1.. SPDX-License-Identifier: BSD-3-Clause 2 Copyright(c) 2010-2014 Intel Corporation. 3 4L3 Forwarding Sample Application 5================================ 6 7The L3 Forwarding application is a simple example of packet processing using 8DPDK to demonstrate usage of poll and event mode packet I/O mechanism. 9The application performs L3 forwarding. 10 11Overview 12-------- 13 14The application demonstrates the use of the hash and LPM libraries in the DPDK 15to implement packet forwarding using poll or event mode PMDs for packet I/O. 16The initialization and run-time paths are very similar to those of the 17:doc:`l2_forward_real_virtual` and :doc:`l2_forward_event`. 18The main difference from the L2 Forwarding sample application is that optionally 19packet can be Rx/Tx from/to eventdev instead of port directly and forwarding 20decision is made based on information read from the input packet. 21 22Eventdev can optionally use S/W or H/W (if supported by platform) scheduler 23implementation for packet I/O based on run time parameters. 24 25The lookup method is either hash-based or LPM-based and is selected at run time. When the selected lookup method is hash-based, 26a hash object is used to emulate the flow classification stage. 27The hash object is used in correlation with a flow table to map each input packet to its flow at runtime. 28 29The hash lookup key is represented by a DiffServ 5-tuple composed of the following fields read from the input packet: 30Source IP Address, Destination IP Address, Protocol, Source Port and Destination Port. 31The ID of the output interface for the input packet is read from the identified flow table entry. 32The set of flows used by the application is statically configured and loaded into the hash at initialization time. 33When the selected lookup method is LPM based, an LPM object is used to emulate the forwarding stage for IPv4 packets. 34The LPM object is used as the routing table to identify the next hop for each input packet at runtime. 35 36The LPM lookup key is represented by the Destination IP Address field read from the input packet. 37The ID of the output interface for the input packet is the next hop returned by the LPM lookup. 38The set of LPM rules used by the application is statically configured and loaded into the LPM object at initialization time. 39 40In the sample application, hash-based forwarding supports IPv4 and IPv6. LPM-based forwarding supports IPv4 only. 41 42Compiling the Application 43------------------------- 44 45To compile the sample application see :doc:`compiling`. 46 47The application is located in the ``l3fwd`` sub-directory. 48 49Running the Application 50----------------------- 51 52The application has a number of command line options:: 53 54 ./dpdk-l3fwd [EAL options] -- -p PORTMASK 55 [-P] 56 [-E] 57 [-L] 58 --config(port,queue,lcore)[,(port,queue,lcore)] 59 [--eth-dest=X,MM:MM:MM:MM:MM:MM] 60 [--enable-jumbo [--max-pkt-len PKTLEN]] 61 [--no-numa] 62 [--hash-entry-num] 63 [--ipv6] 64 [--parse-ptype] 65 [--per-port-pool] 66 [--mode] 67 [--eventq-sched] 68 [--event-eth-rxqs] 69 70Where, 71 72* ``-p PORTMASK:`` Hexadecimal bitmask of ports to configure 73 74* ``-P:`` Optional, sets all ports to promiscuous mode so that packets are accepted regardless of the packet's Ethernet MAC destination address. 75 Without this option, only packets with the Ethernet MAC destination address set to the Ethernet address of the port are accepted. 76 77* ``-E:`` Optional, enable exact match. 78 79* ``-L:`` Optional, enable longest prefix match. 80 81* ``--config (port,queue,lcore)[,(port,queue,lcore)]:`` Determines which queues from which ports are mapped to which cores. 82 83* ``--eth-dest=X,MM:MM:MM:MM:MM:MM:`` Optional, ethernet destination for port X. 84 85* ``--enable-jumbo:`` Optional, enables jumbo frames. 86 87* ``--max-pkt-len:`` Optional, under the premise of enabling jumbo, maximum packet length in decimal (64-9600). 88 89* ``--no-numa:`` Optional, disables numa awareness. 