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** and **Mellanox ConnectX-4 Lx** 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
51 Implementation details
52 ----------------------
54 Besides its dependency on libibverbs (that implies libmlx5 and associated
55 kernel support), librte_pmd_mlx5 relies heavily on system calls for control
56 operations such as querying/updating the MTU and flow control parameters.
58 For security reasons and robustness, this driver only deals with virtual
59 memory addresses. The way resources allocations are handled by the kernel
60 combined with hardware specifications that allow it to handle virtual memory
61 addresses directly ensure that DPDK applications cannot access random
62 physical memory (or memory that does not belong to the current process).
64 This capability allows the PMD to coexist with kernel network interfaces
65 which remain functional, although they stop receiving unicast packets as
66 long as they share the same MAC address.
68 Enabling librte_pmd_mlx5 causes DPDK applications to be linked against
74 - Multiple TX and RX queues.
75 - Support for scattered TX and RX frames.
76 - IPv4, IPv6, TCPv4, TCPv6, UDPv4 and UDPv6 RSS on any number of queues.
77 - Several RSS hash keys, one for each flow type.
78 - Configurable RETA table.
79 - Support for multiple MAC addresses.
83 - Multicast promiscuous mode.
84 - Hardware checksum offloads.
85 - Flow director (RTE_FDIR_MODE_PERFECT and RTE_FDIR_MODE_PERFECT_MAC_VLAN).
86 - Secondary process TX is supported.
91 - KVM and VMware ESX SR-IOV modes are not supported yet.
92 - Inner RSS for VXLAN frames is not supported yet.
93 - Port statistics through software counters only.
94 - Hardware checksum offloads for VXLAN inner header are not supported yet.
95 - Secondary process RX is not supported.
103 These options can be modified in the ``.config`` file.
105 - ``CONFIG_RTE_LIBRTE_MLX5_PMD`` (default **n**)
107 Toggle compilation of librte_pmd_mlx5 itself.
109 - ``CONFIG_RTE_LIBRTE_MLX5_DEBUG`` (default **n**)
111 Toggle debugging code and stricter compilation flags. Enabling this option
112 adds additional run-time checks and debugging messages at the cost of
115 - ``CONFIG_RTE_LIBRTE_MLX5_SGE_WR_N`` (default **4**)
117 Number of scatter/gather elements (SGEs) per work request (WR). Lowering
118 this number improves performance but also limits the ability to receive
119 scattered packets (packets that do not fit a single mbuf). The default
120 value is a safe tradeoff.
122 - ``CONFIG_RTE_LIBRTE_MLX5_MAX_INLINE`` (default **0**)
124 Amount of data to be inlined during TX operations. Improves latency.
125 Can improve PPS performance when PCI backpressure is detected and may be
126 useful for scenarios involving heavy traffic on many queues.
128 Since the additional software logic necessary to handle this mode can
129 lower performance when there is no backpressure, it is not enabled by
132 - ``CONFIG_RTE_LIBRTE_MLX5_TX_MP_CACHE`` (default **8**)
134 Maximum number of cached memory pools (MPs) per TX queue. Each MP from
135 which buffers are to be transmitted must be associated to memory regions
136 (MRs). This is a slow operation that must be cached.
138 This value is always 1 for RX queues since they use a single MP.
140 Environment variables
141 ~~~~~~~~~~~~~~~~~~~~~
143 - ``MLX5_ENABLE_CQE_COMPRESSION``
145 A nonzero value lets ConnectX-4 return smaller completion entries to
146 improve performance when PCI backpressure is detected. It is most useful
147 for scenarios involving heavy traffic on many queues.
149 Since the additional software logic necessary to handle this mode can
150 lower performance when there is no backpressure, it is not enabled by
153 Run-time configuration
154 ~~~~~~~~~~~~~~~~~~~~~~
156 - librte_pmd_mlx5 brings kernel network interfaces up during initialization
157 because it is affected by their state. Forcing them down prevents packets
160 - **ethtool** operations on related kernel interfaces also affect the PMD.
