1 .. SPDX-License-Identifier: BSD-3-Clause
2 Copyright(c) 2019-2021 Intel Corporation.
4 .. include:: <isonum.txt>
6 Packet copying using DMAdev library
7 ===================================
12 This sample is intended as a demonstration of the basic components of a DPDK
13 forwarding application and example of how to use the DMAdev API to make a packet
16 Also while forwarding, the MAC addresses are affected as follows:
18 * The source MAC address is replaced by the TX port MAC address
20 * The destination MAC address is replaced by 02:00:00:00:00:TX_PORT_ID
22 This application can be used to compare performance of using software packet
23 copy with copy done using a DMA device for different sizes of packets.
24 The example will print out statistics each second. The stats shows
25 received/send packets and packets dropped or failed to copy.
27 Compiling the Application
28 -------------------------
30 To compile the sample application see :doc:`compiling`.
32 The application is located in the ``dma`` sub-directory.
35 Running the Application
36 -----------------------
38 In order to run the hardware copy application, the copying device
39 needs to be bound to user-space IO driver.
41 Refer to the "DMAdev library" chapter in the "Programmers guide" for information
44 The application requires a number of command line options:
46 .. code-block:: console
48 ./<build_dir>/examples/dpdk-ioat [EAL options] -- [-p MASK] [-q NQ] [-s RS] [-c <sw|hw>]
49 [--[no-]mac-updating] [-b BS] [-f FS] [-i SI]
53 * p MASK: A hexadecimal bitmask of the ports to configure (default is all)
55 * q NQ: Number of Rx queues used per port equivalent to DMA channels
56 per port (default is 1)
58 * c CT: Performed packet copy type: software (sw) or hardware using
59 DMA (hw) (default is hw)
61 * s RS: Size of dmadev descriptor ring for hardware copy mode or rte_ring for
62 software copy mode (default is 2048)
64 * --[no-]mac-updating: Whether MAC address of packets should be changed
65 or not (default is mac-updating)
67 * b BS: set the DMA batch size
69 * f FS: set the max frame size
71 * i SI: set the interval, in second, between statistics prints (default is 1)
73 The application can be launched in various configurations depending on
74 provided parameters. The app can use up to 2 lcores: one of them receives
75 incoming traffic and makes a copy of each packet. The second lcore then
76 updates MAC address and sends the copy. If one lcore per port is used,
77 both operations are done sequentially. For each configuration an additional
78 lcore is needed since the main lcore does not handle traffic but is
79 responsible for configuration, statistics printing and safe shutdown of
80 all ports and devices.
82 The application can use a maximum of 8 ports.
84 To run the application in a Linux environment with 3 lcores (the main lcore,
85 plus two forwarding cores), a single port (port 0), software copying and MAC
86 updating issue the command:
88 .. code-block:: console
90 $ ./<build_dir>/examples/dpdk-dma -l 0-2 -n 2 -- -p 0x1 --mac-updating -c sw
92 To run the application in a Linux environment with 2 lcores (the main lcore,
93 plus one forwarding core), 2 ports (ports 0 and 1), hardware copying and no MAC
94 updating issue the command:
96 .. code-block:: console
98 $ ./<build_dir>/examples/dpdk-dma -l 0-1 -n 1 -- -p 0x3 --no-mac-updating -c hw
100 Refer to the *DPDK Getting Started Guide* for general information on
101 running applications and the Environment Abstraction Layer (EAL) options.
106 The following sections provide an explanation of the main components of the
109 All DPDK library functions used in the sample code are prefixed with
110 ``rte_`` and are explained in detail in the *DPDK API Documentation*.
116 The ``main()`` function performs the initialization and calls the execution
117 threads for each lcore.
119 The first task is to initialize the Environment Abstraction Layer (EAL).
120 The ``argc`` and ``argv`` arguments are provided to the ``rte_eal_init()``
121 function. The value returned is the number of parsed arguments:
123 .. literalinclude:: ../../../examples/dma/dmafwd.c
125 :start-after: Init EAL. 8<
126 :end-before: >8 End of init EAL.
130 The ``main()`` also allocates a mempool to hold the mbufs (Message Buffers)
131 used by the application:
133 .. literalinclude:: ../../../examples/dma/dmafwd.c
135 :start-after: Allocates mempool to hold the mbufs. 8<
136 :end-before: >8 End of allocates mempool to hold the mbufs.
139 Mbufs are the packet buffer structure used by DPDK. They are explained in
140 detail in the "Mbuf Library" section of the *DPDK Programmer's Guide*.
142 The ``main()`` function also initializes the ports:
144 .. literalinclude:: ../../../examples/dma/dmafwd.c
146 :start-after: Initialize each port. 8<
147 :end-before: >8 End of initializing each port.
150 Each port is configured using ``port_init()`` function. The Ethernet
151 ports are configured with local settings using the ``rte_eth_dev_configure()``
152 function and the ``port_conf`` struct. The RSS is enabled so that
153 multiple Rx queues could be used for packet receiving and copying by
154 multiple DMA channels per port:
156 .. literalinclude:: ../../../examples/dma/dmafwd.c
158 :start-after: Configuring port to use RSS for multiple RX queues. 8<
159 :end-before: >8 End of configuring port to use RSS for multiple RX queues.
162 For this example the ports are set up with the number of Rx queues provided
163 with -q option and 1 Tx queue using the ``rte_eth_rx_queue_setup()``
164 and ``rte_eth_tx_queue_setup()`` functions.
166 The Ethernet port is then started:
168 .. literalinclude:: ../../../examples/dma/dmafwd.c
170 :start-after: Start device. 8<
171 :end-before: >8 End of starting device.
