1 /*- 2 * BSD LICENSE 3 * 4 * Copyright(c) 2013 6WIND 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 11 * * Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * * Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in 15 * the documentation and/or other materials provided with the 16 * distribution. 17 * * Neither the name of 6WIND S.A. nor the names of its 18 * contributors may be used to endorse or promote products derived 19 * from this software without specific prior written permission. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 22 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 23 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR 24 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT 25 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, 26 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 27 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 28 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 29 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 30 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE 31 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 32 * 33 */ 34 35 #include <stdarg.h> 36 #include <string.h> 37 #include <stdio.h> 38 #include <errno.h> 39 #include <stdint.h> 40 #include <unistd.h> 41 #include <inttypes.h> 42 43 #include <sys/queue.h> 44 #include <sys/stat.h> 45 46 #include <rte_common.h> 47 #include <rte_byteorder.h> 48 #include <rte_log.h> 49 #include <rte_debug.h> 50 #include <rte_cycles.h> 51 #include <rte_per_lcore.h> 52 #include <rte_lcore.h> 53 #include <rte_atomic.h> 54 #include <rte_branch_prediction.h> 55 #include <rte_memory.h> 56 #include <rte_mempool.h> 57 #include <rte_mbuf.h> 58 #include <rte_ether.h> 59 #include <rte_ethdev.h> 60 #include <rte_arp.h> 61 #include <rte_ip.h> 62 #include <rte_icmp.h> 63 #include <rte_string_fns.h> 64 #include <rte_flow.h> 65 66 #include "testpmd.h" 67 68 static const char * 69 arp_op_name(uint16_t arp_op) 70 { 71 switch (arp_op ) { 72 case ARP_OP_REQUEST: 73 return "ARP Request"; 74 case ARP_OP_REPLY: 75 return "ARP Reply"; 76 case ARP_OP_REVREQUEST: 77 return "Reverse ARP Request"; 78 case ARP_OP_REVREPLY: 79 return "Reverse ARP Reply"; 80 case ARP_OP_INVREQUEST: 81 return "Peer Identify Request"; 82 case ARP_OP_INVREPLY: 83 return "Peer Identify Reply"; 84 default: 85 break; 86 } 87 return "Unkwown ARP op"; 88 } 89 90 static const char * 91 ip_proto_name(uint16_t ip_proto) 92 { 93 static const char * ip_proto_names[] = { 94 "IP6HOPOPTS", /**< IP6 hop-by-hop options */ 95 "ICMP", /**< control message protocol */ 96 "IGMP", /**< group mgmt protocol */ 97 "GGP", /**< gateway^2 (deprecated) */ 98 "IPv4", /**< IPv4 encapsulation */ 99 100 "UNASSIGNED", 101 "TCP", /**< transport control