2 Copyright(c) 2010-2015 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 Forwarding Sample Application (in Real and Virtualized Environments) with core load statistics.
32 ==================================================================================================
34 The L2 Forwarding sample application is a simple example of packet processing using
35 the Data Plane Development Kit (DPDK) which
36 also takes advantage of Single Root I/O Virtualization (SR-IOV) features in a virtualized environment.
40 This application is a variation of L2 Forwarding sample application. It demonstrate possible
41 scheme of job stats library usage therefore some parts of this document is identical with original
42 L2 forwarding application.
47 The L2 Forwarding sample application, which can operate in real and virtualized environments,
48 performs L2 forwarding for each packet that is received.
49 The destination port is the adjacent port from the enabled portmask, that is,
50 if the first four ports are enabled (portmask 0xf),
51 ports 1 and 2 forward into each other, and ports 3 and 4 forward into each other.
52 Also, the MAC addresses are affected as follows:
54 * The source MAC address is replaced by the TX port MAC address
56 * The destination MAC address is replaced by 02:00:00:00:00:TX_PORT_ID
58 This application can be used to benchmark performance using a traffic-generator, as shown in the :numref:`figure_l2_fwd_benchmark_setup_jobstats`.
60 The application can also be used in a virtualized environment as shown in :numref:`figure_l2_fwd_virtenv_benchmark_setup_jobstats`.
62 The L2 Forwarding application can also be used as a starting point for developing a new application based on the DPDK.
64 .. _figure_l2_fwd_benchmark_setup_jobstats:
66 .. figure:: img/l2_fwd_benchmark_setup.*
68 Performance Benchmark Setup (Basic Environment)
70 .. _figure_l2_fwd_virtenv_benchmark_setup_jobstats:
72 .. figure:: img/l2_fwd_virtenv_benchmark_setup.*
74 Performance Benchmark Setup (Virtualized Environment)
77 Virtual Function Setup Instructions
78 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
80 This application can use the virtual function available in the system and
81 therefore can be used in a virtual machine without passing through
82 the whole Network Device into a guest machine in a virtualized scenario.
83 The virtual functions can be enabled in the host machine or the hypervisor with the respective physical function driver.
85 For example, in a Linux* host machine, it is possible to enable a virtual function using the following command:
87 .. code-block:: console
89 modprobe ixgbe max_vfs=2,2
91 This command enables two Virtual Functions on each of Physical Function of the NIC,
92 with two physical ports in the PCI configuration space.
93 It is important to note that enabled Virtual Function 0 and 2 would belong to Physical Function 0
94 and Virtual Function 1 and 3 would belong to Physical Function 1,
95 in this case enabling a total of four Virtual Functions.
97 Compiling the Application
98 -------------------------
100 #. Go to the example directory:
102 .. code-block:: console
104 export RTE_SDK=/path/to/rte_sdk cd ${RTE_SDK}/examples/l2fwd-jobstats
106 #. Set the target (a default target is used if not specified). For example:
108 .. code-block:: console
110 export RTE_TARGET=x86_64-native-linuxapp-gcc
112 *See the DPDK Getting Started Guide* for possible RTE_TARGET values.
114 #. Build the application:
116 .. code-block:: console
120 Running the Application
121 -----------------------
123 The application requires a number of command line options:
125 .. code-block:: console
127 ./build/l2fwd-jobstats [EAL options] -- -p PORTMASK [-q NQ] [-l]
131 * p PORTMASK: A hexadecimal bitmask of the ports to configure
133 * q NQ: A number of queues (=ports) per lcore (default is 1)
135 * l: Use locale thousands separator when formatting big numbers.
137 To run the application in linuxapp environment with 4 lcores, 16 ports, 8 RX queues per lcore and
138 thousands separator printing, issue the command:
140 .. code-block:: console
142 $ ./build/l2fwd-jobstats -c f -n 4 -- -q 8 -p ffff -l
144 Refer to the *DPDK Getting Started Guide* for general information on running applications
145 and the Environment Abstraction Layer (EAL) options.
