1 /*- 2 * BSD LICENSE 3 * 4 * Copyright(c) 2010-2014 Intel Corporation. All rights reserved. 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 Intel Corporation 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 #include <stdio.h> 35 #include <stdlib.h> 36 #include <stdint.h> 37 #include <inttypes.h> 38 #include <sys/types.h> 39 #include <sys/param.h> 40 #include <string.h> 41 #include <sys/queue.h> 42 #include <stdarg.h> 43 #include <errno.h> 44 #include <getopt.h> 45 46 #include <rte_common.h> 47 #include <rte_byteorder.h> 48 #include <rte_log.h> 49 #include <rte_memory.h> 50 #include <rte_memcpy.h> 51 #include <rte_memzone.h> 52 #include <rte_tailq.h> 53 #include <rte_eal.h> 54 #include <rte_per_lcore.h> 55 #include <rte_launch.h> 56 #include <rte_atomic.h> 57 #include <rte_cycles.h> 58 #include <rte_prefetch.h> 59 #include <rte_lcore.h> 60 #include <rte_per_lcore.h> 61 #include <rte_branch_prediction.h> 62 #include <rte_interrupts.h> 63 #include <rte_pci.h> 64 #include <rte_random.h> 65 #include <rte_debug.h> 66 #include <rte_ether.h> 67 #include <rte_ethdev.h> 68 #include <rte_ring.h> 69 #include <rte_mempool.h> 70 #include <rte_mbuf.h> 71 #include <rte_lpm.h> 72 #include <rte_lpm6.h> 73 #include <rte_ip.h> 74 #include <rte_string_fns.h> 75 76 #include <rte_ip_frag.h> 77 78 #include "main.h" 79 80 #define RTE_LOGTYPE_IP_FRAG RTE_LOGTYPE_USER1 81 82 #define MBUF_SIZE (2048 + sizeof(struct rte_mbuf) + RTE_PKTMBUF_HEADROOM) 83 84 /* allow max jumbo frame 9.5 KB */ 85 #define JUMBO_FRAME_MAX_SIZE 0x2600 86 87 #define ROUNDUP_DIV(a, b) (((a) + (b) - 1) / (b)) 88 89 /* 90 * Default byte size for the IPv6 Maximum Transfer Unit (MTU). 91 * This value includes the size of IPv6 header. 92 */ 93 #define IPV4_MTU_DEFAULT ETHER_MTU 94 #define IPV6_MTU_DEFAULT ETHER_MTU 95 96 /* 97 * Default payload in bytes for the IPv6 packet. 98 */ 99 #define IPV4_DEFAULT_PAYLOAD (IPV4_MTU_DEFAULT - sizeof(struct ipv4_hdr)) 100 #define IPV6_DEFAULT_PAYLOAD (IPV6_MTU_DEFAULT - sizeof(struct ipv6_hdr)) 101 102 /* 103 * Max number of fragments per packet expected - defined by config file. 104 */ 105 #define MAX_PACKET_FRAG RTE_LIBRTE_IP_FRAG_MAX_FRAG 106 107 #define NB_MBUF 8192 108 109 #define MAX_PKT_BURST 32 110 #define BURST_TX_DRAIN_US 100 /* TX drain every ~100us */ 111 112 /* Configure how many packets ahead to prefetch, when reading packets */ 113 #define PREFETCH_OFFSET 3 114 115 /* 116 * Configurable number of RX/TX ring descriptors 117 */ 118 #define RTE_TEST_RX_DESC_DEFAULT 128 119 #define RTE_TEST_TX_DESC_DEFAULT 512 120 static uint16_t nb_rxd = RTE_TEST_RX_DESC_DEFAULT; 121 static uint16_t nb_txd = RTE_TEST_TX_DESC_DEFAULT; 122 123 /* ethernet addresses of ports */ 124 static struct ether_addr ports_eth_addr[RTE_MAX_ETHPORTS]; 125 126 #ifndef IPv4_BYTES 127 #define IPv4_BYTES_FMT "%" PRIu8 ".%" PRIu8 ".%" PRIu8 ".