1 .. SPDX-License-Identifier: BSD-3-Clause
2 Copyright 2015 6WIND S.A.
3 Copyright 2015 Mellanox Technologies, Ltd
8 The MLX5 poll mode driver library (**librte_pmd_mlx5**) provides support
9 for **Mellanox ConnectX-4**, **Mellanox ConnectX-4 Lx** and **Mellanox
10 ConnectX-5** families of 10/25/40/50/100 Gb/s adapters as well as their
11 virtual functions (VF) in SR-IOV context.
13 Information and documentation about these adapters can be found on the
14 `Mellanox website <http://www.mellanox.com>`__. Help is also provided by the
15 `Mellanox community <http://community.mellanox.com/welcome>`__.
17 There is also a `section dedicated to this poll mode driver
18 <http://www.mellanox.com/page/products_dyn?product_family=209&mtag=pmd_for_dpdk>`__.
22 Due to external dependencies, this driver is disabled by default. It must
23 be enabled manually by setting ``CONFIG_RTE_LIBRTE_MLX5_PMD=y`` and
26 Implementation details
27 ----------------------
29 Besides its dependency on libibverbs (that implies libmlx5 and associated
30 kernel support), librte_pmd_mlx5 relies heavily on system calls for control
31 operations such as querying/updating the MTU and flow control parameters.
33 For security reasons and robustness, this driver only deals with virtual
34 memory addresses. The way resources allocations are handled by the kernel
35 combined with hardware specifications that allow it to handle virtual memory
36 addresses directly ensure that DPDK applications cannot access random
37 physical memory (or memory that does not belong to the current process).
39 This capability allows the PMD to coexist with kernel network interfaces
40 which remain functional, although they stop receiving unicast packets as
41 long as they share the same MAC address.
42 This means legacy linux control tools (for example: ethtool, ifconfig and
43 more) can operate on the same network interfaces that owned by the DPDK
46 Enabling librte_pmd_mlx5 causes DPDK applications to be linked against
52 - Multi arch support: x86_64, POWER8, ARMv8.
53 - Multiple TX and RX queues.
54 - Support for scattered TX and RX frames.
55 - IPv4, IPv6, TCPv4, TCPv6, UDPv4 and UDPv6 RSS on any number of queues.
56 - Several RSS hash keys, one for each flow type.
57 - Configurable RETA table.
58 - Support for multiple MAC addresses.
62 - RX CRC stripping configuration.
64 - Multicast promiscuous mode.
65 - Hardware checksum offloads.
66 - Flow director (RTE_FDIR_MODE_PERFECT, RTE_FDIR_MODE_PERFECT_MAC_VLAN and
70 - KVM and VMware ESX SR-IOV modes are supported.
71 - RSS hash result is supported.
73 - Hardware checksum TX offload for VXLAN and GRE.
75 - Statistics query including Basic, Extended and per queue.
81 - Inner RSS for VXLAN frames is not supported yet.
82 - Hardware checksum RX offloads for VXLAN inner header are not supported yet.
83 - For secondary process:
85 - Forked secondary process not supported.
86 - All mempools must be initialized before rte_eth_dev_start().
88 - Flow pattern without any specific vlan will match for vlan packets as well:
90 When VLAN spec is not specified in the pattern, the matching rule will be created with VLAN as a wild card.
91 Meaning, the flow rule::
93 flow create 0 ingress pattern eth / vlan vid is 3 / ipv4 / end ...
95 Will only match vlan packets with vid=3. and the flow rules::
97 flow create 0 ingress pattern eth / ipv4 / end ...
101 flow create 0 ingress pattern eth / vlan / ipv4 / end ...
103 Will match any ipv4 packet (VLAN included).
105 - A multi segment packet must have less than 6 segments in case the Tx burst function
106 is set to multi-packet send or Enhanced multi-packet send. Otherwise it must have
107 less than 50 segments.
108 - Count action for RTE flow is **only supported in Mellanox OFED**.
109 - Flows with a VXLAN Network Identifier equal (or ends to be equal)
110 to 0 are not supported.
111 - VXLAN TSO and checksum offloads are not supported on VM.
116 MLX5 supports various of methods to report statistics:
118 Port statistics can be queried using ``rte_eth_stats_get()``. The port statistics are through SW only and counts the number of packets received or sent successfully by the PMD.
