2 Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
5 Redistribution and use in source and binary forms, with or without
6 modification, are permitted provided that the following conditions
9 * Redistributions of source code must retain the above copyright
10 notice, this list of conditions and the following disclaimer.
11 * Redistributions in binary form must reproduce the above copyright
12 notice, this list of conditions and the following disclaimer in
13 the documentation and/or other materials provided with the
15 * Neither the name of Intel Corporation nor the names of its
16 contributors may be used to endorse or promote products derived
17 from this software without specific prior written permission.
19 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
20 "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
21 LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
22 A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
23 OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
24 SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
25 LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
26 DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
27 THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
28 (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
29 OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
31 .. _l2_fwd_app_real_and_virtual:
33 L2 Forwarding Sample Application (in Real and Virtualized Environments)
34 =======================================================================
36 The L2 Forwarding sample application is a simple example of packet processing using
37 the Data Plane Development Kit (DPDK) which
38 also takes advantage of Single Root I/O Virtualization (SR-IOV) features in a virtualized environment.
42 Please note that previously a separate L2 Forwarding in Virtualized Environments sample application was used,
43 however, in later DPDK versions these sample applications have been merged.
48 The L2 Forwarding sample application, which can operate in real and virtualized environments,
49 performs L2 forwarding for each packet that is received on an RX_PORT.
50 The destination port is the adjacent port from the enabled portmask, that is,
51 if the first four ports are enabled (portmask 0xf),
52 ports 1 and 2 forward into each other, and ports 3 and 4 forward into each other.
53 Also, if MAC addresses updating is enabled, the MAC addresses are affected as follows:
55 * The source MAC address is replaced by the TX_PORT MAC address
57 * The destination MAC address is replaced by 02:00:00:00:00:TX_PORT_ID
59 This application can be used to benchmark performance using a traffic-generator, as shown in the :numref:`figure_l2_fwd_benchmark_setup`,
60 or in a virtualized environment as shown in :numref:`figure_l2_fwd_virtenv_benchmark_setup`.
62 .. _figure_l2_fwd_benchmark_setup:
64 .. figure:: img/l2_fwd_benchmark_setup.*
66 Performance Benchmark Setup (Basic Environment)
68 .. _figure_l2_fwd_virtenv_benchmark_setup:
70 .. figure:: img/l2_fwd_virtenv_benchmark_setup.*
72 Performance Benchmark Setup (Virtualized Environment)
74 This application may be used for basic VM to VM communication as shown in :numref:`figure_l2_fwd_vm2vm`,
75 when MAC addresses updating is disabled.
77 .. _figure_l2_fwd_vm2vm:
79 .. figure:: img/l2_fwd_vm2vm.*
81 Virtual Machine to Virtual Machine communication.
83 The L2 Forwarding application can also be used as a starting point for developing a new application based on the DPDK.
87 Virtual Function Setup Instructions
88 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
90 This application can use the virtual function available in the system and
91 therefore can be used in a virtual machine without passing through
92 the whole Network Device into a guest machine in a virtualized scenario.
93 The virtual functions can be enabled in the host machine or the hypervisor with the respective physical function driver.
95 For example, in a Linux* host machine, it is possible to enable a virtual function using the following command:
97 .. code-block:: console
99 modprobe ixgbe max_vfs=2,2
101 This command enables two Virtual Functions on each of Physical Function of the NIC,
102 with two physical ports in the PCI configuration space.
103 It is important to note that enabled Virtual Function 0 and 2 would belong to Physical Function 0
104 and Virtual Function 1 and 3 would belong to Physical Function 1,
105 in this case enabling a total of four Virtual Functions.
107 Compiling the Application
108 -------------------------
110 #. Go to the example directory:
112 .. code-block:: console
114 export RTE_SDK=/path/to/rte_sdk
115 cd ${RTE_SDK}/examples/l2fwd
117 #. Set the target (a default target is used if not specified). For example:
119 .. code-block:: console
121 export RTE_TARGET=x86_64-native-linuxapp-gcc
123 *See the DPDK Getting Started Guide* for possible RTE_TARGET values.
125 #. Build the application:
127 .. code-block:: console
131 Running the Application
132 -----------------------
134 The application requires a number of command line options:
136 .. code-block:: console
138 ./build/l2fwd [EAL options] -- -p PORTMASK [-q NQ] --[no-]mac-updating
142 * p PORTMASK: A hexadecimal bitmask of the ports to configure
144 * q NQ: A number of queues (=ports) per lcore (default is 1)
146 * --[no-]mac-updating: Enable or disable MAC addresses updating (enabled by default).
