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 The :ref:`flow_isolated_mode` is supported.
51 Compiling librte_pmd_mlx4 causes DPDK to be linked against libibverbs.
59 These options can be modified in the ``.config`` file.
61 - ``CONFIG_RTE_LIBRTE_MLX4_PMD`` (default **n**)
63 Toggle compilation of librte_pmd_mlx4 itself.
65 - ``CONFIG_RTE_IBVERBS_LINK_DLOPEN`` (default **n**)
67 Build PMD with additional code to make it loadable without hard
68 dependencies on **libibverbs** nor **libmlx4**, which may not be installed
71 In this mode, their presence is still required for it to run properly,
72 however their absence won't prevent a DPDK application from starting (with
73 ``CONFIG_RTE_BUILD_SHARED_LIB`` disabled) and they won't show up as
74 missing with ``ldd(1)``.
76 It works by moving these dependencies to a purpose-built rdma-core "glue"
77 plug-in which must either be installed in a directory whose name is based
78 on ``CONFIG_RTE_EAL_PMD_PATH`` suffixed with ``-glue`` if set, or in a
79 standard location for the dynamic linker (e.g. ``/lib``) if left to the
80 default empty string (``""``).
82 This option has no performance impact.
84 - ``CONFIG_RTE_IBVERBS_LINK_STATIC`` (default **n**)
86 Embed static flavour of the dependencies **libibverbs** and **libmlx4**
87 in the PMD shared library or the executable static binary.
89 - ``CONFIG_RTE_LIBRTE_MLX4_DEBUG`` (default **n**)
91 Toggle debugging code and stricter compilation flags. Enabling this option
92 adds additional run-time checks and debugging messages at the cost of
100 A list of directories in which to search for the rdma-core "glue" plug-in,
101 separated by colons or semi-colons.
103 Only matters when compiled with ``CONFIG_RTE_IBVERBS_LINK_DLOPEN``
104 enabled and most useful when ``CONFIG_RTE_EAL_PMD_PATH`` is also set,
105 since ``LD_LIBRARY_PATH`` has no effect in this case.
107 Run-time configuration
108 ~~~~~~~~~~~~~~~~~~~~~~
110 - librte_pmd_mlx4 brings kernel network interfaces up during initialization
111 because it is affected by their state. Forcing them down prevents packets
114 - **ethtool** operations on related kernel interfaces also affect the PMD.
116 - ``port`` parameter [int]
118 This parameter provides a physical port to probe and can be specified multiple
119 times for additional ports. All ports are probed by default if left
122 - ``mr_ext_memseg_en`` parameter [int]
124 A nonzero value enables extending memseg when registering DMA memory. If
125 enabled, the number of entries in MR (Memory Region) lookup table on datapath
126 is minimized and it benefits performance. On the other hand, it worsens memory
127 utilization because registered memory is pinned by kernel driver. Even if a
128 page in the extended chunk is freed, that doesn't become reusable until the
129 entire memory is freed.
133 Kernel module parameters
134 ~~~~~~~~~~~~~~~~~~~~~~~~
136 The **mlx4_core** kernel module has several parameters that affect the
137 behavior and/or the performance of librte_pmd_mlx4. Some of them are described
140 - **num_vfs** (integer or triplet, optionally prefixed by device address
143 Create the given number of VFs on the specified devices.
145 - **log_num_mgm_entry_size** (integer)
147 Device-managed flow steering (DMFS) is required by DPDK applications. It is
148 enabled by using a negative value, the last four bits of which have a
151 - **-1**: force device-managed flow steering (DMFS).
152 - **-7**: configure optimized steering mode to improve performance with the
153 following limitation: VLAN filtering is not supported with this mode.
154 This is the recommended mode in case VLAN filter is not needed.
159 - For secondary process:
161 - Forked secondary process not supported.
162 - All mempools must be initialized before rte_eth_dev_start().
