2 Copyright 2012-2015 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.
30 MLX4 poll mode driver library
31 =============================
33 The MLX4 poll mode driver library (**librte_pmd_mlx4**) implements support
34 for **Mellanox ConnectX-3 EN** 10/40 Gbps adapters as well as their virtual
35 functions (VF) in SR-IOV context.
37 Information and documentation about this family of adapters can be found on
38 the `Mellanox website <http://www.mellanox.com>`_. Help is also provided by
39 the `Mellanox community <http://community.mellanox.com/welcome>`_.
41 There is also a `section dedicated to this poll mode driver
42 <http://www.mellanox.com/page/products_dyn?product_family=209&mtag=pmd_for_dpdk>`_.
46 Due to external dependencies, this driver is disabled by default. It must
47 be enabled manually by setting ``CONFIG_RTE_LIBRTE_MLX4_PMD=y`` and
50 Implementation details
51 ----------------------
53 Most Mellanox ConnectX-3 devices provide two ports but expose a single PCI
54 bus address, thus unlike most drivers, librte_pmd_mlx4 registers itself as a
55 PCI driver that allocates one Ethernet device per detected port.
57 For this reason, one cannot white/blacklist a single port without also
58 white/blacklisting the others on the same device.
60 Besides its dependency on libibverbs (that implies libmlx4 and associated
61 kernel support), librte_pmd_mlx4 relies heavily on system calls for control
62 operations such as querying/updating the MTU and flow control parameters.
64 For security reasons and robustness, this driver only deals with virtual
65 memory addresses. The way resources allocations are handled by the kernel
66 combined with hardware specifications that allow it to handle virtual memory
67 addresses directly ensure that DPDK applications cannot access random
68 physical memory (or memory that does not belong to the current process).
70 This capability allows the PMD to coexist with kernel network interfaces
71 which remain functional, although they stop receiving unicast packets as
72 long as they share the same MAC address.
74 Compiling librte_pmd_mlx4 causes DPDK to be linked against libibverbs.
76 Features and limitations
77 ------------------------
79 - RSS, also known as RCA, is supported. In this mode the number of
80 configured RX queues must be a power of two.
81 - VLAN filtering is supported.
82 - Link state information is provided.
83 - Promiscuous mode is supported.
84 - All multicast mode is supported.
85 - Multiple MAC addresses (unicast, multicast) can be configured.
86 - Scattered packets are supported for TX and RX.
90 - RSS hash key cannot be modified.
91 - Hardware counters are not implemented (they are software counters).
92 - Checksum offloads are not supported yet.
100 These options can be modified in the ``.config`` file.
102 - ``CONFIG_RTE_LIBRTE_MLX4_PMD`` (default **n**)
104 Toggle compilation of librte_pmd_mlx4 itself.
106 - ``CONFIG_RTE_LIBRTE_MLX4_DEBUG`` (default **n**)
108 Toggle debugging code and stricter compilation flags. Enabling this option
109 adds additional run-time checks and debugging messages at the cost of
112 - ``CONFIG_RTE_LIBRTE_MLX4_SGE_WR_N`` (default **4**)
114 Number of scatter/gather elements (SGEs) per work request (WR). Lowering
115 this number improves performance but also limits the ability to receive
116 scattered packets (packets that do not fit a single mbuf). The default
117 value is a safe tradeoff.
119 - ``CONFIG_RTE_LIBRTE_MLX4_MAX_INLINE`` (default **0**)
121 Amount of data to be inlined during TX operations. Improves latency but
124 - ``CONFIG_RTE_LIBRTE_MLX4_TX_MP_CACHE`` (default **8**)
126 Maximum number of cached memory pools (MPs) per TX queue. Each MP from
127 which buffers are to be transmitted must be associated to memory regions
128 (MRs). This is a slow operation that must be cached.
130 This value is always 1 for RX queues since they use a single MP.
132 - ``CONFIG_RTE_LIBRTE_MLX4_SOFT_COUNTERS`` (default **1**)
134 Toggle software counters. No counters are available if this option is
135 disabled since hardware counters are not supported.
137 Environment variables
138 ~~~~~~~~~~~~~~~~~~~~~
140 - ``MLX4_INLINE_RECV_SIZE``
142 A nonzero value enables inline receive for packets up to that size. May
143 significantly improve performance in some cases but lower it in
144 others. Requires careful testing.