90 91* ``--hash-entry-num:`` Optional, specifies the hash entry number in hexadecimal to be setup. 92 93* ``--ipv6:`` Optional, set if running ipv6 packets. 94 95* ``--parse-ptype:`` Optional, set to use software to analyze packet type. Without this option, hardware will check the packet type. 96 97* ``--per-port-pool:`` Optional, set to use independent buffer pools per port. Without this option, single buffer pool is used for all ports. 98 99* ``--mode:`` Optional, Packet transfer mode for I/O, poll or eventdev. 100 101* ``--eventq-sched:`` Optional, Event queue synchronization method, Ordered, Atomic or Parallel. Only valid if --mode=eventdev. 102 103* ``--event-eth-rxqs:`` Optional, Number of ethernet RX queues per device. Only valid if --mode=eventdev. 104 105 106For example, consider a dual processor socket platform with 8 physical cores, where cores 0-7 and 16-23 appear on socket 0, 107while cores 8-15 and 24-31 appear on socket 1. 108 109To enable L3 forwarding between two ports, assuming that both ports are in the same socket, using two cores, cores 1 and 2, 110(which are in the same socket too), use the following command: 111 112.. code-block:: console 113 114 ./<build_dir>/examples/dpdk-l3fwd -l 1,2 -n 4 -- -p 0x3 --config="(0,0,1),(1,0,2)" 115 116In this command: 117 118* The -l option enables cores 1, 2 119 120* The -p option enables ports 0 and 1 121 122* The --config option enables one queue on each port and maps each (port,queue) pair to a specific core. 123 The following table shows the mapping in this example: 124 125+----------+-----------+-----------+-------------------------------------+ 126| **Port** | **Queue** | **lcore** | **Description** | 127| | | | | 128+----------+-----------+-----------+-------------------------------------+ 129| 0 | 0 | 1 | Map queue 0 from port 0 to lcore 1. | 130| | | | | 131+----------+-----------+-----------+-------------------------------------+ 132| 1 | 0 | 2 | Map queue 0 from port 1 to lcore 2. | 133| | | | | 134+----------+-----------+-----------+-------------------------------------+ 135 136To use eventdev mode with sync method **ordered** on above mentioned environment, 137Following is the sample command: 138 139.. code-block:: console 140 141 ./<build_dir>/examples/dpdk-l3fwd -l 0-3 -n 4 -w <event device> -- -p 0x3 --eventq-sched=ordered 142 143or 144 145.. code-block:: console 146 147 ./<build_dir>/examples/dpdk-l3fwd -l 0-3 -n 4 -w <event device> -- -p 0x03 --mode=eventdev --eventq-sched=ordered 148 149In this command: 150 151* -w option whitelist the event device supported by platform. Way to pass this device may vary based on platform. 152 153* The --mode option defines PMD to be used for packet I/O. 154 155* The --eventq-sched option enables synchronization menthod of event queue so that packets will be scheduled accordingly. 156 157If application uses S/W scheduler, it uses following DPDK services: 158 159* Software scheduler 160* Rx adapter service function 161* Tx adapter service function 162 163Application needs service cores to run above mentioned services. Service cores 164must be provided as EAL parameters along with the --vdev=event_sw0 to enable S/W 165scheduler. Following is the sample command: 166 167.. code-block:: console 168 169 ./<build_dir>/examples/dpdk-l3fwd -l 0-7 -s 0xf0000 -n 4 --vdev event_sw0 -- -p 0x3 --mode=eventdev --eventq-sched=ordered 170 171In case of eventdev mode, *--config* option is not used for ethernet port 172configuration. Instead each ethernet port will be configured with mentioned 173setup: 174 175* Single Rx/Tx queue 176 177* Each Rx queue will be connected to event queue via Rx adapter. 178 179* Each Tx queue will be connected via Tx adapter. 180 181Refer to the *DPDK Getting Started Guide* for general information on running applications and 182the Environment Abstraction Layer (EAL) options. 