1 .. SPDX-License-Identifier: BSD-3-Clause
2 Copyright 2015 6WIND S.A.
3 Copyright 2015 Mellanox Technologies, Ltd
5 .. include:: <isonum.txt>
10 The MLX5 poll mode driver library (**librte_net_mlx5**) provides support
11 for **Mellanox ConnectX-4**, **Mellanox ConnectX-4 Lx** , **Mellanox
12 ConnectX-5**, **Mellanox ConnectX-6**, **Mellanox ConnectX-6 Dx**, **Mellanox
13 ConnectX-6 Lx**, **Mellanox BlueField** and **Mellanox BlueField-2** families
14 of 10/25/40/50/100/200 Gb/s adapters as well as their virtual functions (VF)
17 Information and documentation about these adapters can be found on the
18 `Mellanox website <http://www.mellanox.com>`__. Help is also provided by the
19 `Mellanox community <http://community.mellanox.com/welcome>`__.
21 There is also a `section dedicated to this poll mode driver
22 <http://www.mellanox.com/page/products_dyn?product_family=209&mtag=pmd_for_dpdk>`__.
28 Besides its dependency on libibverbs (that implies libmlx5 and associated
29 kernel support), librte_net_mlx5 relies heavily on system calls for control
30 operations such as querying/updating the MTU and flow control parameters.
32 For security reasons and robustness, this driver only deals with virtual
33 memory addresses. The way resources allocations are handled by the kernel,
34 combined with hardware specifications that allow to handle virtual memory
35 addresses directly, ensure that DPDK applications cannot access random
36 physical memory (or memory that does not belong to the current process).
38 This capability allows the PMD to coexist with kernel network interfaces
39 which remain functional, although they stop receiving unicast packets as
40 long as they share the same MAC address.
41 This means legacy linux control tools (for example: ethtool, ifconfig and
42 more) can operate on the same network interfaces that owned by the DPDK
45 The PMD can use libibverbs and libmlx5 to access the device firmware
46 or directly the hardware components.
47 There are different levels of objects and bypassing abilities
48 to get the best performances:
50 - Verbs is a complete high-level generic API
51 - Direct Verbs is a device-specific API
52 - DevX allows to access firmware objects
53 - Direct Rules manages flow steering at low-level hardware layer
55 Enabling librte_net_mlx5 causes DPDK applications to be linked against
61 - Multi arch support: x86_64, POWER8, ARMv8, i686.
62 - Multiple TX and RX queues.
63 - Support for scattered TX frames.
64 - Advanced support for scattered Rx frames with tunable buffer attributes.
65 - IPv4, IPv6, TCPv4, TCPv6, UDPv4 and UDPv6 RSS on any number of queues.
66 - RSS using different combinations of fields: L3 only, L4 only or both,
67 and source only, destination only or both.
68 - Several RSS hash keys, one for each flow type.
69 - Default RSS operation with no hash key specification.
70 - Configurable RETA table.
71 - Link flow control (pause frame).
72 - Support for multiple MAC addresses.
76 - RX CRC stripping configuration.
77 - TX mbuf fast free offload.
78 - Promiscuous mode on PF and VF.
79 - Multicast promiscuous mode on PF and VF.
80 - Hardware checksum offloads.
81 - Flow director (RTE_FDIR_MODE_PERFECT, RTE_FDIR_MODE_PERFECT_MAC_VLAN and
83 - Flow API, including :ref:`flow_isolated_mode`.
85 - KVM and VMware ESX SR-IOV modes are supported.
86 - RSS hash result is supported.
87 - Hardware TSO for generic IP or UDP tunnel, including VXLAN and GRE.
88 - Hardware checksum Tx offload for generic IP or UDP tunnel, including VXLAN and GRE.
90 - Statistics query including Basic, Extended and per queue.
92 - Tunnel types: VXLAN, L3 VXLAN, VXLAN-GPE, GRE, MPLSoGRE, MPLSoUDP, IP-in-IP, Geneve, GTP.
93 - Tunnel HW offloads: packet type, inner/outer RSS, IP and UDP checksum verification.
94 - NIC HW offloads: encapsulation (vxlan, gre, mplsoudp, mplsogre), NAT, routing, TTL
95 increment/decrement, count, drop, mark. For details please see :ref:`mlx5_offloads_support`.
96 - Flow insertion rate of more then million flows per second, when using Direct Rules.
97 - Support for multiple rte_flow groups.
98 - Per packet no-inline hint flag to disable packet data copying into Tx descriptors.
101 - Multiple-thread flow insertion.
102 - Matching on GTP extension header with raw encap/decap action.
103 - Matching on Geneve TLV option header with raw encap/decap action.
104 - RSS support in sample action.
105 - E-Switch mirroring and jump.
106 - E-Switch mirroring and modify.
107 - 21844 flow priorities for ingress or egress flow groups greater than 0 and for any transfer
115 On Windows, the features are limited:
117 - Promiscuous mode is not supported
118 - The following rules are supported:
120 - IPv4/UDP with CVLAN filtering
121 - Unicast MAC filtering
123 - For secondary process:
125 - Forked secondary process not supported.
126 - External memory unregistered in EAL memseg list cannot be used for DMA
127 unless such memory has been registered by ``mlx5_mr_update_ext_mp()`` in
128 primary process and remapped to the same virtual address in secondary
129 process. If the external memory is registered by primary process but has
130 different virtual address in secondary process, unexpected error may happen.
132 - When using Verbs flow engine (``dv_flow_en`` = 0), flow pattern without any
133 specific VLAN will match for VLAN packets as well:
135 When VLAN spec is not specified in the pattern, the matching rule will be created with VLAN as a wild card.
136 Meaning, the flow rule::
138 flow create 0 ingress pattern eth / vlan vid is 3 / ipv4 / end ...
140 Will only match vlan packets with vid=3. and the flow rule::
142 flow create 0 ingress pattern eth / ipv4 / end ...
144 Will match any ipv4 packet (VLAN included).
146 - When using Verbs flow engine (``dv_flow_en`` = 0), multi-tagged(QinQ) match is not supported.
148 - When using DV flow engine (``dv_flow_en`` = 1), flow pattern with any VLAN specification will match only single-tagged packets unless the ETH item ``type`` field is 0x88A8 or the VLAN item ``has_more_vlan`` field is 1.
151 flow create 0 ingress pattern eth / ipv4 / end ...
153 Will match any ipv4 packet.
156 flow create 0 ingress pattern eth / vlan / end ...
157 flow create 0 ingress pattern eth has_vlan is 1 / end ...
158 flow create 0 ingress pattern eth type is 0x8100 / end ...
160 Will match single-tagged packets only, with any VLAN ID value.
163 flow create 0 ingress pattern eth type is 0x88A8 / end ...
164 flow create 0 ingress pattern eth / vlan has_more_vlan is 1 / end ...
166 Will match multi-tagged packets only, with any VLAN ID value.
168 - A flow pattern with 2 sequential VLAN items is not supported.
170 - VLAN pop offload command:
172 - Flow rules having a VLAN pop offload command as one of their actions and
173 are lacking a match on VLAN as one of their items are not supported.
174 - The command is not supported on egress traffic.
176 - VLAN push offload is not supported on ingress traffic.
178 - VLAN set PCP offload is not supported on existing headers.
180 - A multi segment packet must have not more segments than reported by dev_infos_get()
181 in tx_desc_lim.nb_seg_max field. This value depends on maximal supported Tx descriptor
182 size and ``txq_inline_min`` settings and may be from 2 (worst case forced by maximal
183 inline settings) to 58.
185 - Flows with a VXLAN Network Identifier equal (or ends to be equal)
186 to 0 are not supported.