%" PRIu8 128 #define IPv4_BYTES(addr) \ 129 (uint8_t) (((addr) >> 24) & 0xFF),\ 130 (uint8_t) (((addr) >> 16) & 0xFF),\ 131 (uint8_t) (((addr) >> 8) & 0xFF),\ 132 (uint8_t) ((addr) & 0xFF) 133 #endif 134 135 #ifndef IPv6_BYTES 136 #define IPv6_BYTES_FMT "%02x%02x:%02x%02x:%02x%02x:%02x%02x:"\ 137 "%02x%02x:%02x%02x:%02x%02x:%02x%02x" 138 #define IPv6_BYTES(addr) \ 139 addr[0], addr[1], addr[2], addr[3], \ 140 addr[4], addr[5], addr[6], addr[7], \ 141 addr[8], addr[9], addr[10], addr[11],\ 142 addr[12], addr[13],addr[14], addr[15] 143 #endif 144 145 #define IPV6_ADDR_LEN 16 146 147 /* mask of enabled ports */ 148 static int enabled_port_mask = 0; 149 150 static int rx_queue_per_lcore = 1; 151 152 #define MBUF_TABLE_SIZE (2 * MAX(MAX_PKT_BURST, MAX_PACKET_FRAG)) 153 154 struct mbuf_table { 155 uint16_t len; 156 struct rte_mbuf *m_table[MBUF_TABLE_SIZE]; 157 }; 158 159 struct rx_queue { 160 struct rte_mempool *direct_pool; 161 struct rte_mempool *indirect_pool; 162 struct rte_lpm *lpm; 163 struct rte_lpm6 *lpm6; 164 uint8_t portid; 165 }; 166 167 #define MAX_RX_QUEUE_PER_LCORE 16 168 #define MAX_TX_QUEUE_PER_PORT 16 169 struct lcore_queue_conf { 170 uint16_t n_rx_queue; 171 uint16_t tx_queue_id[RTE_MAX_ETHPORTS]; 172 struct rx_queue rx_queue_list[MAX_RX_QUEUE_PER_LCORE]; 173 struct mbuf_table tx_mbufs[RTE_MAX_ETHPORTS]; 174 } __rte_cache_aligned; 175 struct lcore_queue_conf lcore_queue_conf[RTE_MAX_LCORE]; 176 177 static const struct rte_eth_conf port_conf = { 178 .rxmode = { 179 .max_rx_pkt_len = JUMBO_FRAME_MAX_SIZE, 180 .split_hdr_size = 0, 181 .header_split = 0, /**< Header Split disabled */ 182 .hw_ip_checksum = 1, /**< IP checksum offload enabled */ 183 .hw_vlan_filter = 0, /**< VLAN filtering disabled */ 184 .jumbo_frame = 1, /**< Jumbo Frame Support enabled */ 185 .hw_strip_crc = 0, /**< CRC stripped by hardware */ 186 }, 187 .txmode = { 188 .mq_mode = ETH_MQ_TX_NONE, 189 }, 190 }; 191 192 /* 193 * IPv4 forwarding table 194 */ 195 struct l3fwd_ipv4_route { 196 uint32_t ip; 197 uint8_t depth; 198 uint8_t if_out; 199 }; 200 201 struct l3fwd_ipv4_route l3fwd_ipv4_route_array[] = { 202 {IPv4(100,10,0,0), 16, 0}, 203 {IPv4(100,20,0,0), 16, 1}, 204 {IPv4(100,30,0,0), 16, 2}, 205 {IPv4(100,40,0,0), 16, 3}, 206 {IPv4(100,50,0,0), 16, 4}, 207 {IPv4(100,60,0,0), 16, 5}, 208 {IPv4(100,70,0,0), 16, 6}, 209 {IPv4(100,80,0,0), 16, 7}, 210 }; 211 212 /* 213 * IPv6 forwarding table 214 */ 215 216 struct l3fwd_ipv6_route { 217 uint8_t ip[IPV6_ADDR_LEN]; 218 uint8_t depth; 219 uint8_t if_out; 220 }; 221 222 static struct l3fwd_ipv6_route l3fwd_ipv6_route_array[] = { 223 {{1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 0}, 224 {{2,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 1}, 225 {{3,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 2}, 226 {{4,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 3}, 227 {{5,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 4}, 228 {{6,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 5}, 229 {{7,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 6}, 230 {{8,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 7}, 231 }; 232 233 #define LPM_MAX_RULES 1024 234 #define LPM6_MAX_RULES 1024 235 #define LPM6_NUMBER_TBL8S (1 << 16) 236 237 struct rte_lpm6_config lpm6_config = { 238 .max_rules = LPM6_MAX_RULES, 239 .number_tbl8s = LPM6_NUMBER_TBL8S, 240 .flags = 0 241 }; 242 243 static struct rte_mempool *socket_direct_pool[RTE_MAX_NUMA_NODES]; 244 static struct rte_mempool *socket_indirect_pool[RTE_MAX_NUMA_NODES]; 245 static struct rte_lpm *socket_lpm[RTE_MAX_NUMA_NODES]; 246 static