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 - CRC stripping is supported by default and always reported as "true".
150 The ability to enable/disable CRC stripping requires OFED version
151 4.3-1.5.0.0 and above or rdma-core version v18 and above.
156 This driver relies on external libraries and kernel drivers for resources
157 allocations and initialization. The following dependencies are not part of
158 DPDK and must be installed separately:
160 - **libibverbs** (provided by rdma-core package)
162 User space verbs framework used by librte_pmd_mlx4. This library provides
163 a generic interface between the kernel and low-level user space drivers
166 It allows slow and privileged operations (context initialization, hardware
167 resources allocations) to be managed by the kernel and fast operations to
168 never leave user space.
170 - **libmlx4** (provided by rdma-core package)
172 Low-level user space driver library for Mellanox ConnectX-3 devices,
173 it is automatically loaded by libibverbs.
175 This library basically implements send/receive calls to the hardware
180 They provide the kernel-side verbs API and low level device drivers that
181 manage actual hardware initialization and resources sharing with user
184 Unlike most other PMDs, these modules must remain loaded and bound to
187 - mlx4_core: hardware driver managing Mellanox ConnectX-3 devices.
188 - mlx4_en: Ethernet device driver that provides kernel network interfaces.
189 - mlx4_ib: InifiniBand device driver.
190 - ib_uverbs: user space driver for verbs (entry point for libibverbs).
192 - **Firmware update**
194 Mellanox OFED releases include firmware updates for ConnectX-3 adapters.
196 Because each release provides new features, these updates must be applied to
197 match the kernel modules and libraries they come with.
201 Both libraries are BSD and GPL licensed. Linux kernel modules are GPL
204 Depending on system constraints and user preferences either RDMA core library
205 with a recent enough Linux kernel release (recommended) or Mellanox OFED,
206 which provides compatibility with older releases.
208 Current RDMA core package and Linux kernel (recommended)
209 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
211 - Minimal Linux kernel version: 4.14.
212 - Minimal RDMA core version: v15 (see `RDMA core installation documentation`_).
214 .. _`RDMA core installation documentation`: https://raw.githubusercontent.com/linux-rdma/rdma-core/master/README.md
216 .. _Mellanox_OFED_as_a_fallback:
218 Mellanox OFED as a fallback
219 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
221 - `Mellanox OFED`_ version: **4.2, 4.3**.
222 - firmware version: **2.42.5000** and above.
224 .. _`Mellanox OFED`: http://www.mellanox.com/page/products_dyn?product_family=26&mtag=linux_sw_drivers
228 Several versions of Mellanox OFED are available. Installing the version
229 this DPDK release was developed and tested against is strongly
230 recommended. Please check the `prerequisites`_.
232 Installing Mellanox OFED
233 ^^^^^^^^^^^^^^^^^^^^^^^^
235 1. Download latest Mellanox OFED.
237 2. Install the required libraries and kernel modules either by installing
238 only the required set, or by installing the entire Mellanox OFED:
242 .. code-block:: console
244 ./mlnxofedinstall --dpdk --upstream-libs
246 For SR-IOV hypervisors use:
248 .. code-block:: console
250 ./mlnxofedinstall --dpdk --upstream-libs --enable-sriov --hypervisor
252 For SR-IOV virtual machine use:
254 .. code-block:: console
256 ./mlnxofedinstall --dpdk --upstream-libs --guest
258 3. Verify the firmware is the correct one:
260 .. code-block:: console
264 4. Set all ports links to Ethernet, follow instructions on the screen:
266 .. code-block:: console
270 5. Continue with :ref:`section 2 of the Quick Start Guide <QSG_2>`.
275 * Mellanox(R) ConnectX(R)-3 Pro 40G MCX354A-FCC_Ax (2*40G)
282 1. Set all ports links to Ethernet
284 .. code-block:: console
286 PCI=<NIC PCI address>
287 echo eth > "/sys/bus/pci/devices/$PCI/mlx4_port0"
288 echo eth > "/sys/bus/pci/devices/$PCI/mlx4_port1"
292 If using Mellanox OFED one can permanently set the port link
293 to Ethernet using connectx_port_config tool provided by it.
294 :ref:`Mellanox_OFED_as_a_fallback`:
298 2. In case of bare metal or hypervisor, configure optimized steering mode
299 by adding the following line to ``/etc/modprobe.d/mlx4_core.conf``:
301 .. code-block:: console
303 options mlx4_core log_num_mgm_entry_size=-7
307 If VLAN filtering is used, set log_num_mgm_entry_size=-1.
308 Performance degradation can occur on this case.
310 3. Restart the driver:
312 .. code-block:: console
314 /etc/init.d/openibd restart
318 .. code-block:: console
320 service openibd restart
322 4. Compile DPDK and you are ready to go. See instructions on
323 :ref:`Development Kit Build System <Development_Kit_Build_System>`
328 1. Verify the optimized steering mode is configured:
330 .. code-block:: console
332 cat /sys/module/mlx4_core/parameters/log_num_mgm_entry_size
334 2. Use the CPU near local NUMA node to which the PCIe adapter is connected,
335 for better performance. For VMs, verify that the right CPU
336 and NUMA node are pinned according to the above. Run:
338 .. code-block:: console
342 to identify the NUMA node to which the PCIe adapter is connected.