188 - L3 VXLAN and VXLAN-GPE tunnels cannot be supported together with MPLSoGRE and MPLSoUDP.
190 - Match on Geneve header supports the following fields only:
197 - Match on Geneve TLV option is supported on the following fields:
204 Only one Class/Type/Length Geneve TLV option is supported per shared device.
205 Class/Type/Length fields must be specified as well as masks.
206 Class/Type/Length specified masks must be full.
207 Matching Geneve TLV option without specifying data is not supported.
208 Matching Geneve TLV option with ``data & mask == 0`` is not supported.
210 - VF: flow rules created on VF devices can only match traffic targeted at the
211 configured MAC addresses (see ``rte_eth_dev_mac_addr_add()``).
213 - Match on GTP tunnel header item supports the following fields only:
215 - v_pt_rsv_flags: E flag, S flag, PN flag
219 - Match on GTP extension header only for GTP PDU session container (next
220 extension header type = 0x85).
221 - Match on GTP extension header is not supported in group 0.
223 - No Tx metadata go to the E-Switch steering domain for the Flow group 0.
224 The flows within group 0 and set metadata action are rejected by hardware.
228 MAC addresses not already present in the bridge table of the associated
229 kernel network device will be added and cleaned up by the PMD when closing
230 the device. In case of ungraceful program termination, some entries may
231 remain present and should be removed manually by other means.
233 - Buffer split offload is supported with regular Rx burst routine only,
234 no MPRQ feature or vectorized code can be engaged.
236 - When Multi-Packet Rx queue is configured (``mprq_en``), a Rx packet can be
237 externally attached to a user-provided mbuf with having EXT_ATTACHED_MBUF in
238 ol_flags. As the mempool for the external buffer is managed by PMD, all the
239 Rx mbufs must be freed before the device is closed. Otherwise, the mempool of
240 the external buffers will be freed by PMD and the application which still
241 holds the external buffers may be corrupted.
243 - If Multi-Packet Rx queue is configured (``mprq_en``) and Rx CQE compression is
244 enabled (``rxq_cqe_comp_en``) at the same time, RSS hash result is not fully
245 supported. Some Rx packets may not have PKT_RX_RSS_HASH.
247 - IPv6 Multicast messages are not supported on VM, while promiscuous mode
248 and allmulticast mode are both set to off.
249 To receive IPv6 Multicast messages on VM, explicitly set the relevant
250 MAC address using rte_eth_dev_mac_addr_add() API.
252 - To support a mixed traffic pattern (some buffers from local host memory, some
253 buffers from other devices) with high bandwidth, a mbuf flag is used.
255 An application hints the PMD whether or not it should try to inline the
256 given mbuf data buffer. PMD should do the best effort to act upon this request.
258 The hint flag ``RTE_PMD_MLX5_FINE_GRANULARITY_INLINE`` is dynamic,
259 registered by application with rte_mbuf_dynflag_register(). This flag is
260 purely driver-specific and declared in PMD specific header ``rte_pmd_mlx5.h``,
261 which is intended to be used by the application.
263 To query the supported specific flags in runtime,
264 the function ``rte_pmd_mlx5_get_dyn_flag_names`` returns the array of
265 currently (over present hardware and configuration) supported specific flags.
266 The "not inline hint" feature operating flow is the following one:
269 - probe the devices, ports are created
270 - query the port capabilities
271 - if port supporting the feature is found
272 - register dynamic flag ``RTE_PMD_MLX5_FINE_GRANULARITY_INLINE``
273 - application starts the ports
274 - on ``dev_start()`` PMD checks whether the feature flag is registered and
275 enables the feature support in datapath
276 - application might set the registered flag bit in ``ol_flags`` field
277 of mbuf being sent and PMD will handle ones appropriately.
279 - The amount of descriptors in Tx queue may be limited by data inline settings.
280 Inline data require the more descriptor building blocks and overall block
281 amount may exceed the hardware supported limits. The application should
282 reduce the requested Tx size or adjust data inline settings with
283 ``txq_inline_max`` and ``txq_inline_mpw`` devargs keys.
285 - To provide the packet send scheduling on mbuf timestamps the ``tx_pp``
286 parameter should be specified.
287 When PMD sees the RTE_MBUF_DYNFLAG_TX_TIMESTAMP_NAME set on the packet
288 being sent it tries to synchronize the time of packet appearing on
289 the wire with the specified packet timestamp. It the specified one
290 is in the past it should be ignored, if one is in the distant future
291 it should be capped with some reasonable value (in range of seconds).
292 These specific cases ("too late" and "distant future") can be optionally
293 reported via device xstats to assist applications to detect the
294 time-related problems.
296 The timestamp upper "too-distant-future" limit
297 at the moment of invoking the Tx burst routine
298 can be estimated as ``tx_pp`` option (in nanoseconds) multiplied by 2^23.
299 Please note, for the testpmd txonly mode,
300 the limit is deduced from the expression::
302 (n_tx_descriptors / burst_size + 1) * inter_burst_gap
304 There is no any packet reordering according timestamps is supposed,
305 neither within packet burst, nor between packets, it is an entirely
306 application responsibility to generate packets and its timestamps
307 in desired order. The timestamps can be put only in the first packet
308 in the burst providing the entire burst scheduling.
310 - E-Switch decapsulation Flow:
312 - can be applied to PF port only.
313 - must specify VF port action (packet redirection from PF to VF).
314 - optionally may specify tunnel inner source and destination MAC addresses.
316 - E-Switch encapsulation Flow:
318 - can be applied to VF ports only.
319 - must specify PF port action (packet redirection from VF to PF).
323 - The input buffer, used as outer header, is not validated.
327 - The decapsulation is always done up to the outermost tunnel detected by the HW.
328 - The input buffer, providing the removal size, is not validated.
329 - The buffer size must match the length of the headers to be removed.
331 - ICMP(code/type/identifier/sequence number) / ICMP6(code/type) matching, IP-in-IP and MPLS flow matching are all
332 mutually exclusive features which cannot be supported together
333 (see :ref:`mlx5_firmware_config`).
337 - Requires DevX and DV flow to be enabled.
338 - KEEP_CRC offload cannot be supported with LRO.
339 - The first mbuf length, without head-room, must be big enough to include the
341 - Rx queue with LRO offload enabled, receiving a non-LRO packet, can forward
342 it with size limited to max LRO size, not to max RX packet length.
343 - LRO can be used with outer header of TCP packets of the standard format:
344 eth (with or without vlan) / ipv4 or ipv6 / tcp / payload
346 Other TCP packets (e.g. with MPLS label) received on Rx queue with LRO enabled, will be received with bad checksum.
347 - LRO packet aggregation is performed by HW only for packet size larger than
348 ``lro_min_mss_size``. This value is reported on device start, when debug
353 - ``DEV_RX_OFFLOAD_KEEP_CRC`` cannot be supported with decapsulation
354 for some NICs (such as ConnectX-6 Dx, ConnectX-6 Lx, and BlueField-2).
355 The capability bit ``scatter_fcs_w_decap_disable`` shows NIC support.
359 - fast free offload assumes the all mbufs being sent are originated from the
360 same memory pool and there is no any extra references to the mbufs (the
361 reference counter for each mbuf is equal 1 on tx_burst call). The latter
362 means there should be no any externally attached buffers in mbufs. It is
363 an application responsibility to provide the correct mbufs if the fast
364 free offload is engaged. The mlx5 PMD implicitly produces the mbufs with
365 externally attached buffers if MPRQ option is enabled, hence, the fast
366 free offload is neither supported nor advertised if there is MPRQ enabled.