protocol */ 102 "ST", /**< Stream protocol II */ 103 "EGP", /**< exterior gateway protocol */ 104 "PIGP", /**< private interior gateway */ 105 106 "RCC_MON", /**< BBN RCC Monitoring */ 107 "NVPII", /**< network voice protocol*/ 108 "PUP", /**< pup */ 109 "ARGUS", /**< Argus */ 110 "EMCON", /**< EMCON */ 111 112 "XNET", /**< Cross Net Debugger */ 113 "CHAOS", /**< Chaos*/ 114 "UDP", /**< user datagram protocol */ 115 "MUX", /**< Multiplexing */ 116 "DCN_MEAS", /**< DCN Measurement Subsystems */ 117 118 "HMP", /**< Host Monitoring */ 119 "PRM", /**< Packet Radio Measurement */ 120 "XNS_IDP", /**< xns idp */ 121 "TRUNK1", /**< Trunk-1 */ 122 "TRUNK2", /**< Trunk-2 */ 123 124 "LEAF1", /**< Leaf-1 */ 125 "LEAF2", /**< Leaf-2 */ 126 "RDP", /**< Reliable Data */ 127 "IRTP", /**< Reliable Transaction */ 128 "TP4", /**< tp-4 w/ class negotiation */ 129 130 "BLT", /**< Bulk Data Transfer */ 131 "NSP", /**< Network Services */ 132 "INP", /**< Merit Internodal */ 133 "SEP", /**< Sequential Exchange */ 134 "3PC", /**< Third Party Connect */ 135 136 "IDPR", /**< InterDomain Policy Routing */ 137 "XTP", /**< XTP */ 138 "DDP", /**< Datagram Delivery */ 139 "CMTP", /**< Control Message Transport */ 140 "TPXX", /**< TP++ Transport */ 141 142 "ILTP", /**< IL transport protocol */ 143 "IPv6_HDR", /**< IP6 header */ 144 "SDRP", /**< Source Demand Routing */ 145 "IPv6_RTG", /**< IP6 routing header */ 146 "IPv6_FRAG", /**< IP6 fragmentation header */ 147 148 "IDRP", /**< InterDomain Routing*/ 149 "RSVP", /**< resource reservation */ 150 "GRE", /**< General Routing Encap. */ 151 "MHRP", /**< Mobile Host Routing */ 152 "BHA", /**< BHA */ 153 154 "ESP", /**< IP6 Encap Sec. Payload */ 155 "AH", /**< IP6 Auth Header */ 156 "INLSP", /**< Integ. Net Layer Security */ 157 "SWIPE", /**< IP with encryption */ 158 "NHRP", /**< Next Hop Resolution */ 159 160 "UNASSIGNED", 161 "UNASSIGNED", 162 "UNASSIGNED", 163 "ICMPv6", /**< ICMP6 */ 164 "IPv6NONEXT", /**< IP6 no next header */ 165 166 "Ipv6DSTOPTS",/**< IP6 destination option */ 167 "AHIP", /**< any host internal protocol */ 168 "CFTP", /**< CFTP */ 169 "HELLO", /**< "hello" routing protocol */ 170 "SATEXPAK", /**< SATNET/Backroom EXPAK */ 171 172 "KRYPTOLAN", /**< Kryptolan */ 173 "RVD", /**< Remote Virtual Disk */ 174 "IPPC", /**< Pluribus Packet Core */ 175 "ADFS", /**< Any distributed FS */ 176 "SATMON", /**< Satnet Monitoring */ 177 178 "VISA", /**< VISA Protocol */ 179 "IPCV", /**< Packet Core Utility */ 180 "CPNX", /**< Comp. Prot. Net. Executive */ 181 "CPHB", /**< Comp. Prot. HeartBeat */ 182 "WSN", /**< Wang Span Network */ 183 184 "PVP", /**< Packet Video Protocol */ 185 "BRSATMON", /**< BackRoom SATNET Monitoring */ 186 "ND", /**< Sun net disk proto (temp.) */ 187 "WBMON", /**< WIDEBAND Monitoring */ 188 "WBEXPAK", /**< WIDEBAND EXPAK */ 189 190 "EON", /**< ISO cnlp */ 191 "VMTP", /**< VMTP */ 192 "SVMTP", /**< Secure VMTP */ 193 "VINES", /**< Banyon VINES */ 194 "TTP", /**< TTP */ 195 196 "IGP", /**< NSFNET-IGP */ 197 "DGP", /**< dissimilar gateway prot. */ 198 "TCF", /**< TCF */ 199 "IGRP", /**< Cisco/GXS IGRP */ 200 "OSPFIGP", /**< OSPFIGP */ 201 202 "SRPC", /**< Strite RPC protocol */ 203 "LARP", /**< Locus Address Resoloution */ 204 "MTP", /**< Multicast Transport */ 205 "AX25", /**< AX.25 Frames */ 206 "4IN4", /**< IP encapsulated in IP */ 207 208 "MICP", /**< Mobile Int.ing control */ 209 "SCCSP", /**< Semaphore Comm. security */ 210 "ETHERIP", /**< Ethernet IP encapsulation */ 211 "ENCAP", /**< encapsulation header */ 212 "AES", /**< any private encr. scheme */ 213 214 "GMTP", /**< GMTP */ 215 "IPCOMP", /**< payload compression (IPComp) */ 216 "UNASSIGNED", 217 "UNASSIGNED", 218 "PIM", /**< Protocol Independent Mcast */ 219 }; 220 221 if (ip_proto < sizeof(ip_proto_names) / sizeof(ip_proto_names[0])) 222 return ip_proto_names[ip_proto]; 223 switch (ip_proto) { 224 #ifdef IPPROTO_PGM 225 case IPPROTO_PGM: /**< PGM */ 226 return "PGM"; 227 #endif 228 case IPPROTO_SCTP: /**< Stream Control Transport Protocol */ 229 return "SCTP"; 230 #ifdef IPPROTO_DIVERT 231 case IPPROTO_DIVERT: /**< divert pseudo-protocol */ 232 return "DIVERT"; 233 #endif 234 case IPPROTO_RAW: /**< raw IP packet */ 235 return "RAW"; 236 default: 237 break; 238 } 239 return "UNASSIGNED"; 240 } 241 242 static void 243 ipv4_addr_to_dot(uint32_t be_ipv4_addr, char *buf) 244 { 245 uint32_t ipv4_addr; 246 247 ipv4_addr = rte_be_to_cpu_32(be_ipv4_addr); 248 sprintf(buf, "%d.%d.%d.%d", (ipv4_addr >> 24) & 0xFF, 249 (ipv4_addr >> 16) & 0xFF, (ipv4_addr >> 8) & 0xFF, 250 ipv4_addr & 0xFF); 251 } 252 253 static void 254 ether_addr_dump(const char *what, const struct ether_addr *ea) 255 { 256 char buf[ETHER_ADDR_FMT_SIZE]; 257 258 ether_format_addr(buf, ETHER_ADDR_FMT_SIZE, ea); 259 if (what) 260 printf("%s", what); 261 printf("%s", buf); 262 } 263 264 static void 265 ipv4_addr_dump(const char *what, uint32_t be_ipv4_addr) 266 { 267 char buf[16]; 268 269 ipv4_addr_to_dot(be_ipv4_addr, buf); 270 if (what) 271 printf("%s", what); 272 printf("%s", buf); 273 } 274 275 static uint16_t 276 ipv4_hdr_cksum(struct ipv4_hdr *ip_h) 277 { 278 uint16_t *v16_h; 279 uint32_t ip_cksum; 280 281 /* 282 * Compute the sum of successive 16-bit words of the IPv4 header, 283 * skipping the checksum field of the header. 284 */ 285 v16_h = (unaligned_uint16_t *) ip_h; 286 ip_cksum = v16_h[0] + v16_h[1] + v16_h[2] + v16_h[3] + 287 v16_h[4] + v16_h[6] + v16_h[7] + v16_h[8] + v16_h[9]; 288 289 /* reduce 32 bit checksum to 16 bits and complement it */ 290 ip_cksum = (ip_cksum & 0xffff) + (ip_cksum >> 16); 291 ip_cksum = (ip_cksum & 0xffff) + (ip_cksum >> 16); 292 ip_cksum = (~ip_cksum) & 0x0000FFFF; 293 return (ip_cksum == 0) ? 