 vtcfl = 0x6 << 28; |
| 132 | vtcfl |= tc << 20; |
Damjan Marion | 2c29d75 | 2015-12-18 10:26:56 +0100 | [diff] [blame] | 133 | vtcfl |= vnet_buffer(p)->ip.flow_hash & 0x000fffff; |
Ed Warnicke | cb9cada | 2015-12-08 15:45:58 -0700 | [diff] [blame] | 134 | |
| 135 | return (clib_host_to_net_u32(vtcfl)); |
| 136 | } |
| 137 | |
| 138 | static_always_inline bool |
| 139 | ip4_map_ip6_lookup_bypass (vlib_buffer_t *p0, ip4_header_t *ip) |
| 140 | { |
| 141 | #ifdef MAP_SKIP_IP6_LOOKUP |
| 142 | map_main_t *mm = &map_main; |
| 143 | u32 adj_index0 = mm->adj6_index; |
| 144 | if (adj_index0 > 0) { |
| 145 | ip_lookup_main_t *lm6 = &ip6_main.lookup_main; |
| 146 | ip_adjacency_t *adj = ip_get_adjacency(lm6, mm->adj6_index); |
| 147 | if (adj->n_adj > 1) { |
| 148 | u32 hash_c0 = ip4_compute_flow_hash(ip, IP_FLOW_HASH_DEFAULT); |
| 149 | adj_index0 += (hash_c0 & (adj->n_adj - 1)); |
| 150 | } |
| 151 | vnet_buffer(p0)->ip.adj_index[VLIB_TX] = adj_index0; |
| 152 | return (true); |
| 153 | } |
| 154 | #endif |
| 155 | return (false); |
| 156 | } |
| 157 | |
| 158 | /* |
Ole Troan | 366ac6e | 2016-01-06 12:40:28 +0100 | [diff] [blame^] | 159 | * ip4_map_ttl |
| 160 | */ |
| 161 | static inline void |
| 162 | ip4_map_decrement_ttl (ip4_header_t *ip, u8 *error) |
| 163 | { |
| 164 | i32 ttl = ip->ttl; |
| 165 | |
| 166 | /* Input node should have reject packets with ttl 0. */ |
| 167 | ASSERT (ip->ttl > 0); |
| 168 | |
| 169 | u32 checksum = ip->checksum + clib_host_to_net_u16(0x0100); |
| 170 | checksum += checksum >= 0xffff; |
| 171 | ip->checksum = checksum; |
| 172 | ttl -= 1; |
| 173 | ip->ttl = ttl; |
| 174 | *error = ttl <= 0 ? IP4_ERROR_TIME_EXPIRED : *error; |
| 175 | |
| 176 | /* Verify checksum. */ |
| 177 | ASSERT (ip->checksum == ip4_header_checksum(ip)); |
| 178 | } |
| 179 | |
| 180 | /* |
Ed Warnicke | cb9cada | 2015-12-08 15:45:58 -0700 | [diff] [blame] | 181 | * ip4_map |
| 182 | */ |
| 183 | static uword |
| 184 | ip4_map (vlib_main_t *vm, |
| 185 | vlib_node_runtime_t *node, |
| 186 | vlib_frame_t *frame) |
| 187 | { |
| 188 | u32 n_left_from, *from, next_index, *to_next, n_left_to_next; |
| 189 | vlib_node_runtime_t *error_node = vlib_node_get_runtime(vm, ip4_map_node.index); |
| 190 | from = vlib_frame_vector_args(frame); |
| 191 | n_left_from = frame->n_vectors; |
| 192 | next_index = node->cached_next_index; |
| 193 | map_main_t *mm = &map_main; |
| 194 | vlib_combined_counter_main_t *cm = mm->domain_counters; |
| 195 | u32 cpu_index = os_get_cpu_number(); |
| 196 | |
| 197 | while (n_left_from > 0) { |