370 - Supports ``RTE_FLOW_ACTION_TYPE_SAMPLE`` action only within NIC Rx and E-Switch steering domain.
371 - The E-Switch Sample flow must have the eswitch_manager VPORT destination (PF or ECPF) and no additional actions.
372 - For ConnectX-5, the ``RTE_FLOW_ACTION_TYPE_SAMPLE`` is typically used as first action in the E-Switch egress flow if with header modify or encapsulation actions.
376 - Supports the 'set' operation only for ``RTE_FLOW_ACTION_TYPE_MODIFY_FIELD`` action.
377 - Modification of an arbitrary place in a packet via the special ``RTE_FLOW_FIELD_START`` Field ID is not supported.
378 - Encapsulation levels are not supported, can modify outermost header fields only.
379 - Offsets must be 32-bits aligned, cannot skip past the boundary of a field.
381 - IPv6 header item 'proto' field, indicating the next header protocol, should
382 not be set as extension header.
383 In case the next header is an extension header, it should not be specified in
384 IPv6 header item 'proto' field.
385 The last extension header item 'next header' field can specify the following
386 header protocol type.
390 - Hairpin between two ports could only manual binding and explicit Tx flow mode. For single port hairpin, all the combinations of auto/manual binding and explicit/implicit Tx flow mode could be supported.
391 - Hairpin in switchdev SR-IOV mode is not supported till now.
396 MLX5 supports various methods to report statistics:
398 Port statistics can be queried using ``rte_eth_stats_get()``. The received and sent statistics are through SW only and counts the number of packets received or sent successfully by the PMD. The imissed counter is the amount of packets that could not be delivered to SW because a queue was full. Packets not received due to congestion in the bus or on the NIC can be queried via the rx_discards_phy xstats counter.
400 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.
402 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.
410 The ibverbs libraries can be linked with this PMD in a number of ways,
411 configured by the ``ibverbs_link`` build option:
413 - ``shared`` (default): the PMD depends on some .so files.
415 - ``dlopen``: Split the dependencies glue in a separate library
416 loaded when needed by dlopen.
417 It make dependencies on libibverbs and libmlx4 optional,
418 and has no performance impact.
420 - ``static``: Embed static flavor of the dependencies libibverbs and libmlx4
421 in the PMD shared library or the executable static binary.
423 Environment variables
424 ~~~~~~~~~~~~~~~~~~~~~
428 A list of directories in which to search for the rdma-core "glue" plug-in,
429 separated by colons or semi-colons.
431 - ``MLX5_SHUT_UP_BF``
433 Configures HW Tx doorbell register as IO-mapped.
435 By default, the HW Tx doorbell is configured as a write-combining register.
436 The register would be flushed to HW usually when the write-combining buffer
437 becomes full, but it depends on CPU design.
439 Except for vectorized Tx burst routines, a write memory barrier is enforced
440 after updating the register so that the update can be immediately visible to
443 When vectorized Tx burst is called, the barrier is set only if the burst size
444 is not aligned to MLX5_VPMD_TX_MAX_BURST. However, setting this environmental
445 variable will bring better latency even though the maximum throughput can
448 Run-time configuration
449 ~~~~~~~~~~~~~~~~~~~~~~
451 - librte_net_mlx5 brings kernel network interfaces up during initialization
452 because it is affected by their state. Forcing them down prevents packets
455 - **ethtool** operations on related kernel interfaces also affect the PMD.
460 In order to run as a non-root user,
461 some capabilities must be granted to the application::
463 setcap cap_sys_admin,cap_net_admin,cap_net_raw,cap_ipc_lock+ep <dpdk-app>
465 Below are the reasons of the need for each capability:
468 When using physical addresses (PA mode), with Linux >= 4.0,
469 for access to ``/proc/self/pagemap``.
472 For device configuration.
475 For raw ethernet queue allocation through kernel driver.
478 For DMA memory pinning.
483 - ``rxq_cqe_comp_en`` parameter [int]
485 A nonzero value enables the compression of CQE on RX side. This feature
486 allows to save PCI bandwidth and improve performance. Enabled by default.
487 Different compression formats are supported in order to achieve the best
488 performance for different traffic patterns. Default format depends on
489 Multi-Packet Rx queue configuration: Hash RSS format is used in case
490 MPRQ is disabled, Checksum format is used in case MPRQ is enabled.
492 Specifying 2 as a ``rxq_cqe_comp_en`` value selects Flow Tag format for
493 better compression rate in case of RTE Flow Mark traffic.
494 Specifying 3 as a ``rxq_cqe_comp_en`` value selects Checksum format.
495 Specifying 4 as a ``rxq_cqe_comp_en`` value selects L3/L4 Header format for
496 better compression rate in case of mixed TCP/UDP and IPv4/IPv6 traffic.
497 CQE compression format selection requires DevX to be enabled. If there is
498 no DevX enabled/supported the value is reset to 1 by default.
502 - x86_64 with ConnectX-4, ConnectX-4 Lx, ConnectX-5, ConnectX-6, ConnectX-6 Dx,
503 ConnectX-6 Lx, BlueField and BlueField-2.
504 - POWER9 and ARMv8 with ConnectX-4 Lx, ConnectX-5, ConnectX-6, ConnectX-6 Dx,
505 ConnectX-6 Lx, BlueField and BlueField-2.
507 - ``rxq_pkt_pad_en`` parameter [int]
509 A nonzero value enables padding Rx packet to the size of cacheline on PCI
510 transaction. This feature would waste PCI bandwidth but could improve
511 performance by avoiding partial cacheline write which may cause costly
512 read-modify-copy in memory transaction on some architectures. Disabled by
517 - x86_64 with ConnectX-4, ConnectX-4 Lx, ConnectX-5, ConnectX-6, ConnectX-6 Dx,
518 ConnectX-6 Lx, BlueField and BlueField-2.
519 - POWER8 and ARMv8 with ConnectX-4 Lx, ConnectX-5, ConnectX-6, ConnectX-6 Dx,
520 ConnectX-6 Lx, BlueField and BlueField-2.
522 - ``mprq_en`` parameter [int]
524 A nonzero value enables configuring Multi-Packet Rx queues. Rx queue is
525 configured as Multi-Packet RQ if the total number of Rx queues is
526 ``rxqs_min_mprq`` or more. Disabled by default.
528 Multi-Packet Rx Queue (MPRQ a.k.a Striding RQ) can further save PCIe bandwidth
529 by posting a single large buffer for multiple packets. Instead of posting a
530 buffers per a packet, one large buffer is posted in order to receive multiple
531 packets on the buffer. A MPRQ buffer consists of multiple fixed-size strides
532 and each stride receives one packet. MPRQ can improve throughput for
533 small-packet traffic.
535 When MPRQ is enabled, max_rx_pkt_len can be larger than the size of
536 user-provided mbuf even if DEV_RX_OFFLOAD_SCATTER isn't enabled. PMD will
537 configure large stride size enough to accommodate max_rx_pkt_len as long as
538 device allows. Note that this can waste system memory compared to enabling Rx
539 scatter and multi-segment packet.
541 - ``mprq_log_stride_num`` parameter [int]
543 Log 2 of the number of strides for Multi-Packet Rx queue. Configuring more
544 strides can reduce PCIe traffic further. If configured value is not in the
545 range of device capability, the default value will be set with a warning
546 message. The default value is 4 which is 16 strides per a buffer, valid only
547 if ``mprq_en`` is set.
