1 .. SPDX-License-Identifier: BSD-3-Clause
2 Copyright 2012 6WIND S.A.
3 Copyright 2015 Mellanox Technologies, Ltd
5 MLX4 poll mode driver library
6 =============================
8 The MLX4 poll mode driver library (**librte_pmd_mlx4**) implements support
9 for **Mellanox ConnectX-3** and **Mellanox ConnectX-3 Pro** 10/40 Gbps adapters
10 as well as their virtual functions (VF) in SR-IOV context.
12 Information and documentation about this family of adapters can be found on
13 the `Mellanox website <http://www.mellanox.com>`_. Help is also provided by
14 the `Mellanox community <http://community.mellanox.com/welcome>`_.
16 There is also a `section dedicated to this poll mode driver
17 <http://www.mellanox.com/page/products_dyn?product_family=209&mtag=pmd_for_dpdk>`_.
21 Due to external dependencies, this driver is disabled by default. It must
22 be enabled manually by setting ``CONFIG_RTE_LIBRTE_MLX4_PMD=y`` and
25 Implementation details
26 ----------------------
28 Most Mellanox ConnectX-3 devices provide two ports but expose a single PCI
29 bus address, thus unlike most drivers, librte_pmd_mlx4 registers itself as a
30 PCI driver that allocates one Ethernet device per detected port.
32 For this reason, one cannot white/blacklist a single port without also
33 white/blacklisting the others on the same device.
35 Besides its dependency on libibverbs (that implies libmlx4 and associated
36 kernel support), librte_pmd_mlx4 relies heavily on system calls for control
37 operations such as querying/updating the MTU and flow control parameters.
39 For security reasons and robustness, this driver only deals with virtual
40 memory addresses. The way resources allocations are handled by the kernel
41 combined with hardware specifications that allow it to handle virtual memory
42 addresses directly ensure that DPDK applications cannot access random
43 physical memory (or memory that does not belong to the current process).
45 This capability allows the PMD to coexist with kernel network interfaces
46 which remain functional, although they stop receiving unicast packets as
47 long as they share the same MAC address.
49 Compiling librte_pmd_mlx4 causes DPDK to be linked against libibverbs.
57 These options can be modified in the ``.config`` file.
59 - ``CONFIG_RTE_LIBRTE_MLX4_PMD`` (default **n**)
61 Toggle compilation of librte_pmd_mlx4 itself.
63 - ``CONFIG_RTE_LIBRTE_MLX4_DLOPEN_DEPS`` (default **n**)
65 Build PMD with additional code to make it loadable without hard
66 dependencies on **libibverbs** nor **libmlx4**, which may not be installed
69 In this mode, their presence is still required for it to run properly,
70 however their absence won't prevent a DPDK application from starting (with
71 ``CONFIG_RTE_BUILD_SHARED_LIB`` disabled) and they won't show up as
72 missing with ``ldd(1)``.
74 It works by moving these dependencies to a purpose-built rdma-core "glue"
75 plug-in which must either be installed in a directory whose name is based
76 on ``CONFIG_RTE_EAL_PMD_PATH`` suffixed with ``-glue`` if set, or in a
77 standard location for the dynamic linker (e.g. ``/lib``) if left to the
78 default empty string (``""``).
80 This option has no performance impact.
82 - ``CONFIG_RTE_LIBRTE_MLX4_DEBUG`` (default **n**)
84 Toggle debugging code and stricter compilation flags. Enabling this option
85 adds additional run-time checks and debugging messages at the cost of
88 - ``CONFIG_RTE_LIBRTE_MLX4_TX_MP_CACHE`` (default **8**)
90 Maximum number of cached memory pools (MPs) per TX queue. Each MP from
91 which buffers are to be transmitted must be associated to memory regions
92 (MRs). This is a slow operation that must be cached.
94 This value is always 1 for RX queues since they use a single MP.
101 A list of directories in which to search for the rdma-core "glue" plug-in,
102 separated by colons or semi-colons.
