1 /* SPDX-License-Identifier: BSD-3-Clause 2 * Copyright(c) 2010-2014 Intel Corporation 3 */ 4 5 #include <stdio.h> 6 #include <stdlib.h> 7 #include <stdint.h> 8 #include <inttypes.h> 9 #include <sys/types.h> 10 #include <sys/param.h> 11 #include <string.h> 12 #include <sys/queue.h> 13 #include <stdarg.h> 14 #include <errno.h> 15 #include <getopt.h> 16 17 #include <rte_common.h> 18 #include <rte_byteorder.h> 19 #include <rte_log.h> 20 #include <rte_memory.h> 21 #include <rte_memcpy.h> 22 #include <rte_eal.h> 23 #include <rte_launch.h> 24 #include <rte_cycles.h> 25 #include <rte_prefetch.h> 26 #include <rte_lcore.h> 27 #include <rte_per_lcore.h> 28 #include <rte_branch_prediction.h> 29 #include <rte_interrupts.h> 30 #include <rte_random.h> 31 #include <rte_debug.h> 32 #include <rte_ether.h> 33 #include <rte_ethdev.h> 34 #include <rte_mempool.h> 35 #include <rte_mbuf.h> 36 #include <rte_lpm.h> 37 #include <rte_lpm6.h> 38 #include <rte_ip.h> 39 #include <rte_string_fns.h> 40 41 #include <rte_ip_frag.h> 42 43 #define RTE_LOGTYPE_IP_FRAG RTE_LOGTYPE_USER1 44 45 /* allow max jumbo frame 9.5 KB */ 46 #define JUMBO_FRAME_MAX_SIZE 0x2600 47 48 #define ROUNDUP_DIV(a, b) (((a) + (b) - 1) / (b)) 49 50 /* 51 * Default byte size for the IPv6 Maximum Transfer Unit (MTU). 52 * This value includes the size of IPv6 header. 53 */ 54 #define IPV4_MTU_DEFAULT RTE_ETHER_MTU 55 #define IPV6_MTU_DEFAULT RTE_ETHER_MTU 56 57 /* 58 * The overhead from max frame size to MTU. 59 * We have to consider the max possible overhead. 60 */ 61 #define MTU_OVERHEAD \ 62 (RTE_ETHER_HDR_LEN + RTE_ETHER_CRC_LEN + \ 63 2 * sizeof(struct rte_vlan_hdr)) 64 65 /* 66 * Default payload in bytes for the IPv6 packet. 67 */ 68 #define IPV4_DEFAULT_PAYLOAD (IPV4_MTU_DEFAULT - sizeof(struct rte_ipv4_hdr)) 69 #define IPV6_DEFAULT_PAYLOAD (IPV6_MTU_DEFAULT - sizeof(struct rte_ipv6_hdr)) 70 71 /* 72 * Max number of fragments per packet expected - defined by config file. 73 */ 74 #define MAX_PACKET_FRAG RTE_LIBRTE_IP_FRAG_MAX_FRAG 75 76 #define NB_MBUF 8192 77 78 #define MAX_PKT_BURST 32 79 #define BURST_TX_DRAIN_US 100 /* TX drain every ~100us */ 80 81 /* Configure how many packets ahead to prefetch, when reading packets */ 82 #define PREFETCH_OFFSET 3 83 84 /* 85 * Configurable number of RX/TX ring descriptors 86 */ 87 #define RTE_TEST_RX_DESC_DEFAULT 1024 88 #define RTE_TEST_TX_DESC_DEFAULT 1024 89 static uint16_t nb_rxd = RTE_TEST_RX_DESC_DEFAULT; 90 static uint16_t nb_txd = RTE_TEST_TX_DESC_DEFAULT; 91 92 /* ethernet addresses of ports */ 93 static struct rte_ether_addr ports_eth_addr[RTE_MAX_ETHPORTS]; 94 95 #ifndef IPv4_BYTES 96 #define IPv4_BYTES_FMT "%" PRIu8 ".%" PRIu8 ".%" PRIu8 ".%" PRIu8 97 #define IPv4_BYTES(addr) \ 98 (uint8_t) (((addr) >> 24) & 0xFF),\ 99 (uint8_t) (((addr) >> 16) & 0xFF),\ 100 (uint8_t) (((addr) >> 8) & 0xFF),\ 101 (uint8_t) ((addr) & 0xFF) 102 #endif 103 104 #ifndef IPv6_BYTES 105 #define IPv6_BYTES_FMT "%02x%02x:%02x%02x:%02x%02x:%02x%02x:"\ 106 "%02x%02x:%02x%02x:%02x%02x:%02x%02x" 107 #define IPv6_BYTES(addr) \ 108 addr[0], addr[1], addr[2], addr[3], \ 109 addr[4], addr[5], addr[6], addr[7], \ 110 addr[8], addr[9], addr[10], addr[11],\ 111 addr[12], addr[13],addr[14], addr[15] 112 #endif 113 114 #define IPV6_ADDR_LEN 16 115 116 /* mask of enabled ports */ 117 static int enabled_port_mask = 0; 118 119 static int rx_queue_per_lcore = 1; 120 121 #define MBUF_TABLE_SIZE (2 * MAX(MAX_PKT_BURST, MAX_PACKET_FRAG)) 122 123 struct mbuf_table { 124 uint16_t len; 125 struct rte_mbuf *m_table[MBUF_TABLE_SIZE]; 126 }; 127 128 struct rx_queue { 129 struct rte_mempool *direct_pool; 130 struct rte_mempool *indirect_pool; 131 struct rte_lpm *lpm; 132 struct rte_lpm6 *lpm6; 133 uint16_t portid; 