struct rte_lpm6 *socket_lpm6[RTE_MAX_NUMA_NODES]; 247 248 /* Send burst of packets on an output interface */ 249 static inline int 250 send_burst(struct lcore_queue_conf *qconf, uint16_t n, uint8_t port) 251 { 252 struct rte_mbuf **m_table; 253 int ret; 254 uint16_t queueid; 255 256 queueid = qconf->tx_queue_id[port]; 257 m_table = (struct rte_mbuf **)qconf->tx_mbufs[port].m_table; 258 259 ret = rte_eth_tx_burst(port, queueid, m_table, n); 260 if (unlikely(ret < n)) { 261 do { 262 rte_pktmbuf_free(m_table[ret]); 263 } while (++ret < n); 264 } 265 266 return 0; 267 } 268 269 static inline void 270 l3fwd_simple_forward(struct rte_mbuf *m, struct lcore_queue_conf *qconf, 271 uint8_t queueid, uint8_t port_in) 272 { 273 struct rx_queue *rxq; 274 uint32_t i, len; 275 uint8_t next_hop, port_out, ipv6; 276 int32_t len2; 277 278 ipv6 = 0; 279 rxq = &qconf->rx_queue_list[queueid]; 280 281 /* by default, send everything back to the source port */ 282 port_out = port_in; 283 284 /* Remove the Ethernet header and trailer from the input packet */ 285 rte_pktmbuf_adj(m, (uint16_t)sizeof(struct ether_hdr)); 286 287 /* Build transmission burst */ 288 len = qconf->tx_mbufs[port_out].len; 289 290 /* if this is an IPv4 packet */ 291 if (m->ol_flags & PKT_RX_IPV4_HDR) { 292 struct ipv4_hdr *ip_hdr; 293 uint32_t ip_dst; 294 /* Read the lookup key (i.e. ip_dst) from the input packet */ 295 ip_hdr = rte_pktmbuf_mtod(m, struct ipv4_hdr *); 296 ip_dst = rte_be_to_cpu_32(ip_hdr->dst_addr); 297 298 /* Find destination port */ 299 if (rte_lpm_lookup(rxq->lpm, ip_dst, &next_hop) == 0 && 300 (enabled_port_mask & 1 << next_hop) != 0) { 301 port_out = next_hop; 302 303 /* Build transmission burst for new port */ 304 len = qconf->tx_mbufs[port_out].len; 305 } 306 307 /* if we don't need to do any fragmentation */ 308 if (likely (IPV4_MTU_DEFAULT >= m->pkt_len)) { 309 qconf->tx_mbufs[port_out].m_table[len] = m; 310 len2 = 1; 311 } else { 312 len2 = rte_ipv4_fragment_packet(m, 313 &qconf->tx_mbufs[port_out].m_table[len], 314 (uint16_t)(MBUF_TABLE_SIZE - len), 315 IPV4_MTU_DEFAULT, 316 rxq->direct_pool, rxq->indirect_pool); 317 318 /* Free input packet */ 319 rte_pktmbuf_free(m); 320 321 /* If we fail to fragment the packet */ 322 if (unlikely (len2 < 0)) 323 return; 324 } 325 } 326 /* if this is an IPv6 packet */ 327 else if (m->ol_flags & PKT_RX_IPV6_HDR) { 328 struct ipv6_hdr *ip_hdr; 329 330 ipv6 = 1; 331 332 /* Read the lookup key (i.e. ip_dst) from the input packet */ 333 ip_hdr = rte_pktmbuf_mtod(m, struct ipv6_hdr *); 334 335 /* Find destination port */ 336 if (rte_lpm6_lookup(rxq->lpm6, ip_hdr->dst_addr, &next_hop) == 0 && 337 (enabled_port_mask & 1 << next_hop) != 0) { 338 port_out = next_hop; 339 340 /* Build transmission burst for new port */ 341 len = qconf->tx_mbufs[port_out].len; 342 } 343 344 /* if we don't need to do any fragmentation */ 345 if (likely (IPV6_MTU_DEFAULT >= m->pkt_len)) { 346 qconf->tx_mbufs[port_out].m_table[len] = m; 347 len2 = 1; 348 } else { 349 len2 = rte_ipv6_fragment_packet(m, 350 &qconf->tx_mbufs[port_out].m_table[len], 351 (uint16_t)(MBUF_TABLE_SIZE - len), 352 IPV6_MTU_DEFAULT, 353 rxq->direct_pool, rxq->indirect_pool); 354 355 /* Free input packet */ 356 