344 3. If more than one adapter is used, and root complex capabilities allow
345 to put both adapters on the same NUMA node without PCI bandwidth degradation,
346 it is recommended to locate both adapters on the same NUMA node.
347 This in order to forward packets from one to the other without
348 NUMA performance penalty.
350 4. Disable pause frames:
352 .. code-block:: console
354 ethtool -A <netdev> rx off tx off
356 5. Verify IO non-posted prefetch is disabled by default. This can be checked
357 via the BIOS configuration. Please contact you server provider for more
358 information about the settings.
362 On some machines, depends on the machine integrator, it is beneficial
363 to set the PCI max read request parameter to 1K. This can be
364 done in the following way:
366 To query the read request size use:
368 .. code-block:: console
370 setpci -s <NIC PCI address> 68.w
372 If the output is different than 3XXX, set it by:
374 .. code-block:: console
376 setpci -s <NIC PCI address> 68.w=3XXX
378 The XXX can be different on different systems. Make sure to configure
379 according to the setpci output.
384 This section demonstrates how to launch **testpmd** with Mellanox ConnectX-3
385 devices managed by librte_pmd_mlx4.
387 #. Load the kernel modules:
389 .. code-block:: console
391 modprobe -a ib_uverbs mlx4_en mlx4_core mlx4_ib
393 Alternatively if MLNX_OFED is fully installed, the following script can
396 .. code-block:: console
398 /etc/init.d/openibd restart
402 User space I/O kernel modules (uio and igb_uio) are not used and do
403 not have to be loaded.
405 #. Make sure Ethernet interfaces are in working order and linked to kernel
406 verbs. Related sysfs entries should be present:
408 .. code-block:: console
410 ls -d /sys/class/net/*/device/infiniband_verbs/uverbs* | cut -d / -f 5
414 .. code-block:: console
421 #. Optionally, retrieve their PCI bus addresses for whitelisting:
423 .. code-block:: console
426 for intf in eth2 eth3 eth4 eth5;
428 (cd "/sys/class/net/${intf}/device/" && pwd -P);
431 sed -n 's,.*/\(.*\),-w \1,p'
435 .. code-block:: console
444 There are only two distinct PCI bus addresses because the Mellanox
445 ConnectX-3 adapters installed on this system are dual port.
447 #. Request huge pages:
449 .. code-block:: console
451 echo 1024 > /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages/nr_hugepages
453 #. Start testpmd with basic parameters:
455 .. code-block:: console
457 testpmd -l 8-15 -n 4 -w 0000:83:00.0 -w 0000:84:00.0 -- --rxq=2 --txq=2 -i
461 .. code-block:: console
464 EAL: PCI device 0000:83:00.0 on NUMA socket 1
465 EAL: probe driver: 15b3:1007 librte_pmd_mlx4
466 PMD: librte_pmd_mlx4: PCI information matches, using device "mlx4_0" (VF: false)
467 PMD: librte_pmd_mlx4: 2 port(s) detected
468 PMD: librte_pmd_mlx4: port 1 MAC address is 00:02:c9:b5:b7:50
469 PMD: librte_pmd_mlx4: port 2 MAC address is 00:02:c9:b5:b7:51
470 EAL: PCI device 0000:84:00.0 on NUMA socket 1
471 EAL: probe driver: 15b3:1007 librte_pmd_mlx4
472 PMD: librte_pmd_mlx4: PCI information matches, using device "mlx4_1" (VF: false)
473 PMD: librte_pmd_mlx4: 2 port(s) detected
474 PMD: librte_pmd_mlx4: port 1 MAC address is 00:02:c9:b5:ba:b0
475 PMD: librte_pmd_mlx4: port 2 MAC address is 00:02:c9:b5:ba:b1
476 Interactive-mode selected
477 Configuring Port 0 (socket 0)
478 PMD: librte_pmd_mlx4: 0x867d60: TX queues number update: 0 -> 2
479 PMD: librte_pmd_mlx4: 0x867d60: RX queues number update: 0 -> 2
480 Port 0: 00:02:C9:B5:B7:50
481 Configuring Port 1 (socket 0)
482 PMD: librte_pmd_mlx4: 0x867da0: TX queues number update: 0 -> 2
483 PMD: librte_pmd_mlx4: 0x867da0: RX queues number update: 0 -> 2
484 Port 1: 00:02:C9:B5:B7:51
485 Configuring Port 2 (socket 0)
486 PMD: librte_pmd_mlx4: 0x867de0: TX queues number update: 0 -> 2
487 PMD: librte_pmd_mlx4: 0x867de0: RX queues number update: 0 -> 2
488 Port 2: 00:02:C9:B5:BA:B0
489 Configuring Port 3 (socket 0)
490 PMD: librte_pmd_mlx4: 0x867e20: TX queues number update: 0 -> 2
491 PMD: librte_pmd_mlx4: 0x867e20: RX queues number update: 0 -> 2
492 Port 3: 00:02:C9:B5:BA:B1
493 Checking link statuses...
494 Port 0 Link Up - speed 10000 Mbps - full-duplex
495 Port 1 Link Up - speed 40000 Mbps - full-duplex
496 Port 2 Link Up - speed 10000 Mbps - full-duplex
497 Port 3 Link Up - speed 40000 Mbps - full-duplex