134 }; 135 136 #define MAX_RX_QUEUE_PER_LCORE 16 137 #define MAX_TX_QUEUE_PER_PORT 16 138 struct lcore_queue_conf { 139 uint16_t n_rx_queue; 140 uint16_t tx_queue_id[RTE_MAX_ETHPORTS]; 141 struct rx_queue rx_queue_list[MAX_RX_QUEUE_PER_LCORE]; 142 struct mbuf_table tx_mbufs[RTE_MAX_ETHPORTS]; 143 } __rte_cache_aligned; 144 struct lcore_queue_conf lcore_queue_conf[RTE_MAX_LCORE]; 145 146 static struct rte_eth_conf port_conf = { 147 .rxmode = { 148 .mtu = JUMBO_FRAME_MAX_SIZE - RTE_ETHER_HDR_LEN - 149 RTE_ETHER_CRC_LEN, 150 .split_hdr_size = 0, 151 .offloads = (RTE_ETH_RX_OFFLOAD_CHECKSUM | 152 RTE_ETH_RX_OFFLOAD_SCATTER), 153 }, 154 .txmode = { 155 .mq_mode = RTE_ETH_MQ_TX_NONE, 156 .offloads = (RTE_ETH_TX_OFFLOAD_IPV4_CKSUM | 157 RTE_ETH_TX_OFFLOAD_MULTI_SEGS), 158 }, 159 }; 160 161 /* 162 * IPv4 forwarding table 163 */ 164 struct l3fwd_ipv4_route { 165 uint32_t ip; 166 uint8_t depth; 167 uint8_t if_out; 168 }; 169 170 /* Default l3fwd_ipv4_route_array table. 8< */ 171 struct l3fwd_ipv4_route l3fwd_ipv4_route_array[] = { 172 {RTE_IPV4(100,10,0,0), 16, 0}, 173 {RTE_IPV4(100,20,0,0), 16, 1}, 174 {RTE_IPV4(100,30,0,0), 16, 2}, 175 {RTE_IPV4(100,40,0,0), 16, 3}, 176 {RTE_IPV4(100,50,0,0), 16, 4}, 177 {RTE_IPV4(100,60,0,0), 16, 5}, 178 {RTE_IPV4(100,70,0,0), 16, 6}, 179 {RTE_IPV4(100,80,0,0), 16, 7}, 180 }; 181 /* >8 End of default l3fwd_ipv4_route_array table */ 182 183 /* 184 * IPv6 forwarding table 185 */ 186 187 struct l3fwd_ipv6_route { 188 uint8_t ip[IPV6_ADDR_LEN]; 189 uint8_t depth; 190 uint8_t if_out; 191 }; 192 193 /* Default l3fwd_ipv6_route_array table. 8< */ 194 static struct l3fwd_ipv6_route l3fwd_ipv6_route_array[] = { 195 {{1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 0}, 196 {{2,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 1}, 197 {{3,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 2}, 198 {{4,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 3}, 199 {{5,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 4}, 200 {{6,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 5}, 201 {{7,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 6}, 202 {{8,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 7}, 203 }; 204 /* >8 End of default l3fwd_ipv6_route_array table. */ 205 206 #define LPM_MAX_RULES 1024 207 #define LPM6_MAX_RULES 1024 208 #define LPM6_NUMBER_TBL8S (1 << 16) 209 210 struct rte_lpm6_config lpm6_config = { 211 .max_rules = LPM6_MAX_RULES, 212 .number_tbl8s = LPM6_NUMBER_TBL8S, 213 .flags = 0 214 }; 215 216 static struct rte_mempool *socket_direct_pool[RTE_MAX_NUMA_NODES]; 217 static struct rte_mempool *socket_indirect_pool[RTE_MAX_NUMA_NODES]; 218 static struct rte_lpm *socket_lpm[RTE_MAX_NUMA_NODES]; 219 static struct rte_lpm6 *socket_lpm6[RTE_MAX_NUMA_NODES]; 220 221 /* Send burst of packets on an output interface */ 222 static inline int 223 send_burst(struct lcore_queue_conf *qconf, uint16_t n, uint16_t port) 224 { 225 struct rte_mbuf **m_table; 226 int ret; 227 uint16_t queueid; 228 229 queueid = qconf->tx_queue_id[port]; 230 m_table = (struct rte_mbuf **)qconf->tx_mbufs[port].m_table; 231 232 ret = rte_eth_tx_burst(port, queueid, m_table, n); 233 if (unlikely(ret < n)) { 234 do { 235 rte_pktmbuf_free(m_table[ret]); 236 } while (++ret < n); 237 } 238 239 return 0; 240 } 241 242 static inline void 243 l3fwd_simple_forward(struct rte_mbuf *m, struct lcore_queue_conf *qconf, 244 uint8_t queueid, uint16_t port_in) 245 { 246 struct rx_queue *rxq; 247 uint32_t i, len, next_hop; 248 uint16_t port_out, ether_type; 249 int32_t len2; 250 uint64_t ol_flags; 251 const struct rte_ether_hdr *eth; 252 253 ol_flags = 0; 254 rxq = &qconf->rx_queue_list[queueid]; 255 256 /* by default, send everything back to the source