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
82 These options can be modified in the ``.config`` file.
84 - ``CONFIG_RTE_LIBRTE_MLX5_PMD`` (default **n**)
86 Toggle compilation of librte_pmd_mlx5 itself.
88 - ``CONFIG_RTE_LIBRTE_MLX5_DEBUG`` (default **n**)
90 Toggle debugging code and stricter compilation flags. Enabling this option
91 adds additional run-time checks and debugging messages at the cost of
94 - ``CONFIG_RTE_LIBRTE_MLX5_SGE_WR_N`` (default **4**)
96 Number of scatter/gather elements (SGEs) per work request (WR). Lowering
97 this number improves performance but also limits the ability to receive
98 scattered packets (packets that do not fit a single mbuf). The default
99 value is a safe tradeoff.
101 - ``CONFIG_RTE_LIBRTE_MLX5_MAX_INLINE`` (default **0**)
103 Amount of data to be inlined during TX operations. Improves latency but
106 - ``CONFIG_RTE_LIBRTE_MLX5_TX_MP_CACHE`` (default **8**)
108 Maximum number of cached memory pools (MPs) per TX queue. Each MP from
109 which buffers are to be transmitted must be associated to memory regions
110 (MRs). This is a slow operation that must be cached.
112 This value is always 1 for RX queues since they use a single MP.
114 Run-time configuration
115 ~~~~~~~~~~~~~~~~~~~~~~
117 - librte_pmd_mlx5 brings kernel network interfaces up during initialization
118 because it is affected by their state. Forcing them down prevents packets
121 - **ethtool** operations on related kernel interfaces also affect the PMD.
126 This driver relies on external libraries and kernel drivers for resources
127 allocations and initialization. The following dependencies are not part of
128 DPDK and must be installed separately:
132 User space Verbs framework used by librte_pmd_mlx5. This library provides
133 a generic interface between the kernel and low-level user space drivers
136 It allows slow and privileged operations (context initialization, hardware
137 resources allocations) to be managed by the kernel and fast operations to
138 never leave user space.
142 Low-level user space driver library for Mellanox ConnectX-4 devices,
143 it is automatically loaded by libibverbs.
145 This library basically implements send/receive calls to the hardware
148 - **Kernel modules** (mlnx-ofed-kernel)
150 They provide the kernel-side Verbs API and low level device drivers that
151 manage actual hardware initialization and resources sharing with user
154 Unlike most other PMDs, these modules must remain loaded and bound to
157 - mlx5_core: hardware driver managing Mellanox ConnectX-4 devices and
158 related Ethernet kernel network devices.
159 - mlx5_ib: InifiniBand device driver.
160 - ib_uverbs: user space driver for Verbs (entry point for libibverbs).
162 - **Firmware update**
164 Mellanox OFED releases include firmware updates for ConnectX-4 adapters.
166 Because each release provides new features, these updates must be applied to
167 match the kernel modules and libraries they come with.
171 Both libraries are BSD and GPL licensed. Linux kernel modules are GPL
174 Getting Mellanox OFED
175 ~~~~~~~~~~~~~~~~~~~~~
177 While these libraries and kernel modules are available on OpenFabrics
178 Alliance's `website <https://www.openfabrics.org/>`__ and provided by package
179 managers on most distributions, this PMD requires Ethernet extensions that
180 may not be supported at the moment (this is a work in progress).
183 <http://www.mellanox.com/page/products_dyn?product_family=26&mtag=linux>`__
184 includes the necessary support and should be used in the meantime. For DPDK,
185 only libibverbs, libmlx5, mlnx-ofed-kernel packages and firmware updates are
186 required from that distribution.
190 Several versions of Mellanox OFED are available. Installing the version
191 this DPDK release was developed and tested against is strongly
192 recommended. Please check the `prerequisites`_.
197 This section demonstrates how to launch **testpmd** with Mellanox ConnectX-4
198 devices managed by librte_pmd_mlx5.
200 #. Load the kernel modules:
202 .. code-block:: console
204 modprobe -a ib_uverbs mlx5_core mlx5_ib
208 User space I/O kernel modules (uio and igb_uio) are not used and do
209 not have to be loaded.
