1 .. SPDX-License-Identifier: BSD-3-Clause
2 Copyright(c) 2010-2014 Intel Corporation.
6 L2 Forwarding Eventdev Sample Application
7 =========================================
9 The L2 Forwarding eventdev sample application is a simple example of packet
10 processing using the Data Plane Development Kit (DPDK) to demonstrate usage of
11 poll and event mode packet I/O mechanism.
16 The L2 Forwarding eventdev sample application, performs L2 forwarding for each
17 packet that is received on an RX_PORT. The destination port is the adjacent port
18 from the enabled portmask, that is, if the first four ports are enabled (portmask=0x0f),
19 ports 1 and 2 forward into each other, and ports 3 and 4 forward into each other.
20 Also, if MAC addresses updating is enabled, the MAC addresses are affected as follows:
22 * The source MAC address is replaced by the TX_PORT MAC address
24 * The destination MAC address is replaced by 02:00:00:00:00:TX_PORT_ID
26 Application receives packets from RX_PORT using below mentioned methods:
30 * Eventdev mode (default)
32 This application can be used to benchmark performance using a traffic-generator,
33 as shown in the :numref:`figure_l2fwd_event_benchmark_setup`.
35 .. _figure_l2fwd_event_benchmark_setup:
37 .. figure:: img/l2_fwd_benchmark_setup.*
39 Performance Benchmark Setup (Basic Environment)
41 Compiling the Application
42 -------------------------
44 To compile the sample application see :doc:`compiling`.
46 The application is located in the ``l2fwd-event`` sub-directory.
48 Running the Application
49 -----------------------
51 The application requires a number of command line options:
53 .. code-block:: console
55 ./<build_dir>/examples/dpdk-l2fwd-event [EAL options] -- -p PORTMASK
59 [--eventq-sched=SCHED_MODE]
60 [--event-vector [--event-vector-size SIZE] [--event-vector-tmo NS]]
64 * p PORTMASK: A hexadecimal bitmask of the ports to configure
66 * q NQ: A number of queues (=ports) per lcore (default is 1)
68 * --[no-]mac-updating: Enable or disable MAC addresses updating (enabled by default).
70 * --mode=MODE: Packet transfer mode for I/O, poll or eventdev. Eventdev by default.
72 * --eventq-sched=SCHED_MODE: Event queue schedule mode, Ordered, Atomic or Parallel. Atomic by default.
74 * --config: Configure forwarding port pair mapping. Alternate port pairs by default.
76 * --event-vector: Enable event vectorization. Only valid if --mode=eventdev.
78 * --event-vector-size: Max vector size if event vectorization is enabled.
80 * --event-vector-tmo: Max timeout to form vector in nanoseconds if event vectorization is enabled.
82 Sample usage commands are given below to run the application into different mode:
84 Poll mode with 4 lcores, 16 ports and 8 RX queues per lcore and MAC address updating enabled,
87 .. code-block:: console
89 ./<build_dir>/examples/dpdk-l2fwd-event -l 0-3 -n 4 -- -q 8 -p ffff --mode=poll
91 Eventdev mode with 4 lcores, 16 ports , sched method ordered and MAC address updating enabled,
94 .. code-block:: console
96 ./<build_dir>/examples/dpdk-l2fwd-event -l 0-3 -n 4 -- -p ffff --eventq-sched=ordered
100 .. code-block:: console
102 ./<build_dir>/examples/dpdk-l2fwd-event -l 0-3 -n 4 -- -q 8 -p ffff --mode=eventdev --eventq-sched=ordered
104 Refer to the *DPDK Getting Started Guide* for general information on running
105 applications and the Environment Abstraction Layer (EAL) options.