120 Extended statistics can be queried using ``rte_eth_xstats_get()``. The extended statistics expose a wider set of counters counted by the device. The extended port statistics counts the number of packets received or sent successfully by the port. As Mellanox NICs are using the :ref:`Bifurcated Linux Driver <linux_gsg_linux_drivers>` those counters counts also packet received or sent by the Linux kernel. The counters with ``_phy`` suffix counts the total events on the physical port, therefore not valid for VF.
122 Finally per-flow statistics can by queried using ``rte_flow_query`` when attaching a count action for specific flow. The flow counter counts the number of packets received successfully by the port and match the specific flow.
130 These options can be modified in the ``.config`` file.
132 - ``CONFIG_RTE_LIBRTE_MLX5_PMD`` (default **n**)
134 Toggle compilation of librte_pmd_mlx5 itself.
136 - ``CONFIG_RTE_LIBRTE_MLX5_DLOPEN_DEPS`` (default **n**)
138 Build PMD with additional code to make it loadable without hard
139 dependencies on **libibverbs** nor **libmlx5**, which may not be installed
140 on the target system.
142 In this mode, their presence is still required for it to run properly,
143 however their absence won't prevent a DPDK application from starting (with
144 ``CONFIG_RTE_BUILD_SHARED_LIB`` disabled) and they won't show up as
145 missing with ``ldd(1)``.
147 It works by moving these dependencies to a purpose-built rdma-core "glue"
148 plug-in which must either be installed in a directory whose name is based
149 on ``CONFIG_RTE_EAL_PMD_PATH`` suffixed with ``-glue`` if set, or in a
150 standard location for the dynamic linker (e.g. ``/lib``) if left to the
151 default empty string (``""``).
153 This option has no performance impact.
155 - ``CONFIG_RTE_LIBRTE_MLX5_DEBUG`` (default **n**)
157 Toggle debugging code and stricter compilation flags. Enabling this option
158 adds additional run-time checks and debugging messages at the cost of
161 - ``CONFIG_RTE_LIBRTE_MLX5_TX_MP_CACHE`` (default **8**)
163 Maximum number of cached memory pools (MPs) per TX queue. Each MP from
164 which buffers are to be transmitted must be associated to memory regions
165 (MRs). This is a slow operation that must be cached.
167 This value is always 1 for RX queues since they use a single MP.
169 Environment variables
170 ~~~~~~~~~~~~~~~~~~~~~
174 A list of directories in which to search for the rdma-core "glue" plug-in,
175 separated by colons or semi-colons.
177 Only matters when compiled with ``CONFIG_RTE_LIBRTE_MLX5_DLOPEN_DEPS``
178 enabled and most useful when ``CONFIG_RTE_EAL_PMD_PATH`` is also set,
179 since ``LD_LIBRARY_PATH`` has no effect in this case.
181 - ``MLX5_PMD_ENABLE_PADDING``
183 Enables HW packet padding in PCI bus transactions.
185 When packet size is cache aligned and CRC stripping is enabled, 4 fewer
186 bytes are written to the PCI bus. Enabling padding makes such packets
189 In cases where PCI bandwidth is the bottleneck, padding can improve
192 This is disabled by default since this can also decrease performance for
193 unaligned packet sizes.
195 - ``MLX5_SHUT_UP_BF``
197 Configures HW Tx doorbell register as IO-mapped.
199 By default, the HW Tx doorbell is configured as a write-combining register.
200 The register would be flushed to HW usually when the write-combining buffer
201 becomes full, but it depends on CPU design.
203 Except for vectorized Tx burst routines, a write memory barrier is enforced
204 after updating the register so that the update can be immediately visible to
207 When vectorized Tx burst is called, the barrier is set only if the burst size
208 is not aligned to MLX5_VPMD_TX_MAX_BURST. However, setting this environmental
209 variable will bring better latency even though the maximum throughput can
212 Run-time configuration
213 ~~~~~~~~~~~~~~~~~~~~~~
215 - librte_pmd_mlx5 brings kernel network interfaces up during initialization
216 because it is affected by their state. Forcing them down prevents packets
219 - **ethtool** operations on related kernel interfaces also affect the PMD.
221 - ``rxq_cqe_comp_en`` parameter [int]
223 A nonzero value enables the compression of CQE on RX side. This feature
224 allows to save PCI bandwidth and improve performance. Enabled by default.
228 - x86_64 with ConnectX-4, ConnectX-4 LX and ConnectX-5.