549 The size of Rx queue should be bigger than the number of strides.
551 - ``mprq_log_stride_size`` parameter [int]
553 Log 2 of the size of a stride for Multi-Packet Rx queue. Configuring a smaller
554 stride size can save some memory and reduce probability of a depletion of all
555 available strides due to unreleased packets by an application. If configured
556 value is not in the range of device capability, the default value will be set
557 with a warning message. The default value is 11 which is 2048 bytes per a
558 stride, valid only if ``mprq_en`` is set. With ``mprq_log_stride_size`` set
559 it is possible for a packet to span across multiple strides. This mode allows
560 support of jumbo frames (9K) with MPRQ. The memcopy of some packets (or part
561 of a packet if Rx scatter is configured) may be required in case there is no
562 space left for a head room at the end of a stride which incurs some
565 - ``mprq_max_memcpy_len`` parameter [int]
567 The maximum length of packet to memcpy in case of Multi-Packet Rx queue. Rx
568 packet is mem-copied to a user-provided mbuf if the size of Rx packet is less
569 than or equal to this parameter. Otherwise, PMD will attach the Rx packet to
570 the mbuf by external buffer attachment - ``rte_pktmbuf_attach_extbuf()``.
571 A mempool for external buffers will be allocated and managed by PMD. If Rx
572 packet is externally attached, ol_flags field of the mbuf will have
573 EXT_ATTACHED_MBUF and this flag must be preserved. ``RTE_MBUF_HAS_EXTBUF()``
574 checks the flag. The default value is 128, valid only if ``mprq_en`` is set.
576 - ``rxqs_min_mprq`` parameter [int]
578 Configure Rx queues as Multi-Packet RQ if the total number of Rx queues is
579 greater or equal to this value. The default value is 12, valid only if
582 - ``txq_inline`` parameter [int]
584 Amount of data to be inlined during TX operations. This parameter is
585 deprecated and converted to the new parameter ``txq_inline_max`` providing
586 partial compatibility.
588 - ``txqs_min_inline`` parameter [int]
590 Enable inline data send only when the number of TX queues is greater or equal
593 This option should be used in combination with ``txq_inline_max`` and
594 ``txq_inline_mpw`` below and does not affect ``txq_inline_min`` settings above.
596 If this option is not specified the default value 16 is used for BlueField
597 and 8 for other platforms
599 The data inlining consumes the CPU cycles, so this option is intended to
600 auto enable inline data if we have enough Tx queues, which means we have
601 enough CPU cores and PCI bandwidth is getting more critical and CPU
602 is not supposed to be bottleneck anymore.
604 The copying data into WQE improves latency and can improve PPS performance
605 when PCI back pressure is detected and may be useful for scenarios involving
606 heavy traffic on many queues.
608 Because additional software logic is necessary to handle this mode, this
609 option should be used with care, as it may lower performance when back
610 pressure is not expected.
612 If inline data are enabled it may affect the maximal size of Tx queue in
613 descriptors because the inline data increase the descriptor size and
614 queue size limits supported by hardware may be exceeded.
616 - ``txq_inline_min`` parameter [int]
618 Minimal amount of data to be inlined into WQE during Tx operations. NICs
619 may require this minimal data amount to operate correctly. The exact value
620 may depend on NIC operation mode, requested offloads, etc. It is strongly
621 recommended to omit this parameter and use the default values. Anyway,
622 applications using this parameter should take into consideration that
623 specifying an inconsistent value may prevent the NIC from sending packets.
625 If ``txq_inline_min`` key is present the specified value (may be aligned
626 by the driver in order not to exceed the limits and provide better descriptor
627 space utilization) will be used by the driver and it is guaranteed that
628 requested amount of data bytes are inlined into the WQE beside other inline
629 settings. This key also may update ``txq_inline_max`` value (default
630 or specified explicitly in devargs) to reserve the space for inline data.
632 If ``txq_inline_min`` key is not present, the value may be queried by the
633 driver from the NIC via DevX if this feature is available. If there is no DevX
634 enabled/supported the value 18 (supposing L2 header including VLAN) is set
635 for ConnectX-4 and ConnectX-4 Lx, and 0 is set by default for ConnectX-5
636 and newer NICs. If packet is shorter the ``txq_inline_min`` value, the entire
639 For ConnectX-4 NIC, driver does not allow specifying value below 18
640 (minimal L2 header, including VLAN), error will be raised.
642 For ConnectX-4 Lx NIC, it is allowed to specify values below 18, but
643 it is not recommended and may prevent NIC from sending packets over
646 Please, note, this minimal data inlining disengages eMPW feature (Enhanced
647 Multi-Packet Write), because last one does not support partial packet inlining.
648 This is not very critical due to minimal data inlining is mostly required
649 by ConnectX-4 and ConnectX-4 Lx, these NICs do not support eMPW feature.
651 - ``txq_inline_max`` parameter [int]
653 Specifies the maximal packet length to be completely inlined into WQE
654 Ethernet Segment for ordinary SEND method. If packet is larger than specified
655 value, the packet data won't be copied by the driver at all, data buffer
656 is addressed with a pointer. If packet length is less or equal all packet
657 data will be copied into WQE. This may improve PCI bandwidth utilization for
658 short packets significantly but requires the extra CPU cycles.
660 The data inline feature is controlled by number of Tx queues, if number of Tx
661 queues is larger than ``txqs_min_inline`` key parameter, the inline feature
662 is engaged, if there are not enough Tx queues (which means not enough CPU cores
663 and CPU resources are scarce), data inline is not performed by the driver.
664 Assigning ``txqs_min_inline`` with zero always enables the data inline.
666 The default ``txq_inline_max`` value is 290. The specified value may be adjusted
667 by the driver in order not to exceed the limit (930 bytes) and to provide better
668 WQE space filling without gaps, the adjustment is reflected in the debug log.
669 Also, the default value (290) may be decreased in run-time if the large transmit
670 queue size is requested and hardware does not support enough descriptor
671 amount, in this case warning is emitted. If ``txq_inline_max`` key is
672 specified and requested inline settings can not be satisfied then error
675 - ``txq_inline_mpw`` parameter [int]
677 Specifies the maximal packet length to be completely inlined into WQE for
678 Enhanced MPW method. If packet is large the specified value, the packet data
679 won't be copied, and data buffer is addressed with pointer. If packet length
680 is less or equal, all packet data will be copied into WQE. This may improve PCI
681 bandwidth utilization for short packets significantly but requires the extra
684 The data inline feature is controlled by number of TX queues, if number of Tx
685 queues is larger than ``txqs_min_inline`` key parameter, the inline feature
686 is engaged, if there are not enough Tx queues (which means not enough CPU cores
687 and CPU resources are scarce), data inline is not performed by the driver.
688 Assigning ``txqs_min_inline`` with zero always enables the data inline.
690 The default ``txq_inline_mpw`` value is 268. The specified value may be adjusted
691 by the driver in order not to exceed the limit (930 bytes) and to provide better
692 WQE space filling without gaps, the adjustment is reflected in the debug log.
693 Due to multiple packets may be included to the same WQE with Enhanced Multi
694 Packet Write Method and overall WQE size is limited it is not recommended to
695 specify large values for the ``txq_inline_mpw``. Also, the default value (268)
696 may be decreased in run-time if the large transmit queue size is requested
697 and hardware does not support enough descriptor amount, in this case warning
698 is emitted. If ``txq_inline_mpw`` key is specified and requested inline
699 settings can not be satisfied then error will be raised.
701 - ``txqs_max_vec`` parameter [int]
703 Enable vectorized Tx only when the number of TX queues is less than or
704 equal to this value. This parameter is deprecated and ignored, kept
705 for compatibility issue to not prevent driver from probing.