163 - External memory unregistered in EAL memseg list cannot be used for DMA
164 unless such memory has been registered by ``mlx4_mr_update_ext_mp()`` in
165 primary process and remapped to the same virtual address in secondary
166 process. If the external memory is registered by primary process but has
167 different virtual address in secondary process, unexpected error may happen.
169 - CRC stripping is supported by default and always reported as "true".
170 The ability to enable/disable CRC stripping requires OFED version
171 4.3-1.5.0.0 and above or rdma-core version v18 and above.
173 - TSO (Transmit Segmentation Offload) is supported in OFED version
179 This driver relies on external libraries and kernel drivers for resources
180 allocations and initialization. The following dependencies are not part of
181 DPDK and must be installed separately:
183 - **libibverbs** (provided by rdma-core package)
185 User space verbs framework used by librte_pmd_mlx4. This library provides
186 a generic interface between the kernel and low-level user space drivers
189 It allows slow and privileged operations (context initialization, hardware
190 resources allocations) to be managed by the kernel and fast operations to
191 never leave user space.
193 - **libmlx4** (provided by rdma-core package)
195 Low-level user space driver library for Mellanox ConnectX-3 devices,
196 it is automatically loaded by libibverbs.
198 This library basically implements send/receive calls to the hardware
203 They provide the kernel-side verbs API and low level device drivers that
204 manage actual hardware initialization and resources sharing with user
207 Unlike most other PMDs, these modules must remain loaded and bound to
210 - mlx4_core: hardware driver managing Mellanox ConnectX-3 devices.
211 - mlx4_en: Ethernet device driver that provides kernel network interfaces.
212 - mlx4_ib: InifiniBand device driver.
213 - ib_uverbs: user space driver for verbs (entry point for libibverbs).
215 - **Firmware update**
217 Mellanox OFED releases include firmware updates for ConnectX-3 adapters.
219 Because each release provides new features, these updates must be applied to
220 match the kernel modules and libraries they come with.
224 Both libraries are BSD and GPL licensed. Linux kernel modules are GPL
227 Depending on system constraints and user preferences either RDMA core library
228 with a recent enough Linux kernel release (recommended) or Mellanox OFED,
229 which provides compatibility with older releases.
231 Current RDMA core package and Linux kernel (recommended)
232 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
234 - Minimal Linux kernel version: 4.14.
235 - Minimal RDMA core version: v15 (see `RDMA core installation documentation`_).
237 - Starting with rdma-core v21, static libraries can be built::
240 CFLAGS=-fPIC cmake -DIN_PLACE=1 -DENABLE_STATIC=1 -GNinja ..
243 .. _`RDMA core installation documentation`: https://raw.githubusercontent.com/linux-rdma/rdma-core/master/README.md
245 If rdma-core libraries are built but not installed, DPDK makefile can link them,
246 thanks to these environment variables:
248 - ``EXTRA_CFLAGS=-I/path/to/rdma-core/build/include``
249 - ``EXTRA_LDFLAGS=-L/path/to/rdma-core/build/lib``
250 - ``PKG_CONFIG_PATH=/path/to/rdma-core/build/lib/pkgconfig``
252 .. _Mellanox_OFED_as_a_fallback:
254 Mellanox OFED as a fallback
255 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
257 - `Mellanox OFED`_ version: **4.4, 4.5**.
258 - firmware version: **2.42.5000** and above.
260 .. _`Mellanox OFED`: http://www.mellanox.com/page/products_dyn?product_family=26&mtag=linux_sw_drivers
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`_.
268 Installing Mellanox OFED
269 ^^^^^^^^^^^^^^^^^^^^^^^^
271 1. Download latest Mellanox OFED.