| 198 | vlib_get_next_frame(vm, node, next_index, to_next, n_left_to_next); |
| 199 | |
| 200 | /* Dual loop */ |
Ole Troan | a5d2c70 | 2016-01-05 21:26:17 +0100 | [diff] [blame] | 201 | while (n_left_from >= 4 && n_left_to_next >= 2) { |
Ed Warnicke | cb9cada | 2015-12-08 15:45:58 -0700 | [diff] [blame] | 202 | u32 pi0, pi1; |
| 203 | vlib_buffer_t *p0, *p1; |
| 204 | map_domain_t *d0, *d1; |
| 205 | u8 error0 = MAP_ERROR_NONE, error1 = MAP_ERROR_NONE; |
| 206 | ip4_header_t *ip40, *ip41; |
| 207 | u16 port0 = 0, port1 = 0; |
| 208 | ip6_header_t *ip6h0, *ip6h1; |
| 209 | u32 map_domain_index0 = ~0, map_domain_index1 = ~0; |
| 210 | u32 next0 = IP4_MAP_NEXT_IP6_LOOKUP, next1 = IP4_MAP_NEXT_IP6_LOOKUP; |
| 211 | |
| 212 | /* Prefetch next iteration. */ |
| 213 | { |
| 214 | vlib_buffer_t *p2, *p3; |
| 215 | |
| 216 | p2 = vlib_get_buffer(vm, from[2]); |
| 217 | p3 = vlib_get_buffer(vm, from[3]); |
| 218 | |
| 219 | vlib_prefetch_buffer_header(p2, STORE); |
| 220 | vlib_prefetch_buffer_header(p3, STORE); |
| 221 | /* IPv4 + 8 = 28. possibly plus -40 */ |
| 222 | CLIB_PREFETCH (p2->data-40, 68, STORE); |
| 223 | CLIB_PREFETCH (p3->data-40, 68, STORE); |
| 224 | } |
| 225 | |
| 226 | pi0 = to_next[0] = from[0]; |
| 227 | pi1 = to_next[1] = from[1]; |
| 228 | from += 2; |
| 229 | n_left_from -= 2; |
| 230 | to_next +=2; |
| 231 | n_left_to_next -= 2; |
| 232 | |
| 233 | p0 = vlib_get_buffer(vm, pi0); |
| 234 | p1 = vlib_get_buffer(vm, pi1); |
| 235 | ip40 = vlib_buffer_get_current(p0); |
| 236 | ip41 = vlib_buffer_get_current(p1); |
| 237 | p0->current_length = clib_net_to_host_u16(ip40->length); |
| 238 | p1->current_length = clib_net_to_host_u16(ip41->length); |
| 239 | d0 = ip4_map_get_domain(vnet_buffer(p0)->ip.adj_index[VLIB_TX], &map_domain_index0); |
| 240 | d1 = ip4_map_get_domain(vnet_buffer(p1)->ip.adj_index[VLIB_TX], &map_domain_index1); |
| 241 | ASSERT(d0); |
| 242 | ASSERT(d1); |
| 243 | |
| 244 | /* |
| 245 | * Shared IPv4 address |
| 246 | */ |
| 247 | port0 = ip4_map_port_and_security_check(d0, ip40, &next0, &error0); |
| 248 | port1 = ip4_map_port_and_security_check(d1, ip41, &next1, &error1); |
| 249 | |
| 250 | /* MAP calc */ |
| 251 | u32 da40 = clib_net_to_host_u32(ip40->dst_address.as_u32); |
| 252 | u32 da41 = clib_net_to_host_u32(ip41->dst_address.as_u32); |
| 253 | u16 dp40 = clib_net_to_host_u16(port0); |
| 254 | u16 dp41 = clib_net_to_host_u16(port1); |
| 255 | u64 dal60 = map_get_pfx(d0, da40, dp40); |
| 256 | u64 dal61 = map_get_pfx(d1, da41, dp41); |