148 To run the application in linuxapp environment with 4 lcores, 16 ports and 8 RX queues per lcore and MAC address
149 updating enabled, issue the command:
151 .. code-block:: console
153 $ ./build/l2fwd -c f -n 4 -- -q 8 -p ffff
155 Refer to the *DPDK Getting Started Guide* for general information on running applications
156 and the Environment Abstraction Layer (EAL) options.
161 The following sections provide some explanation of the code.
163 .. _l2_fwd_app_cmd_arguments:
165 Command Line Arguments
166 ~~~~~~~~~~~~~~~~~~~~~~
168 The L2 Forwarding sample application takes specific parameters,
169 in addition to Environment Abstraction Layer (EAL) arguments.
170 The preferred way to parse parameters is to use the getopt() function,
171 since it is part of a well-defined and portable library.
173 The parsing of arguments is done in the l2fwd_parse_args() function.
174 The method of argument parsing is not described here.
175 Refer to the *glibc getopt(3)* man page for details.
177 EAL arguments are parsed first, then application-specific arguments.
178 This is done at the beginning of the main() function:
184 ret = rte_eal_init(argc, argv);
186 rte_exit(EXIT_FAILURE, "Invalid EAL arguments\n");
191 /* parse application arguments (after the EAL ones) */
193 ret = l2fwd_parse_args(argc, argv);
195 rte_exit(EXIT_FAILURE, "Invalid L2FWD arguments\n");
197 .. _l2_fwd_app_mbuf_init:
199 Mbuf Pool Initialization
200 ~~~~~~~~~~~~~~~~~~~~~~~~
202 Once the arguments are parsed, the mbuf pool is created.
203 The mbuf pool contains a set of mbuf objects that will be used by the driver
204 and the application to store network packet data:
208 /* create the mbuf pool */
210 l2fwd_pktmbuf_pool = rte_mempool_create("mbuf_pool", NB_MBUF, MBUF_SIZE, 32, sizeof(struct rte_pktmbuf_pool_private),
211 rte_pktmbuf_pool_init, NULL, rte_pktmbuf_init, NULL, SOCKET0, 0);
213 if (l2fwd_pktmbuf_pool == NULL)
214 rte_panic("Cannot init mbuf pool\n");
216 The rte_mempool is a generic structure used to handle pools of objects.
217 In this case, it is necessary to create a pool that will be used by the driver,
218 which expects to have some reserved space in the mempool structure,
219 sizeof(struct rte_pktmbuf_pool_private) bytes.
220 The number of allocated pkt mbufs is NB_MBUF, with a size of MBUF_SIZE each.
221 A per-lcore cache of 32 mbufs is kept.
222 The memory is allocated in NUMA socket 0,
223 but it is possible to extend this code to allocate one mbuf pool per socket.
225 Two callback pointers are also given to the rte_mempool_create() function:
227 * The first callback pointer is to rte_pktmbuf_pool_init() and is used
228 to initialize the private data of the mempool, which is needed by the driver.
229 This function is provided by the mbuf API, but can be copied and extended by the developer.
231 * The second callback pointer given to rte_mempool_create() is the mbuf initializer.
232 The default is used, that is, rte_pktmbuf_init(), which is provided in the rte_mbuf library.
233 If a more complex application wants to extend the rte_pktmbuf structure for its own needs,
234 a new function derived from rte_pktmbuf_init( ) can be created.
236 .. _l2_fwd_app_dvr_init:
238 Driver Initialization
239 ~~~~~~~~~~~~~~~~~~~~~
241 The main part of the code in the main() function relates to the initialization of the driver.
242 To fully understand this code, it is recommended to study the chapters that related to the Poll Mode Driver
243 in the *DPDK Programmer's Guide* - Rel 1.4 EAR and the *DPDK API Reference*.
247 if (rte_eal_pci_probe() < 0)
248 rte_exit(EXIT_FAILURE, "Cannot probe PCI\n");
250 nb_ports = rte_eth_dev_count();
253 rte_exit(EXIT_FAILURE, "No Ethernet ports - bye\n");
255 /* reset l2fwd_dst_ports */
257 for (portid = 0; portid < RTE_MAX_ETHPORTS; portid++)
258 l2fwd_dst_ports[portid] = 0;
263 * Each logical core is assigned a dedicated TX queue on each port.