rte_pktmbuf_free(m); 357 358 /* If we fail to fragment the packet */ 359 if (unlikely (len2 < 0)) 360 return; 361 } 362 } 363 /* else, just forward the packet */ 364 else { 365 qconf->tx_mbufs[port_out].m_table[len] = m; 366 len2 = 1; 367 } 368 369 for (i = len; i < len + len2; i ++) { 370 void *d_addr_bytes; 371 372 m = qconf->tx_mbufs[port_out].m_table[i]; 373 struct ether_hdr *eth_hdr = (struct ether_hdr *) 374 rte_pktmbuf_prepend(m, (uint16_t)sizeof(struct ether_hdr)); 375 if (eth_hdr == NULL) { 376 rte_panic("No headroom in mbuf.\n"); 377 } 378 379 m->l2_len = sizeof(struct ether_hdr); 380 381 /* 02:00:00:00:00:xx */ 382 d_addr_bytes = ð_hdr->d_addr.addr_bytes[0]; 383 *((uint64_t *)d_addr_bytes) = 0x000000000002 + ((uint64_t)port_out << 40); 384 385 /* src addr */ 386 ether_addr_copy(&ports_eth_addr[port_out], ð_hdr->s_addr); 387 if (ipv6) 388 eth_hdr->ether_type = rte_be_to_cpu_16(ETHER_TYPE_IPv6); 389 else 390 eth_hdr->ether_type = rte_be_to_cpu_16(ETHER_TYPE_IPv4); 391 } 392 393 len += len2; 394 395 if (likely(len < MAX_PKT_BURST)) { 396 qconf->tx_mbufs[port_out].len = (uint16_t)len; 397 return; 398 } 399 400 /* Transmit packets */ 401 send_burst(qconf, (uint16_t)len, port_out); 402 qconf->tx_mbufs[port_out].len = 0; 403 } 404 405 /* main processing loop */ 406 static int 407 main_loop(__attribute__((unused)) void *dummy) 408 { 409 struct rte_mbuf *pkts_burst[MAX_PKT_BURST]; 410 unsigned lcore_id; 411 uint64_t prev_tsc, diff_tsc, cur_tsc; 412 int i, j, nb_rx; 413 uint8_t portid; 414 struct lcore_queue_conf *qconf; 415 const uint64_t drain_tsc = (rte_get_tsc_hz() + US_PER_S - 1) / US_PER_S * BURST_TX_DRAIN_US; 416 417 prev_tsc = 0; 418 419 lcore_id = rte_lcore_id(); 420 qconf = &lcore_queue_conf[lcore_id]; 421 422 if (qconf->n_rx_queue == 0) { 423 RTE_LOG(INFO, IP_FRAG, "lcore %u has nothing to do\n", lcore_id); 424 return 0; 425 } 426 427 RTE_LOG(INFO, IP_FRAG, "entering main loop on lcore %u\n", lcore_id); 428 429 for (i = 0; i < qconf->n_rx_queue; i++) { 430 431 portid = qconf->rx_queue_list[i].portid; 432 RTE_LOG(INFO, IP_FRAG, " -- lcoreid=%u portid=%d\n", lcore_id, 433 (int) portid); 434 } 435 436 while (1) { 437 438 cur_tsc = rte_rdtsc(); 439 440 /* 441 * TX burst queue drain 442 */ 443 diff_tsc = cur_tsc - prev_tsc; 444 if (unlikely(diff_tsc > drain_tsc)) { 445 446 /* 447 * This could be optimized (use queueid instead of 448 * portid), but it is not called so often 449 */ 450 for (portid = 0; portid < RTE_MAX_ETHPORTS; portid++) { 451 if (qconf->tx_mbufs[portid].len == 0) 452 continue; 453 send_burst(&lcore_queue_conf[lcore_id], 454 qconf->tx_mbufs[portid].len, 455 portid); 456 qconf->tx_mbufs[portid].len = 0; 457 } 458 459 prev_tsc = cur_tsc; 460 } 461 462 /* 463 * Read packet from RX queues 464 */ 465 for (i = 0; i < qconf->n_rx_queue; i++) { 466 467 portid = qconf->rx_queue_list[i].portid; 468 nb_rx = rte_eth_rx_burst(portid, 0, pkts_burst, 469 MAX_PKT_BURST); 470 471 /* Prefetch first packets */ 472 for (j = 0; j < PREFETCH_OFFSET && j < nb_rx; j++) { 473 rte_prefetch0(rte_pktmbuf_mtod( 474 pkts_burst[j], void *)); 475 } 476 477 /* Prefetch and forward already prefetched packets */ 478 for (j = 0; j < (nb_rx - PREFETCH_OFFSET); j++) { 479 