| 257 | u64 dar60 = map_get_sfx(d0, da40, dp40); |
| 258 | u64 dar61 = map_get_sfx(d1, da41, dp41); |
| 259 | if (dal60 == 0 && dar60 == 0) error0 = MAP_ERROR_UNKNOWN; |
| 260 | if (dal61 == 0 && dar61 == 0) error1 = MAP_ERROR_UNKNOWN; |
| 261 | |
| 262 | /* construct ipv6 header */ |
| 263 | vlib_buffer_advance(p0, - sizeof(ip6_header_t)); |
| 264 | vlib_buffer_advance(p1, - sizeof(ip6_header_t)); |
| 265 | ip6h0 = vlib_buffer_get_current(p0); |
| 266 | ip6h1 = vlib_buffer_get_current(p1); |
| 267 | vnet_buffer(p0)->sw_if_index[VLIB_TX] = (u32)~0; |
| 268 | vnet_buffer(p1)->sw_if_index[VLIB_TX] = (u32)~0; |
| 269 | |
| 270 | ip6h0->ip_version_traffic_class_and_flow_label = ip4_map_vtcfl(ip40, p0); |
| 271 | ip6h1->ip_version_traffic_class_and_flow_label = ip4_map_vtcfl(ip41, p1); |
| 272 | ip6h0->payload_length = ip40->length; |
| 273 | ip6h1->payload_length = ip41->length; |
| 274 | ip6h0->protocol = IP_PROTOCOL_IP_IN_IP; |
| 275 | ip6h1->protocol = IP_PROTOCOL_IP_IN_IP; |
| 276 | ip6h0->hop_limit = 0x40; |
| 277 | ip6h1->hop_limit = 0x40; |
| 278 | ip6h0->src_address = d0->ip6_src; |
| 279 | ip6h1->src_address = d1->ip6_src; |
| 280 | ip6h0->dst_address.as_u64[0] = clib_host_to_net_u64(dal60); |
| 281 | ip6h0->dst_address.as_u64[1] = clib_host_to_net_u64(dar60); |
| 282 | ip6h1->dst_address.as_u64[0] = clib_host_to_net_u64(dal61); |
| 283 | ip6h1->dst_address.as_u64[1] = clib_host_to_net_u64(dar61); |
| 284 | |
| 285 | /* |
| 286 | * Determine next node. Can be one of: |
| 287 | * ip6-lookup, ip6-rewrite, ip4-fragment, ip4-virtreass, error-drop |
| 288 | */ |
| 289 | if (PREDICT_TRUE(error0 == MAP_ERROR_NONE)) { |
| 290 | if (PREDICT_FALSE(d0->mtu && (clib_net_to_host_u16(ip6h0->payload_length) + sizeof(*ip6h0) > d0->mtu))) { |
| 291 | vnet_buffer(p0)->ip_frag.header_offset = sizeof(*ip6h0); |
| 292 | vnet_buffer(p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP6_LOOKUP; |
| 293 | vnet_buffer(p0)->ip_frag.mtu = d0->mtu; |
| 294 | vnet_buffer(p0)->ip_frag.flags = IP_FRAG_FLAG_IP6_HEADER; |
| 295 | next0 = IP4_MAP_NEXT_FRAGMENT; |
| 296 | } else { |
| 297 | next0 = ip4_map_ip6_lookup_bypass(p0, ip40) ? IP4_MAP_NEXT_IP6_REWRITE : next0; |
| 298 | vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_TX, cpu_index, map_domain_index0, 1, |
| 299 | clib_net_to_host_u16(ip6h0->payload_length) + 40); |
| 300 | } |
| 301 | } else { |
| 302 | next0 = IP4_MAP_NEXT_DROP; |
| 303 | } |
| 304 | |
| 305 | /* |
| 306 | * Determine next node. Can be one of: |
| 307 | * ip6-lookup, ip6-rewrite, ip4-fragment, ip4-virtreass, error-drop |