229 - POWER8 and ARMv8 with ConnectX-4 LX and ConnectX-5.
231 - ``txq_inline`` parameter [int]
233 Amount of data to be inlined during TX operations. Improves latency.
234 Can improve PPS performance when PCI back pressure is detected and may be
235 useful for scenarios involving heavy traffic on many queues.
237 Because additional software logic is necessary to handle this mode, this
238 option should be used with care, as it can lower performance when back
239 pressure is not expected.
241 - ``txqs_min_inline`` parameter [int]
243 Enable inline send only when the number of TX queues is greater or equal
246 This option should be used in combination with ``txq_inline`` above.
248 On ConnectX-4, ConnectX-4 LX and ConnectX-5 without Enhanced MPW:
250 - Disabled by default.
251 - In case ``txq_inline`` is set recommendation is 4.
253 On ConnectX-5 with Enhanced MPW:
255 - Set to 8 by default.
257 - ``txq_mpw_en`` parameter [int]
259 A nonzero value enables multi-packet send (MPS) for ConnectX-4 Lx and
260 enhanced multi-packet send (Enhanced MPS) for ConnectX-5. MPS allows the
261 TX burst function to pack up multiple packets in a single descriptor
262 session in order to save PCI bandwidth and improve performance at the
263 cost of a slightly higher CPU usage. When ``txq_inline`` is set along
264 with ``txq_mpw_en``, TX burst function tries to copy entire packet data
265 on to TX descriptor instead of including pointer of packet only if there
266 is enough room remained in the descriptor. ``txq_inline`` sets
267 per-descriptor space for either pointers or inlined packets. In addition,
268 Enhanced MPS supports hybrid mode - mixing inlined packets and pointers
269 in the same descriptor.
271 This option cannot be used with certain offloads such as ``DEV_TX_OFFLOAD_TCP_TSO,
272 DEV_TX_OFFLOAD_VXLAN_TNL_TSO, DEV_TX_OFFLOAD_GRE_TNL_TSO, DEV_TX_OFFLOAD_VLAN_INSERT``.
273 When those offloads are requested the MPS send function will not be used.
275 It is currently only supported on the ConnectX-4 Lx and ConnectX-5
276 families of adapters. Enabled by default.
278 - ``txq_mpw_hdr_dseg_en`` parameter [int]
280 A nonzero value enables including two pointers in the first block of TX
281 descriptor. This can be used to lessen CPU load for memory copy.
283 Effective only when Enhanced MPS is supported. Disabled by default.
285 - ``txq_max_inline_len`` parameter [int]
287 Maximum size of packet to be inlined. This limits the size of packet to
288 be inlined. If the size of a packet is larger than configured value, the
289 packet isn't inlined even though there's enough space remained in the
290 descriptor. Instead, the packet is included with pointer.
292 Effective only when Enhanced MPS is supported. The default value is 256.
294 - ``tx_vec_en`` parameter [int]
296 A nonzero value enables Tx vector on ConnectX-5 only NIC if the number of
297 global Tx queues on the port is lesser than MLX5_VPMD_MIN_TXQS.
299 This option cannot be used with certain offloads such as ``DEV_TX_OFFLOAD_TCP_TSO,
300 DEV_TX_OFFLOAD_VXLAN_TNL_TSO, DEV_TX_OFFLOAD_GRE_TNL_TSO, DEV_TX_OFFLOAD_VLAN_INSERT``.
301 When those offloads are requested the MPS send function will not be used.
303 Enabled by default on ConnectX-5.
305 - ``rx_vec_en`` parameter [int]
307 A nonzero value enables Rx vector if the port is not configured in
308 multi-segment otherwise this parameter is ignored.
315 This driver relies on external libraries and kernel drivers for resources
316 allocations and initialization. The following dependencies are not part of
317 DPDK and must be installed separately:
321 User space Verbs framework used by librte_pmd_mlx5. This library provides
322 a generic interface between the kernel and low-level user space drivers
325 It allows slow and privileged operations (context initialization, hardware
326 resources allocations) to be managed by the kernel and fast operations to
327 never leave user space.
331 Low-level user space driver library for Mellanox ConnectX-4/ConnectX-5
332 devices, it is automatically loaded by libibverbs.
334 This library basically implements send/receive calls to the hardware
339 They provide the kernel-side Verbs API and low level device drivers that
340 manage actual hardware initialization and resources sharing with user
343 Unlike most other PMDs, these modules must remain loaded and bound to
346 - mlx5_core: hardware driver managing Mellanox ConnectX-4/ConnectX-5
347 devices and related Ethernet kernel network devices.