273 2. Install the required libraries and kernel modules either by installing
274 only the required set, or by installing the entire Mellanox OFED:
278 .. code-block:: console
280 ./mlnxofedinstall --dpdk --upstream-libs
282 For SR-IOV hypervisors use:
284 .. code-block:: console
286 ./mlnxofedinstall --dpdk --upstream-libs --enable-sriov --hypervisor
288 For SR-IOV virtual machine use:
290 .. code-block:: console
292 ./mlnxofedinstall --dpdk --upstream-libs --guest
294 3. Verify the firmware is the correct one:
296 .. code-block:: console
300 4. Set all ports links to Ethernet, follow instructions on the screen:
302 .. code-block:: console
306 5. Continue with :ref:`section 2 of the Quick Start Guide <QSG_2>`.
311 * Mellanox(R) ConnectX(R)-3 Pro 40G MCX354A-FCC_Ax (2*40G)
318 1. Set all ports links to Ethernet
320 .. code-block:: console
322 PCI=<NIC PCI address>
323 echo eth > "/sys/bus/pci/devices/$PCI/mlx4_port0"
324 echo eth > "/sys/bus/pci/devices/$PCI/mlx4_port1"
328 If using Mellanox OFED one can permanently set the port link
329 to Ethernet using connectx_port_config tool provided by it.
330 :ref:`Mellanox_OFED_as_a_fallback`:
334 2. In case of bare metal or hypervisor, configure optimized steering mode
335 by adding the following line to ``/etc/modprobe.d/mlx4_core.conf``:
337 .. code-block:: console
339 options mlx4_core log_num_mgm_entry_size=-7
343 If VLAN filtering is used, set log_num_mgm_entry_size=-1.
344 Performance degradation can occur on this case.
346 3. Restart the driver:
348 .. code-block:: console
350 /etc/init.d/openibd restart
354 .. code-block:: console
356 service openibd restart
358 4. Compile DPDK and you are ready to go. See instructions on
359 :ref:`Development Kit Build System <Development_Kit_Build_System>`
364 1. Verify the optimized steering mode is configured:
366 .. code-block:: console
368 cat /sys/module/mlx4_core/parameters/log_num_mgm_entry_size
370 2. Use the CPU near local NUMA node to which the PCIe adapter is connected,
371 for better performance. For VMs, verify that the right CPU
372 and NUMA node are pinned according to the above. Run:
374 .. code-block:: console
378 to identify the NUMA node to which the PCIe adapter is connected.
380 3. If more than one adapter is used, and root complex capabilities allow
381 to put both adapters on the same NUMA node without PCI bandwidth degradation,
382 it is recommended to locate both adapters on the same NUMA node.
383 This in order to forward packets from one to the other without
384 NUMA performance penalty.
386 4. Disable pause frames:
388 .. code-block:: console
390 ethtool -A <netdev> rx off tx off
392 5. Verify IO non-posted prefetch is disabled by default. This can be checked
393 via the BIOS configuration. Please contact you server provider for more
394 information about the settings.
398 On some machines, depends on the machine integrator, it is beneficial
399 to set the PCI max read request parameter to 1K. This can be
400 done in the following way:
402 To query the read request size use:
404 .. code-block:: console
406 setpci -s <NIC PCI address> 68.w
408 If the output is different than 3XXX, set it by:
410 .. code-block:: console
412 setpci -s <NIC PCI address> 68.w=3XXX
414 The XXX can be different on different systems. Make sure to configure
415 according to the setpci output.
417 6. To minimize overhead of searching Memory Regions:
419 - '--socket-mem' is recommended to pin memory by predictable amount.
420 - Configure per-lcore cache when creating Mempools for packet buffer.
421 - Refrain from dynamically allocating/freeing memory in run-time.
426 This section demonstrates how to launch **testpmd** with Mellanox ConnectX-3
427 devices managed by librte_pmd_mlx4.
429 #. Load the kernel modules:
431 .. code-block:: console
433 modprobe -a ib_uverbs mlx4_en mlx4_core mlx4_ib
435 Alternatively if MLNX_OFED is fully installed, the following script can
438 .. code-block:: console
440 /etc/init.d/openibd restart
444 User space I/O kernel modules (uio and igb_uio) are not used and do
445 not have to be loaded.