port */ 257 port_out = port_in; 258 259 /* save ether type of the incoming packet */ 260 eth = rte_pktmbuf_mtod(m, const struct rte_ether_hdr *); 261 ether_type = eth->ether_type; 262 263 /* Remove the Ethernet header and trailer from the input packet */ 264 rte_pktmbuf_adj(m, (uint16_t)sizeof(struct rte_ether_hdr)); 265 266 /* Build transmission burst */ 267 len = qconf->tx_mbufs[port_out].len; 268 269 /* if this is an IPv4 packet */ 270 if (RTE_ETH_IS_IPV4_HDR(m->packet_type)) { 271 struct rte_ipv4_hdr *ip_hdr; 272 uint32_t ip_dst; 273 /* Read the lookup key (i.e. ip_dst) from the input packet */ 274 ip_hdr = rte_pktmbuf_mtod(m, struct rte_ipv4_hdr *); 275 ip_dst = rte_be_to_cpu_32(ip_hdr->dst_addr); 276 277 /* Find destination port */ 278 if (rte_lpm_lookup(rxq->lpm, ip_dst, &next_hop) == 0 && 279 (enabled_port_mask & 1 << next_hop) != 0) { 280 port_out = next_hop; 281 282 /* Build transmission burst for new port */ 283 len = qconf->tx_mbufs[port_out].len; 284 } 285 286 /* if we don't need to do any fragmentation */ 287 if (likely (IPV4_MTU_DEFAULT >= m->pkt_len)) { 288 qconf->tx_mbufs[port_out].m_table[len] = m; 289 len2 = 1; 290 } else { 291 len2 = rte_ipv4_fragment_packet(m, 292 &qconf->tx_mbufs[port_out].m_table[len], 293 (uint16_t)(MBUF_TABLE_SIZE - len), 294 IPV4_MTU_DEFAULT, 295 rxq->direct_pool, rxq->indirect_pool); 296 297 /* Free input packet */ 298 rte_pktmbuf_free(m); 299 300 /* request HW to regenerate IPv4 cksum */ 301 ol_flags |= (RTE_MBUF_F_TX_IPV4 | RTE_MBUF_F_TX_IP_CKSUM); 302 303 /* If we fail to fragment the packet */ 304 if (unlikely (len2 < 0)) 305 return; 306 } 307 } else if (RTE_ETH_IS_IPV6_HDR(m->packet_type)) { 308 /* if this is an IPv6 packet */ 309 struct rte_ipv6_hdr *ip_hdr; 310 311 /* Read the lookup key (i.e. ip_dst) from the input packet */ 312 ip_hdr = rte_pktmbuf_mtod(m, struct rte_ipv6_hdr *); 313 314 /* Find destination port */ 315 if (rte_lpm6_lookup(rxq->lpm6, ip_hdr->dst_addr, 316 &next_hop) == 0 && 317 (enabled_port_mask & 1 << next_hop) != 0) { 318 port_out = next_hop; 319 320 /* Build transmission burst for new port */ 321 len = qconf->tx_mbufs[port_out].len; 322 } 323 324 /* if we don't need to do any fragmentation */ 325 if (likely (IPV6_MTU_DEFAULT >= m->pkt_len)) { 326 qconf->tx_mbufs[port_out].m_table[len] = m; 327 len2 = 1; 328 } else { 329 len2 = rte_ipv6_fragment_packet(m, 330 &qconf->tx_mbufs[port_out].m_table[len], 331 (uint16_t)(MBUF_TABLE_SIZE - len), 332 IPV6_MTU_DEFAULT, 333 rxq->direct_pool, rxq->indirect_pool); 334 335 /* Free input packet */ 336 rte_pktmbuf_free(m); 337 338 /* If we fail to fragment the packet */ 339 if (unlikely (len2 < 0)) 340 return; 341 } 342 } 343 /* else, just forward the packet */ 344 else { 345 qconf->tx_mbufs[port_out].m_table[len] = m; 346 len2 = 1; 347 } 348 349 for (i = len; i < len + len2; i ++) { 350 void *d_addr_bytes; 351 352 m = qconf->tx_mbufs[port_out].m_table[i]; 353 struct rte_ether_hdr *eth_hdr = (struct rte_ether_hdr *) 354 rte_pktmbuf_prepend(m, 355 (uint16_t)sizeof(struct rte_ether_hdr)); 356 if (eth_hdr == NULL) { 357 rte_panic("No headroom in mbuf.\n"); 358 } 359 360 m->ol_flags |= ol_flags; 361 m->l2_len = sizeof(struct rte_ether_hdr); 362 363 /* 02:00:00:00:00:xx */ 364 d_addr_bytes = ð_hdr->dst_addr.addr_bytes[0]; 365 *((uint64_t *)d_addr_bytes) = 0x000000000002 + 366 ((uint64_t)port_out << 40); 367 368 /* src addr */ 369 rte_ether_addr_copy(&ports_eth_addr[port_out], 370 ð_hdr->src_addr); 371 eth_hdr->ether_type = ether_type; 372 } 373 374 len += len2; 375 376 if (likely(len < MAX_PKT_BURST)) { 377 qconf->tx_mbufs[port_out].len = (uint16_t)len; 378 return; 379 } 380 381 /* Transmit packets */ 382 send_burst(qconf, (uint16_t)len, port_out); 383 