183 184.. _l3_fwd_explanation: 185 186Explanation 187----------- 188 189The following sections provide some explanation of the sample application code. As mentioned in the overview section, 190the initialization and run-time paths are very similar to those of the :doc:`l2_forward_real_virtual` and :doc:`l2_forward_event`. 191The following sections describe aspects that are specific to the L3 Forwarding sample application. 192 193Hash Initialization 194~~~~~~~~~~~~~~~~~~~ 195 196The hash object is created and loaded with the pre-configured entries read from a global array, 197and then generate the expected 5-tuple as key to keep consistence with those of real flow 198for the convenience to execute hash performance test on 4M/8M/16M flows. 199 200.. note:: 201 202 The Hash initialization will setup both ipv4 and ipv6 hash table, 203 and populate the either table depending on the value of variable ipv6. 204 To support the hash performance test with up to 8M single direction flows/16M bi-direction flows, 205 populate_ipv4_many_flow_into_table() function will populate the hash table with specified hash table entry number(default 4M). 206 207.. note:: 208 209 Value of global variable ipv6 can be specified with --ipv6 in the command line. 210 Value of global variable hash_entry_number, 211 which is used to specify the total hash entry number for all used ports in hash performance test, 212 can be specified with --hash-entry-num VALUE in command line, being its default value 4. 213 214.. code-block:: c 215 216 #if (APP_LOOKUP_METHOD == APP_LOOKUP_EXACT_MATCH) 217 218 static void 219 setup_hash(int socketid) 220 { 221 // ... 222 223 if (hash_entry_number != HASH_ENTRY_NUMBER_DEFAULT) { 224 if (ipv6 == 0) { 225 /* populate the ipv4 hash */ 226 populate_ipv4_many_flow_into_table(ipv4_l3fwd_lookup_struct[socketid], hash_entry_number); 227 } else { 228 /* populate the ipv6 hash */ 229 populate_ipv6_many_flow_into_table( ipv6_l3fwd_lookup_struct[socketid], hash_entry_number); 230 } 231 } else 232 if (ipv6 == 0) { 233 /* populate the ipv4 hash */ 234 populate_ipv4_few_flow_into_table(ipv4_l3fwd_lookup_struct[socketid]); 235 } else { 236 /* populate the ipv6 hash */ 237 populate_ipv6_few_flow_into_table(ipv6_l3fwd_lookup_struct[socketid]); 238 } 239 } 240 } 241 #endif 242 243LPM Initialization 244~~~~~~~~~~~~~~~~~~ 245 246The LPM object is created and loaded with the pre-configured entries read from a global array. 247 248.. code-block:: c 249 250 #if (APP_LOOKUP_METHOD == APP_LOOKUP_LPM) 251 252 static void 253 setup_lpm(int socketid) 254 { 255 unsigned i; 256 int ret; 257 char s[64]; 258 259 /* create the LPM table */ 260 261 snprintf(s, sizeof(s), "IPV4_L3FWD_LPM_%d", socketid); 262 263 ipv4_l3fwd_lookup_struct[socketid] = rte_lpm_create(s, socketid, IPV4_L3FWD_LPM_MAX_RULES, 0); 264 265 if (ipv4_l3fwd_lookup_struct[socketid] == NULL) 266 rte_exit(EXIT_FAILURE, "Unable to create the l3fwd LPM table" 267 " on socket %d\n", socketid); 268 269 /* populate the LPM table */ 270 271 for (i = 0; i < IPV4_L3FWD_NUM_ROUTES; i++) { 272 /* skip unused ports */ 273 274 if ((1 << ipv4_l3fwd_route_array[i].if_out & enabled_port_mask) == 0) 275 continue; 276 277 ret = rte_lpm_add(ipv4_l3fwd_lookup_struct[socketid], ipv4_l3fwd_route_array[i].ip, 278 ipv4_l3fwd_route_array[i].depth, ipv4_l3fwd_route_array[i].if_out); 279 280 if (ret < 0) { 281 rte_exit(EXIT_FAILURE, "Unable to add entry %u to the " 282 "l3fwd LPM table on socket %d\n", i, socketid); 283 } 284 285 printf("LPM: Adding route 0x%08x / %d (%d)\n", 286 (unsigned)ipv4_l3fwd_route_array[i].ip, ipv4_l3fwd_route_array[i].depth, ipv4_l3fwd_route_array[i].if_out); 287 } 288 } 289 #endif 290 291Packet Forwarding for Hash-based Lookups 292~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 293 294For each input packet, the packet forwarding operation is done by the l3fwd_simple_forward() 295or simple_ipv4_fwd_4pkts() function for IPv4 packets or the simple_ipv6_fwd_4pkts() function for IPv6 packets. 