rte_prefetch0(rte_pktmbuf_mtod(pkts_burst[ 480 j + PREFETCH_OFFSET], void *)); 481 l3fwd_simple_forward(pkts_burst[j], qconf, i, portid); 482 } 483 484 /* Forward remaining prefetched packets */ 485 for (; j < nb_rx; j++) { 486 l3fwd_simple_forward(pkts_burst[j], qconf, i, portid); 487 } 488 } 489 } 490 } 491 492 /* display usage */ 493 static void 494 print_usage(const char *prgname) 495 { 496 printf("%s [EAL options] -- -p PORTMASK [-q NQ]\n" 497 " -p PORTMASK: hexadecimal bitmask of ports to configure\n" 498 " -q NQ: number of queue (=ports) per lcore (default is 1)\n", 499 prgname); 500 } 501 502 static int 503 parse_portmask(const char *portmask) 504 { 505 char *end = NULL; 506 unsigned long pm; 507 508 /* parse hexadecimal string */ 509 pm = strtoul(portmask, &end, 16); 510 if ((portmask[0] == '\0') || (end == NULL) || (*end != '\0')) 511 return -1; 512 513 if (pm == 0) 514 return -1; 515 516 return pm; 517 } 518 519 static int 520 parse_nqueue(const char *q_arg) 521 { 522 char *end = NULL; 523 unsigned long n; 524 525 /* parse hexadecimal string */ 526 n = strtoul(q_arg, &end, 10); 527 if ((q_arg[0] == '\0') || (end == NULL) || (*end != '\0')) 528 return -1; 529 if (n == 0) 530 return -1; 531 if (n >= MAX_RX_QUEUE_PER_LCORE) 532 return -1; 533 534 return n; 535 } 536 537 /* Parse the argument given in the command line of the application */ 538 static int 539 parse_args(int argc, char **argv) 540 { 541 int opt, ret; 542 char **argvopt; 543 int option_index; 544 char *prgname = argv[0]; 545 static struct option lgopts[] = { 546 {NULL, 0, 0, 0} 547 }; 548 549 argvopt = argv; 550 551 while ((opt = getopt_long(argc, argvopt, "p:q:", 552 lgopts, &option_index)) != EOF) { 553 554 switch (opt) { 555 /* portmask */ 556 case 'p': 557 enabled_port_mask = parse_portmask(optarg); 558 if (enabled_port_mask < 0) { 559 printf("invalid portmask\n"); 560 print_usage(prgname); 561 return -1; 562 } 563 break; 564 565 /* nqueue */ 566 case 'q': 567 rx_queue_per_lcore = parse_nqueue(optarg); 568 if (rx_queue_per_lcore < 0) { 569 printf("invalid queue number\n"); 570 print_usage(prgname); 571 return -1; 572 } 573 break; 574 575 /* long options */ 576 case 0: 577 print_usage(prgname); 578 return -1; 579 580 default: 581 print_usage(prgname); 582 return -1; 583 } 584 } 585 586 if (enabled_port_mask == 0) { 587 printf("portmask not specified\n"); 588 print_usage(prgname); 589 return -1; 590 } 591 592 if (optind >= 0) 593 argv[optind-1] = prgname; 594 595 ret = optind-1; 596 optind = 0; /* reset getopt lib */ 597 return ret; 598 } 599 600 static void 601 print_ethaddr(const char *name, struct ether_addr *eth_addr) 602 { 603 char buf[ETHER_ADDR_FMT_SIZE]; 604 ether_format_addr(buf, ETHER_ADDR_FMT_SIZE, eth_addr); 605 printf("%s%s", name, buf); 606 } 607 608 /* Check the link status of all ports in up to 9s, and print them finally */ 609 static void 610 check_all_ports_link_status(uint8_t port_num, uint32_t port_mask) 611 { 612 #define CHECK_INTERVAL 100 /* 100ms */ 613 #define MAX_CHECK_TIME 90 /* 9s (90 * 100ms) in total */ 614 uint8_t portid, count, all_ports_up, print_flag = 0; 615 struct rte_eth_link link; 616 617 printf("\nChecking link status"); 618 fflush(stdout); 619 for (count = 0; count <= MAX_CHECK_TIME; count++) { 620 all_ports_up = 1; 621 for (portid = 0; portid < port_num; portid++) { 622 if ((port_mask & (1 << portid)) == 0) 623 continue; 624 memset(&link, 0, sizeof(link)); 625 rte_eth_link_get_nowait(portid, &link); 626 /* print link status if flag set */ 627 if (print_flag == 1) { 628 if (link.link_status) 629 printf("Port %d Link Up - speed %u " 630 "Mbps - %s\n", (uint8_t)portid, 631 (unsigned)link.link_speed, 632 (link.link_duplex == ETH_LINK_FULL_DUPLEX) ? 