2 Copyright 2015 6WIND S.A.
4 Redistribution and use in source and binary forms, with or without
5 modification, are permitted provided that the following conditions
8 * Redistributions of source code must retain the above copyright
9 notice, this list of conditions and the following disclaimer.
10 * Redistributions in binary form must reproduce the above copyright
11 notice, this list of conditions and the following disclaimer in
12 the documentation and/or other materials provided with the
14 * Neither the name of 6WIND S.A. nor the names of its
15 contributors may be used to endorse or promote products derived
16 from this software without specific prior written permission.
18 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
19 "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
20 LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
21 A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
22 OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
23 SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
24 LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
25 DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
26 THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
27 (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
28 OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
33 The MLX5 poll mode driver library (**librte_pmd_mlx5**) provides support for
34 **Mellanox ConnectX-4 EN** and **Mellanox ConnectX-4 Lx EN** families of
35 10/25/40/50/100 Gb/s adapters as well as their virtual functions (VF) in
38 Information and documentation about these adapters can be found on the
39 `Mellanox website <http://www.mellanox.com>`__. Help is also provided by the
40 `Mellanox community <http://community.mellanox.com/welcome>`__.
42 There is also a `section dedicated to this poll mode driver
43 <http://www.mellanox.com/page/products_dyn?product_family=209&mtag=pmd_for_dpdk>`__.
47 Due to external dependencies, this driver is disabled by default. It must
48 be enabled manually by setting ``CONFIG_RTE_LIBRTE_MLX5_PMD=y`` and
53 ``CONFIG_RTE_BUILD_COMBINE_LIBS`` with ``CONFIG_RTE_BUILD_SHARED_LIB``
54 is not supported and thus the compilation will fail with this configuration.
56 Implementation details
57 ----------------------
59 Besides its dependency on libibverbs (that implies libmlx5 and associated
60 kernel support), librte_pmd_mlx5 relies heavily on system calls for control
61 operations such as querying/updating the MTU and flow control parameters.
63 For security reasons and robustness, this driver only deals with virtual
64 memory addresses. The way resources allocations are handled by the kernel
65 combined with hardware specifications that allow it to handle virtual memory
66 addresses directly ensure that DPDK applications cannot access random
67 physical memory (or memory that does not belong to the current process).
69 This capability allows the PMD to coexist with kernel network interfaces
70 which remain functional, although they stop receiving unicast packets as
71 long as they share the same MAC address.
73 Enabling librte_pmd_mlx5 causes DPDK applications to be linked against
79 - Multiple TX and RX queues.
80 - Support for scattered TX and RX frames.
81 - IPv4, TCPv4 and UDPv4 RSS on any number of queues.
82 - Several RSS hash keys, one for each flow type.
83 - Support for multiple MAC addresses.
90 - IPv6 and inner VXLAN RSS are not supported yet.
91 - Port statistics through software counters only.
92 - No allmulticast mode.
93 - Hardware checksum offloads are not supported yet.
101 These options can be modified in the ``.config`` file.
103 - ``CONFIG_RTE_LIBRTE_MLX5_PMD`` (default **n**)
105 Toggle compilation of librte_pmd_mlx5 itself.
107 - ``CONFIG_RTE_LIBRTE_MLX5_DEBUG`` (default **n**)
109 Toggle debugging code and stricter compilation flags. Enabling this option
110 adds additional run-time checks and debugging messages at the cost of
113 - ``CONFIG_RTE_LIBRTE_MLX5_SGE_WR_N`` (default **4**)
115 Number of scatter/gather elements (SGEs) per work request (WR). Lowering
116 this number improves performance but also limits the ability to receive
117 scattered packets (packets that do not fit a single mbuf). The default
118 value is a safe tradeoff.
120 - ``CONFIG_RTE_LIBRTE_MLX5_MAX_INLINE`` (default **0**)
122 Amount of data to be inlined during TX operations. Improves latency but
125 - ``CONFIG_RTE_LIBRTE_MLX5_TX_MP_CACHE`` (default **8**)
127 Maximum number of cached memory pools (MPs) per TX queue. Each MP from
128 which buffers are to be transmitted must be associated to memory regions
129 (MRs). This is a slow operation that must be cached.