150 The following sections provide some explanation of the code.
152 Command Line Arguments
153 ~~~~~~~~~~~~~~~~~~~~~~
155 The L2 Forwarding sample application takes specific parameters,
156 in addition to Environment Abstraction Layer (EAL) arguments (see Section 9.3).
157 The preferred way to parse parameters is to use the getopt() function,
158 since it is part of a well-defined and portable library.
160 The parsing of arguments is done in the l2fwd_parse_args() function.
161 The method of argument parsing is not described here.
162 Refer to the *glibc getopt(3)* man page for details.
164 EAL arguments are parsed first, then application-specific arguments.
165 This is done at the beginning of the main() function:
171 ret = rte_eal_init(argc, argv);
173 rte_exit(EXIT_FAILURE, "Invalid EAL arguments\n");
178 /* parse application arguments (after the EAL ones) */
180 ret = l2fwd_parse_args(argc, argv);
182 rte_exit(EXIT_FAILURE, "Invalid L2FWD arguments\n");
184 Mbuf Pool Initialization
185 ~~~~~~~~~~~~~~~~~~~~~~~~
187 Once the arguments are parsed, the mbuf pool is created.
188 The mbuf pool contains a set of mbuf objects that will be used by the driver
189 and the application to store network packet data:
193 /* create the mbuf pool */
195 rte_mempool_create("mbuf_pool", NB_MBUF,
197 sizeof(struct rte_pktmbuf_pool_private),
198 rte_pktmbuf_pool_init, NULL,
199 rte_pktmbuf_init, NULL,
202 if (l2fwd_pktmbuf_pool == NULL)
203 rte_exit(EXIT_FAILURE, "Cannot init mbuf pool\n");
205 The rte_mempool is a generic structure used to handle pools of objects.
206 In this case, it is necessary to create a pool that will be used by the driver,
207 which expects to have some reserved space in the mempool structure,
208 sizeof(struct rte_pktmbuf_pool_private) bytes.
209 The number of allocated pkt mbufs is NB_MBUF, with a size of MBUF_SIZE each.
210 A per-lcore cache of 32 mbufs is kept.
211 The memory is allocated in rte_socket_id() socket,
212 but it is possible to extend this code to allocate one mbuf pool per socket.
214 Two callback pointers are also given to the rte_mempool_create() function:
216 * The first callback pointer is to rte_pktmbuf_pool_init() and is used
217 to initialize the private data of the mempool, which is needed by the driver.
218 This function is provided by the mbuf API, but can be copied and extended by the developer.
220 * The second callback pointer given to rte_mempool_create() is the mbuf initializer.
221 The default is used, that is, rte_pktmbuf_init(), which is provided in the rte_mbuf library.
222 If a more complex application wants to extend the rte_pktmbuf structure for its own needs,
223 a new function derived from rte_pktmbuf_init( ) can be created.
225 Driver Initialization
226 ~~~~~~~~~~~~~~~~~~~~~
228 The main part of the code in the main() function relates to the initialization of the driver.
229 To fully understand this code, it is recommended to study the chapters that related to the Poll Mode Driver
230 in the *DPDK Programmer's Guide* and the *DPDK API Reference*.
234 nb_ports = rte_eth_dev_count();
237 rte_exit(EXIT_FAILURE, "No Ethernet ports - bye\n");
239 if (nb_ports > RTE_MAX_ETHPORTS)
240 nb_ports = RTE_MAX_ETHPORTS;
242 /* reset l2fwd_dst_ports */
244 for (portid = 0; portid < RTE_MAX_ETHPORTS; portid++)
245 l2fwd_dst_ports[portid] = 0;
250 * Each logical core is assigned a dedicated TX queue on each port.