633 ("full-duplex") : ("half-duplex\n")); 634 else 635 printf("Port %d Link Down\n", 636 (uint8_t)portid); 637 continue; 638 } 639 /* clear all_ports_up flag if any link down */ 640 if (link.link_status == 0) { 641 all_ports_up = 0; 642 break; 643 } 644 } 645 /* after finally printing all link status, get out */ 646 if (print_flag == 1) 647 break; 648 649 if (all_ports_up == 0) { 650 printf("."); 651 fflush(stdout); 652 rte_delay_ms(CHECK_INTERVAL); 653 } 654 655 /* set the print_flag if all ports up or timeout */ 656 if (all_ports_up == 1 || count == (MAX_CHECK_TIME - 1)) { 657 print_flag = 1; 658 printf("\ndone\n"); 659 } 660 } 661 } 662 663 static int 664 init_routing_table(void) 665 { 666 struct rte_lpm *lpm; 667 struct rte_lpm6 *lpm6; 668 int socket, ret; 669 unsigned i; 670 671 for (socket = 0; socket < RTE_MAX_NUMA_NODES; socket++) { 672 if (socket_lpm[socket]) { 673 lpm = socket_lpm[socket]; 674 /* populate the LPM table */ 675 for (i = 0; i < RTE_DIM(l3fwd_ipv4_route_array); i++) { 676 ret = rte_lpm_add(lpm, 677 l3fwd_ipv4_route_array[i].ip, 678 l3fwd_ipv4_route_array[i].depth, 679 l3fwd_ipv4_route_array[i].if_out); 680 681 if (ret < 0) { 682 RTE_LOG(ERR, IP_FRAG, "Unable to add entry %i to the l3fwd " 683 "LPM table\n", i); 684 return -1; 685 } 686 687 RTE_LOG(INFO, IP_FRAG, "Socket %i: adding route " IPv4_BYTES_FMT 688 "/%d (port %d)\n", 689 socket, 690 IPv4_BYTES(l3fwd_ipv4_route_array[i].ip), 691 l3fwd_ipv4_route_array[i].depth, 692 l3fwd_ipv4_route_array[i].if_out); 693 } 694 } 695 696 if (socket_lpm6[socket]) { 697 lpm6 = socket_lpm6[socket]; 698 /* populate the LPM6 table */ 699 for (i = 0; i < RTE_DIM(l3fwd_ipv6_route_array); i++) { 700 ret = rte_lpm6_add(lpm6, 701 l3fwd_ipv6_route_array[i].ip, 702 l3fwd_ipv6_route_array[i].depth, 703 l3fwd_ipv6_route_array[i].if_out); 704 705 if (ret < 0) { 706 RTE_LOG(ERR, IP_FRAG, "Unable to add entry %i to the l3fwd " 707 "LPM6 table\n", i); 708 return -1; 709 } 710 711 RTE_LOG(INFO, IP_FRAG, "Socket %i: adding route " IPv6_BYTES_FMT 712 "/%d (port %d)\n", 713 socket, 714 IPv6_BYTES(l3fwd_ipv6_route_array[i].ip), 715 l3fwd_ipv6_route_array[i].depth, 716 l3fwd_ipv6_route_array[i].if_out); 717 } 718 } 719 } 720 return 0; 721 } 722 723 static int 724 init_mem(void) 725 { 726 char buf[PATH_MAX]; 727 struct rte_mempool *mp; 728 struct rte_lpm *lpm; 729 struct rte_lpm6 *lpm6; 730 int socket; 731 unsigned lcore_id; 732 733 /* traverse through lcores and initialize structures on each socket */ 734 735 for (lcore_id = 0; lcore_id < RTE_MAX_LCORE; lcore_id++) { 736 737 if (rte_lcore_is_enabled(lcore_id) == 0) 738 continue; 739 740 socket = rte_lcore_to_socket_id(lcore_id); 741 742 if (socket == SOCKET_ID_ANY) 743 socket = 0; 744 745 if (socket_direct_pool[socket] == NULL) { 