165 This driver relies on external libraries and kernel drivers for resources
166 allocations and initialization. The following dependencies are not part of
167 DPDK and must be installed separately:
171 User space Verbs framework used by librte_pmd_mlx5. This library provides
172 a generic interface between the kernel and low-level user space drivers
175 It allows slow and privileged operations (context initialization, hardware
176 resources allocations) to be managed by the kernel and fast operations to
177 never leave user space.
181 Low-level user space driver library for Mellanox ConnectX-4 devices,
182 it is automatically loaded by libibverbs.
184 This library basically implements send/receive calls to the hardware
187 - **Kernel modules** (mlnx-ofed-kernel)
189 They provide the kernel-side Verbs API and low level device drivers that
190 manage actual hardware initialization and resources sharing with user
193 Unlike most other PMDs, these modules must remain loaded and bound to
196 - mlx5_core: hardware driver managing Mellanox ConnectX-4 devices and
197 related Ethernet kernel network devices.
198 - mlx5_ib: InifiniBand device driver.
199 - ib_uverbs: user space driver for Verbs (entry point for libibverbs).
201 - **Firmware update**
203 Mellanox OFED releases include firmware updates for ConnectX-4 adapters.
205 Because each release provides new features, these updates must be applied to
206 match the kernel modules and libraries they come with.
210 Both libraries are BSD and GPL licensed. Linux kernel modules are GPL
213 Currently supported by DPDK:
215 - Mellanox OFED **3.1-1.0.3**, **3.1-1.5.7.1** or **3.2-2.0.0.0** depending
218 The following features are supported with version **3.1-1.5.7.1** and
221 - IPv6, UPDv6, TCPv6 RSS.
222 - RX checksum offloads.
225 The following features are supported with version **3.2-2.0.0.0** and
231 - Minimum firmware version:
233 With MLNX_OFED **3.1-1.0.3**:
235 - ConnectX-4: **12.12.1240**
236 - ConnectX-4 Lx: **14.12.1100**
238 With MLNX_OFED **3.1-1.5.7.1**:
240 - ConnectX-4: **12.13.0144**
241 - ConnectX-4 Lx: **14.13.0144**
243 With MLNX_OFED **3.2-2.0.0.0**:
245 - ConnectX-4: **12.14.2036**
246 - ConnectX-4 Lx: **14.14.2036**
248 Getting Mellanox OFED
249 ~~~~~~~~~~~~~~~~~~~~~
251 While these libraries and kernel modules are available on OpenFabrics
252 Alliance's `website <https://www.openfabrics.org/>`__ and provided by package
253 managers on most distributions, this PMD requires Ethernet extensions that
254 may not be supported at the moment (this is a work in progress).
257 <http://www.mellanox.com/page/products_dyn?product_family=26&mtag=linux>`__
258 includes the necessary support and should be used in the meantime. For DPDK,
259 only libibverbs, libmlx5, mlnx-ofed-kernel packages and firmware updates are
260 required from that distribution.
264 Several versions of Mellanox OFED are available. Installing the version
265 this DPDK release was developed and tested against is strongly
266 recommended. Please check the `prerequisites`_.
271 Compared to librte_pmd_mlx4 that implements a single RSS configuration per
272 port, librte_pmd_mlx5 supports per-protocol RSS configuration.
274 Since ``testpmd`` defaults to IP RSS mode and there is currently no
275 command-line parameter to enable additional protocols (UDP and TCP as well
276 as IP), the following commands must be entered from its CLI to get the same
277 behavior as librte_pmd_mlx4:
279 .. code-block:: console
282 > port config all rss all
288 This section demonstrates how to launch **testpmd** with Mellanox ConnectX-4
289 devices managed by librte_pmd_mlx5.
291 #. Load the kernel modules:
293 .. code-block:: console
295 modprobe -a ib_uverbs mlx5_core mlx5_ib
297 Alternatively if MLNX_OFED is fully installed, the following script can
300 .. code-block:: console
302 /etc/init.d/openibd restart
306 User space I/O kernel modules (uio and igb_uio) are not used and do
307 not have to be loaded.