| 308 | */ |
| 309 | if (PREDICT_TRUE(error1 == MAP_ERROR_NONE)) { |
| 310 | if (PREDICT_FALSE(d1->mtu && (clib_net_to_host_u16(ip6h1->payload_length) + sizeof(*ip6h1) > d1->mtu))) { |
| 311 | vnet_buffer(p1)->ip_frag.header_offset = sizeof(*ip6h1); |
| 312 | vnet_buffer(p1)->ip_frag.next_index = IP4_FRAG_NEXT_IP6_LOOKUP; |
| 313 | vnet_buffer(p1)->ip_frag.mtu = d1->mtu; |
| 314 | vnet_buffer(p1)->ip_frag.flags = IP_FRAG_FLAG_IP6_HEADER; |
| 315 | next1 = IP4_MAP_NEXT_FRAGMENT; |
| 316 | } else { |
| 317 | next1 = ip4_map_ip6_lookup_bypass(p1, ip41) ? IP4_MAP_NEXT_IP6_REWRITE : next1; |
| 318 | vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_TX, cpu_index, map_domain_index1, 1, |
| 319 | clib_net_to_host_u16(ip6h1->payload_length) + 40); |
| 320 | } |
| 321 | } else { |
| 322 | next1 = IP4_MAP_NEXT_DROP; |
| 323 | } |
| 324 | |
| 325 | if (PREDICT_FALSE(p0->flags & VLIB_BUFFER_IS_TRACED)) { |
| 326 | map_trace_t *tr = vlib_add_trace(vm, node, p0, sizeof(*tr)); |
| 327 | tr->map_domain_index = map_domain_index0; |
| 328 | tr->port = port0; |
| 329 | } |
| 330 | if (PREDICT_FALSE(p1->flags & VLIB_BUFFER_IS_TRACED)) { |
| 331 | map_trace_t *tr = vlib_add_trace(vm, node, p1, sizeof(*tr)); |
| 332 | tr->map_domain_index = map_domain_index1; |
| 333 | tr->port = port1; |
| 334 | } |
| 335 | |
| 336 | p0->error = error_node->errors[error0]; |
| 337 | p1->error = error_node->errors[error1]; |
| 338 | |
| 339 | vlib_validate_buffer_enqueue_x2(vm, node, next_index, to_next, n_left_to_next, pi0, pi1, next0, next1); |
| 340 | } |
| 341 | |
| 342 | while (n_left_from > 0 && n_left_to_next > 0) { |
| 343 | u32 pi0; |
| 344 | vlib_buffer_t *p0; |
| 345 | map_domain_t *d0; |
| 346 | u8 error0 = MAP_ERROR_NONE; |
| 347 | ip4_header_t *ip40; |
| 348 | u16 port0 = 0; |
| 349 | ip6_header_t *ip6h0; |
| 350 | u32 next0 = IP4_MAP_NEXT_IP6_LOOKUP; |
| 351 | u32 map_domain_index0 = ~0; |
| 352 | |
| 353 | pi0 = to_next[0] = from[0]; |
| 354 | from += 1; |
| 355 | n_left_from -= 1; |
| 356 | to_next +=1; |
| 357 | n_left_to_next -= 1; |
| 358 | |
| 359 | p0 = vlib_get_buffer(vm, pi0); |
| 360 | ip40 = vlib_buffer_get_current(p0); |
| 361 | p0->current_length = clib_net_to_host_u16(ip40->length); |
| 362 | d0 = ip4_map_get_domain(vnet_buffer(p0)->ip.adj_index[VLIB_TX], &map_domain_index0); |
| 363 | ASSERT(d0); |
| 364 | |
| 365 | /* |
| 366 | * Shared IPv4 address |
| 367 | */ |
| 368 | port0 = ip4_map_port_and_security_check(d0, ip40, &next0, &error0); |
| 369 | |