746 RTE_LOG(INFO, IP_FRAG, "Creating direct mempool on socket %i\n", 747 socket); 748 snprintf(buf, sizeof(buf), "pool_direct_%i", socket); 749 750 mp = rte_mempool_create(buf, NB_MBUF, 751 MBUF_SIZE, 32, 752 sizeof(struct rte_pktmbuf_pool_private), 753 rte_pktmbuf_pool_init, NULL, 754 rte_pktmbuf_init, NULL, 755 socket, 0); 756 if (mp == NULL) { 757 RTE_LOG(ERR, IP_FRAG, "Cannot create direct mempool\n"); 758 return -1; 759 } 760 socket_direct_pool[socket] = mp; 761 } 762 763 if (socket_indirect_pool[socket] == NULL) { 764 RTE_LOG(INFO, IP_FRAG, "Creating indirect mempool on socket %i\n", 765 socket); 766 snprintf(buf, sizeof(buf), "pool_indirect_%i", socket); 767 768 mp = rte_mempool_create(buf, NB_MBUF, 769 sizeof(struct rte_mbuf), 32, 770 0, 771 NULL, NULL, 772 rte_pktmbuf_init, NULL, 773 socket, 0); 774 if (mp == NULL) { 775 RTE_LOG(ERR, IP_FRAG, "Cannot create indirect mempool\n"); 776 return -1; 777 } 778 socket_indirect_pool[socket] = mp; 779 } 780 781 if (socket_lpm[socket] == NULL) { 782 RTE_LOG(INFO, IP_FRAG, "Creating LPM table on socket %i\n", socket); 783 snprintf(buf, sizeof(buf), "IP_FRAG_LPM_%i", socket); 784 785 lpm = rte_lpm_create(buf, socket, LPM_MAX_RULES, 0); 786 if (lpm == NULL) { 787 RTE_LOG(ERR, IP_FRAG, "Cannot create LPM table\n"); 788 return -1; 789 } 790 socket_lpm[socket] = lpm; 791 } 792 793 if (socket_lpm6[socket] == NULL) { 794 RTE_LOG(INFO, IP_FRAG, "Creating LPM6 table on socket %i\n", socket); 795 snprintf(buf, sizeof(buf), "IP_FRAG_LPM_%i", socket); 796 797 lpm6 = rte_lpm6_create("IP_FRAG_LPM6", socket, &lpm6_config); 798 if (lpm6 == NULL) { 799 RTE_LOG(ERR, IP_FRAG, "Cannot create LPM table\n"); 800 return -1; 801 } 802 socket_lpm6[socket] = lpm6; 803 } 804 } 805 806 return 0; 807 } 808 809 int 810 MAIN(int argc, char **argv) 811 { 812 struct lcore_queue_conf *qconf; 813 struct rte_eth_dev_info dev_info; 814 struct rte_eth_txconf *txconf; 815 struct rx_queue *rxq; 816 int socket, ret; 817 unsigned nb_ports; 818 uint16_t queueid = 0; 819 unsigned lcore_id = 0, rx_lcore_id = 0; 820 uint32_t n_tx_queue, nb_lcores; 821 uint8_t portid; 822 823 /* init EAL */ 824 ret = rte_eal_init(argc, argv); 825 if (ret < 0) 826 rte_exit(EXIT_FAILURE, "rte_eal_init failed"); 827 argc -= ret; 828 argv += ret; 829 830 /* parse application arguments (after the EAL ones) */ 831 ret = parse_args(argc, argv); 832 if (ret < 0) 833 rte_exit(EXIT_FAILURE, "Invalid arguments"); 834 835 nb_ports = rte_eth_dev_count(); 836 if (nb_ports > RTE_MAX_ETHPORTS) 837 nb_ports = RTE_MAX_ETHPORTS; 838 else if (nb_ports == 0) 839 rte_exit(EXIT_FAILURE, "No ports found!\n"); 840 841 nb_lcores = rte_lcore_count(); 842 843 /* initialize structures (mempools, lpm etc.) */ 844 if (init_mem() < 0) 845 rte_panic("Cannot initialize memory structures!\n"); 846 847 /* check if portmask has non-existent ports */ 848 if (enabled_port_mask & ~(RTE_LEN2MASK(nb_ports, unsigned))) 849 rte_exit(EXIT_FAILURE, "Non-existent ports in portmask!