296The l3fwd_simple_forward() function provides the basic functionality for both IPv4 and IPv6 packet forwarding 297for any number of burst packets received, 298and the packet forwarding decision (that is, the identification of the output interface for the packet) 299for hash-based lookups is done by the get_ipv4_dst_port() or get_ipv6_dst_port() function. 300The get_ipv4_dst_port() function is shown below: 301 302.. code-block:: c 303 304 static inline uint8_t 305 get_ipv4_dst_port(void *ipv4_hdr, uint16_t portid, lookup_struct_t *ipv4_l3fwd_lookup_struct) 306 { 307 int ret = 0; 308 union ipv4_5tuple_host key; 309 310 ipv4_hdr = (uint8_t *)ipv4_hdr + offsetof(struct rte_ipv4_hdr, time_to_live); 311 312 m128i data = _mm_loadu_si128(( m128i*)(ipv4_hdr)); 313 314 /* Get 5 tuple: dst port, src port, dst IP address, src IP address and protocol */ 315 316 key.xmm = _mm_and_si128(data, mask0); 317 318 /* Find destination port */ 319 320 ret = rte_hash_lookup(ipv4_l3fwd_lookup_struct, (const void *)&key); 321 322 return (uint8_t)((ret < 0)? portid : ipv4_l3fwd_out_if[ret]); 323 } 324 325The get_ipv6_dst_port() function is similar to the get_ipv4_dst_port() function. 326 327The simple_ipv4_fwd_4pkts() and simple_ipv6_fwd_4pkts() function are optimized for continuous 4 valid ipv4 and ipv6 packets, 328they leverage the multiple buffer optimization to boost the performance of forwarding packets with the exact match on hash table. 329The key code snippet of simple_ipv4_fwd_4pkts() is shown below: 330 331.. code-block:: c 332 333 static inline void 334 simple_ipv4_fwd_4pkts(struct rte_mbuf* m[4], uint16_t portid, struct lcore_conf *qconf) 335 { 336 // ... 337 338 data[0] = _mm_loadu_si128(( m128i*)(rte_pktmbuf_mtod(m[0], unsigned char *) + sizeof(struct rte_ether_hdr) + offsetof(struct rte_ipv4_hdr, time_to_live))); 339 data[1] = _mm_loadu_si128(( m128i*)(rte_pktmbuf_mtod(m[1], unsigned char *) + sizeof(struct rte_ether_hdr) + offsetof(struct rte_ipv4_hdr, time_to_live))); 340 data[2] = _mm_loadu_si128(( m128i*)(rte_pktmbuf_mtod(m[2], unsigned char *) + sizeof(struct rte_ether_hdr) + offsetof(struct rte_ipv4_hdr, time_to_live))); 341 data[3] = _mm_loadu_si128(( m128i*)(rte_pktmbuf_mtod(m[3], unsigned char *) + sizeof(struct rte_ether_hdr) + offsetof(struct rte_ipv4_hdr, time_to_live))); 342 343 key[0].xmm = _mm_and_si128(data[0], mask0); 344 key[1].xmm = _mm_and_si128(data[1], mask0); 345 key[2].xmm = _mm_and_si128(data[2], mask0); 346 key[3].xmm = _mm_and_si128(data[3], mask0); 347 348 const void *key_array[4] = {&key[0], &key[1], &key[2],&key[3]}; 349 350 rte_hash_lookup_bulk(qconf->ipv4_lookup_struct, &key_array[0], 4, ret); 351 352 dst_port[0] = (ret[0] < 0)? portid:ipv4_l3fwd_out_if[ret[0]]; 353 dst_port[1] = (ret[1] < 0)? portid:ipv4_l3fwd_out_if[ret[1]]; 354 dst_port[2] = (ret[2] < 0)? portid:ipv4_l3fwd_out_if[ret[2]]; 355 dst_port[3] = (ret[3] < 0)? portid:ipv4_l3fwd_out_if[ret[3]]; 356 357 // ... 358 } 359 360The simple_ipv6_fwd_4pkts() function is similar to the simple_ipv4_fwd_4pkts() function. 361 362Known issue: IP packets with extensions or IP packets which are not TCP/UDP cannot work well at this mode. 363 364Packet Forwarding for LPM-based Lookups 365~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 366 367For each input packet, the packet forwarding operation is done by the l3fwd_simple_forward() function, 368but the packet forwarding decision (that is, the identification of the output interface for the packet) 369for LPM-based lookups is done by the get_ipv4_dst_port() function below: 370 371.. code-block:: c 372 373 static inline uint16_t 374 get_ipv4_dst_port(struct rte_ipv4_hdr *ipv4_hdr, uint16_t portid, lookup_struct_t *ipv4_l3fwd_lookup_struct) 375 { 376 uint8_t next_hop; 377 378 return ((rte_lpm_lookup(ipv4_l3fwd_lookup_struct, rte_be_to_cpu_32(ipv4_hdr->dst_addr), &next_hop) == 0)? next_hop : portid); 379 } 380 381Eventdev Driver Initialization 382~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 383Eventdev driver initialization is same as L2 forwarding eventdev application. 384Refer :doc:`l2_forward_event` for more details. 385