252 for (portid = 0; portid < nb_ports; portid++) {
253 /* skip ports that are not enabled */
254 if ((l2fwd_enabled_port_mask & (1 << portid)) == 0)
257 if (nb_ports_in_mask % 2) {
258 l2fwd_dst_ports[portid] = last_port;
259 l2fwd_dst_ports[last_port] = portid;
266 rte_eth_dev_info_get((uint8_t) portid, &dev_info);
269 The next step is to configure the RX and TX queues.
270 For each port, there is only one RX queue (only one lcore is able to poll a given port).
271 The number of TX queues depends on the number of available lcores.
272 The rte_eth_dev_configure() function is used to configure the number of queues for a port:
276 ret = rte_eth_dev_configure((uint8_t)portid, 1, 1, &port_conf);
278 rte_exit(EXIT_FAILURE, "Cannot configure device: "
282 The global configuration is stored in a static structure:
286 static const struct rte_eth_conf port_conf = {
289 .header_split = 0, /**< Header Split disabled */
290 .hw_ip_checksum = 0, /**< IP checksum offload disabled */
291 .hw_vlan_filter = 0, /**< VLAN filtering disabled */
292 .jumbo_frame = 0, /**< Jumbo Frame Support disabled */
293 .hw_strip_crc= 0, /**< CRC stripped by hardware */
297 .mq_mode = ETH_DCB_NONE
301 RX Queue Initialization
302 ~~~~~~~~~~~~~~~~~~~~~~~
304 The application uses one lcore to poll one or several ports, depending on the -q option,
305 which specifies the number of queues per lcore.
307 For example, if the user specifies -q 4, the application is able to poll four ports with one lcore.
308 If there are 16 ports on the target (and if the portmask argument is -p ffff ),
309 the application will need four lcores to poll all the ports.
313 ret = rte_eth_rx_queue_setup(portid, 0, nb_rxd,
314 rte_eth_dev_socket_id(portid),
319 rte_exit(EXIT_FAILURE, "rte_eth_rx_queue_setup:err=%d, port=%u\n",
320 ret, (unsigned) portid);
322 The list of queues that must be polled for a given lcore is stored in a private structure called struct lcore_queue_conf.
326 struct lcore_queue_conf {
328 unsigned rx_port_list[MAX_RX_QUEUE_PER_LCORE];
329 truct mbuf_table tx_mbufs[RTE_MAX_ETHPORTS];
331 struct rte_timer rx_timers[MAX_RX_QUEUE_PER_LCORE];
332 struct rte_jobstats port_fwd_jobs[MAX_RX_QUEUE_PER_LCORE];
334 struct rte_timer flush_timer;
335 struct rte_jobstats flush_job;
336 struct rte_jobstats idle_job;
337 struct rte_jobstats_context jobs_context;
339 rte_atomic16_t stats_read_pending;
341 } __rte_cache_aligned;
343 Values of struct lcore_queue_conf:
345 * n_rx_port and rx_port_list[] are used in the main packet processing loop
346 (see Section 9.4.6 "Receive, Process and Transmit Packets" later in this chapter).
348 * rx_timers and flush_timer are used to ensure forced TX on low packet rate.
350 * flush_job, idle_job and jobs_context are librte_jobstats objects used for managing l2fwd jobs.
352 * stats_read_pending and lock are used during job stats read phase.
354 TX Queue Initialization
355 ~~~~~~~~~~~~~~~~~~~~~~~
357 Each lcore should be able to transmit on any port. For every port, a single TX queue is initialized.
361 /* init one TX queue on each port */
364 ret = rte_eth_tx_queue_setup(portid, 0, nb_txd,
365 rte_eth_dev_socket_id(portid),
368 rte_exit(EXIT_FAILURE, "rte_eth_tx_queue_setup:err=%d, port=%u\n",
369 ret, (unsigned) portid);
371 Jobs statistics initialization
372 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
373 There are several statistics objects available:
375 * Flush job statistics
379 rte_jobstats_init(&qconf->flush_job, "flush", drain_tsc, drain_tsc,
382 rte_timer_init(&qconf->flush_timer);
383 ret = rte_timer_reset(&qconf->flush_timer, drain_tsc, PERIODICAL,
384 lcore_id, &l2fwd_flush_job, NULL);
387 rte_exit(1, "Failed to reset flush job timer for lcore %u: %s",
388 lcore_id, rte_strerror(-ret));
391 * Statistics per RX port
395 rte_jobstats_init(job, name, 0, drain_tsc, 0, MAX_PKT_BURST);
396 rte_jobstats_set_update_period_function(job, l2fwd_job_update_cb);
398 rte_timer_init(&qconf->rx_timers[i]);
399 ret = rte_timer_reset(&qconf->rx_timers[i], 0, PERIODICAL, lcore_id,
400 l2fwd_fwd_job, (void *)(uintptr_t)i);
403 rte_exit(1, "Failed to reset lcore %u port %u job timer: %s",
404 lcore_id, qconf->rx_port_list[i], rte_strerror(-ret));
407 Following parameters are passed to rte_jobstats_init():
409 * 0 as minimal poll period
411 * drain_tsc as maximum poll period
413 * MAX_PKT_BURST as desired target value (RX burst size)