qconf->tx_mbufs[port_out].len = 0; 384 } 385 386 /* main processing loop */ 387 static int 388 main_loop(__rte_unused void *dummy) 389 { 390 struct rte_mbuf *pkts_burst[MAX_PKT_BURST]; 391 unsigned lcore_id; 392 uint64_t prev_tsc, diff_tsc, cur_tsc; 393 int i, j, nb_rx; 394 uint16_t portid; 395 struct lcore_queue_conf *qconf; 396 const uint64_t drain_tsc = (rte_get_tsc_hz() + US_PER_S - 1) / US_PER_S * BURST_TX_DRAIN_US; 397 398 prev_tsc = 0; 399 400 lcore_id = rte_lcore_id(); 401 qconf = &lcore_queue_conf[lcore_id]; 402 403 if (qconf->n_rx_queue == 0) { 404 RTE_LOG(INFO, IP_FRAG, "lcore %u has nothing to do\n", lcore_id); 405 return 0; 406 } 407 408 RTE_LOG(INFO, IP_FRAG, "entering main loop on lcore %u\n", lcore_id); 409 410 for (i = 0; i < qconf->n_rx_queue; i++) { 411 412 portid = qconf->rx_queue_list[i].portid; 413 RTE_LOG(INFO, IP_FRAG, " -- lcoreid=%u portid=%d\n", lcore_id, 414 portid); 415 } 416 417 while (1) { 418 419 cur_tsc = rte_rdtsc(); 420 421 /* 422 * TX burst queue drain 423 */ 424 diff_tsc = cur_tsc - prev_tsc; 425 if (unlikely(diff_tsc > drain_tsc)) { 426 427 /* 428 * This could be optimized (use queueid instead of 429 * portid), but it is not called so often 430 */ 431 for (portid = 0; portid < RTE_MAX_ETHPORTS; portid++) { 432 if (qconf->tx_mbufs[portid].len == 0) 433 continue; 434 send_burst(&lcore_queue_conf[lcore_id], 435 qconf->tx_mbufs[portid].len, 436 portid); 437 qconf->tx_mbufs[portid].len = 0; 438 } 439 440 prev_tsc = cur_tsc; 441 } 442 443 /* 444 * Read packet from RX queues 445 */ 446 for (i = 0; i < qconf->n_rx_queue; i++) { 447 448 portid = qconf->rx_queue_list[i].portid; 449 nb_rx = rte_eth_rx_burst(portid, 0, pkts_burst, 450 MAX_PKT_BURST); 451 452 /* Prefetch first packets */ 453 for (j = 0; j < PREFETCH_OFFSET && j < nb_rx; j++) { 454 rte_prefetch0(rte_pktmbuf_mtod( 455 pkts_burst[j], void *)); 456 } 457 458 /* Prefetch and forward already prefetched packets */ 459 for (j = 0; j < (nb_rx - PREFETCH_OFFSET); j++) { 460 rte_prefetch0(rte_pktmbuf_mtod(pkts_burst[ 461 j + PREFETCH_OFFSET], void *)); 462 l3fwd_simple_forward(pkts_burst[j], qconf, i, portid); 463 } 464 465 /* Forward remaining prefetched packets */ 466 for (; j < nb_rx; j++) { 467 l3fwd_simple_forward(pkts_burst[j], qconf, i, portid); 468 } 469 } 470 } 471 } 472 473 /* display usage */ 474 static void 475 print_usage(const char *prgname) 476 { 477 printf("%s [EAL options] -- -p PORTMASK [-q NQ]\n" 478 " -p PORTMASK: hexadecimal bitmask of ports to configure\n" 479 " -q NQ: number of queue (=ports) per lcore (default is 1)\n", 480 prgname); 481 } 482 483 static int 484 parse_portmask(const char *portmask) 485 { 486 char *end = NULL; 487 unsigned long pm; 488 489 /* parse hexadecimal string */ 490 pm = strtoul(portmask, &end, 16); 491 if ((portmask[0] == '\0') || (end == NULL) || (*end != '\0')) 492 return -1; 493 494 if (pm == 0) 495 return -1; 496 497 return pm; 498 } 499 500 static int 501 parse_nqueue(const char *q_arg) 502 { 503 char *end = NULL; 504 unsigned long n; 505 506 /* parse hexadecimal string */ 507 n = strtoul(q_arg, &end, 10); 508 if ((q_arg[0] == '\0') || (end == NULL) || (*end != '\0')) 509 return -1; 510 if (n == 0) 511 return -1; 512 if (n >= MAX_RX_QUEUE_PER_LCORE) 513 return -1; 514 515 return n; 516 } 517 518 /* Parse the argument given in the command line of the application */ 519 static int 520 parse_args(int argc, char **argv) 521 { 522 int opt, ret; 523 char **argvopt; 524 int option_index; 525 char *prgname = argv[0]; 526 static struct option lgopts[] = { 527 {NULL, 0, 0, 0} 528 }; 529 530 argvopt = argv; 531 532 while ((opt = getopt_long(argc, argvopt, "p:q:", 533 lgopts, &option_index)) != EOF) { 534 535 switch (opt) { 536 /* portmask */ 