Ole Troan | 366ac6e | 2016-01-06 12:40:28 +0100 | [diff] [blame^] | 370 | /* Decrement IPv4 TTL */ |
| 371 | ip4_map_decrement_ttl(ip40, &error0); |
| 372 | |
Ed Warnicke | cb9cada | 2015-12-08 15:45:58 -0700 | [diff] [blame] | 373 | /* MAP calc */ |
| 374 | u32 da40 = clib_net_to_host_u32(ip40->dst_address.as_u32); |
| 375 | u16 dp40 = clib_net_to_host_u16(port0); |
| 376 | u64 dal60 = map_get_pfx(d0, da40, dp40); |
| 377 | u64 dar60 = map_get_sfx(d0, da40, dp40); |
| 378 | if (dal60 == 0 && dar60 == 0 && error0 == MAP_ERROR_NONE) error0 = MAP_ERROR_UNKNOWN; |
| 379 | |
| 380 | /* construct ipv6 header */ |
| 381 | vlib_buffer_advance(p0, - (sizeof(ip6_header_t))); |
| 382 | ip6h0 = vlib_buffer_get_current(p0); |
| 383 | vnet_buffer(p0)->sw_if_index[VLIB_TX] = (u32)~0; |
| 384 | |
| 385 | ip6h0->ip_version_traffic_class_and_flow_label = ip4_map_vtcfl(ip40, p0); |
| 386 | ip6h0->payload_length = ip40->length; |
| 387 | ip6h0->protocol = IP_PROTOCOL_IP_IN_IP; |
| 388 | ip6h0->hop_limit = 0x40; |
| 389 | ip6h0->src_address = d0->ip6_src; |
| 390 | ip6h0->dst_address.as_u64[0] = clib_host_to_net_u64(dal60); |
| 391 | ip6h0->dst_address.as_u64[1] = clib_host_to_net_u64(dar60); |
| 392 | |
| 393 | /* |
| 394 | * Determine next node. Can be one of: |
| 395 | * ip6-lookup, ip6-rewrite, ip4-fragment, ip4-virtreass, error-drop |
| 396 | */ |
| 397 | if (PREDICT_TRUE(error0 == MAP_ERROR_NONE)) { |
| 398 | if (PREDICT_FALSE(d0->mtu && (clib_net_to_host_u16(ip6h0->payload_length) + sizeof(*ip6h0) > d0->mtu))) { |
| 399 | vnet_buffer(p0)->ip_frag.header_offset = sizeof(*ip6h0); |
| 400 | vnet_buffer(p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP6_LOOKUP; |
| 401 | vnet_buffer(p0)->ip_frag.mtu = d0->mtu; |
| 402 | vnet_buffer(p0)->ip_frag.flags = IP_FRAG_FLAG_IP6_HEADER; |
| 403 | next0 = IP4_MAP_NEXT_FRAGMENT; |
| 404 | } else { |
| 405 | next0 = ip4_map_ip6_lookup_bypass(p0, ip40) ? IP4_MAP_NEXT_IP6_REWRITE : next0; |
| 406 | vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_TX, cpu_index, map_domain_index0, 1, |
| 407 | clib_net_to_host_u16(ip6h0->payload_length) + 40); |
| 408 | } |
| 409 | } else { |
| 410 | next0 = IP4_MAP_NEXT_DROP; |
| 411 | } |
| 412 | |
| 413 | if (PREDICT_FALSE(p0->flags & VLIB_BUFFER_IS_TRACED)) { |
| 414 | map_trace_t *tr = vlib_add_trace(vm, node, p0, sizeof(*tr)); |
| 415 | tr->map_domain_index = map_domain_index0; |
| 416 | tr->port = port0; |
| 417 | } |
| 418 | |
| 419 | p0->error = error_node->errors[error0]; |
| 420 | vlib_validate_buffer_enqueue_x1(vm, node, next_index, to_next, n_left_to_next, pi0, next0); |
| 421 | } |