0xFFFF : (uint16_t) ip_cksum; 294 } 295 296 #define is_multicast_ipv4_addr(ipv4_addr) \ 297 (((rte_be_to_cpu_32((ipv4_addr)) >> 24) & 0x000000FF) == 0xE0) 298 299 /* 300 * Receive a burst of packets, lookup for ICMP echo requets, and, if any, 301 * send back ICMP echo replies. 302 */ 303 static void 304 reply_to_icmp_echo_rqsts(struct fwd_stream *fs) 305 { 306 struct rte_mbuf *pkts_burst[MAX_PKT_BURST]; 307 struct rte_mbuf *pkt; 308 struct ether_hdr *eth_h; 309 struct vlan_hdr *vlan_h; 310 struct arp_hdr *arp_h; 311 struct ipv4_hdr *ip_h; 312 struct icmp_hdr *icmp_h; 313 struct ether_addr eth_addr; 314 uint32_t retry; 315 uint32_t ip_addr; 316 uint16_t nb_rx; 317 uint16_t nb_tx; 318 uint16_t nb_replies; 319 uint16_t eth_type; 320 uint16_t vlan_id; 321 uint16_t arp_op; 322 uint16_t arp_pro; 323 uint32_t cksum; 324 uint8_t i; 325 int l2_len; 326 #ifdef RTE_TEST_PMD_RECORD_CORE_CYCLES 327 uint64_t start_tsc; 328 uint64_t end_tsc; 329 uint64_t core_cycles; 330 #endif 331 332 #ifdef RTE_TEST_PMD_RECORD_CORE_CYCLES 333 start_tsc = rte_rdtsc(); 334 #endif 335 336 /* 337 * First, receive a burst of packets. 338 */ 339 nb_rx = rte_eth_rx_burst(fs->rx_port, fs->rx_queue, pkts_burst, 340 nb_pkt_per_burst); 341 if (unlikely(nb_rx == 0)) 342 return; 343 344 #ifdef RTE_TEST_PMD_RECORD_BURST_STATS 345 fs->rx_burst_stats.pkt_burst_spread[nb_rx]++; 346 #endif 347 fs->rx_packets += nb_rx; 348 nb_replies = 0; 349 for (i = 0; i < nb_rx; i++) { 350 if (likely(i < nb_rx - 1)) 351 rte_prefetch0(rte_pktmbuf_mtod(pkts_burst[i + 1], 352 void *)); 353 pkt = pkts_burst[i]; 354 eth_h = rte_pktmbuf_mtod(pkt, struct ether_hdr *); 355 eth_type = RTE_BE_TO_CPU_16(eth_h->ether_type); 356 l2_len = sizeof(struct ether_hdr); 357 if (verbose_level > 0) { 358 printf("\nPort %d pkt-len=%u nb-segs=%u\n", 359 fs->rx_port, pkt->pkt_len, pkt->nb_segs); 360 ether_addr_dump(" ETH: src=", ð_h->s_addr); 361 ether_addr_dump(" dst=", ð_h->d_addr); 362 } 363 if (eth_type == ETHER_TYPE_VLAN) { 364 vlan_h = (struct vlan_hdr *) 365 ((char *)eth_h + sizeof(struct ether_hdr)); 366 l2_len += sizeof(struct vlan_hdr); 367 eth_type = rte_be_to_cpu_16(vlan_h->eth_proto); 368 if (verbose_level > 0) { 369 vlan_id = rte_be_to_cpu_16(vlan_h->vlan_tci) 370 & 0xFFF; 371 printf(" [vlan id=%u]", vlan_id); 372 } 373 } 374 if (verbose_level > 0) { 375 printf(" type=0x%04x\n", eth_type); 376 } 377 378 /* Reply to ARP requests */ 379 if (eth_type == ETHER_TYPE_ARP) { 380 arp_h = (struct arp_hdr *) ((char *)eth_h + l2_len); 381 arp_op = RTE_BE_TO_CPU_16(arp_h->arp_op); 382 arp_pro = RTE_BE_TO_CPU_16(arp_h->arp_pro); 383 if (verbose_level > 0) { 384 printf(" ARP: hrd=%d proto=0x%04x hln=%d " 385 "pln=%d op=%u (%s)\n", 386 RTE_BE_TO_CPU_16(arp_h->arp_hrd), 387 arp_pro, arp_h->arp_hln, 388 arp_h->arp_pln, arp_op, 389 arp_op_name(arp_op)); 390 } 391 if ((RTE_BE_TO_CPU_16(arp_h->arp_hrd) != 392 ARP_HRD_ETHER) || 393 (arp_pro != ETHER_TYPE_IPv4) || 394 (arp_h->arp_hln != 6) || 395 (arp_h->arp_pln != 4) 396 ) { 397 rte_pktmbuf_free(pkt); 398 if (verbose_level > 0) 399 printf("\n"); 400 continue; 401 } 402 if (verbose_level > 0) { 403 ether_addr_copy(&arp_h->arp_data.arp_sha, ð_addr); 404 ether_addr_dump(" sha=", ð_addr); 405 ip_addr = arp_h->arp_data.arp_sip; 406 ipv4_addr_dump(" sip=", ip_addr); 407 printf("\n"); 408 ether_addr_copy(&arp_h->arp_data.arp_tha, ð_addr); 409 ether_addr_dump(" tha=", ð_addr); 410 ip_addr = arp_h->arp_data.arp_tip; 411 ipv4_addr_dump(" tip=", ip_addr); 412 printf("\n"); 413 } 414 if (arp_op != ARP_OP_REQUEST) { 415 rte_pktmbuf_free(pkt); 416 continue; 417 } 418 419 /* 420 * Build ARP reply. 421 */ 422 423 /* Use source MAC address as destination MAC address. */ 424 ether_addr_copy(ð_h->s_addr, ð_h->d_addr); 425 /* Set source MAC address with MAC address of TX port */ 426 ether_addr_copy(&ports[fs->tx_port].eth_addr, 427 ð_h->s_addr); 428 429 arp_h->arp_op = rte_cpu_to_be_16(ARP_OP_REPLY); 430 ether_addr_copy(&arp_h->arp_data.arp_tha, ð_addr); 431 ether_addr_copy(&arp_h->arp_data.arp_sha, &arp_h->arp_data.arp_tha); 432 ether_addr_copy(ð_h->s_addr, &arp_h->arp_data.arp_sha); 433 434 /* Swap IP addresses in ARP payload */ 435 ip_addr = arp_h->arp_data.arp_sip; 436 arp_h->arp_data.arp_sip = arp_h->arp_data.arp_tip; 437 arp_h->arp_data.arp_tip = ip_addr; 438 pkts_burst[nb_replies++] = pkt; 439 continue; 440 } 441 442 if (eth_type != ETHER_TYPE_IPv4) { 443 rte_pktmbuf_free(pkt); 444 continue; 445 } 446 ip_h = (struct ipv4_hdr *) ((char *)eth_h + l2_len); 447 if (verbose_level > 0) { 448 ipv4_addr_dump(" IPV4: src=", ip_h->src_addr); 449 ipv4_addr_dump(" dst=", ip_h->dst_addr); 450 printf(" proto=%d (%s)\n", 451 ip_h->next_proto_id, 452 ip_proto_name(ip_h->next_proto_id)); 453 } 454 455 /* 456 * Check if packet is a ICMP echo request. 457 */ 458 icmp_h = (struct icmp_hdr *) ((char *)ip_h + 459 sizeof(struct ipv4_hdr)); 460 if (! ((ip_h->next_proto_id == IPPROTO_ICMP) && 461 (icmp_h->icmp_type == IP_ICMP_ECHO_REQUEST) && 462 (icmp_h->icmp_code == 0))) { 463 rte_pktmbuf_free(pkt); 464 continue; 465 } 466 467 if (verbose_level > 0) 468 printf(" ICMP: echo request seq id=%d\n", 469 rte_be_to_cpu_16(icmp_h->icmp_seq_nb)); 470 471 /* 472 * Prepare ICMP echo reply to be sent back. 473 * - switch ethernet source and destinations addresses, 474 * - use the request IP source address as the reply IP 475 * destination address, 476 * - if the request IP destination address is a multicast 477 * address: 478 * - choose a reply IP source address different from the 479 * request IP source address, 480 * - re-compute the IP header checksum. 