537 case 'p': 538 enabled_port_mask = parse_portmask(optarg); 539 if (enabled_port_mask < 0) { 540 printf("invalid portmask\n"); 541 print_usage(prgname); 542 return -1; 543 } 544 break; 545 546 /* nqueue */ 547 case 'q': 548 rx_queue_per_lcore = parse_nqueue(optarg); 549 if (rx_queue_per_lcore < 0) { 550 printf("invalid queue number\n"); 551 print_usage(prgname); 552 return -1; 553 } 554 break; 555 556 /* long options */ 557 case 0: 558 print_usage(prgname); 559 return -1; 560 561 default: 562 print_usage(prgname); 563 return -1; 564 } 565 } 566 567 if (enabled_port_mask == 0) { 568 printf("portmask not specified\n"); 569 print_usage(prgname); 570 return -1; 571 } 572 573 if (optind >= 0) 574 argv[optind-1] = prgname; 575 576 ret = optind-1; 577 optind = 1; /* reset getopt lib */ 578 return ret; 579 } 580 581 static void 582 print_ethaddr(const char *name, struct rte_ether_addr *eth_addr) 583 { 584 char buf[RTE_ETHER_ADDR_FMT_SIZE]; 585 rte_ether_format_addr(buf, RTE_ETHER_ADDR_FMT_SIZE, eth_addr); 586 printf("%s%s", name, buf); 587 } 588 589 /* Check the link status of all ports in up to 9s, and print them finally */ 590 static void 591 check_all_ports_link_status(uint32_t port_mask) 592 { 593 #define CHECK_INTERVAL 100 /* 100ms */ 594 #define MAX_CHECK_TIME 90 /* 9s (90 * 100ms) in total */ 595 uint16_t portid; 596 uint8_t count, all_ports_up, print_flag = 0; 597 struct rte_eth_link link; 598 int ret; 599 char link_status_text[RTE_ETH_LINK_MAX_STR_LEN]; 600 601 printf("\nChecking link status"); 602 fflush(stdout); 603 for (count = 0; count <= MAX_CHECK_TIME; count++) { 604 all_ports_up = 1; 605 RTE_ETH_FOREACH_DEV(portid) { 606 if ((port_mask & (1 << portid)) == 0) 607 continue; 608 memset(&link, 0, sizeof(link)); 609 ret = rte_eth_link_get_nowait(portid, &link); 610 if (ret < 0) { 611 all_ports_up = 0; 612 if (print_flag == 1) 613 printf("Port %u link get failed: %s\n", 614 portid, rte_strerror(-ret)); 615 continue; 616 } 617 /* print link status if flag set */ 618 if (print_flag == 1) { 619 rte_eth_link_to_str(link_status_text, 620 sizeof(link_status_text), &link); 621 printf("Port %d %s\n", portid, 622 link_status_text); 623 continue; 624 } 625 /* clear all_ports_up flag if any link down */ 626 if (link.link_status == RTE_ETH_LINK_DOWN) { 627 all_ports_up = 0; 628 break; 629 } 630 } 631 /* after finally printing all link status, get out */ 632 if (print_flag == 1) 633 break; 634 635 if (all_ports_up == 0) { 636 printf("."); 637 fflush(stdout); 638 rte_delay_ms(CHECK_INTERVAL); 639 } 640 641 /* set the print_flag if all ports up or timeout */ 642 if (all_ports_up == 1 || count == (MAX_CHECK_TIME - 1)) { 643 print_flag = 1; 644 printf("\ndone\n"); 645 } 646 } 647 } 648 649 /* Check L3 packet type detection capability of the NIC port */ 650 static int 651 check_ptype(int portid) 652 { 653 int i, ret; 654 int ptype_l3_ipv4 = 0, ptype_l3_ipv6 = 0; 655 uint32_t ptype_mask = RTE_PTYPE_L3_MASK; 656 657 ret = rte_eth_dev_get_supported_ptypes(portid, ptype_mask, NULL, 0); 658 if (ret <= 0) 659 return 0; 660 661 uint32_t ptypes[ret]; 662 663 ret = rte_eth_dev_get_supported_ptypes(portid, ptype_mask, ptypes, ret); 664 for (i = 0; i < ret; ++i) { 665 if (ptypes[i] & RTE_PTYPE_L3_IPV4) 666 ptype_l3_ipv4 = 1; 667 if (ptypes[i] & RTE_PTYPE_L3_IPV6) 668 ptype_l3_ipv6 = 1; 669 } 670 671 if (ptype_l3_ipv4 == 0) 672 printf("port %d cannot parse RTE_PTYPE_L3_IPV4\n", portid); 673 674 if (ptype_l3_ipv6 == 0) 675 printf("port %d cannot parse RTE_PTYPE_L3_IPV6\n", portid); 676 677 if (ptype_l3_ipv4 && ptype_l3_ipv6) 678 return 1; 679 680 return 0; 681 682 } 683 684 /* Parse packet type of a packet by SW */ 685 static inline void 686 parse_ptype(struct rte_mbuf *m) 687 { 688 