348 - mlx5_ib: InifiniBand device driver.
349 - ib_uverbs: user space driver for Verbs (entry point for libibverbs).
351 - **Firmware update**
353 Mellanox OFED releases include firmware updates for ConnectX-4/ConnectX-5
356 Because each release provides new features, these updates must be applied to
357 match the kernel modules and libraries they come with.
361 Both libraries are BSD and GPL licensed. Linux kernel modules are GPL
367 Either RDMA Core library with a recent enough Linux kernel release
368 (recommended) or Mellanox OFED, which provides compatibility with older
371 RMDA Core with Linux Kernel
372 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
374 - Minimal kernel version : v4.14 or the most recent 4.14-rc (see `Linux installation documentation`_)
375 - Minimal rdma-core version: v15+ commit 0c5f5765213a ("Merge pull request #227 from yishaih/tm")
376 (see `RDMA Core installation documentation`_)
378 .. _`Linux installation documentation`: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/plain/Documentation/admin-guide/README.rst
379 .. _`RDMA Core installation documentation`: https://raw.githubusercontent.com/linux-rdma/rdma-core/master/README.md
384 - Mellanox OFED version: **4.2, 4.3**.
387 - ConnectX-4: **12.21.1000** and above.
388 - ConnectX-4 Lx: **14.21.1000** and above.
389 - ConnectX-5: **16.21.1000** and above.
390 - ConnectX-5 Ex: **16.21.1000** and above.
392 While these libraries and kernel modules are available on OpenFabrics
393 Alliance's `website <https://www.openfabrics.org/>`__ and provided by package
394 managers on most distributions, this PMD requires Ethernet extensions that
395 may not be supported at the moment (this is a work in progress).
398 <http://www.mellanox.com/page/products_dyn?product_family=26&mtag=linux>`__
399 includes the necessary support and should be used in the meantime. For DPDK,
400 only libibverbs, libmlx5, mlnx-ofed-kernel packages and firmware updates are
401 required from that distribution.
405 Several versions of Mellanox OFED are available. Installing the version
406 this DPDK release was developed and tested against is strongly
407 recommended. Please check the `prerequisites`_.
412 * Mellanox(R) ConnectX(R)-4 10G MCX4111A-XCAT (1x10G)
413 * Mellanox(R) ConnectX(R)-4 10G MCX4121A-XCAT (2x10G)
414 * Mellanox(R) ConnectX(R)-4 25G MCX4111A-ACAT (1x25G)
415 * Mellanox(R) ConnectX(R)-4 25G MCX4121A-ACAT (2x25G)
416 * Mellanox(R) ConnectX(R)-4 40G MCX4131A-BCAT (1x40G)
417 * Mellanox(R) ConnectX(R)-4 40G MCX413A-BCAT (1x40G)
418 * Mellanox(R) ConnectX(R)-4 40G MCX415A-BCAT (1x40G)
419 * Mellanox(R) ConnectX(R)-4 50G MCX4131A-GCAT (1x50G)
420 * Mellanox(R) ConnectX(R)-4 50G MCX413A-GCAT (1x50G)
421 * Mellanox(R) ConnectX(R)-4 50G MCX414A-BCAT (2x50G)
422 * Mellanox(R) ConnectX(R)-4 50G MCX415A-GCAT (2x50G)
423 * Mellanox(R) ConnectX(R)-4 50G MCX416A-BCAT (2x50G)
424 * Mellanox(R) ConnectX(R)-4 50G MCX416A-GCAT (2x50G)
425 * Mellanox(R) ConnectX(R)-4 50G MCX415A-CCAT (1x100G)
426 * Mellanox(R) ConnectX(R)-4 100G MCX416A-CCAT (2x100G)
427 * Mellanox(R) ConnectX(R)-4 Lx 10G MCX4121A-XCAT (2x10G)
428 * Mellanox(R) ConnectX(R)-4 Lx 25G MCX4121A-ACAT (2x25G)
429 * Mellanox(R) ConnectX(R)-5 100G MCX556A-ECAT (2x100G)
430 * Mellanox(R) ConnectX(R)-5 Ex EN 100G MCX516A-CDAT (2x100G)
432 Quick Start Guide on OFED
433 -------------------------
435 1. Download latest Mellanox OFED. For more info check the `prerequisites`_.