309 #. Make sure Ethernet interfaces are in working order and linked to kernel
310 verbs. Related sysfs entries should be present:
312 .. code-block:: console
314 ls -d /sys/class/net/*/device/infiniband_verbs/uverbs* | cut -d / -f 5
318 .. code-block:: console
325 #. Optionally, retrieve their PCI bus addresses for whitelisting:
327 .. code-block:: console
330 for intf in eth2 eth3 eth4 eth5;
332 (cd "/sys/class/net/${intf}/device/" && pwd -P);
335 sed -n 's,.*/\(.*\),-w \1,p'
339 .. code-block:: console
346 #. Request huge pages:
348 .. code-block:: console
350 echo 1024 > /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages/nr_hugepages
352 #. Start testpmd with basic parameters:
354 .. code-block:: console
356 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
360 .. code-block:: console
363 EAL: PCI device 0000:05:00.0 on NUMA socket 0
364 EAL: probe driver: 15b3:1013 librte_pmd_mlx5
365 PMD: librte_pmd_mlx5: PCI information matches, using device "mlx5_0" (VF: false)
366 PMD: librte_pmd_mlx5: 1 port(s) detected
367 PMD: librte_pmd_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:fe
368 EAL: PCI device 0000:05:00.1 on NUMA socket 0
369 EAL: probe driver: 15b3:1013 librte_pmd_mlx5
370 PMD: librte_pmd_mlx5: PCI information matches, using device "mlx5_1" (VF: false)
371 PMD: librte_pmd_mlx5: 1 port(s) detected
372 PMD: librte_pmd_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:ff
373 EAL: PCI device 0000:06:00.0 on NUMA socket 0
374 EAL: probe driver: 15b3:1013 librte_pmd_mlx5
375 PMD: librte_pmd_mlx5: PCI information matches, using device "mlx5_2" (VF: false)
376 PMD: librte_pmd_mlx5: 1 port(s) detected
377 PMD: librte_pmd_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:fa
378 EAL: PCI device 0000:06:00.1 on NUMA socket 0
379 EAL: probe driver: 15b3:1013 librte_pmd_mlx5
380 PMD: librte_pmd_mlx5: PCI information matches, using device "mlx5_3" (VF: false)
381 PMD: librte_pmd_mlx5: 1 port(s) detected
382 PMD: librte_pmd_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:fb
383 Interactive-mode selected
384 Configuring Port 0 (socket 0)
385 PMD: librte_pmd_mlx5: 0x8cba80: TX queues number update: 0 -> 2
386 PMD: librte_pmd_mlx5: 0x8cba80: RX queues number update: 0 -> 2
387 Port 0: E4:1D:2D:E7:0C:FE
388 Configuring Port 1 (socket 0)
389 PMD: librte_pmd_mlx5: 0x8ccac8: TX queues number update: 0 -> 2
390 PMD: librte_pmd_mlx5: 0x8ccac8: RX queues number update: 0 -> 2
391 Port 1: E4:1D:2D:E7:0C:FF
392 Configuring Port 2 (socket 0)
393 PMD: librte_pmd_mlx5: 0x8cdb10: TX queues number update: 0 -> 2
394 PMD: librte_pmd_mlx5: 0x8cdb10: RX queues number update: 0 -> 2
395 Port 2: E4:1D:2D:E7:0C:FA
396 Configuring Port 3 (socket 0)
397 PMD: librte_pmd_mlx5: 0x8ceb58: TX queues number update: 0 -> 2
398 PMD: librte_pmd_mlx5: 0x8ceb58: RX queues number update: 0 -> 2
399 Port 3: E4:1D:2D:E7:0C:FB
400 Checking link statuses...
401 Port 0 Link Up - speed 40000 Mbps - full-duplex
402 Port 1 Link Up - speed 40000 Mbps - full-duplex
403 Port 2 Link Up - speed 10000 Mbps - full-duplex
404 Port 3 Link Up - speed 10000 Mbps - full-duplex