struct rte_ether_hdr *eth_hdr; 689 uint32_t packet_type = RTE_PTYPE_UNKNOWN; 690 uint16_t ether_type; 691 692 eth_hdr = rte_pktmbuf_mtod(m, struct rte_ether_hdr *); 693 ether_type = eth_hdr->ether_type; 694 if (ether_type == rte_cpu_to_be_16(RTE_ETHER_TYPE_IPV4)) 695 packet_type |= RTE_PTYPE_L3_IPV4_EXT_UNKNOWN; 696 else if (ether_type == rte_cpu_to_be_16(RTE_ETHER_TYPE_IPV6)) 697 packet_type |= RTE_PTYPE_L3_IPV6_EXT_UNKNOWN; 698 699 m->packet_type = packet_type; 700 } 701 702 /* callback function to detect packet type for a queue of a port */ 703 static uint16_t 704 cb_parse_ptype(uint16_t port __rte_unused, uint16_t queue __rte_unused, 705 struct rte_mbuf *pkts[], uint16_t nb_pkts, 706 uint16_t max_pkts __rte_unused, 707 void *user_param __rte_unused) 708 { 709 uint16_t i; 710 711 for (i = 0; i < nb_pkts; ++i) 712 parse_ptype(pkts[i]); 713 714 return nb_pkts; 715 } 716 717 static int 718 init_routing_table(void) 719 { 720 struct rte_lpm *lpm; 721 struct rte_lpm6 *lpm6; 722 int socket, ret; 723 unsigned i; 724 725 for (socket = 0; socket < RTE_MAX_NUMA_NODES; socket++) { 726 if (socket_lpm[socket]) { 727 lpm = socket_lpm[socket]; 728 /* populate the LPM table */ 729 for (i = 0; i < RTE_DIM(l3fwd_ipv4_route_array); i++) { 730 ret = rte_lpm_add(lpm, 731 l3fwd_ipv4_route_array[i].ip, 732 l3fwd_ipv4_route_array[i].depth, 733 l3fwd_ipv4_route_array[i].if_out); 734 735 if (ret < 0) { 736 RTE_LOG(ERR, IP_FRAG, "Unable to add entry %i to the l3fwd " 737 "LPM table\n", i); 738 return -1; 739 } 740 741 RTE_LOG(INFO, IP_FRAG, "Socket %i: adding route " IPv4_BYTES_FMT 742 "/%d (port %d)\n", 743 socket, 744 IPv4_BYTES(l3fwd_ipv4_route_array[i].ip), 745 l3fwd_ipv4_route_array[i].depth, 746 l3fwd_ipv4_route_array[i].if_out); 747 } 748 } 749 750 if (socket_lpm6[socket]) { 751 lpm6 = socket_lpm6[socket]; 752 /* populate the LPM6 table */ 753 for (i = 0; i < RTE_DIM(l3fwd_ipv6_route_array); i++) { 754 ret = rte_lpm6_add(lpm6, 755 l3fwd_ipv6_route_array[i].ip, 756 l3fwd_ipv6_route_array[i].depth, 757 l3fwd_ipv6_route_array[i].if_out); 758 759 if (ret < 0) { 760 RTE_LOG(ERR, IP_FRAG, "Unable to add entry %i to the l3fwd " 761 "LPM6 table\n", i); 762 return -1; 763 } 764 765 RTE_LOG(INFO, IP_FRAG, "Socket %i: adding route " IPv6_BYTES_FMT 766 "/%d (port %d)\n", 767 socket, 768 IPv6_BYTES(l3fwd_ipv6_route_array[i].ip), 769 l3fwd_ipv6_route_array[i].depth, 770 l3fwd_ipv6_route_array[i].if_out); 771 } 772 } 773 } 774 return 0; 775 } 776 777 static int 778 init_mem(void) 779 { 780 char buf[PATH_MAX]; 781 struct rte_mempool *mp; 782 struct rte_lpm *lpm; 783 struct rte_lpm6 *lpm6; 784 struct rte_lpm_config lpm_config; 785 int socket; 786 unsigned lcore_id; 787 788 /* traverse through lcores and initialize structures on each socket */ 789 790 for (lcore_id = 0; lcore_id < RTE_MAX_LCORE; lcore_id++) { 791 792 if (rte_lcore_is_enabled(lcore_id) == 0) 793 continue; 794 795 socket = rte_lcore_to_socket_id(lcore_id); 796 797 if (socket == SOCKET_ID_ANY) 798 socket = 0; 799 800 if (socket_direct_pool[socket] == NULL) { 801 RTE_LOG(INFO, IP_FRAG, "Creating direct mempool on socket %i\n", 802 socket); 803 snprintf(buf, sizeof(buf), "pool_direct_%i", socket); 804 805 mp = rte_pktmbuf_pool_create(buf, NB_MBUF, 32, 806 0, RTE_MBUF_DEFAULT_BUF_SIZE, socket); 807 if (mp == NULL) { 808 RTE_LOG(ERR, IP_FRAG, "Cannot create direct mempool\n"); 809 return -1; 810 } 811 socket_direct_pool[socket] = mp; 812 } 813 814 if (socket_indirect_pool[socket] == NULL) { 815 RTE_LOG(INFO, IP_FRAG, "Creating indirect mempool on socket %i\n", 816 socket); 817 snprintf(buf, sizeof(buf), "pool_indirect_%i", socket); 818 819 mp = rte_pktmbuf_pool_create(buf, NB_MBUF, 32, 0, 0, 820 socket); 821 if (mp == NULL) { 822 