| 422 | vlib_put_next_frame(vm, node, next_index, n_left_to_next); |
| 423 | } |
| 424 | |
| 425 | return frame->n_vectors; |
| 426 | } |
| 427 | |
| 428 | /* |
| 429 | * ip4_map_reass |
| 430 | */ |
| 431 | static uword |
| 432 | ip4_map_reass (vlib_main_t *vm, |
| 433 | vlib_node_runtime_t *node, |
| 434 | vlib_frame_t *frame) |
| 435 | { |
| 436 | u32 n_left_from, *from, next_index, *to_next, n_left_to_next; |
| 437 | vlib_node_runtime_t *error_node = vlib_node_get_runtime(vm, ip4_map_reass_node.index); |
| 438 | from = vlib_frame_vector_args(frame); |
| 439 | n_left_from = frame->n_vectors; |
| 440 | next_index = node->cached_next_index; |
| 441 | map_main_t *mm = &map_main; |
| 442 | vlib_combined_counter_main_t *cm = mm->domain_counters; |
| 443 | u32 cpu_index = os_get_cpu_number(); |
| 444 | u32 *fragments_to_drop = NULL; |
| 445 | u32 *fragments_to_loopback = NULL; |
| 446 | |
| 447 | while (n_left_from > 0) { |
| 448 | vlib_get_next_frame(vm, node, next_index, to_next, n_left_to_next); |
| 449 | |
| 450 | while (n_left_from > 0 && n_left_to_next > 0) { |
| 451 | u32 pi0; |
| 452 | vlib_buffer_t *p0; |
| 453 | map_domain_t *d0; |
| 454 | u8 error0 = MAP_ERROR_NONE; |
| 455 | ip4_header_t *ip40; |
| 456 | i32 port0 = 0; |
| 457 | ip6_header_t *ip60; |
| 458 | u32 next0 = IP4_MAP_REASS_NEXT_IP6_LOOKUP; |
| 459 | u32 map_domain_index0; |
| 460 | u8 cached = 0; |
| 461 | |
| 462 | pi0 = to_next[0] = from[0]; |
| 463 | from += 1; |
| 464 | n_left_from -= 1; |
| 465 | to_next +=1; |
| 466 | n_left_to_next -= 1; |
| 467 | |
| 468 | p0 = vlib_get_buffer(vm, pi0); |
| 469 | ip60 = vlib_buffer_get_current(p0); |
| 470 | ip40 = (ip4_header_t *)(ip60 + 1); |
| 471 | d0 = ip4_map_get_domain(vnet_buffer(p0)->ip.adj_index[VLIB_TX], &map_domain_index0); |
| 472 | |
| 473 | map_ip4_reass_lock(); |
| 474 | map_ip4_reass_t *r = map_ip4_reass_get(ip40->src_address.as_u32, ip40->dst_address.as_u32, |
| 475 | ip40->fragment_id, ip40->protocol, &fragments_to_drop); |
| 476 | if (PREDICT_FALSE(!r)) { |
| 477 | // Could not create a caching entry |
| 478 | error0 = MAP_ERROR_FRAGMENT_MEMORY; |
| 479 | } else if (PREDICT_TRUE(ip4_get_fragment_offset(ip40))) { |
| 480 | if (r->port >= 0) { |
| 481 | // We know the port already |
| 482 | port0 = r->port; |
| 483 | } else if (map_ip4_reass_add_fragment(r, pi0)) { |
| 484 | // Not enough space for caching |
| 485 | error0 = MAP_ERROR_FRAGMENT_MEMORY; |
| 486 | map_ip4_reass_free(r, &fragments_to_drop); |
| 487 | } else { |
| 488 | cached = 1; |
| 489 | } |