707 - ``txq_mpw_hdr_dseg_en`` parameter [int]
709 A nonzero value enables including two pointers in the first block of TX
710 descriptor. The parameter is deprecated and ignored, kept for compatibility
713 - ``txq_max_inline_len`` parameter [int]
715 Maximum size of packet to be inlined. This limits the size of packet to
716 be inlined. If the size of a packet is larger than configured value, the
717 packet isn't inlined even though there's enough space remained in the
718 descriptor. Instead, the packet is included with pointer. This parameter
719 is deprecated and converted directly to ``txq_inline_mpw`` providing full
720 compatibility. Valid only if eMPW feature is engaged.
722 - ``txq_mpw_en`` parameter [int]
724 A nonzero value enables Enhanced Multi-Packet Write (eMPW) for ConnectX-5,
725 ConnectX-6, ConnectX-6 Dx, ConnectX-6 Lx, BlueField, BlueField-2.
726 eMPW allows the Tx burst function to pack up multiple packets
727 in a single descriptor session in order to save PCI bandwidth
728 and improve performance at the cost of a slightly higher CPU usage.
729 When ``txq_inline_mpw`` is set along with ``txq_mpw_en``,
730 Tx burst function copies entire packet data on to Tx descriptor
731 instead of including pointer of packet.
733 The Enhanced Multi-Packet Write feature is enabled by default if NIC supports
734 it, can be disabled by explicit specifying 0 value for ``txq_mpw_en`` option.
735 Also, if minimal data inlining is requested by non-zero ``txq_inline_min``
736 option or reported by the NIC, the eMPW feature is disengaged.
738 - ``tx_db_nc`` parameter [int]
740 The rdma core library can map doorbell register in two ways, depending on the
741 environment variable "MLX5_SHUT_UP_BF":
743 - As regular cached memory (usually with write combining attribute), if the
744 variable is either missing or set to zero.
745 - As non-cached memory, if the variable is present and set to not "0" value.
747 The type of mapping may slightly affect the Tx performance, the optimal choice
748 is strongly relied on the host architecture and should be deduced practically.
750 If ``tx_db_nc`` is set to zero, the doorbell is forced to be mapped to regular
751 memory (with write combining), the PMD will perform the extra write memory barrier
752 after writing to doorbell, it might increase the needed CPU clocks per packet
753 to send, but latency might be improved.
755 If ``tx_db_nc`` is set to one, the doorbell is forced to be mapped to non
756 cached memory, the PMD will not perform the extra write memory barrier
757 after writing to doorbell, on some architectures it might improve the
760 If ``tx_db_nc`` is set to two, the doorbell is forced to be mapped to regular
761 memory, the PMD will use heuristics to decide whether write memory barrier
762 should be performed. For bursts with size multiple of recommended one (64 pkts)
763 it is supposed the next burst is coming and no need to issue the extra memory
764 barrier (it is supposed to be issued in the next coming burst, at least after
765 descriptor writing). It might increase latency (on some hosts till next
766 packets transmit) and should be used with care.
768 If ``tx_db_nc`` is omitted or set to zero, the preset (if any) environment
769 variable "MLX5_SHUT_UP_BF" value is used. If there is no "MLX5_SHUT_UP_BF",
770 the default ``tx_db_nc`` value is zero for ARM64 hosts and one for others.
772 - ``tx_pp`` parameter [int]
774 If a nonzero value is specified the driver creates all necessary internal
775 objects to provide accurate packet send scheduling on mbuf timestamps.
776 The positive value specifies the scheduling granularity in nanoseconds,
777 the packet send will be accurate up to specified digits. The allowed range is
778 from 500 to 1 million of nanoseconds. The negative value specifies the module
779 of granularity and engages the special test mode the check the schedule rate.
780 By default (if the ``tx_pp`` is not specified) send scheduling on timestamps
783 - ``tx_skew`` parameter [int]
785 The parameter adjusts the send packet scheduling on timestamps and represents
786 the average delay between beginning of the transmitting descriptor processing
787 by the hardware and appearance of actual packet data on the wire. The value
788 should be provided in nanoseconds and is valid only if ``tx_pp`` parameter is
789 specified. The default value is zero.
791 - ``tx_vec_en`` parameter [int]
793 A nonzero value enables Tx vector on ConnectX-5, ConnectX-6, ConnectX-6 Dx,
794 ConnectX-6 Lx, BlueField and BlueField-2 NICs
795 if the number of global Tx queues on the port is less than ``txqs_max_vec``.
796 The parameter is deprecated and ignored.
798 - ``rx_vec_en`` parameter [int]
800 A nonzero value enables Rx vector if the port is not configured in
801 multi-segment otherwise this parameter is ignored.
805 - ``vf_nl_en`` parameter [int]
807 A nonzero value enables Netlink requests from the VF to add/remove MAC
808 addresses or/and enable/disable promiscuous/all multicast on the Netdevice.
809 Otherwise the relevant configuration must be run with Linux iproute2 tools.
810 This is a prerequisite to receive this kind of traffic.
812 Enabled by default, valid only on VF devices ignored otherwise.
814 - ``l3_vxlan_en`` parameter [int]
816 A nonzero value allows L3 VXLAN and VXLAN-GPE flow creation. To enable
817 L3 VXLAN or VXLAN-GPE, users has to configure firmware and enable this
818 parameter. This is a prerequisite to receive this kind of traffic.
822 - ``dv_xmeta_en`` parameter [int]
824 A nonzero value enables extensive flow metadata support if device is
825 capable and driver supports it. This can enable extensive support of
826 ``MARK`` and ``META`` item of ``rte_flow``. The newly introduced
827 ``SET_TAG`` and ``SET_META`` actions do not depend on ``dv_xmeta_en``.
829 There are some possible configurations, depending on parameter value:
831 - 0, this is default value, defines the legacy mode, the ``MARK`` and
832 ``META`` related actions and items operate only within NIC Tx and
833 NIC Rx steering domains, no ``MARK`` and ``META`` information crosses
834 the domain boundaries. The ``MARK`` item is 24 bits wide, the ``META``
835 item is 32 bits wide and match supported on egress only.
837 - 1, this engages extensive metadata mode, the ``MARK`` and ``META``
838 related actions and items operate within all supported steering domains,
839 including FDB, ``MARK`` and ``META`` information may cross the domain
840 boundaries. The ``MARK`` item is 24 bits wide, the ``META`` item width
841 depends on kernel and firmware configurations and might be 0, 16 or
842 32 bits. Within NIC Tx domain ``META`` data width is 32 bits for
843 compatibility, the actual width of data transferred to the FDB domain
844 depends on kernel configuration and may be vary. The actual supported
845 width can be retrieved in runtime by series of rte_flow_validate()
848 - 2, this engages extensive metadata mode, the ``MARK`` and ``META``
849 related actions and items operate within all supported steering domains,
850 including FDB, ``MARK`` and ``META`` information may cross the domain
851 boundaries. The ``META`` item is 32 bits wide, the ``MARK`` item width
852 depends on kernel and firmware configurations and might be 0, 16 or
853 24 bits. The actual supported width can be retrieved in runtime by
854 series of rte_flow_validate() trials.
856 - 3, this engages tunnel offload mode. In E-Switch configuration, that
857 mode implicitly activates ``dv_xmeta_en=1``.
859 +------+-----------+-----------+-------------+-------------+
860 | Mode | ``MARK`` | ``META`` | ``META`` Tx | FDB/Through |
861 +======+===========+===========+=============+=============+
862 | 0 | 24 bits | 32 bits | 32 bits | no |
863 +------+-----------+-----------+-------------+-------------+
864 | 1 | 24 bits | vary 0-32 | 32 bits | yes |
865 +------+-----------+-----------+-------------+-------------+
866 | 2 | vary 0-24 | 32 bits | 32 bits | yes |
867 +------+-----------+-----------+-------------+-------------+
869 If there is no E-Switch configuration the ``dv_xmeta_en`` parameter is
870 ignored and the device is configured to operate in legacy mode (0).