131 This value is always 1 for RX queues since they use a single MP.
133 Environment variables
134 ~~~~~~~~~~~~~~~~~~~~~
136 - ``MLX5_ENABLE_CQE_COMPRESSION``
138 A nonzero value lets ConnectX-4 return smaller completion entries to
139 improve performance when PCI backpressure is detected. It is most useful
140 for scenarios involving heavy traffic on many queues.
142 Since the additional software logic necessary to handle this mode can
143 lower performance when there is no backpressure, it is not enabled by
146 Run-time configuration
147 ~~~~~~~~~~~~~~~~~~~~~~
149 - librte_pmd_mlx5 brings kernel network interfaces up during initialization
150 because it is affected by their state. Forcing them down prevents packets
153 - **ethtool** operations on related kernel interfaces also affect the PMD.
158 This driver relies on external libraries and kernel drivers for resources
159 allocations and initialization. The following dependencies are not part of
160 DPDK and must be installed separately:
164 User space Verbs framework used by librte_pmd_mlx5. This library provides
165 a generic interface between the kernel and low-level user space drivers
168 It allows slow and privileged operations (context initialization, hardware
169 resources allocations) to be managed by the kernel and fast operations to
170 never leave user space.
174 Low-level user space driver library for Mellanox ConnectX-4 devices,
175 it is automatically loaded by libibverbs.
177 This library basically implements send/receive calls to the hardware
180 - **Kernel modules** (mlnx-ofed-kernel)
182 They provide the kernel-side Verbs API and low level device drivers that
183 manage actual hardware initialization and resources sharing with user
186 Unlike most other PMDs, these modules must remain loaded and bound to
189 - mlx5_core: hardware driver managing Mellanox ConnectX-4 devices and
190 related Ethernet kernel network devices.
191 - mlx5_ib: InifiniBand device driver.
192 - ib_uverbs: user space driver for Verbs (entry point for libibverbs).
194 - **Firmware update**
196 Mellanox OFED releases include firmware updates for ConnectX-4 adapters.
198 Because each release provides new features, these updates must be applied to
199 match the kernel modules and libraries they come with.
203 Both libraries are BSD and GPL licensed. Linux kernel modules are GPL
206 Currently supported by DPDK:
208 - Mellanox OFED **3.1**.
209 - Minimum firmware version:
210 - ConnectX-4: **12.12.0780**.
211 - ConnectX-4 Lx: **14.12.0780**.
213 Getting Mellanox OFED
214 ~~~~~~~~~~~~~~~~~~~~~
216 While these libraries and kernel modules are available on OpenFabrics
217 Alliance's `website <https://www.openfabrics.org/>`__ and provided by package
218 managers on most distributions, this PMD requires Ethernet extensions that
219 may not be supported at the moment (this is a work in progress).
222 <http://www.mellanox.com/page/products_dyn?product_family=26&mtag=linux>`__
223 includes the necessary support and should be used in the meantime. For DPDK,
224 only libibverbs, libmlx5, mlnx-ofed-kernel packages and firmware updates are
225 required from that distribution.
229 Several versions of Mellanox OFED are available. Installing the version
230 this DPDK release was developed and tested against is strongly
231 recommended. Please check the `prerequisites`_.
236 This section demonstrates how to launch **testpmd** with Mellanox ConnectX-4
237 devices managed by librte_pmd_mlx5.
239 #. Load the kernel modules:
241 .. code-block:: console
243 modprobe -a ib_uverbs mlx5_core mlx5_ib
247 User space I/O kernel modules (uio and igb_uio) are not used and do
248 not have to be loaded.