\n"); 850 851 /* initialize all ports */ 852 for (portid = 0; portid < nb_ports; portid++) { 853 /* skip ports that are not enabled */ 854 if ((enabled_port_mask & (1 << portid)) == 0) { 855 printf("Skipping disabled port %d\n", portid); 856 continue; 857 } 858 859 qconf = &lcore_queue_conf[rx_lcore_id]; 860 861 /* get the lcore_id for this port */ 862 while (rte_lcore_is_enabled(rx_lcore_id) == 0 || 863 qconf->n_rx_queue == (unsigned)rx_queue_per_lcore) { 864 865 rx_lcore_id ++; 866 if (rx_lcore_id >= RTE_MAX_LCORE) 867 rte_exit(EXIT_FAILURE, "Not enough cores\n"); 868 869 qconf = &lcore_queue_conf[rx_lcore_id]; 870 } 871 872 socket = (int) rte_lcore_to_socket_id(rx_lcore_id); 873 if (socket == SOCKET_ID_ANY) 874 socket = 0; 875 876 rxq = &qconf->rx_queue_list[qconf->n_rx_queue]; 877 rxq->portid = portid; 878 rxq->direct_pool = socket_direct_pool[socket]; 879 rxq->indirect_pool = socket_indirect_pool[socket]; 880 rxq->lpm = socket_lpm[socket]; 881 rxq->lpm6 = socket_lpm6[socket]; 882 qconf->n_rx_queue++; 883 884 /* init port */ 885 printf("Initializing port %d on lcore %u...", portid, 886 rx_lcore_id); 887 fflush(stdout); 888 889 n_tx_queue = nb_lcores; 890 if (n_tx_queue > MAX_TX_QUEUE_PER_PORT) 891 n_tx_queue = MAX_TX_QUEUE_PER_PORT; 892 ret = rte_eth_dev_configure(portid, 1, (uint16_t)n_tx_queue, 893 &port_conf); 894 if (ret < 0) { 895 printf("\n"); 896 rte_exit(EXIT_FAILURE, "Cannot configure device: " 897 "err=%d, port=%d\n", 898 ret, portid); 899 } 900 901 /* init one RX queue */ 902 ret = rte_eth_rx_queue_setup(portid, 0, nb_rxd, 903 socket, NULL, 904 socket_direct_pool[socket]); 905 if (ret < 0) { 906 printf("\n"); 907 rte_exit(EXIT_FAILURE, "rte_eth_rx_queue_setup: " 908 "err=%d, port=%d\n", 909 ret, portid); 910 } 911 912 rte_eth_macaddr_get(portid, &ports_eth_addr[portid]); 913 print_ethaddr(" Address:", &ports_eth_addr[portid]); 914 printf("\n"); 915 916 /* init one TX queue per couple (lcore,port) */ 917 queueid = 0; 918 for (lcore_id = 0; lcore_id < RTE_MAX_LCORE; lcore_id++) { 919 if (rte_lcore_is_enabled(lcore_id) == 0) 920 continue; 921 922 socket = (int) rte_lcore_to_socket_id(lcore_id); 923 printf("txq=%u,%d ", lcore_id, queueid); 924 fflush(stdout); 925 926 rte_eth_dev_info_get(portid, &dev_info); 927 txconf = &dev_info.default_txconf; 928 txconf->txq_flags = 0; 929 ret = rte_eth_tx_queue_setup(portid, queueid, nb_txd, 930 socket, txconf); 931 if (ret < 0) { 932 printf("\n"); 933 rte_exit(EXIT_FAILURE, "rte_eth_tx_queue_setup: " 934 "err=%d, port=%d\n", ret, portid); 935 } 936 937 qconf = &lcore_queue_conf[lcore_id]; 938 qconf->tx_queue_id[portid] = queueid; 939 queueid++; 940 } 941 942 printf("\n"); 943 } 944 945 printf("\n"); 946 947 /* start ports */ 948 for (portid = 0; portid < nb_ports; portid++) { 949 if ((enabled_port_mask & (1 << portid)) == 0) { 950 continue; 951 } 952 /* Start device */ 953 ret = rte_eth_dev_start(portid); 954 if (ret < 0) 955 rte_exit(EXIT_FAILURE, "rte_eth_dev_start: err=%d, port=%d\n", 956 ret, portid); 957 958 rte_eth_promiscuous_enable(portid); 959 } 960 961 if (init_routing_table() < 0) 962 rte_exit(EXIT_FAILURE, "Cannot init routing table\n"); 963 964 check_all_ports_link_status((uint8_t)nb_ports, enabled_port_mask); 965 966 /* launch per-lcore init on every lcore */ 967 rte_eal_mp_remote_launch(main_loop, NULL, CALL_MASTER); 968 RTE_LCORE_FOREACH_SLAVE(lcore_id) { 969 if (rte_eal_wait_lcore(lcore_id) < 0) 970 return -1; 971 } 972 973 return 0; 974 } 975