107 To run application with S/W scheduler, it uses following DPDK services:
110 * Rx adapter service function
111 * Tx adapter service function
113 Application needs service cores to run above mentioned services. Service cores
114 must be provided as EAL parameters along with the --vdev=event_sw0 to enable S/W
115 scheduler. Following is the sample command:
117 .. code-block:: console
119 ./<build_dir>/examples/dpdk-l2fwd-event -l 0-7 -s 0-3 -n 4 --vdev event_sw0 -- -q 8 -p ffff --mode=eventdev --eventq-sched=ordered
124 The following sections provide some explanation of the code.
126 .. _l2_fwd_event_app_cmd_arguments:
128 Command Line Arguments
129 ~~~~~~~~~~~~~~~~~~~~~~
131 The L2 Forwarding eventdev sample application takes specific parameters,
132 in addition to Environment Abstraction Layer (EAL) arguments.
133 The preferred way to parse parameters is to use the getopt() function,
134 since it is part of a well-defined and portable library.
136 The parsing of arguments is done in the **l2fwd_parse_args()** function for non
137 eventdev parameters and in **parse_eventdev_args()** for eventdev parameters.
138 The method of argument parsing is not described here. Refer to the
139 *glibc getopt(3)* man page for details.
141 EAL arguments are parsed first, then application-specific arguments.
142 This is done at the beginning of the main() function and eventdev parameters
143 are parsed in eventdev_resource_setup() function during eventdev setup:
145 .. literalinclude:: ../../../examples/l2fwd-event/main.c
147 :start-after: Init EAL. 8<
148 :end-before: >8 End of init EAL.
151 Mbuf Pool Initialization
152 ~~~~~~~~~~~~~~~~~~~~~~~~
154 Once the arguments are parsed, the mbuf pool is created.
155 The mbuf pool contains a set of mbuf objects that will be used by the driver
156 and the application to store network packet data:
158 .. literalinclude:: ../../../examples/l2fwd-event/main.c
160 :start-after: Create the mbuf pool. 8<
161 :end-before: >8 End of creation of mbuf pool.
164 The rte_mempool is a generic structure used to handle pools of objects.
165 In this case, it is necessary to create a pool that will be used by the driver.
166 The number of allocated pkt mbufs is NB_MBUF, with a data room size of
167 RTE_MBUF_DEFAULT_BUF_SIZE each.
168 A per-lcore cache of 32 mbufs is kept.
169 The memory is allocated in NUMA socket 0,
170 but it is possible to extend this code to allocate one mbuf pool per socket.
172 The rte_pktmbuf_pool_create() function uses the default mbuf pool and mbuf
173 initializers, respectively rte_pktmbuf_pool_init() and rte_pktmbuf_init().
174 An advanced application may want to use the mempool API to create the
175 mbuf pool with more control.
177 .. _l2_fwd_event_app_drv_init:
179 Driver Initialization
180 ~~~~~~~~~~~~~~~~~~~~~
182 The main part of the code in the main() function relates to the initialization
183 of the driver. To fully understand this code, it is recommended to study the
184 chapters that related to the Poll Mode and Event mode Driver in the
185 *DPDK Programmer's Guide* - Rel 1.4 EAR and the *DPDK API Reference*.
187 .. literalinclude:: ../../../examples/l2fwd-event/main.c
189 :start-after: Reset l2fwd_dst_ports. 8<
190 :end-before: >8 End of reset l2fwd_dst_ports.
193 The next step is to configure the RX and TX queues. For each port, there is only
194 one RX queue (only one lcore is able to poll a given port). The number of TX
195 queues depends on the number of available lcores. The rte_eth_dev_configure()
196 function is used to configure the number of queues for a port:
198 .. literalinclude:: ../../../examples/l2fwd-event/l2fwd_common.c
200 :start-after: Configure RX and TX queue. 8<
201 :end-before: >8 End of configuration RX and TX queue.
204 RX Queue Initialization
205 ~~~~~~~~~~~~~~~~~~~~~~~
207 The application uses one lcore to poll one or several ports, depending on the -q
208 option, which specifies the number of queues per lcore.
210 For example, if the user specifies -q 4, the application is able to poll four
211 ports with one lcore. If there are 16 ports on the target (and if the portmask
212 argument is -p ffff ), the application will need four lcores to poll all the
215 .. literalinclude:: ../../../examples/l2fwd-event/l2fwd_common.c
217 :start-after: Using lcore to poll one or several ports. 8<
218 :end-before: >8 End of using lcore to poll one or several ports.