146 Run-time configuration
147 ~~~~~~~~~~~~~~~~~~~~~~
149 - The only constraint when RSS mode is requested is to make sure the number
150 of RX queues is a power of two. This is a hardware requirement.
152 - librte_pmd_mlx4 brings kernel network interfaces up during initialization
153 because it is affected by their state. Forcing them down prevents packets
156 - **ethtool** operations on related kernel interfaces also affect the PMD.
158 Kernel module parameters
159 ~~~~~~~~~~~~~~~~~~~~~~~~
161 The **mlx4_core** kernel module has several parameters that affect the
162 behavior and/or the performance of librte_pmd_mlx4. Some of them are described
165 - **num_vfs** (integer or triplet, optionally prefixed by device address
168 Create the given number of VFs on the specified devices.
170 - **log_num_mgm_entry_size** (integer)
172 Device-managed flow steering (DMFS) is required by DPDK applications. It is
173 enabled by using a negative value, the last four bits of which have a
176 - **-1**: force device-managed flow steering (DMFS).
177 - **-7**: configure optimized steering mode to improve performance with the
178 following limitation: Ethernet frames with the port MAC address as the
179 destination cannot be received, even in promiscuous mode. Additional MAC
180 addresses can still be set by ``rte_eth_dev_mac_addr_addr()``.
185 This driver relies on external libraries and kernel drivers for resources
186 allocations and initialization. The following dependencies are not part of
187 DPDK and must be installed separately:
191 User space verbs framework used by librte_pmd_mlx4. This library provides
192 a generic interface between the kernel and low-level user space drivers
195 It allows slow and privileged operations (context initialization, hardware
196 resources allocations) to be managed by the kernel and fast operations to
197 never leave user space.
201 Low-level user space driver library for Mellanox ConnectX-3 devices,
202 it is automatically loaded by libibverbs.
204 This library basically implements send/receive calls to the hardware
207 - **Kernel modules** (mlnx-ofed-kernel)
209 They provide the kernel-side verbs API and low level device drivers that
210 manage actual hardware initialization and resources sharing with user
213 Unlike most other PMDs, these modules must remain loaded and bound to
216 - mlx4_core: hardware driver managing Mellanox ConnectX-3 devices.
217 - mlx4_en: Ethernet device driver that provides kernel network interfaces.
218 - mlx4_ib: InifiniBand device driver.
219 - ib_uverbs: user space driver for verbs (entry point for libibverbs).
221 - **Firmware update**
223 Mellanox OFED releases include firmware updates for ConnectX-3 adapters.
225 Because each release provides new features, these updates must be applied to
226 match the kernel modules and libraries they come with.
230 Both libraries are BSD and GPL licensed. Linux kernel modules are GPL
233 Currently supported by DPDK:
235 - Mellanox OFED **2.4-1**.
236 - Firmware version **2.33.5000** and higher.
238 Getting Mellanox OFED
239 ~~~~~~~~~~~~~~~~~~~~~
241 While these libraries and kernel modules are available on OpenFabrics
242 Alliance's `website <https://www.openfabrics.org/>`_ and provided by package
243 managers on most distributions, this PMD requires Ethernet extensions that
244 may not be supported at the moment (this is a work in progress).
247 <http://www.mellanox.com/page/products_dyn?product_family=26&mtag=linux_sw_drivers>`_
248 includes the necessary support and should be used in the meantime. For DPDK,
249 only libibverbs, libmlx4, mlnx-ofed-kernel packages and firmware updates are
250 required from that distribution.
254 Several versions of Mellanox OFED are available. Installing the version
255 this DPDK release was developed and tested against is strongly
256 recommended. Please check the `prerequisites`_.
258 Getting libibverbs and libmlx4 from DPDK.org
259 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
261 Based on Mellanox OFED, optimized libibverbs and libmlx4 versions can be
262 optionally downloaded from DPDK.org:
264 `<http://www.dpdk.org/download/mlx4>`_
266 Some enhancements are done for better performance with DPDK applications and
267 are not merged upstream yet.
269 Since it is partly achieved by tuning compilation options to disable features
270 not needed by DPDK, linking these libraries statically and avoid system-wide
271 installation is the preferred method.