872 Disabled by default (set to 0).
874 The Direct Verbs/Rules (engaged with ``dv_flow_en`` = 1) supports all
875 of the extensive metadata features. The legacy Verbs supports FLAG and
876 MARK metadata actions over NIC Rx steering domain only.
878 Setting META value to zero in flow action means there is no item provided
879 and receiving datapath will not report in mbufs the metadata are present.
880 Setting MARK value to zero in flow action means the zero FDIR ID value
881 will be reported on packet receiving.
883 For the MARK action the last 16 values in the full range are reserved for
884 internal PMD purposes (to emulate FLAG action). The valid range for the
885 MARK action values is 0-0xFFEF for the 16-bit mode and 0-xFFFFEF
886 for the 24-bit mode, the flows with the MARK action value outside
887 the specified range will be rejected.
889 - ``dv_flow_en`` parameter [int]
891 A nonzero value enables the DV flow steering assuming it is supported
892 by the driver (RDMA Core library version is rdma-core-24.0 or higher).
894 Enabled by default if supported.
896 - ``dv_esw_en`` parameter [int]
898 A nonzero value enables E-Switch using Direct Rules.
900 Enabled by default if supported.
902 - ``lacp_by_user`` parameter [int]
904 A nonzero value enables the control of LACP traffic by the user application.
905 When a bond exists in the driver, by default it should be managed by the
906 kernel and therefore LACP traffic should be steered to the kernel.
907 If this devarg is set to 1 it will allow the user to manage the bond by
908 itself and not steer LACP traffic to the kernel.
910 Disabled by default (set to 0).
912 - ``mr_ext_memseg_en`` parameter [int]
914 A nonzero value enables extending memseg when registering DMA memory. If
915 enabled, the number of entries in MR (Memory Region) lookup table on datapath
916 is minimized and it benefits performance. On the other hand, it worsens memory
917 utilization because registered memory is pinned by kernel driver. Even if a
918 page in the extended chunk is freed, that doesn't become reusable until the
919 entire memory is freed.
923 - ``representor`` parameter [list]
925 This parameter can be used to instantiate DPDK Ethernet devices from
926 existing port (or VF) representors configured on the device.
928 It is a standard parameter whose format is described in
929 :ref:`ethernet_device_standard_device_arguments`.
931 For instance, to probe port representors 0 through 2::
935 - ``max_dump_files_num`` parameter [int]
937 The maximum number of files per PMD entity that may be created for debug information.
938 The files will be created in /var/log directory or in current directory.
940 set to 128 by default.
942 - ``lro_timeout_usec`` parameter [int]
944 The maximum allowed duration of an LRO session, in micro-seconds.
945 PMD will set the nearest value supported by HW, which is not bigger than
946 the input ``lro_timeout_usec`` value.
947 If this parameter is not specified, by default PMD will set
948 the smallest value supported by HW.
950 - ``hp_buf_log_sz`` parameter [int]
952 The total data buffer size of a hairpin queue (logarithmic form), in bytes.
953 PMD will set the data buffer size to 2 ** ``hp_buf_log_sz``, both for RX & TX.
954 The capacity of the value is specified by the firmware and the initialization
955 will get a failure if it is out of scope.
956 The range of the value is from 11 to 19 right now, and the supported frame
957 size of a single packet for hairpin is from 512B to 128KB. It might change if
958 different firmware release is being used. By using a small value, it could
959 reduce memory consumption but not work with a large frame. If the value is
960 too large, the memory consumption will be high and some potential performance
961 degradation will be introduced.
962 By default, the PMD will set this value to 16, which means that 9KB jumbo
963 frames will be supported.
965 - ``reclaim_mem_mode`` parameter [int]
967 Cache some resources in flow destroy will help flow recreation more efficient.
968 While some systems may require the all the resources can be reclaimed after
970 The parameter ``reclaim_mem_mode`` provides the option for user to configure
971 if the resource cache is needed or not.
973 There are three options to choose:
975 - 0. It means the flow resources will be cached as usual. The resources will
976 be cached, helpful with flow insertion rate.
978 - 1. It will only enable the DPDK PMD level resources reclaim.
980 - 2. Both DPDK PMD level and rdma-core low level will be configured as
983 By default, the PMD will set this value to 0.
985 - ``sys_mem_en`` parameter [int]
987 A non-zero value enables the PMD memory management allocating memory
988 from system by default, without explicit rte memory flag.
990 By default, the PMD will set this value to 0.
992 - ``decap_en`` parameter [int]
994 Some devices do not support FCS (frame checksum) scattering for
995 tunnel-decapsulated packets.
996 If set to 0, this option forces the FCS feature and rejects tunnel
997 decapsulation in the flow engine for such devices.
999 By default, the PMD will set this value to 1.
1001 .. _mlx5_firmware_config:
1003 Firmware configuration
1004 ~~~~~~~~~~~~~~~~~~~~~~
1006 Firmware features can be configured as key/value pairs.
1008 The command to set a value is::
1010 mlxconfig -d <device> set <key>=<value>
1012 The command to query a value is::
1014 mlxconfig -d <device> query | grep <key>
1016 The device name for the command ``mlxconfig`` can be either the PCI address,
1017 or the mst device name found with::
1021 Below are some firmware configurations listed.
1027 value: 1=Infiniband 2=Ethernet 3=VPI(auto-sense)
1033 - maximum number of SR-IOV virtual functions::
1037 - enable DevX (required by Direct Rules and other features)::
1041 - aggressive CQE zipping::
1045 - L3 VXLAN and VXLAN-GPE destination UDP port::
1048 IP_OVER_VXLAN_PORT=<udp dport>
1050 - enable VXLAN-GPE tunnel flow matching::
1052 FLEX_PARSER_PROFILE_ENABLE=0
1054 FLEX_PARSER_PROFILE_ENABLE=2
1056 - enable IP-in-IP tunnel flow matching::
1058 FLEX_PARSER_PROFILE_ENABLE=0
1060 - enable MPLS flow matching::
1062 FLEX_PARSER_PROFILE_ENABLE=1
1064 - enable ICMP(code/type/identifier/sequence number) / ICMP6(code/type) fields matching::
1066 FLEX_PARSER_PROFILE_ENABLE=2
1068 - enable Geneve flow matching::
1070 FLEX_PARSER_PROFILE_ENABLE=0
1072 FLEX_PARSER_PROFILE_ENABLE=1
1074 - enable Geneve TLV option flow matching::
1076 FLEX_PARSER_PROFILE_ENABLE=0
1078 - enable GTP flow matching::
1080 FLEX_PARSER_PROFILE_ENABLE=3
1082 - enable eCPRI flow matching::
1084 FLEX_PARSER_PROFILE_ENABLE=4
1090 This driver relies on external libraries and kernel drivers for resources
1091 allocations and initialization. The following dependencies are not part of
1092 DPDK and must be installed separately:
1096 User space Verbs framework used by librte_net_mlx5. This library provides
1097 a generic interface between the kernel and low-level user space drivers
1100 It allows slow and privileged operations (context initialization, hardware
1101 resources allocations) to be managed by the kernel and fast operations to
1102 never leave user space.