104 Only matters when compiled with ``CONFIG_RTE_LIBRTE_MLX4_DLOPEN_DEPS``
105 enabled and most useful when ``CONFIG_RTE_EAL_PMD_PATH`` is also set,
106 since ``LD_LIBRARY_PATH`` has no effect in this case.
108 Run-time configuration
109 ~~~~~~~~~~~~~~~~~~~~~~
111 - librte_pmd_mlx4 brings kernel network interfaces up during initialization
112 because it is affected by their state. Forcing them down prevents packets
115 - **ethtool** operations on related kernel interfaces also affect the PMD.
117 - ``port`` parameter [int]
119 This parameter provides a physical port to probe and can be specified multiple
120 times for additional ports. All ports are probed by default if left
123 Kernel module parameters
124 ~~~~~~~~~~~~~~~~~~~~~~~~
126 The **mlx4_core** kernel module has several parameters that affect the
127 behavior and/or the performance of librte_pmd_mlx4. Some of them are described
130 - **num_vfs** (integer or triplet, optionally prefixed by device address
133 Create the given number of VFs on the specified devices.
135 - **log_num_mgm_entry_size** (integer)
137 Device-managed flow steering (DMFS) is required by DPDK applications. It is
138 enabled by using a negative value, the last four bits of which have a
141 - **-1**: force device-managed flow steering (DMFS).
142 - **-7**: configure optimized steering mode to improve performance with the
143 following limitation: VLAN filtering is not supported with this mode.
144 This is the recommended mode in case VLAN filter is not needed.
149 This driver relies on external libraries and kernel drivers for resources
150 allocations and initialization. The following dependencies are not part of
151 DPDK and must be installed separately:
153 - **libibverbs** (provided by rdma-core package)
155 User space verbs framework used by librte_pmd_mlx4. This library provides
156 a generic interface between the kernel and low-level user space drivers
159 It allows slow and privileged operations (context initialization, hardware
160 resources allocations) to be managed by the kernel and fast operations to
161 never leave user space.
163 - **libmlx4** (provided by rdma-core package)
165 Low-level user space driver library for Mellanox ConnectX-3 devices,
166 it is automatically loaded by libibverbs.
168 This library basically implements send/receive calls to the hardware
173 They provide the kernel-side verbs API and low level device drivers that
174 manage actual hardware initialization and resources sharing with user
177 Unlike most other PMDs, these modules must remain loaded and bound to
180 - mlx4_core: hardware driver managing Mellanox ConnectX-3 devices.
181 - mlx4_en: Ethernet device driver that provides kernel network interfaces.
182 - mlx4_ib: InifiniBand device driver.
183 - ib_uverbs: user space driver for verbs (entry point for libibverbs).
185 - **Firmware update**
187 Mellanox OFED releases include firmware updates for ConnectX-3 adapters.
189 Because each release provides new features, these updates must be applied to
190 match the kernel modules and libraries they come with.
194 Both libraries are BSD and GPL licensed. Linux kernel modules are GPL
197 Depending on system constraints and user preferences either RDMA core library
198 with a recent enough Linux kernel release (recommended) or Mellanox OFED,
199 which provides compatibility with older releases.
201 Current RDMA core package and Linux kernel (recommended)
202 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
204 - Minimal Linux kernel version: 4.14.
205 - Minimal RDMA core version: v15 (see `RDMA core installation documentation`_).
207 .. _`RDMA core installation documentation`: https://raw.githubusercontent.com/linux-rdma/rdma-core/master/README.md
209 .. _Mellanox_OFED_as_a_fallback:
211 Mellanox OFED as a fallback
212 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
214 - `Mellanox OFED`_ version: **4.2, 4.3**.
215 - firmware version: **2.42.5000** and above.
217 .. _`Mellanox OFED`: http://www.mellanox.com/page/products_dyn?product_family=26&mtag=linux_sw_drivers
221 Several versions of Mellanox OFED are available. Installing the version
222 this DPDK release was developed and tested against is strongly
223 recommended. Please check the `prerequisites`_.