418 The forwarding path is reworked comparing to original L2 Forwarding application.
419 In the l2fwd_main_loop() function three loops are placed.
424 rte_spinlock_lock(&qconf->lock);
427 rte_jobstats_context_start(&qconf->jobs_context);
430 * - Read stats_read_pending flag
431 * - check if some real job need to be executed
433 rte_jobstats_start(&qconf->jobs_context, &qconf->idle_job);
437 uint64_t now = rte_get_timer_cycles();
439 need_manage = qconf->flush_timer.expire < now;
440 /* Check if we was esked to give a stats. */
442 rte_atomic16_read(&qconf->stats_read_pending);
443 need_manage |= stats_read_pending;
445 for (i = 0; i < qconf->n_rx_port && !need_manage; i++)
446 need_manage = qconf->rx_timers[i].expire < now;
448 } while (!need_manage);
449 rte_jobstats_finish(&qconf->idle_job, qconf->idle_job.target);
452 rte_jobstats_context_finish(&qconf->jobs_context);
453 } while (likely(stats_read_pending == 0));
455 rte_spinlock_unlock(&qconf->lock);
459 First infinite for loop is to minimize impact of stats reading. Lock is only locked/unlocked when asked.
461 Second inner while loop do the whole jobs management. When any job is ready, the use rte_timer_manage() is used to call the job handler.
462 In this place functions l2fwd_fwd_job() and l2fwd_flush_job() are called when needed.
463 Then rte_jobstats_context_finish() is called to mark loop end - no other jobs are ready to execute. By this time stats are ready to be read
464 and if stats_read_pending is set, loop breaks allowing stats to be read.
466 Third do-while loop is the idle job (idle stats counter). Its only purpose is monitoring if any job is ready or stats job read is pending
467 for this lcore. Statistics from this part of code is considered as the headroom available for additional processing.
469 Receive, Process and Transmit Packets
470 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
472 The main task of l2fwd_fwd_job() function is to read ingress packets from the RX queue of particular port and forward it.
473 This is done using the following code:
477 total_nb_rx = rte_eth_rx_burst((uint8_t) portid, 0, pkts_burst,
480 for (j = 0; j < total_nb_rx; j++) {
482 rte_prefetch0(rte_pktmbuf_mtod(m, void *));
483 l2fwd_simple_forward(m, portid);
486 Packets are read in a burst of size MAX_PKT_BURST.
487 Then, each mbuf in the table is processed by the l2fwd_simple_forward() function.
488 The processing is very simple: process the TX port from the RX port, then replace the source and destination MAC addresses.
490 The rte_eth_rx_burst() function writes the mbuf pointers in a local table and returns the number of available mbufs in the table.
492 After first read second try is issued.
496 if (total_nb_rx == MAX_PKT_BURST) {
497 const uint16_t nb_rx = rte_eth_rx_burst((uint8_t) portid, 0, pkts_burst,
500 total_nb_rx += nb_rx;
501 for (j = 0; j < nb_rx; j++) {
503 rte_prefetch0(rte_pktmbuf_mtod(m, void *));
504 l2fwd_simple_forward(m, portid);
508 This second read is important to give job stats library a feedback how many packets was processed.
512 /* Adjust period time in which we are running here. */
513 if (rte_jobstats_finish(job, total_nb_rx) != 0) {
514 rte_timer_reset(&qconf->rx_timers[port_idx], job->period, PERIODICAL,
515 lcore_id, l2fwd_fwd_job, arg);
518 To maximize performance exactly MAX_PKT_BURST is expected (the target value) to be read for each l2fwd_fwd_job() call.