1106 Low-level user space driver library for Mellanox
1107 ConnectX-4/ConnectX-5/ConnectX-6/BlueField devices, it is automatically loaded
1110 This library basically implements send/receive calls to the hardware
1113 - **Kernel modules**
1115 They provide the kernel-side Verbs API and low level device drivers that
1116 manage actual hardware initialization and resources sharing with user
1119 Unlike most other PMDs, these modules must remain loaded and bound to
1122 - mlx5_core: hardware driver managing Mellanox
1123 ConnectX-4/ConnectX-5/ConnectX-6/BlueField devices and related Ethernet kernel
1125 - mlx5_ib: InifiniBand device driver.
1126 - ib_uverbs: user space driver for Verbs (entry point for libibverbs).
1128 - **Firmware update**
1130 Mellanox OFED/EN releases include firmware updates for
1131 ConnectX-4/ConnectX-5/ConnectX-6/BlueField adapters.
1133 Because each release provides new features, these updates must be applied to
1134 match the kernel modules and libraries they come with.
1138 Both libraries are BSD and GPL licensed. Linux kernel modules are GPL
1144 Either RDMA Core library with a recent enough Linux kernel release
1145 (recommended) or Mellanox OFED/EN, which provides compatibility with older
1148 RDMA Core with Linux Kernel
1149 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
1151 - Minimal kernel version : v4.14 or the most recent 4.14-rc (see `Linux installation documentation`_)
1152 - Minimal rdma-core version: v15+ commit 0c5f5765213a ("Merge pull request #227 from yishaih/tm")
1153 (see `RDMA Core installation documentation`_)
1154 - When building for i686 use:
1156 - rdma-core version 18.0 or above built with 32bit support.
1157 - Kernel version 4.14.41 or above.
1159 - Starting with rdma-core v21, static libraries can be built::
1162 CFLAGS=-fPIC cmake -DIN_PLACE=1 -DENABLE_STATIC=1 -GNinja ..
1165 .. _`Linux installation documentation`: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/plain/Documentation/admin-guide/README.rst
1166 .. _`RDMA Core installation documentation`: https://raw.githubusercontent.com/linux-rdma/rdma-core/master/README.md
1172 - Mellanox OFED version: **4.5** and above /
1173 Mellanox EN version: **4.5** and above
1176 - ConnectX-4: **12.21.1000** and above.
1177 - ConnectX-4 Lx: **14.21.1000** and above.
1178 - ConnectX-5: **16.21.1000** and above.
1179 - ConnectX-5 Ex: **16.21.1000** and above.
1180 - ConnectX-6: **20.27.0090** and above.
1181 - ConnectX-6 Dx: **22.27.0090** and above.
1182 - BlueField: **18.25.1010** and above.
1184 While these libraries and kernel modules are available on OpenFabrics
1185 Alliance's `website <https://www.openfabrics.org/>`__ and provided by package
1186 managers on most distributions, this PMD requires Ethernet extensions that
1187 may not be supported at the moment (this is a work in progress).
1190 <http://www.mellanox.com/page/products_dyn?product_family=26&mtag=linux>`__ and
1192 <http://www.mellanox.com/page/products_dyn?product_family=27&mtag=linux>`__
1193 include the necessary support and should be used in the meantime. For DPDK,
1194 only libibverbs, libmlx5, mlnx-ofed-kernel packages and firmware updates are
1195 required from that distribution.
1199 Several versions of Mellanox OFED/EN are available. Installing the version
1200 this DPDK release was developed and tested against is strongly
1201 recommended. Please check the `linux prerequisites`_.
1203 Windows Prerequisites
1204 ---------------------
1206 This driver relies on external libraries and kernel drivers for resources
1207 allocations and initialization. The dependencies in the following sub-sections
1208 are not part of DPDK, and must be installed separately.
1210 Compilation Prerequisites
1211 ~~~~~~~~~~~~~~~~~~~~~~~~~
1213 DevX SDK installation
1214 ^^^^^^^^^^^^^^^^^^^^^
1216 The DevX SDK must be installed on the machine building the Windows PMD.
1217 Additional information can be found at
1218 `How to Integrate Windows DevX in Your Development Environment
1219 <https://docs.mellanox.com/display/winof2v250/RShim+Drivers+and+Usage#RShimDriversandUsage-DevXInterface>`__.
1221 Runtime Prerequisites
1222 ~~~~~~~~~~~~~~~~~~~~~
1224 WinOF2 version 2.60 or higher must be installed on the machine.
1229 The driver can be downloaded from the following site:
1231 <https://www.mellanox.com/products/adapter-software/ethernet/windows/winof-2>`__
1236 DevX for Windows must be enabled in the Windows registry.
1237 The keys ``DevxEnabled`` and ``DevxFsRules`` must be set.
1238 Additional information can be found in the WinOF2 user manual.
1243 The following Mellanox device families are supported by the same mlx5 driver:
1255 Below are detailed device names:
1257 * Mellanox\ |reg| ConnectX\ |reg|-4 10G MCX4111A-XCAT (1x10G)
1258 * Mellanox\ |reg| ConnectX\ |reg|-4 10G MCX412A-XCAT (2x10G)
1259 * Mellanox\ |reg| ConnectX\ |reg|-4 25G MCX4111A-ACAT (1x25G)
1260 * Mellanox\ |reg| ConnectX\ |reg|-4 25G MCX412A-ACAT (2x25G)
1261 * Mellanox\ |reg| ConnectX\ |reg|-4 40G MCX413A-BCAT (1x40G)
1262 * Mellanox\ |reg| ConnectX\ |reg|-4 40G MCX4131A-BCAT (1x40G)
1263 * Mellanox\ |reg| ConnectX\ |reg|-4 40G MCX415A-BCAT (1x40G)
1264 * Mellanox\ |reg| ConnectX\ |reg|-4 50G MCX413A-GCAT (1x50G)
1265 * Mellanox\ |reg| ConnectX\ |reg|-4 50G MCX4131A-GCAT (1x50G)
1266 * Mellanox\ |reg| ConnectX\ |reg|-4 50G MCX414A-BCAT (2x50G)
1267 * Mellanox\ |reg| ConnectX\ |reg|-4 50G MCX415A-GCAT (1x50G)
1268 * Mellanox\ |reg| ConnectX\ |reg|-4 50G MCX416A-BCAT (2x50G)
1269 * Mellanox\ |reg| ConnectX\ |reg|-4 50G MCX416A-GCAT (2x50G)
1270 * Mellanox\ |reg| ConnectX\ |reg|-4 50G MCX415A-CCAT (1x100G)
1271 * Mellanox\ |reg| ConnectX\ |reg|-4 100G MCX416A-CCAT (2x100G)
1272 * Mellanox\ |reg| ConnectX\ |reg|-4 Lx 10G MCX4111A-XCAT (1x10G)
1273 * Mellanox\ |reg| ConnectX\ |reg|-4 Lx 10G MCX4121A-XCAT (2x10G)
1274 * Mellanox\ |reg| ConnectX\ |reg|-4 Lx 25G MCX4111A-ACAT (1x25G)
1275 * Mellanox\ |reg| ConnectX\ |reg|-4 Lx 25G MCX4121A-ACAT (2x25G)
1276 * Mellanox\ |reg| ConnectX\ |reg|-4 Lx 40G MCX4131A-BCAT (1x40G)
1277 * Mellanox\ |reg| ConnectX\ |reg|-5 100G MCX556A-ECAT (2x100G)
1278 * Mellanox\ |reg| ConnectX\ |reg|-5 Ex EN 100G MCX516A-CDAT (2x100G)
1279 * Mellanox\ |reg| ConnectX\ |reg|-6 200G MCX654106A-HCAT (2x200G)
1280 * Mellanox\ |reg| ConnectX\ |reg|-6 Dx EN 100G MCX623106AN-CDAT (2x100G)
1281 * Mellanox\ |reg| ConnectX\ |reg|-6 Dx EN 200G MCX623105AN-VDAT (1x200G)
1282 * Mellanox\ |reg| ConnectX\ |reg|-6 Lx EN 25G MCX631102AN-ADAT (2x25G)
1284 Quick Start Guide on OFED/EN
1285 ----------------------------
1287 1. Download latest Mellanox OFED/EN. For more info check the `linux prerequisites`_.