225 Installing Mellanox OFED
226 ^^^^^^^^^^^^^^^^^^^^^^^^
228 1. Download latest Mellanox OFED.
230 2. Install the required libraries and kernel modules either by installing
231 only the required set, or by installing the entire Mellanox OFED:
235 .. code-block:: console
237 ./mlnxofedinstall --dpdk --upstream-libs
239 For SR-IOV hypervisors use:
241 .. code-block:: console
243 ./mlnxofedinstall --dpdk --upstream-libs --enable-sriov --hypervisor
245 For SR-IOV virtual machine use:
247 .. code-block:: console
249 ./mlnxofedinstall --dpdk --upstream-libs --guest
251 3. Verify the firmware is the correct one:
253 .. code-block:: console
257 4. Set all ports links to Ethernet, follow instructions on the screen:
259 .. code-block:: console
263 5. Continue with :ref:`section 2 of the Quick Start Guide <QSG_2>`.
268 * Mellanox(R) ConnectX(R)-3 Pro 40G MCX354A-FCC_Ax (2*40G)
275 1. Set all ports links to Ethernet
277 .. code-block:: console
279 PCI=<NIC PCI address>
280 echo eth > "/sys/bus/pci/devices/$PCI/mlx4_port0"
281 echo eth > "/sys/bus/pci/devices/$PCI/mlx4_port1"
285 If using Mellanox OFED one can permanently set the port link
286 to Ethernet using connectx_port_config tool provided by it.
287 :ref:`Mellanox_OFED_as_a_fallback`:
291 2. In case of bare metal or hypervisor, configure optimized steering mode
292 by adding the following line to ``/etc/modprobe.d/mlx4_core.conf``:
294 .. code-block:: console
296 options mlx4_core log_num_mgm_entry_size=-7
300 If VLAN filtering is used, set log_num_mgm_entry_size=-1.
301 Performance degradation can occur on this case.
303 3. Restart the driver:
305 .. code-block:: console
307 /etc/init.d/openibd restart
311 .. code-block:: console
313 service openibd restart
315 4. Compile DPDK and you are ready to go. See instructions on
316 :ref:`Development Kit Build System <Development_Kit_Build_System>`
321 1. Verify the optimized steering mode is configured:
323 .. code-block:: console
325 cat /sys/module/mlx4_core/parameters/log_num_mgm_entry_size
327 2. Use the CPU near local NUMA node to which the PCIe adapter is connected,
328 for better performance. For VMs, verify that the right CPU
329 and NUMA node are pinned according to the above. Run:
331 .. code-block:: console
335 to identify the NUMA node to which the PCIe adapter is connected.
337 3. If more than one adapter is used, and root complex capabilities allow
338 to put both adapters on the same NUMA node without PCI bandwidth degradation,
339 it is recommended to locate both adapters on the same NUMA node.
340 This in order to forward packets from one to the other without
341 NUMA performance penalty.
343 4. Disable pause frames:
345 .. code-block:: console
347 ethtool -A <netdev> rx off tx off
349 5. Verify IO non-posted prefetch is disabled by default. This can be checked
350 via the BIOS configuration. Please contact you server provider for more
351 information about the settings.
355 On some machines, depends on the machine integrator, it is beneficial
356 to set the PCI max read request parameter to 1K. This can be
357 done in the following way:
359 To query the read request size use:
361 .. code-block:: console
363 setpci -s <NIC PCI address> 68.w
365 If the output is different than 3XXX, set it by:
367 .. code-block:: console
369 setpci -s <NIC PCI address> 68.w=3XXX
371 The XXX can be different on different systems. Make sure to configure
372 according to the setpci output.
377 This section demonstrates how to launch **testpmd** with Mellanox ConnectX-3
378 devices managed by librte_pmd_mlx4.