266 for (portid = 0; portid < nb_ports; portid++) {
267 /* skip ports that are not enabled */
269 if ((l2fwd_enabled_port_mask & (1 << portid)) == 0)
272 if (nb_ports_in_mask % 2) {
273 l2fwd_dst_ports[portid] = last_port;
274 l2fwd_dst_ports[last_port] = portid;
281 rte_eth_dev_info_get((uint8_t) portid, &dev_info);
286 * rte_igb_pmd_init_all() simultaneously registers the driver as a PCI driver and as an Ethernet* Poll Mode Driver.
288 * rte_eal_pci_probe() parses the devices on the PCI bus and initializes recognized devices.
290 The next step is to configure the RX and TX queues.
291 For each port, there is only one RX queue (only one lcore is able to poll a given port).
292 The number of TX queues depends on the number of available lcores.
293 The rte_eth_dev_configure() function is used to configure the number of queues for a port:
297 ret = rte_eth_dev_configure((uint8_t)portid, 1, 1, &port_conf);
299 rte_exit(EXIT_FAILURE, "Cannot configure device: "
303 The global configuration is stored in a static structure:
307 static const struct rte_eth_conf port_conf = {
310 .header_split = 0, /**< Header Split disabled */
311 .hw_ip_checksum = 0, /**< IP checksum offload disabled */
312 .hw_vlan_filter = 0, /**< VLAN filtering disabled */
313 .jumbo_frame = 0, /**< Jumbo Frame Support disabled */
314 .hw_strip_crc= 0, /**< CRC stripped by hardware */
318 .mq_mode = ETH_DCB_NONE
322 .. _l2_fwd_app_rx_init:
324 RX Queue Initialization
325 ~~~~~~~~~~~~~~~~~~~~~~~
327 The application uses one lcore to poll one or several ports, depending on the -q option,
328 which specifies the number of queues per lcore.
330 For example, if the user specifies -q 4, the application is able to poll four ports with one lcore.
331 If there are 16 ports on the target (and if the portmask argument is -p ffff ),
332 the application will need four lcores to poll all the ports.
336 ret = rte_eth_rx_queue_setup((uint8_t) portid, 0, nb_rxd, SOCKET0, &rx_conf, l2fwd_pktmbuf_pool);
339 rte_exit(EXIT_FAILURE, "rte_eth_rx_queue_setup: "
343 The list of queues that must be polled for a given lcore is stored in a private structure called struct lcore_queue_conf.
347 struct lcore_queue_conf {
349 unsigned rx_port_list[MAX_RX_QUEUE_PER_LCORE];
350 struct mbuf_table tx_mbufs[L2FWD_MAX_PORTS];
353 struct lcore_queue_conf lcore_queue_conf[RTE_MAX_LCORE];
355 The values n_rx_port and rx_port_list[] are used in the main packet processing loop
356 (see :ref:`l2_fwd_app_rx_tx_packets`).
358 The global configuration for the RX queues is stored in a static structure:
362 static const struct rte_eth_rxconf rx_conf = {
364 .pthresh = RX_PTHRESH,
365 .hthresh = RX_HTHRESH,
366 .wthresh = RX_WTHRESH,
370 .. _l2_fwd_app_tx_init:
372 TX Queue Initialization
373 ~~~~~~~~~~~~~~~~~~~~~~~
375 Each lcore should be able to transmit on any port. For every port, a single TX queue is initialized.
379 /* init one TX queue on each port */
383 ret = rte_eth_tx_queue_setup((uint8_t) portid, 0, nb_txd, rte_eth_dev_socket_id(portid), &tx_conf);
385 rte_exit(EXIT_FAILURE, "rte_eth_tx_queue_setup:err=%d, port=%u\n", ret, (unsigned) portid);
387 The global configuration for TX queues is stored in a static structure:
391 static const struct rte_eth_txconf tx_conf = {
393 .pthresh = TX_PTHRESH,
394 .hthresh = TX_HTHRESH,
395 .wthresh = TX_WTHRESH,
397 .tx_free_thresh = RTE_TEST_TX_DESC_DEFAULT + 1, /* disable feature */
400 .. _l2_fwd_app_rx_tx_packets:
402 Receive, Process and Transmit Packets
403 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
405 In the l2fwd_main_loop() function, the main task is to read ingress packets from the RX queues.
406 This is done using the following code:
411 * Read packet from RX queues
414 for (i = 0; i < qconf->n_rx_port; i++) {
415 portid = qconf->rx_port_list[i];
416 nb_rx = rte_eth_rx_burst((uint8_t) portid, 0, pkts_burst, MAX_PKT_BURST);
418 for (j = 0; j < nb_rx; j++) {
420 rte_prefetch0[rte_pktmbuf_mtod(m, void *)); l2fwd_simple_forward(m, portid);
424 Packets are read in a burst of size MAX_PKT_BURST.
425 The rte_eth_rx_burst() function writes the mbuf pointers in a local table and returns the number of available mbufs in the table.
427 Then, each mbuf in the table is processed by the l2fwd_simple_forward() function.
428 The processing is very simple: process the TX port from the RX port, then replace the source and destination MAC addresses if MAC
429 addresses updating is enabled.
433 In the following code, one line for getting the output port requires some explanation.
435 During the initialization process, a static array of destination ports (l2fwd_dst_ports[]) is filled such that for each source port,
436 a destination port is assigned that is either the next or previous enabled port from the portmask.
437 Naturally, the number of ports in the portmask must be even, otherwise, the application exits.
442 l2fwd_simple_forward(struct rte_mbuf *m, unsigned portid)
444 struct ether_hdr *eth;
448 dst_port = l2fwd_dst_ports[portid];
450 eth = rte_pktmbuf_mtod(m, struct ether_hdr *);
452 /* 02:00:00:00:00:xx */
454 tmp = ð->d_addr.addr_bytes[0];
456 *((uint64_t *)tmp) = 0x000000000002 + ((uint64_t) dst_port << 40);
460 ether_addr_copy(&l2fwd_ports_eth_addr[dst_port], ð->s_addr);
462 l2fwd_send_packet(m, (uint8_t) dst_port);
465 Then, the packet is sent using the l2fwd_send_packet (m, dst_port) function.
466 For this test application, the processing is exactly the same for all packets arriving on the same RX port.
467 Therefore, it would have been possible to call the l2fwd_send_burst() function directly from the main loop
468 to send all the received packets on the same TX port,
469 using the burst-oriented send function, which is more efficient.
471 However, in real-life applications (such as, L3 routing),
472 packet N is not necessarily forwarded on the same port as packet N-1.
473 The application is implemented to illustrate that, so the same approach can be reused in a more complex application.
475 The l2fwd_send_packet() function stores the packet in a per-lcore and per-txport table.
476 If the table is full, the whole packets table is transmitted using the l2fwd_send_burst() function:
480 /* Send the packet on an output interface */
483 l2fwd_send_packet(struct rte_mbuf *m, uint8_t port)
485 unsigned lcore_id, len;
486 struct lcore_queue_conf *qconf;
488 lcore_id = rte_lcore_id();
489 qconf = &lcore_queue_conf[lcore_id];
490 len = qconf->tx_mbufs[port].len;
491 qconf->tx_mbufs[port].m_table[len] = m;
494 /* enough pkts to be sent */
496 if (unlikely(len == MAX_PKT_BURST)) {
497 l2fwd_send_burst(qconf, MAX_PKT_BURST, port);
501 qconf->tx_mbufs[port].len = len; return 0;
504 To ensure that no packets remain in the tables, each lcore does a draining of TX queue in its main loop.
505 This technique introduces some latency when there are not many packets to send,
506 however it improves performance:
510 cur_tsc = rte_rdtsc();
513 * TX burst queue drain
516 diff_tsc = cur_tsc - prev_tsc;
518 if (unlikely(diff_tsc > drain_tsc)) {
519 for (portid = 0; portid < RTE_MAX_ETHPORTS; portid++) {
520 if (qconf->tx_mbufs[portid].len == 0)
523 l2fwd_send_burst(&lcore_queue_conf[lcore_id], qconf->tx_mbufs[portid].len, (uint8_t) portid);
525 qconf->tx_mbufs[portid].len = 0;
528 /* if timer is enabled */
530 if (timer_period > 0) {
531 /* advance the timer */
533 timer_tsc += diff_tsc;
535 /* if timer has reached its timeout */
537 if (unlikely(timer_tsc >= (uint64_t) timer_period)) {
538 /* do this only on master core */
540 if (lcore_id == rte_get_master_lcore()) {
543 /* reset the timer */