| 490 | } else if ((port0 = ip4_get_port(ip40, MAP_RECEIVER, p0->current_length)) < 0) { |
| 491 | // Could not find port. We'll free the reassembly. |
| 492 | error0 = MAP_ERROR_BAD_PROTOCOL; |
| 493 | port0 = 0; |
| 494 | map_ip4_reass_free(r, &fragments_to_drop); |
| 495 | } else { |
| 496 | r->port = port0; |
| 497 | map_ip4_reass_get_fragments(r, &fragments_to_loopback); |
| 498 | } |
| 499 | |
| 500 | #ifdef MAP_IP4_REASS_COUNT_BYTES |
| 501 | if (!cached && r) { |
| 502 | r->forwarded += clib_host_to_net_u16(ip40->length) - 20; |
| 503 | if (!ip4_get_fragment_more(ip40)) |
| 504 | r->expected_total = ip4_get_fragment_offset(ip40) * 8 + clib_host_to_net_u16(ip40->length) - 20; |
| 505 | if(r->forwarded >= r->expected_total) |
| 506 | map_ip4_reass_free(r, &fragments_to_drop); |
| 507 | } |
| 508 | #endif |
| 509 | |
| 510 | map_ip4_reass_unlock(); |
| 511 | |
| 512 | // NOTE: Most operations have already been performed by ip4_map |
| 513 | // All we need is the right destination address |
| 514 | ip60->dst_address.as_u64[0] = map_get_pfx_net(d0, ip40->dst_address.as_u32, port0); |
| 515 | ip60->dst_address.as_u64[1] = map_get_sfx_net(d0, ip40->dst_address.as_u32, port0); |
| 516 | |
| 517 | if (PREDICT_FALSE(d0->mtu && (clib_net_to_host_u16(ip60->payload_length) + sizeof(*ip60) > d0->mtu))) { |
| 518 | vnet_buffer(p0)->ip_frag.header_offset = sizeof(*ip60); |
| 519 | vnet_buffer(p0)->ip_frag.next_index = IP4_FRAG_NEXT_IP6_LOOKUP; |
| 520 | vnet_buffer(p0)->ip_frag.mtu = d0->mtu; |
| 521 | vnet_buffer(p0)->ip_frag.flags = IP_FRAG_FLAG_IP6_HEADER; |
| 522 | next0 = IP4_MAP_REASS_NEXT_IP4_FRAGMENT; |
| 523 | } |
| 524 | |
| 525 | if (PREDICT_FALSE(p0->flags & VLIB_BUFFER_IS_TRACED)) { |
| 526 | map_ip4_map_reass_trace_t *tr = vlib_add_trace(vm, node, p0, sizeof(*tr)); |
| 527 | tr->map_domain_index = map_domain_index0; |
| 528 | tr->port = port0; |
| 529 | tr->cached = cached; |
| 530 | } |
| 531 | |
| 532 | if(cached) { |
| 533 | //Dequeue the packet |
| 534 | n_left_to_next++; |
| 535 | to_next--; |
| 536 | } else { |
| 537 | if (error0 == MAP_ERROR_NONE) |
| 538 | vlib_increment_combined_counter(cm + MAP_DOMAIN_COUNTER_TX, cpu_index, map_domain_index0, 1, |
| 539 | clib_net_to_host_u16(ip60->payload_length) + 40); |
| 540 | next0 = (error0 == MAP_ERROR_NONE) ? next0 : IP4_MAP_REASS_NEXT_DROP; |
| 541 | p0->error = error_node->errors[error0]; |
| 542 | vlib_validate_buffer_enqueue_x1(vm, node, next_index, to_next, n_left_to_next, pi0, next0); |
| 543 | } |
| 544 | |
| 545 | //Loopback when we reach the end of the inpu vector |