438 2. Install the required libraries and kernel modules either by installing
439 only the required set, or by installing the entire Mellanox OFED:
441 .. code-block:: console
443 ./mlnxofedinstall --upstream-libs --dpdk
445 3. Verify the firmware is the correct one:
447 .. code-block:: console
451 4. Verify all ports links are set to Ethernet:
453 .. code-block:: console
455 mlxconfig -d <mst device> query | grep LINK_TYPE
459 Link types may have to be configured to Ethernet:
461 .. code-block:: console
463 mlxconfig -d <mst device> set LINK_TYPE_P1/2=1/2/3
465 * LINK_TYPE_P1=<1|2|3> , 1=Infiniband 2=Ethernet 3=VPI(auto-sense)
467 For hypervisors verify SR-IOV is enabled on the NIC:
469 .. code-block:: console
471 mlxconfig -d <mst device> query | grep SRIOV_EN
474 If needed, set enable the set the relevant fields:
476 .. code-block:: console
478 mlxconfig -d <mst device> set SRIOV_EN=1 NUM_OF_VFS=16
479 mlxfwreset -d <mst device> reset
481 5. Restart the driver:
483 .. code-block:: console
485 /etc/init.d/openibd restart
489 .. code-block:: console
491 service openibd restart
493 If link type was changed, firmware must be reset as well:
495 .. code-block:: console
497 mlxfwreset -d <mst device> reset
499 For hypervisors, after reset write the sysfs number of virtual functions
502 To dynamically instantiate a given number of virtual functions (VFs):
504 .. code-block:: console
506 echo [num_vfs] > /sys/class/infiniband/mlx5_0/device/sriov_numvfs
508 6. Compile DPDK and you are ready to go. See instructions on
509 :ref:`Development Kit Build System <Development_Kit_Build_System>`
514 1. Configure aggressive CQE Zipping for maximum performance:
516 .. code-block:: console
518 mlxconfig -d <mst device> s CQE_COMPRESSION=1
520 To set it back to the default CQE Zipping mode use:
522 .. code-block:: console
524 mlxconfig -d <mst device> s CQE_COMPRESSION=0
526 2. In case of virtualization:
528 - Make sure that hypervisor kernel is 3.16 or newer.
529 - Configure boot with ``iommu=pt``.
531 - Make sure to allocate a VM on huge pages.
532 - Make sure to set CPU pinning.
534 3. Use the CPU near local NUMA node to which the PCIe adapter is connected,
535 for better performance. For VMs, verify that the right CPU
536 and NUMA node are pinned according to the above. Run:
538 .. code-block:: console
542 to identify the NUMA node to which the PCIe adapter is connected.
544 4. If more than one adapter is used, and root complex capabilities allow
545 to put both adapters on the same NUMA node without PCI bandwidth degradation,
546 it is recommended to locate both adapters on the same NUMA node.
547 This in order to forward packets from one to the other without
548 NUMA performance penalty.
550 5. Disable pause frames:
552 .. code-block:: console
554 ethtool -A <netdev> rx off tx off
556 6. Verify IO non-posted prefetch is disabled by default. This can be checked
557 via the BIOS configuration. Please contact you server provider for more
558 information about the settings.
562 On some machines, depends on the machine integrator, it is beneficial
563 to set the PCI max read request parameter to 1K. This can be
564 done in the following way:
566 To query the read request size use:
568 .. code-block:: console
570 setpci -s <NIC PCI address> 68.w
572 If the output is different than 3XXX, set it by:
574 .. code-block:: console
576 setpci -s <NIC PCI address> 68.w=3XXX
578 The XXX can be different on different systems. Make sure to configure
579 according to the setpci output.
584 Compared to librte_pmd_mlx4 that implements a single RSS configuration per
585 port, librte_pmd_mlx5 supports per-protocol RSS configuration.
587 Since ``testpmd`` defaults to IP RSS mode and there is currently no
588 command-line parameter to enable additional protocols (UDP and TCP as well
589 as IP), the following commands must be entered from its CLI to get the same
590 behavior as librte_pmd_mlx4:
592 .. code-block:: console
595 > port config all rss all
601 This section demonstrates how to launch **testpmd** with Mellanox
602 ConnectX-4/ConnectX-5 devices managed by librte_pmd_mlx5.
604 #. Load the kernel modules:
606 .. code-block:: console
608 modprobe -a ib_uverbs mlx5_core mlx5_ib
610 Alternatively if MLNX_OFED is fully installed, the following script can
613 .. code-block:: console
615 /etc/init.d/openibd restart
619 User space I/O kernel modules (uio and igb_uio) are not used and do
620 not have to be loaded.