519 If total_nb_rx is smaller than target value job->period will be increased. If it is greater the period will be decreased.
523 In the following code, one line for getting the output port requires some explanation.
525 During the initialization process, a static array of destination ports (l2fwd_dst_ports[]) is filled such that for each source port,
526 a destination port is assigned that is either the next or previous enabled port from the portmask.
527 Naturally, the number of ports in the portmask must be even, otherwise, the application exits.
532 l2fwd_simple_forward(struct rte_mbuf *m, unsigned portid)
534 struct ether_hdr *eth;
538 dst_port = l2fwd_dst_ports[portid];
540 eth = rte_pktmbuf_mtod(m, struct ether_hdr *);
542 /* 02:00:00:00:00:xx */
544 tmp = ð->d_addr.addr_bytes[0];
546 *((uint64_t *)tmp) = 0x000000000002 + ((uint64_t) dst_port << 40);
550 ether_addr_copy(&l2fwd_ports_eth_addr[dst_port], ð->s_addr);
552 l2fwd_send_packet(m, (uint8_t) dst_port);
555 Then, the packet is sent using the l2fwd_send_packet (m, dst_port) function.
556 For this test application, the processing is exactly the same for all packets arriving on the same RX port.
557 Therefore, it would have been possible to call the l2fwd_send_burst() function directly from the main loop
558 to send all the received packets on the same TX port,
559 using the burst-oriented send function, which is more efficient.
561 However, in real-life applications (such as, L3 routing),
562 packet N is not necessarily forwarded on the same port as packet N-1.
563 The application is implemented to illustrate that, so the same approach can be reused in a more complex application.
565 The l2fwd_send_packet() function stores the packet in a per-lcore and per-txport table.
566 If the table is full, the whole packets table is transmitted using the l2fwd_send_burst() function:
570 /* Send the packet on an output interface */
573 l2fwd_send_packet(struct rte_mbuf *m, uint8_t port)
575 unsigned lcore_id, len;
576 struct lcore_queue_conf *qconf;
578 lcore_id = rte_lcore_id();
579 qconf = &lcore_queue_conf[lcore_id];
580 len = qconf->tx_mbufs[port].len;
581 qconf->tx_mbufs[port].m_table[len] = m;
584 /* enough pkts to be sent */
586 if (unlikely(len == MAX_PKT_BURST)) {
587 l2fwd_send_burst(qconf, MAX_PKT_BURST, port);
591 qconf->tx_mbufs[port].len = len; return 0;
594 To ensure that no packets remain in the tables, the flush job exists. The l2fwd_flush_job()
595 is called periodically to for each lcore draining TX queue of each port.
596 This technique introduces some latency when there are not many packets to send,
597 however it improves performance:
602 l2fwd_flush_job(__rte_unused struct rte_timer *timer, __rte_unused void *arg)
606 struct lcore_queue_conf *qconf;
607 struct mbuf_table *m_table;
610 lcore_id = rte_lcore_id();
611 qconf = &lcore_queue_conf[lcore_id];
613 rte_jobstats_start(&qconf->jobs_context, &qconf->flush_job);
615 now = rte_get_timer_cycles();
616 lcore_id = rte_lcore_id();
617 qconf = &lcore_queue_conf[lcore_id];
618 for (portid = 0; portid < RTE_MAX_ETHPORTS; portid++) {
619 m_table = &qconf->tx_mbufs[portid];
620 if (m_table->len == 0 || m_table->next_flush_time <= now)
623 l2fwd_send_burst(qconf, portid);
627 /* Pass target to indicate that this job is happy of time interval
628 * in which it was called. */
629 rte_jobstats_finish(&qconf->flush_job, qconf->flush_job.target);