250 #. Make sure Ethernet interfaces are in working order and linked to kernel
251 verbs. Related sysfs entries should be present:
253 .. code-block:: console
255 ls -d /sys/class/net/*/device/infiniband_verbs/uverbs* | cut -d / -f 5
259 .. code-block:: console
266 #. Optionally, retrieve their PCI bus addresses for whitelisting:
268 .. code-block:: console
271 for intf in eth2 eth3 eth4 eth5;
273 (cd "/sys/class/net/${intf}/device/" && pwd -P);
276 sed -n 's,.*/\(.*\),-w \1,p'
280 .. code-block:: console
287 #. Request huge pages:
289 .. code-block:: console
291 echo 1024 > /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages/nr_hugepages
293 #. Start testpmd with basic parameters:
295 .. code-block:: console
297 testpmd -c 0xff00 -n 4 -w 05:00.0 -w 05:00.1 -w 06:00.0 -w 06:00.1 -- --rxq=2 --txq=2 -i
301 .. code-block:: console
304 EAL: PCI device 0000:05:00.0 on NUMA socket 0
305 EAL: probe driver: 15b3:1013 librte_pmd_mlx5
306 PMD: librte_pmd_mlx5: PCI information matches, using device "mlx5_0" (VF: false)
307 PMD: librte_pmd_mlx5: 1 port(s) detected
308 PMD: librte_pmd_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:fe
309 EAL: PCI device 0000:05:00.1 on NUMA socket 0
310 EAL: probe driver: 15b3:1013 librte_pmd_mlx5
311 PMD: librte_pmd_mlx5: PCI information matches, using device "mlx5_1" (VF: false)
312 PMD: librte_pmd_mlx5: 1 port(s) detected
313 PMD: librte_pmd_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:ff
314 EAL: PCI device 0000:06:00.0 on NUMA socket 0
315 EAL: probe driver: 15b3:1013 librte_pmd_mlx5
316 PMD: librte_pmd_mlx5: PCI information matches, using device "mlx5_2" (VF: false)
317 PMD: librte_pmd_mlx5: 1 port(s) detected
318 PMD: librte_pmd_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:fa
319 EAL: PCI device 0000:06:00.1 on NUMA socket 0
320 EAL: probe driver: 15b3:1013 librte_pmd_mlx5
321 PMD: librte_pmd_mlx5: PCI information matches, using device "mlx5_3" (VF: false)
322 PMD: librte_pmd_mlx5: 1 port(s) detected
323 PMD: librte_pmd_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:fb
324 Interactive-mode selected
325 Configuring Port 0 (socket 0)
326 PMD: librte_pmd_mlx5: 0x8cba80: TX queues number update: 0 -> 2
327 PMD: librte_pmd_mlx5: 0x8cba80: RX queues number update: 0 -> 2
328 Port 0: E4:1D:2D:E7:0C:FE
329 Configuring Port 1 (socket 0)
330 PMD: librte_pmd_mlx5: 0x8ccac8: TX queues number update: 0 -> 2
331 PMD: librte_pmd_mlx5: 0x8ccac8: RX queues number update: 0 -> 2
332 Port 1: E4:1D:2D:E7:0C:FF
333 Configuring Port 2 (socket 0)
334 PMD: librte_pmd_mlx5: 0x8cdb10: TX queues number update: 0 -> 2
335 PMD: librte_pmd_mlx5: 0x8cdb10: RX queues number update: 0 -> 2
336 Port 2: E4:1D:2D:E7:0C:FA
337 Configuring Port 3 (socket 0)
338 PMD: librte_pmd_mlx5: 0x8ceb58: TX queues number update: 0 -> 2
339 PMD: librte_pmd_mlx5: 0x8ceb58: RX queues number update: 0 -> 2
340 Port 3: E4:1D:2D:E7:0C:FB
341 Checking link statuses...
342 Port 0 Link Up - speed 40000 Mbps - full-duplex
343 Port 1 Link Up - speed 40000 Mbps - full-duplex
344 Port 2 Link Up - speed 10000 Mbps - full-duplex
345 Port 3 Link Up - speed 10000 Mbps - full-duplex