622 #. Make sure Ethernet interfaces are in working order and linked to kernel
623 verbs. Related sysfs entries should be present:
625 .. code-block:: console
627 ls -d /sys/class/net/*/device/infiniband_verbs/uverbs* | cut -d / -f 5
631 .. code-block:: console
638 #. Optionally, retrieve their PCI bus addresses for whitelisting:
640 .. code-block:: console
643 for intf in eth2 eth3 eth4 eth5;
645 (cd "/sys/class/net/${intf}/device/" && pwd -P);
648 sed -n 's,.*/\(.*\),-w \1,p'
652 .. code-block:: console
659 #. Request huge pages:
661 .. code-block:: console
663 echo 1024 > /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages/nr_hugepages
665 #. Start testpmd with basic parameters:
667 .. code-block:: console
669 testpmd -l 8-15 -n 4 -w 05:00.0 -w 05:00.1 -w 06:00.0 -w 06:00.1 -- --rxq=2 --txq=2 -i
673 .. code-block:: console
676 EAL: PCI device 0000:05:00.0 on NUMA socket 0
677 EAL: probe driver: 15b3:1013 librte_pmd_mlx5
678 PMD: librte_pmd_mlx5: PCI information matches, using device "mlx5_0" (VF: false)
679 PMD: librte_pmd_mlx5: 1 port(s) detected
680 PMD: librte_pmd_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:fe
681 EAL: PCI device 0000:05:00.1 on NUMA socket 0
682 EAL: probe driver: 15b3:1013 librte_pmd_mlx5
683 PMD: librte_pmd_mlx5: PCI information matches, using device "mlx5_1" (VF: false)
684 PMD: librte_pmd_mlx5: 1 port(s) detected
685 PMD: librte_pmd_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:ff
686 EAL: PCI device 0000:06:00.0 on NUMA socket 0
687 EAL: probe driver: 15b3:1013 librte_pmd_mlx5
688 PMD: librte_pmd_mlx5: PCI information matches, using device "mlx5_2" (VF: false)
689 PMD: librte_pmd_mlx5: 1 port(s) detected
690 PMD: librte_pmd_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:fa
691 EAL: PCI device 0000:06:00.1 on NUMA socket 0
692 EAL: probe driver: 15b3:1013 librte_pmd_mlx5
693 PMD: librte_pmd_mlx5: PCI information matches, using device "mlx5_3" (VF: false)
694 PMD: librte_pmd_mlx5: 1 port(s) detected
695 PMD: librte_pmd_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:fb
696 Interactive-mode selected
697 Configuring Port 0 (socket 0)
698 PMD: librte_pmd_mlx5: 0x8cba80: TX queues number update: 0 -> 2
699 PMD: librte_pmd_mlx5: 0x8cba80: RX queues number update: 0 -> 2
700 Port 0: E4:1D:2D:E7:0C:FE
701 Configuring Port 1 (socket 0)
702 PMD: librte_pmd_mlx5: 0x8ccac8: TX queues number update: 0 -> 2
703 PMD: librte_pmd_mlx5: 0x8ccac8: RX queues number update: 0 -> 2
704 Port 1: E4:1D:2D:E7:0C:FF
705 Configuring Port 2 (socket 0)
706 PMD: librte_pmd_mlx5: 0x8cdb10: TX queues number update: 0 -> 2
707 PMD: librte_pmd_mlx5: 0x8cdb10: RX queues number update: 0 -> 2
708 Port 2: E4:1D:2D:E7:0C:FA
709 Configuring Port 3 (socket 0)
710 PMD: librte_pmd_mlx5: 0x8ceb58: TX queues number update: 0 -> 2
711 PMD: librte_pmd_mlx5: 0x8ceb58: RX queues number update: 0 -> 2
712 Port 3: E4:1D:2D:E7:0C:FB
713 Checking link statuses...
714 Port 0 Link Up - speed 40000 Mbps - full-duplex
715 Port 1 Link Up - speed 40000 Mbps - full-duplex
716 Port 2 Link Up - speed 10000 Mbps - full-duplex
717 Port 3 Link Up - speed 10000 Mbps - full-duplex