RTE_LOG(ERR, IP_FRAG, "Cannot create indirect mempool\n"); 823 return -1; 824 } 825 socket_indirect_pool[socket] = mp; 826 } 827 828 if (socket_lpm[socket] == NULL) { 829 RTE_LOG(INFO, IP_FRAG, "Creating LPM table on socket %i\n", socket); 830 snprintf(buf, sizeof(buf), "IP_FRAG_LPM_%i", socket); 831 832 lpm_config.max_rules = LPM_MAX_RULES; 833 lpm_config.number_tbl8s = 256; 834 lpm_config.flags = 0; 835 836 lpm = rte_lpm_create(buf, socket, &lpm_config); 837 if (lpm == NULL) { 838 RTE_LOG(ERR, IP_FRAG, "Cannot create LPM table\n"); 839 return -1; 840 } 841 socket_lpm[socket] = lpm; 842 } 843 844 if (socket_lpm6[socket] == NULL) { 845 RTE_LOG(INFO, IP_FRAG, "Creating LPM6 table on socket %i\n", socket); 846 snprintf(buf, sizeof(buf), "IP_FRAG_LPM_%i", socket); 847 848 lpm6 = rte_lpm6_create(buf, socket, &lpm6_config); 849 if (lpm6 == NULL) { 850 RTE_LOG(ERR, IP_FRAG, "Cannot create LPM table\n"); 851 return -1; 852 } 853 socket_lpm6[socket] = lpm6; 854 } 855 } 856 857 return 0; 858 } 859 860 int 861 main(int argc, char **argv) 862 { 863 struct lcore_queue_conf *qconf; 864 struct rte_eth_dev_info dev_info; 865 struct rte_eth_txconf *txconf; 866 struct rx_queue *rxq; 867 int socket, ret; 868 uint16_t nb_ports; 869 uint16_t queueid = 0; 870 unsigned lcore_id = 0, rx_lcore_id = 0; 871 uint32_t n_tx_queue, nb_lcores; 872 uint16_t portid; 873 874 /* init EAL */ 875 ret = rte_eal_init(argc, argv); 876 if (ret < 0) 877 rte_exit(EXIT_FAILURE, "rte_eal_init failed"); 878 argc -= ret; 879 argv += ret; 880 881 /* parse application arguments (after the EAL ones) */ 882 ret = parse_args(argc, argv); 883 if (ret < 0) 884 rte_exit(EXIT_FAILURE, "Invalid arguments"); 885 886 nb_ports = rte_eth_dev_count_avail(); 887 if (nb_ports == 0) 888 rte_exit(EXIT_FAILURE, "No ports found!\n"); 889 890 nb_lcores = rte_lcore_count(); 891 892 /* initialize structures (mempools, lpm etc.) */ 893 if (init_mem() < 0) 894 rte_panic("Cannot initialize memory structures!\n"); 895 896 /* check if portmask has non-existent ports */ 897 if (enabled_port_mask & ~(RTE_LEN2MASK(nb_ports, unsigned))) 898 rte_exit(EXIT_FAILURE, "Non-existent ports in portmask!\n"); 899 900 /* initialize all ports */ 901 RTE_ETH_FOREACH_DEV(portid) { 902 struct rte_eth_conf local_port_conf = port_conf; 903 struct rte_eth_rxconf rxq_conf; 904 905 /* skip ports that are not enabled */ 906 if ((enabled_port_mask & (1 << portid)) == 0) { 907 printf("Skipping disabled port %d\n", portid); 908 continue; 909 } 910 911 qconf = &lcore_queue_conf[rx_lcore_id]; 912 913 /* limit the frame size to the maximum supported by NIC */ 914 ret = rte_eth_dev_info_get(portid, &dev_info); 915 if (ret != 0) 916 rte_exit(EXIT_FAILURE, 917 "Error during getting device (port %u) info: %s\n", 918 portid, strerror(-ret)); 919 920 local_port_conf.rxmode.mtu = RTE_MIN( 921 dev_info.max_mtu, 922 local_port_conf.rxmode.mtu); 923 924 /* get the lcore_id for this port */ 925 while (rte_lcore_is_enabled(rx_lcore_id) == 0 || 926 qconf->n_rx_queue == (unsigned)rx_queue_per_lcore) { 927 928 rx_lcore_id ++; 929 if (rx_lcore_id >= RTE_MAX_LCORE) 930 rte_exit(EXIT_FAILURE, "Not enough cores\n"); 931 932 qconf = &lcore_queue_conf[rx_lcore_id]; 933 } 934 935 socket = (int) rte_lcore_to_socket_id(rx_lcore_id); 936 if (socket == SOCKET_ID_ANY) 937 socket = 0; 938 939 rxq = &qconf->rx_queue_list[qconf->n_rx_queue]; 940 rxq->portid = portid; 941 rxq->direct_pool = socket_direct_pool[socket]; 942 rxq->indirect_pool = socket_indirect_pool[socket]; 943 rxq->lpm = socket_lpm[socket]; 944 rxq->lpm6 = socket_lpm6[socket]; 945 qconf->n_rx_queue++; 946 947 /* init port */ 948 printf("Initializing port %d on lcore %u...", portid, 949 rx_lcore_id); 950 fflush(stdout); 951 952 n_tx_queue = nb_lcores; 953 if (n_tx_queue > MAX_TX_QUEUE_PER_PORT) 954 n_tx_queue = MAX_TX_QUEUE_PER_PORT; 955 ret = rte_eth_dev_configure(portid, 1, (uint16_t)n_tx_queue, 956 &local_port_conf); 957 if (ret < 0) { 958 printf("\n"); 959 rte_exit(EXIT_FAILURE, "Cannot configure device: " 960 "err=%d, port=%d\n", 961 ret, portid); 962 } 963 964 /* set the mtu to the maximum received packet size */ 965 ret = rte_eth_dev_set_mtu(portid, local_port_conf.rxmode.mtu); 966 if (ret < 0) { 967 printf("\n"); 968 rte_exit(EXIT_FAILURE, "Set MTU failed: " 969 "err=%d, port=%d\n", 970 ret, portid); 971 } 972 973 ret = rte_eth_dev_adjust_nb_rx_tx_desc(portid, &nb_rxd, 974 &nb_txd); 975 if (ret < 0) { 976 printf("\n"); 977 rte_exit(EXIT_FAILURE, "Cannot adjust number of " 978 "descriptors: err=%d, port=%d\n", ret, portid); 979 } 980 981 /* init one RX queue */ 982 rxq_conf = dev_info.default_rxconf; 983 rxq_conf.offloads = local_port_conf.rxmode.offloads; 984 ret = rte_eth_rx_queue_setup(portid, 0, nb_rxd, 985 socket, &rxq_conf, 986 socket_direct_pool[socket]); 987 if (ret < 0) { 988 printf("\n"); 989 rte_exit(EXIT_FAILURE, "rte_eth_rx_queue_setup: " 990 "err=%d, port=%d\n", 991 ret, portid); 992 } 993 994 ret = rte_eth_macaddr_get(portid, &ports_eth_addr[portid]); 995 if (ret < 0) { 996 printf("\n"); 997 rte_exit(EXIT_FAILURE, 998 "rte_eth_macaddr_get: err=%d, port=%d\n", 999 ret, portid); 1000 } 1001 1002 print_ethaddr(" Address:", &ports_eth_addr[portid]); 1003 printf("\n"); 1004 1005 /* init one TX queue per couple (lcore,port) */ 1006 ret = rte_eth_dev_info_get(portid, &dev_info); 1007 if (ret != 0) 1008 rte_exit(EXIT_FAILURE, 1009 "Error during getting device (port %u) info: %s\n", 1010 portid, strerror(-ret)); 1011 1012 queueid = 0; 1013 for (lcore_id = 0; lcore_id < RTE_MAX_LCORE; lcore_id++) { 1014 if (rte_lcore_is_enabled(lcore_id) == 0) 1015 continue; 1016 1017 if (queueid >= dev_info.nb_tx_queues) 1018 break; 1019 1020 socket = (int) rte_lcore_to_socket_id(lcore_id); 1021 printf("txq=%u,%d ", lcore_id, queueid); 1022 fflush(stdout); 1023 1024 txconf = &dev_info.default_txconf; 1025 txconf->offloads = local_port_conf.txmode.offloads; 1026 ret = rte_eth_tx_queue_setup(portid, queueid, nb_txd, 1027 socket, txconf); 1028 if (ret < 0) { 1029 printf("\n"); 1030 rte_exit(EXIT_FAILURE, "rte_eth_tx_queue_setup: " 1031 "err=%d, port=%d\n", ret, portid); 1032 } 1033 1034 qconf = &lcore_queue_conf[lcore_id]; 1035 qconf->tx_queue_id[portid] = queueid; 1036 queueid++; 1037 } 1038 1039 printf("\n"); 1040 } 1041 1042 printf("\n"); 1043 1044 /* start ports */ 1045 RTE_ETH_FOREACH_DEV(portid) { 1046 if ((enabled_port_mask & (1 << portid)) == 0) { 1047 continue; 1048 } 1049 /* Start device */ 1050 ret = rte_eth_dev_start(portid); 1051 if (ret < 0) 1052 rte_exit(EXIT_FAILURE, "rte_eth_dev_start: err=%d, port=%d\n", 1053 ret, portid); 1054 1055 ret = rte_eth_promiscuous_enable(portid); 1056 if (ret != 0) 1057 rte_exit(EXIT_FAILURE, 1058 "rte_eth_promiscuous_enable: err=%s, port=%d\n", 1059 rte_strerror(-ret), portid); 1060 1061 if (check_ptype(portid) == 0) { 1062 rte_eth_add_rx_callback(portid, 0, cb_parse_ptype, NULL); 1063 printf("Add Rx callback function to detect L3 packet type by SW :" 1064 " port = %d\n", portid); 1065 } 1066 } 1067 1068 if (init_routing_table() < 0) 1069 rte_exit(EXIT_FAILURE, "Cannot init routing table\n"); 1070 1071 check_all_ports_link_status(enabled_port_mask); 1072 1073 /* launch per-lcore init on every lcore */ 1074 rte_eal_mp_remote_launch(main_loop, NULL, CALL_MAIN); 1075 RTE_LCORE_FOREACH_WORKER(lcore_id) { 1076 if (rte_eal_wait_lcore(lcore_id) < 0) 1077 return -1; 1078 } 1079 1080 /* clean up the EAL */ 1081 rte_eal_cleanup(); 1082 1083 return 0; 1084 } 1085