| 546 | if(n_left_from == 0 && vec_len(fragments_to_loopback)) { |
| 547 | from = vlib_frame_vector_args(frame); |
| 548 | u32 len = vec_len(fragments_to_loopback); |
| 549 | if(len <= VLIB_FRAME_SIZE) { |
| 550 | memcpy(from, fragments_to_loopback, sizeof(u32)*len); |
| 551 | n_left_from = len; |
| 552 | vec_reset_length(fragments_to_loopback); |
| 553 | } else { |
| 554 | memcpy(from, fragments_to_loopback + (len - VLIB_FRAME_SIZE), sizeof(u32)*VLIB_FRAME_SIZE); |
| 555 | n_left_from = VLIB_FRAME_SIZE; |
| 556 | _vec_len(fragments_to_loopback) = len - VLIB_FRAME_SIZE; |
| 557 | } |
| 558 | } |
| 559 | } |
| 560 | vlib_put_next_frame(vm, node, next_index, n_left_to_next); |
| 561 | } |
| 562 | |
| 563 | map_send_all_to_node(vm, fragments_to_drop, node, |
| 564 | &error_node->errors[MAP_ERROR_FRAGMENT_DROPPED], |
| 565 | IP4_MAP_REASS_NEXT_DROP); |
| 566 | |
| 567 | vec_free(fragments_to_drop); |
| 568 | vec_free(fragments_to_loopback); |
| 569 | return frame->n_vectors; |
| 570 | } |
| 571 | |
| 572 | static char *map_error_strings[] = { |
| 573 | #define _(sym,string) string, |
| 574 | foreach_map_error |
| 575 | #undef _ |
| 576 | }; |
| 577 | |
| 578 | VLIB_REGISTER_NODE(ip4_map_node) = { |
| 579 | .function = ip4_map, |
| 580 | .name = "ip4-map", |
| 581 | .vector_size = sizeof(u32), |
| 582 | .format_trace = format_map_trace, |
| 583 | .type = VLIB_NODE_TYPE_INTERNAL, |
| 584 | |
| 585 | .n_errors = MAP_N_ERROR, |
| 586 | .error_strings = map_error_strings, |
| 587 | |
| 588 | .n_next_nodes = IP4_MAP_N_NEXT, |
| 589 | .next_nodes = { |
| 590 | [IP4_MAP_NEXT_IP6_LOOKUP] = "ip6-lookup", |
| 591 | #ifdef MAP_SKIP_IP6_LOOKUP |
| 592 | [IP4_MAP_NEXT_IP6_REWRITE] = "ip6-rewrite", |
| 593 | #endif |
| 594 | [IP4_MAP_NEXT_FRAGMENT] = "ip4-frag", |
| 595 | [IP4_MAP_NEXT_REASS] = "ip4-map-reass", |
| 596 | [IP4_MAP_NEXT_DROP] = "error-drop", |
| 597 | }, |
| 598 | }; |
| 599 | |
| 600 | VLIB_REGISTER_NODE(ip4_map_reass_node) = { |
| 601 | .function = ip4_map_reass, |
| 602 | .name = "ip4-map-reass", |
| 603 | .vector_size = sizeof(u32), |
| 604 | .format_trace = format_ip4_map_reass_trace, |
| 605 | .type = VLIB_NODE_TYPE_INTERNAL, |
| 606 | |
| 607 | .n_errors = MAP_N_ERROR, |
| 608 | .error_strings = map_error_strings, |
| 609 | |
| 610 | .n_next_nodes = IP4_MAP_REASS_N_NEXT, |
| 611 | .next_nodes = { |
| 612 | [IP4_MAP_REASS_NEXT_IP6_LOOKUP] = "ip6-lookup", |
| 613 | [IP4_MAP_REASS_NEXT_IP4_FRAGMENT] = "ip4-frag", |
| 614 | [IP4_MAP_REASS_NEXT_DROP] = "error-drop", |
| 615 | }, |
| 616 | }; |