447 #. Make sure Ethernet interfaces are in working order and linked to kernel
448 verbs. Related sysfs entries should be present:
450 .. code-block:: console
452 ls -d /sys/class/net/*/device/infiniband_verbs/uverbs* | cut -d / -f 5
456 .. code-block:: console
463 #. Optionally, retrieve their PCI bus addresses for whitelisting:
465 .. code-block:: console
468 for intf in eth2 eth3 eth4 eth5;
470 (cd "/sys/class/net/${intf}/device/" && pwd -P);
473 sed -n 's,.*/\(.*\),-w \1,p'
477 .. code-block:: console
486 There are only two distinct PCI bus addresses because the Mellanox
487 ConnectX-3 adapters installed on this system are dual port.
489 #. Request huge pages:
491 .. code-block:: console
493 echo 1024 > /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages/nr_hugepages
495 #. Start testpmd with basic parameters:
497 .. code-block:: console
499 testpmd -l 8-15 -n 4 -w 0000:83:00.0 -w 0000:84:00.0 -- --rxq=2 --txq=2 -i
503 .. code-block:: console
506 EAL: PCI device 0000:83:00.0 on NUMA socket 1
507 EAL: probe driver: 15b3:1007 librte_pmd_mlx4
508 PMD: librte_pmd_mlx4: PCI information matches, using device "mlx4_0" (VF: false)
509 PMD: librte_pmd_mlx4: 2 port(s) detected
510 PMD: librte_pmd_mlx4: port 1 MAC address is 00:02:c9:b5:b7:50
511 PMD: librte_pmd_mlx4: port 2 MAC address is 00:02:c9:b5:b7:51
512 EAL: PCI device 0000:84:00.0 on NUMA socket 1
513 EAL: probe driver: 15b3:1007 librte_pmd_mlx4
514 PMD: librte_pmd_mlx4: PCI information matches, using device "mlx4_1" (VF: false)
515 PMD: librte_pmd_mlx4: 2 port(s) detected
516 PMD: librte_pmd_mlx4: port 1 MAC address is 00:02:c9:b5:ba:b0
517 PMD: librte_pmd_mlx4: port 2 MAC address is 00:02:c9:b5:ba:b1
518 Interactive-mode selected
519 Configuring Port 0 (socket 0)
520 PMD: librte_pmd_mlx4: 0x867d60: TX queues number update: 0 -> 2
521 PMD: librte_pmd_mlx4: 0x867d60: RX queues number update: 0 -> 2
522 Port 0: 00:02:C9:B5:B7:50
523 Configuring Port 1 (socket 0)
524 PMD: librte_pmd_mlx4: 0x867da0: TX queues number update: 0 -> 2
525 PMD: librte_pmd_mlx4: 0x867da0: RX queues number update: 0 -> 2
526 Port 1: 00:02:C9:B5:B7:51
527 Configuring Port 2 (socket 0)
528 PMD: librte_pmd_mlx4: 0x867de0: TX queues number update: 0 -> 2
529 PMD: librte_pmd_mlx4: 0x867de0: RX queues number update: 0 -> 2
530 Port 2: 00:02:C9:B5:BA:B0
531 Configuring Port 3 (socket 0)
532 PMD: librte_pmd_mlx4: 0x867e20: TX queues number update: 0 -> 2
533 PMD: librte_pmd_mlx4: 0x867e20: RX queues number update: 0 -> 2
534 Port 3: 00:02:C9:B5:BA:B1
535 Checking link statuses...
536 Port 0 Link Up - speed 10000 Mbps - full-duplex
537 Port 1 Link Up - speed 40000 Mbps - full-duplex
538 Port 2 Link Up - speed 10000 Mbps - full-duplex
539 Port 3 Link Up - speed 40000 Mbps - full-duplex