221 The list of queues that must be polled for a given lcore is stored in a private
222 structure called struct lcore_queue_conf.
224 .. literalinclude:: ../../../examples/l2fwd/main.c
226 :start-after: List of queues to be polled for a given lcore. 8<
227 :end-before: >8 End of list of queues to be polled for a given lcore.
229 The values n_rx_port and rx_port_list[] are used in the main packet processing
230 loop (see :ref:`l2_fwd_event_app_rx_tx_packets`).
232 .. _l2_fwd_event_app_tx_init:
234 TX Queue Initialization
235 ~~~~~~~~~~~~~~~~~~~~~~~
237 Each lcore should be able to transmit on any port. For every port, a single TX
238 queue is initialized.
240 .. literalinclude:: ../../../examples/l2fwd-event/l2fwd_common.c
242 :start-after: Init one TX queue on each port. 8<
243 :end-before: >8 End of init one TX queue on each port.
246 To configure eventdev support, application setups following components:
254 .. _l2_fwd_event_app_event_dev_init:
256 Event device Initialization
257 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
258 Application can use either H/W or S/W based event device scheduler
259 implementation and supports single instance of event device. It configures event
260 device as per below configuration
262 .. literalinclude:: ../../../examples/l2fwd-event/l2fwd_event_generic.c
264 :start-after: Configures event device as per below configuration. 8<
265 :end-before: >8 End of configuration event device as per below configuration.
268 In case of S/W scheduler, application runs eventdev scheduler service on service
269 core. Application retrieves service id and finds the best possible service core to
272 .. literalinclude:: ../../../examples/l2fwd-event/l2fwd_event.c
274 :start-after: Running eventdev scheduler service on service core. 8<
275 :end-before: >8 End of running eventdev scheduler service on service core.
278 Event queue Initialization
279 ~~~~~~~~~~~~~~~~~~~~~~~~~~
280 Each Ethernet device is assigned a dedicated event queue which will be linked
281 to all available event ports i.e. each lcore can dequeue packets from any of the
284 .. literalinclude:: ../../../examples/l2fwd-event/l2fwd_event_generic.c
286 :start-after: Event queue initialization. 8<
287 :end-before: >8 End of event queue initialization.
290 In case of S/W scheduler, an extra event queue is created which will be used for
291 Tx adapter service function for enqueue operation.
293 .. _l2_fwd_app_event_port_init:
295 Event port Initialization
296 ~~~~~~~~~~~~~~~~~~~~~~~~~
297 Each worker thread is assigned a dedicated event port for enq/deq operations
298 to/from an event device. All event ports are linked with all available event
301 .. literalinclude:: ../../../examples/l2fwd-event/l2fwd_event_generic.c
303 :start-after: Event port initialization. 8<
304 :end-before: >8 End of event port initialization.
307 In case of S/W scheduler, an extra event port is created by DPDK library which
308 is retrieved by the application and same will be used by Tx adapter service.
310 .. literalinclude:: ../../../examples/l2fwd-event/l2fwd_event_generic.c
312 :start-after: Extra port created. 8<
313 :end-before: >8 End of extra port created.
316 Rx/Tx adapter Initialization
317 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
318 Each Ethernet port is assigned a dedicated Rx/Tx adapter for H/W scheduler. Each
319 Ethernet port's Rx queues are connected to its respective event queue at
320 priority 0 via Rx adapter configuration and Ethernet port's tx queues are
321 connected via Tx adapter.
323 .. literalinclude:: ../../../examples/l2fwd-event/l2fwd_event_internal_port.c
325 :start-after: Assigned ethernet port. 8<
326 :end-before: >8 End of assigned ethernet port.
329 For S/W scheduler instead of dedicated adapters, common Rx/Tx adapters are
330 configured which will be shared among all the Ethernet ports. Also DPDK library
331 need service cores to run internal services for Rx/Tx adapters. Application gets
332 service id for Rx/Tx adapters and after successful setup it runs the services
333 on dedicated service cores.