481 * Otherwise: 482 * - switch the request IP source and destination 483 * addresses in the reply IP header, 484 * - keep the IP header checksum unchanged. 485 * - set IP_ICMP_ECHO_REPLY in ICMP header. 486 * ICMP checksum is computed by assuming it is valid in the 487 * echo request and not verified. 488 */ 489 ether_addr_copy(ð_h->s_addr, ð_addr); 490 ether_addr_copy(ð_h->d_addr, ð_h->s_addr); 491 ether_addr_copy(ð_addr, ð_h->d_addr); 492 ip_addr = ip_h->src_addr; 493 if (is_multicast_ipv4_addr(ip_h->dst_addr)) { 494 uint32_t ip_src; 495 496 ip_src = rte_be_to_cpu_32(ip_addr); 497 if ((ip_src & 0x00000003) == 1) 498 ip_src = (ip_src & 0xFFFFFFFC) | 0x00000002; 499 else 500 ip_src = (ip_src & 0xFFFFFFFC) | 0x00000001; 501 ip_h->src_addr = rte_cpu_to_be_32(ip_src); 502 ip_h->dst_addr = ip_addr; 503 ip_h->hdr_checksum = ipv4_hdr_cksum(ip_h); 504 } else { 505 ip_h->src_addr = ip_h->dst_addr; 506 ip_h->dst_addr = ip_addr; 507 } 508 icmp_h->icmp_type = IP_ICMP_ECHO_REPLY; 509 cksum = ~icmp_h->icmp_cksum & 0xffff; 510 cksum += ~htons(IP_ICMP_ECHO_REQUEST << 8) & 0xffff; 511 cksum += htons(IP_ICMP_ECHO_REPLY << 8); 512 cksum = (cksum & 0xffff) + (cksum >> 16); 513 cksum = (cksum & 0xffff) + (cksum >> 16); 514 icmp_h->icmp_cksum = ~cksum; 515 pkts_burst[nb_replies++] = pkt; 516 } 517 518 /* Send back ICMP echo replies, if any. */ 519 if (nb_replies > 0) { 520 nb_tx = rte_eth_tx_burst(fs->tx_port, fs->tx_queue, pkts_burst, 521 nb_replies); 522 /* 523 * Retry if necessary 524 */ 525 if (unlikely(nb_tx < nb_replies) && fs->retry_enabled) { 526 retry = 0; 527 while (nb_tx < nb_replies && 528 retry++ < burst_tx_retry_num) { 529 rte_delay_us(burst_tx_delay_time); 530 nb_tx += rte_eth_tx_burst(fs->tx_port, 531 fs->tx_queue, 532 &pkts_burst[nb_tx], 533 nb_replies - nb_tx); 534 } 535 } 536 fs->tx_packets += nb_tx; 537 #ifdef RTE_TEST_PMD_RECORD_BURST_STATS 538 fs->tx_burst_stats.pkt_burst_spread[nb_tx]++; 539 #endif 540 if (unlikely(nb_tx < nb_replies)) { 541 fs->fwd_dropped += (nb_replies - nb_tx); 542 do { 543 rte_pktmbuf_free(pkts_burst[nb_tx]); 544 } while (++nb_tx < nb_replies); 545 } 546 } 547 548 #ifdef RTE_TEST_PMD_RECORD_CORE_CYCLES 549 end_tsc = rte_rdtsc(); 550 core_cycles = (end_tsc - start_tsc); 551 fs->core_cycles = (uint64_t) (fs->core_cycles + core_cycles); 552 #endif 553 } 554 555 struct fwd_engine icmp_echo_engine = { 556 .fwd_mode_name = "icmpecho", 557 .port_fwd_begin = NULL, 558 .port_fwd_end = NULL, 559 .packet_fwd = reply_to_icmp_echo_rqsts, 560 }; 561