211 #. Make sure Ethernet interfaces are in working order and linked to kernel
212 verbs. Related sysfs entries should be present:
214 .. code-block:: console
216 ls -d /sys/class/net/*/device/infiniband_verbs/uverbs* | cut -d / -f 5
220 .. code-block:: console
227 #. Optionally, retrieve their PCI bus addresses for whitelisting:
229 .. code-block:: console
232 for intf in eth2 eth3 eth4 eth5;
234 (cd "/sys/class/net/${intf}/device/" && pwd -P);
237 sed -n 's,.*/\(.*\),-w \1,p'
241 .. code-block:: console
248 #. Request huge pages:
250 .. code-block:: console
252 echo 1024 > /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages/nr_hugepages
254 #. Start testpmd with basic parameters:
256 .. code-block:: console
258 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
262 .. code-block:: console
265 EAL: PCI device 0000:05:00.0 on NUMA socket 0
266 EAL: probe driver: 15b3:1013 librte_pmd_mlx5
267 PMD: librte_pmd_mlx5: PCI information matches, using device "mlx5_0" (VF: false)
268 PMD: librte_pmd_mlx5: 1 port(s) detected
269 PMD: librte_pmd_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:fe
270 EAL: PCI device 0000:05:00.1 on NUMA socket 0
271 EAL: probe driver: 15b3:1013 librte_pmd_mlx5
272 PMD: librte_pmd_mlx5: PCI information matches, using device "mlx5_1" (VF: false)
273 PMD: librte_pmd_mlx5: 1 port(s) detected
274 PMD: librte_pmd_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:ff
275 EAL: PCI device 0000:06:00.0 on NUMA socket 0
276 EAL: probe driver: 15b3:1013 librte_pmd_mlx5
277 PMD: librte_pmd_mlx5: PCI information matches, using device "mlx5_2" (VF: false)
278 PMD: librte_pmd_mlx5: 1 port(s) detected
279 PMD: librte_pmd_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:fa
280 EAL: PCI device 0000:06:00.1 on NUMA socket 0
281 EAL: probe driver: 15b3:1013 librte_pmd_mlx5
282 PMD: librte_pmd_mlx5: PCI information matches, using device "mlx5_3" (VF: false)
283 PMD: librte_pmd_mlx5: 1 port(s) detected
284 PMD: librte_pmd_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:fb
285 Interactive-mode selected
286 Configuring Port 0 (socket 0)
287 PMD: librte_pmd_mlx5: 0x8cba80: TX queues number update: 0 -> 2
288 PMD: librte_pmd_mlx5: 0x8cba80: RX queues number update: 0 -> 2
289 Port 0: E4:1D:2D:E7:0C:FE
290 Configuring Port 1 (socket 0)
291 PMD: librte_pmd_mlx5: 0x8ccac8: TX queues number update: 0 -> 2
292 PMD: librte_pmd_mlx5: 0x8ccac8: RX queues number update: 0 -> 2
293 Port 1: E4:1D:2D:E7:0C:FF
294 Configuring Port 2 (socket 0)
295 PMD: librte_pmd_mlx5: 0x8cdb10: TX queues number update: 0 -> 2
296 PMD: librte_pmd_mlx5: 0x8cdb10: RX queues number update: 0 -> 2
297 Port 2: E4:1D:2D:E7:0C:FA
298 Configuring Port 3 (socket 0)
299 PMD: librte_pmd_mlx5: 0x8ceb58: TX queues number update: 0 -> 2
300 PMD: librte_pmd_mlx5: 0x8ceb58: RX queues number update: 0 -> 2
301 Port 3: E4:1D:2D:E7:0C:FB
302 Checking link statuses...
303 Port 0 Link Up - speed 40000 Mbps - full-duplex
304 Port 1 Link Up - speed 40000 Mbps - full-duplex
305 Port 2 Link Up - speed 10000 Mbps - full-duplex
306 Port 3 Link Up - speed 10000 Mbps - full-duplex