335 .. literalinclude:: ../../../examples/l2fwd-event/l2fwd_event.c
337 :start-after: Gets service ID for RX/TX adapters. 8<
338 :end-before: >8 End of get service ID for RX/TX adapters.
341 .. _l2_fwd_event_app_rx_tx_packets:
343 Receive, Process and Transmit Packets
344 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
346 In the **l2fwd_main_loop()** function, the main task is to read ingress packets from
347 the RX queues. This is done using the following code:
349 .. literalinclude:: ../../../examples/l2fwd-event/l2fwd_poll.c
351 :start-after: Reading ingress packets. 8<
352 :end-before: >8 End of reading ingress packets.
355 Packets are read in a burst of size MAX_PKT_BURST. The rte_eth_rx_burst()
356 function writes the mbuf pointers in a local table and returns the number of
357 available mbufs in the table.
359 Then, each mbuf in the table is processed by the l2fwd_simple_forward()
360 function. The processing is very simple: process the TX port from the RX port,
361 then replace the source and destination MAC addresses if MAC addresses updating
364 During the initialization process, a static array of destination ports
365 (l2fwd_dst_ports[]) is filled such that for each source port, a destination port
366 is assigned that is either the next or previous enabled port from the portmask.
367 If number of ports are odd in portmask then packet from last port will be
368 forwarded to first port i.e. if portmask=0x07, then forwarding will take place
369 like p0--->p1, p1--->p2, p2--->p0.
371 Also to optimize enqueue operation, l2fwd_simple_forward() stores incoming mbufs
372 up to MAX_PKT_BURST. Once it reaches up to limit, all packets are transmitted to
375 .. literalinclude:: ../../../examples/l2fwd/main.c
377 :start-after: Simple forward. 8<
378 :end-before: >8 End of simple forward.
380 For this test application, the processing is exactly the same for all packets
381 arriving on the same RX port. Therefore, it would have been possible to call
382 the rte_eth_tx_buffer() function directly from the main loop to send all the
383 received packets on the same TX port, using the burst-oriented send function,
384 which is more efficient.
386 However, in real-life applications (such as, L3 routing),
387 packet N is not necessarily forwarded on the same port as packet N-1.
388 The application is implemented to illustrate that, so the same approach can be
389 reused in a more complex application.
391 To ensure that no packets remain in the tables, each lcore does a draining of TX
392 queue in its main loop. This technique introduces some latency when there are
393 not many packets to send, however it improves performance:
395 .. literalinclude:: ../../../examples/l2fwd-event/l2fwd_poll.c
397 :start-after: Draining TX queue in main loop. 8<
398 :end-before: >8 End of draining TX queue in main loop.
401 In the **l2fwd_event_loop()** function, the main task is to read ingress
402 packets from the event ports. This is done using the following code:
404 .. literalinclude:: ../../../examples/l2fwd-event/l2fwd_event.c
406 :start-after: Read packet from eventdev. 8<
407 :end-before: >8 End of reading packets from eventdev.
411 Before reading packets, deq_len is fetched to ensure correct allowed deq length
413 The rte_event_dequeue_burst() function writes the mbuf pointers in a local table
414 and returns the number of available mbufs in the table.
416 Then, each mbuf in the table is processed by the l2fwd_eventdev_forward()
417 function. The processing is very simple: process the TX port from the RX port,
418 then replace the source and destination MAC addresses if MAC addresses updating
421 During the initialization process, a static array of destination ports
422 (l2fwd_dst_ports[]) is filled such that for each source port, a destination port
423 is assigned that is either the next or previous enabled port from the portmask.
424 If number of ports are odd in portmask then packet from last port will be
425 forwarded to first port i.e. if portmask=0x07, then forwarding will take place
426 like p0--->p1, p1--->p2, p2--->p0.
428 l2fwd_eventdev_forward() does not stores incoming mbufs. Packet will forwarded
429 be to destination ports via Tx adapter or generic event dev enqueue API
430 depending H/W or S/W scheduler is used.
432 .. literalinclude:: ../../../examples/l2fwd-event/l2fwd_event.c
434 :start-after: Read packet from eventdev. 8<
435 :end-before: >8 End of reading packets from eventdev.