380 #. Load the kernel modules:
382 .. code-block:: console
384 modprobe -a ib_uverbs mlx4_en mlx4_core mlx4_ib
386 Alternatively if MLNX_OFED is fully installed, the following script can
389 .. code-block:: console
391 /etc/init.d/openibd restart
395 User space I/O kernel modules (uio and igb_uio) are not used and do
396 not have to be loaded.
398 #. Make sure Ethernet interfaces are in working order and linked to kernel
399 verbs. Related sysfs entries should be present:
401 .. code-block:: console
403 ls -d /sys/class/net/*/device/infiniband_verbs/uverbs* | cut -d / -f 5
407 .. code-block:: console
414 #. Optionally, retrieve their PCI bus addresses for whitelisting:
416 .. code-block:: console
419 for intf in eth2 eth3 eth4 eth5;
421 (cd "/sys/class/net/${intf}/device/" && pwd -P);
424 sed -n 's,.*/\(.*\),-w \1,p'
428 .. code-block:: console
437 There are only two distinct PCI bus addresses because the Mellanox
438 ConnectX-3 adapters installed on this system are dual port.
440 #. Request huge pages:
442 .. code-block:: console
444 echo 1024 > /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages/nr_hugepages
446 #. Start testpmd with basic parameters:
448 .. code-block:: console
450 testpmd -l 8-15 -n 4 -w 0000:83:00.0 -w 0000:84:00.0 -- --rxq=2 --txq=2 -i
454 .. code-block:: console
457 EAL: PCI device 0000:83:00.0 on NUMA socket 1
458 EAL: probe driver: 15b3:1007 librte_pmd_mlx4
459 PMD: librte_pmd_mlx4: PCI information matches, using device "mlx4_0" (VF: false)
460 PMD: librte_pmd_mlx4: 2 port(s) detected
461 PMD: librte_pmd_mlx4: port 1 MAC address is 00:02:c9:b5:b7:50
462 PMD: librte_pmd_mlx4: port 2 MAC address is 00:02:c9:b5:b7:51
463 EAL: PCI device 0000:84:00.0 on NUMA socket 1
464 EAL: probe driver: 15b3:1007 librte_pmd_mlx4
465 PMD: librte_pmd_mlx4: PCI information matches, using device "mlx4_1" (VF: false)
466 PMD: librte_pmd_mlx4: 2 port(s) detected
467 PMD: librte_pmd_mlx4: port 1 MAC address is 00:02:c9:b5:ba:b0
468 PMD: librte_pmd_mlx4: port 2 MAC address is 00:02:c9:b5:ba:b1
469 Interactive-mode selected
470 Configuring Port 0 (socket 0)
471 PMD: librte_pmd_mlx4: 0x867d60: TX queues number update: 0 -> 2
472 PMD: librte_pmd_mlx4: 0x867d60: RX queues number update: 0 -> 2
473 Port 0: 00:02:C9:B5:B7:50
474 Configuring Port 1 (socket 0)
475 PMD: librte_pmd_mlx4: 0x867da0: TX queues number update: 0 -> 2
476 PMD: librte_pmd_mlx4: 0x867da0: RX queues number update: 0 -> 2
477 Port 1: 00:02:C9:B5:B7:51
478 Configuring Port 2 (socket 0)
479 PMD: librte_pmd_mlx4: 0x867de0: TX queues number update: 0 -> 2
480 PMD: librte_pmd_mlx4: 0x867de0: RX queues number update: 0 -> 2
481 Port 2: 00:02:C9:B5:BA:B0
482 Configuring Port 3 (socket 0)
483 PMD: librte_pmd_mlx4: 0x867e20: TX queues number update: 0 -> 2
484 PMD: librte_pmd_mlx4: 0x867e20: RX queues number update: 0 -> 2
485 Port 3: 00:02:C9:B5:BA:B1
486 Checking link statuses...
487 Port 0 Link Up - speed 10000 Mbps - full-duplex
488 Port 1 Link Up - speed 40000 Mbps - full-duplex
489 Port 2 Link Up - speed 10000 Mbps - full-duplex
490 Port 3 Link Up - speed 40000 Mbps - full-duplex