175 Finally the Rx port is set in promiscuous mode:
177 .. literalinclude:: ../../../examples/dma/dmafwd.c
179 :start-after: RX port is set in promiscuous mode. 8<
180 :end-before: >8 End of RX port is set in promiscuous mode.
184 After that each port application assigns resources needed.
186 .. literalinclude:: ../../../examples/dma/dmafwd.c
188 :start-after: Assigning each port resources. 8<
189 :end-before: >8 End of assigning each port resources.
192 Ring structures are assigned for exchanging packets between lcores for both SW
195 .. literalinclude:: ../../../examples/dma/dmafwd.c
197 :start-after: Assign ring structures for packet exchanging. 8<
198 :end-before: >8 End of assigning ring structures for packet exchanging.
202 When using hardware copy each Rx queue of the port is assigned a DMA device
203 (``assign_dmadevs()``) using DMAdev library API functions:
205 .. literalinclude:: ../../../examples/dma/dmafwd.c
207 :start-after: Using dmadev API functions. 8<
208 :end-before: >8 End of using dmadev API functions.
212 The initialization of hardware device is done by ``rte_dma_configure()`` and
213 ``rte_dma_vchan_setup()`` functions using the ``rte_dma_conf`` and
214 ``rte_dma_vchan_conf`` structs. After configuration the device is started
215 using ``rte_dma_start()`` function. Each of the above operations is done in
216 ``configure_dmadev_queue()``.
218 .. literalinclude:: ../../../examples/dma/dmafwd.c
220 :start-after: Configuration of device. 8<
221 :end-before: >8 End of configuration of device.
224 If initialization is successful, memory for hardware device
225 statistics is allocated.
227 Finally ``main()`` function starts all packet handling lcores and starts
228 printing stats in a loop on the main lcore. The application can be
229 interrupted and closed using ``Ctrl-C``. The main lcore waits for
230 all worker lcores to finish, deallocates resources and exits.
232 The processing lcores launching function are described below.
234 The Lcores Launching Functions
235 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
237 As described above, ``main()`` function invokes ``start_forwarding_cores()``
238 function in order to start processing for each lcore:
240 .. literalinclude:: ../../../examples/dma/dmafwd.c
242 :start-after: Start processing for each lcore. 8<
243 :end-before: >8 End of starting to processfor each lcore.
246 The function launches Rx/Tx processing functions on configured lcores
247 using ``rte_eal_remote_launch()``. The configured ports, their number
248 and number of assigned lcores are stored in user-defined
249 ``rxtx_transmission_config`` struct:
251 .. literalinclude:: ../../../examples/dma/dmafwd.c
253 :start-after: Configuring ports and number of assigned lcores in struct. 8<
254 :end-before: >8 End of configuration of ports and number of assigned lcores.
257 The structure is initialized in 'main()' function with the values
258 corresponding to ports and lcores configuration provided by the user.
260 The Lcores Processing Functions
261 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
263 For receiving packets on each port, the ``dma_rx_port()`` function is used.
264 The function receives packets on each configured Rx queue. Depending on the
265 mode the user chose, it will enqueue packets to DMA channels and
266 then invoke copy process (hardware copy), or perform software copy of each
267 packet using ``pktmbuf_sw_copy()`` function and enqueue them to an rte_ring:
269 .. literalinclude:: ../../../examples/dma/dmafwd.c
271 :start-after: Receive packets on one port and enqueue to dmadev or rte_ring. 8<
272 :end-before: >8 End of receive packets on one port and enqueue to dmadev or rte_ring.
275 The packets are received in burst mode using ``rte_eth_rx_burst()``
276 function. When using hardware copy mode the packets are enqueued in
277 copying device's buffer using ``dma_enqueue_packets()`` which calls
278 ``rte_dma_copy()``. When all received packets are in the
279 buffer the copy operations are started by calling ``rte_dma_submit()``.
280 Function ``rte_dma_copy()`` operates on physical address of
281 the packet. Structure ``rte_mbuf`` contains only physical address to
282 start of the data buffer (``buf_iova``). Thus the ``rte_pktmbuf_iova()`` API is
283 used to get the address of the start of the data within the mbuf.
285 .. literalinclude:: ../../../examples/dma/dmafwd.c
287 :start-after: Receive packets on one port and enqueue to dmadev or rte_ring. 8<
288 :end-before: >8 End of receive packets on one port and enqueue to dmadev or rte_ring.
292 Once the copies have been completed (this includes gathering the completions in
293 HW copy mode), the copied packets are enqueued to the ``rx_to_tx_ring``, which
294 is used to pass the packets to the TX function.
296 All completed copies are processed by ``dma_tx_port()`` function. This function
297 dequeues copied packets from the ``rx_to_tx_ring``. Then each packet MAC address is changed
298 if it was enabled. After that copies are sent in burst mode using ``rte_eth_tx_burst()``.
301 .. literalinclude:: ../../../examples/dma/dmafwd.c
303 :start-after: Transmit packets from dmadev/rte_ring for one port. 8<
304 :end-before: >8 End of transmitting packets from dmadev.
307 The Packet Copying Functions
308 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
310 In order to perform SW packet copy, there are user-defined functions to first copy
311 the packet metadata (``pktmbuf_metadata_copy()``) and then the packet data
312 (``pktmbuf_sw_copy()``):
314 .. literalinclude:: ../../../examples/dma/dmafwd.c
316 :start-after: Perform packet copy there is a user-defined function. 8<
317 :end-before: >8 End of perform packet copy there is a user-defined function.
320 The metadata in this example is copied from ``rx_descriptor_fields1`` marker of
321 ``rte_mbuf`` struct up to ``buf_len`` member.
323 In order to understand why software packet copying is done as shown
324 above please refer to the "Mbuf Library" section of the
325 *DPDK Programmer's Guide*.