273 Installation documentation is available from the above link.
278 This section demonstrates how to launch **testpmd** with Mellanox ConnectX-3
279 devices managed by librte_pmd_mlx4.
281 #. Load the kernel modules:
283 .. code-block:: console
285 modprobe -a ib_uverbs mlx4_en mlx4_core mlx4_ib
289 User space I/O kernel modules (uio and igb_uio) are not used and do
290 not have to be loaded.
292 #. Make sure Ethernet interfaces are in working order and linked to kernel
293 verbs. Related sysfs entries should be present:
295 .. code-block:: console
297 ls -d /sys/class/net/*/device/infiniband_verbs/uverbs* | cut -d / -f 5
301 .. code-block:: console
308 #. Optionally, retrieve their PCI bus addresses for whitelisting:
310 .. code-block:: console
313 for intf in eth2 eth3 eth4 eth5;
315 (cd "/sys/class/net/${intf}/device/" && pwd -P);
318 sed -n 's,.*/\(.*\),-w \1,p'
322 .. code-block:: console
331 There are only two distinct PCI bus addresses because the Mellanox
332 ConnectX-3 adapters installed on this system are dual port.
334 #. Request huge pages:
336 .. code-block:: console
338 echo 1024 > /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages/nr_hugepages
340 #. Start testpmd with basic parameters:
342 .. code-block:: console
344 testpmd -c 0xff00 -n 4 -w 0000:83:00.0 -w 0000:84:00.0 -- --rxq=2 --txq=2 -i
348 .. code-block:: console
351 EAL: PCI device 0000:83:00.0 on NUMA socket 1
352 EAL: probe driver: 15b3:1007 librte_pmd_mlx4
353 PMD: librte_pmd_mlx4: PCI information matches, using device "mlx4_0" (VF: false)
354 PMD: librte_pmd_mlx4: 2 port(s) detected
355 PMD: librte_pmd_mlx4: port 1 MAC address is 00:02:c9:b5:b7:50
356 PMD: librte_pmd_mlx4: port 2 MAC address is 00:02:c9:b5:b7:51
357 EAL: PCI device 0000:84:00.0 on NUMA socket 1
358 EAL: probe driver: 15b3:1007 librte_pmd_mlx4
359 PMD: librte_pmd_mlx4: PCI information matches, using device "mlx4_1" (VF: false)
360 PMD: librte_pmd_mlx4: 2 port(s) detected
361 PMD: librte_pmd_mlx4: port 1 MAC address is 00:02:c9:b5:ba:b0
362 PMD: librte_pmd_mlx4: port 2 MAC address is 00:02:c9:b5:ba:b1
363 Interactive-mode selected
364 Configuring Port 0 (socket 0)
365 PMD: librte_pmd_mlx4: 0x867d60: TX queues number update: 0 -> 2
366 PMD: librte_pmd_mlx4: 0x867d60: RX queues number update: 0 -> 2
367 Port 0: 00:02:C9:B5:B7:50
368 Configuring Port 1 (socket 0)
369 PMD: librte_pmd_mlx4: 0x867da0: TX queues number update: 0 -> 2
370 PMD: librte_pmd_mlx4: 0x867da0: RX queues number update: 0 -> 2
371 Port 1: 00:02:C9:B5:B7:51
372 Configuring Port 2 (socket 0)
373 PMD: librte_pmd_mlx4: 0x867de0: TX queues number update: 0 -> 2
374 PMD: librte_pmd_mlx4: 0x867de0: RX queues number update: 0 -> 2
375 Port 2: 00:02:C9:B5:BA:B0
376 Configuring Port 3 (socket 0)
377 PMD: librte_pmd_mlx4: 0x867e20: TX queues number update: 0 -> 2
378 PMD: librte_pmd_mlx4: 0x867e20: RX queues number update: 0 -> 2
379 Port 3: 00:02:C9:B5:BA:B1
380 Checking link statuses...
381 Port 0 Link Up - speed 10000 Mbps - full-duplex
382 Port 1 Link Up - speed 40000 Mbps - full-duplex
383 Port 2 Link Up - speed 10000 Mbps - full-duplex
384 Port 3 Link Up - speed 40000 Mbps - full-duplex