1290 2. Install the required libraries and kernel modules either by installing
1291 only the required set, or by installing the entire Mellanox OFED/EN::
1293 ./mlnxofedinstall --upstream-libs --dpdk
1295 3. Verify the firmware is the correct one::
1299 4. Verify all ports links are set to Ethernet::
1301 mlxconfig -d <mst device> query | grep LINK_TYPE
1305 Link types may have to be configured to Ethernet::
1307 mlxconfig -d <mst device> set LINK_TYPE_P1/2=1/2/3
1309 * LINK_TYPE_P1=<1|2|3> , 1=Infiniband 2=Ethernet 3=VPI(auto-sense)
1311 For hypervisors, verify SR-IOV is enabled on the NIC::
1313 mlxconfig -d <mst device> query | grep SRIOV_EN
1316 If needed, configure SR-IOV::
1318 mlxconfig -d <mst device> set SRIOV_EN=1 NUM_OF_VFS=16
1319 mlxfwreset -d <mst device> reset
1321 5. Restart the driver::
1323 /etc/init.d/openibd restart
1327 service openibd restart
1329 If link type was changed, firmware must be reset as well::
1331 mlxfwreset -d <mst device> reset
1333 For hypervisors, after reset write the sysfs number of virtual functions
1336 To dynamically instantiate a given number of virtual functions (VFs)::
1338 echo [num_vfs] > /sys/class/infiniband/mlx5_0/device/sriov_numvfs
1340 6. Install DPDK and you are ready to go.
1341 See :doc:`compilation instructions <../linux_gsg/build_dpdk>`.
1343 Enable switchdev mode
1344 ---------------------
1346 Switchdev mode is a mode in E-Switch, that binds between representor and VF.
1347 Representor is a port in DPDK that is connected to a VF in such a way
1348 that assuming there are no offload flows, each packet that is sent from the VF
1349 will be received by the corresponding representor. While each packet that is
1350 sent to a representor will be received by the VF.
1351 This is very useful in case of SRIOV mode, where the first packet that is sent
1352 by the VF will be received by the DPDK application which will decide if this
1353 flow should be offloaded to the E-Switch. After offloading the flow packet
1354 that the VF that are matching the flow will not be received any more by
1355 the DPDK application.
1357 1. Enable SRIOV mode::
1359 mlxconfig -d <mst device> set SRIOV_EN=true
1361 2. Configure the max number of VFs::
1363 mlxconfig -d <mst device> set NUM_OF_VFS=<num of vfs>
1367 mlxfwreset -d <mst device> reset
1369 3. Configure the actual number of VFs::
1371 echo <num of vfs > /sys/class/net/<net device>/device/sriov_numvfs
1373 4. Unbind the device (can be rebind after the switchdev mode)::
1375 echo -n "<device pci address" > /sys/bus/pci/drivers/mlx5_core/unbind
1377 5. Enbale switchdev mode::
1379 echo switchdev > /sys/class/net/<net device>/compat/devlink/mode
1384 1. Configure aggressive CQE Zipping for maximum performance::
1386 mlxconfig -d <mst device> s CQE_COMPRESSION=1
1388 To set it back to the default CQE Zipping mode use::
1390 mlxconfig -d <mst device> s CQE_COMPRESSION=0
1392 2. In case of virtualization:
1394 - Make sure that hypervisor kernel is 3.16 or newer.
1395 - Configure boot with ``iommu=pt``.
1396 - Use 1G huge pages.
1397 - Make sure to allocate a VM on huge pages.
1398 - Make sure to set CPU pinning.
1400 3. Use the CPU near local NUMA node to which the PCIe adapter is connected,
1401 for better performance. For VMs, verify that the right CPU
1402 and NUMA node are pinned according to the above. Run::
1406 to identify the NUMA node to which the PCIe adapter is connected.
1408 4. If more than one adapter is used, and root complex capabilities allow
1409 to put both adapters on the same NUMA node without PCI bandwidth degradation,
1410 it is recommended to locate both adapters on the same NUMA node.
1411 This in order to forward packets from one to the other without
1412 NUMA performance penalty.
1414 5. Disable pause frames::
1416 ethtool -A <netdev> rx off tx off
1418 6. Verify IO non-posted prefetch is disabled by default. This can be checked
1419 via the BIOS configuration. Please contact you server provider for more
1420 information about the settings.
1424 On some machines, depends on the machine integrator, it is beneficial
1425 to set the PCI max read request parameter to 1K. This can be
1426 done in the following way:
1428 To query the read request size use::
1430 setpci -s <NIC PCI address> 68.w
1432 If the output is different than 3XXX, set it by::
1434 setpci -s <NIC PCI address> 68.w=3XXX
1436 The XXX can be different on different systems. Make sure to configure
1437 according to the setpci output.
1439 7. To minimize overhead of searching Memory Regions:
1441 - '--socket-mem' is recommended to pin memory by predictable amount.
1442 - Configure per-lcore cache when creating Mempools for packet buffer.
1443 - Refrain from dynamically allocating/freeing memory in run-time.
1448 There are multiple Rx burst functions with different advantages and limitations.
1450 .. table:: Rx burst functions
1452 +-------------------+------------------------+---------+-----------------+------+-------+
1453 || Function Name || Enabler || Scatter|| Error Recovery || CQE || Large|
1454 | | | | || comp|| MTU |
1455 +===================+========================+=========+=================+======+=======+
1456 | rx_burst | rx_vec_en=0 | Yes | Yes | Yes | Yes |
1457 +-------------------+------------------------+---------+-----------------+------+-------+
1458 | rx_burst_vec | rx_vec_en=1 (default) | No | if CQE comp off | Yes | No |
1459 +-------------------+------------------------+---------+-----------------+------+-------+
1460 | rx_burst_mprq || mprq_en=1 | No | Yes | Yes | Yes |
1461 | || RxQs >= rxqs_min_mprq | | | | |
1462 +-------------------+------------------------+---------+-----------------+------+-------+
1463 | rx_burst_mprq_vec || rx_vec_en=1 (default) | No | if CQE comp off | Yes | Yes |
1464 | || mprq_en=1 | | | | |
1465 | || RxQs >= rxqs_min_mprq | | | | |
1466 +-------------------+------------------------+---------+-----------------+------+-------+
1468 .. _mlx5_offloads_support:
1470 Supported hardware offloads
1471 ---------------------------
1473 .. table:: Minimal SW/HW versions for queue offloads
1475 ============== ===== ===== ========= ===== ========== =============
1476 Offload DPDK Linux rdma-core OFED firmware hardware
1477 ============== ===== ===== ========= ===== ========== =============
1478 common base 17.11 4.14 16 4.2-1 12.21.1000 ConnectX-4
1479 checksums 17.11 4.14 16 4.2-1 12.21.1000 ConnectX-4
1480 Rx timestamp 17.11 4.14 16 4.2-1 12.21.1000 ConnectX-4
1481 TSO 17.11 4.14 16 4.2-1 12.21.1000 ConnectX-4
1482 LRO 19.08 N/A N/A 4.6-4 16.25.6406 ConnectX-5
1483 Tx scheduling 20.08 N/A N/A 5.1-2 22.28.2006 ConnectX-6 Dx
1484 Buffer Split 20.11 N/A N/A 5.1-2 16.28.2006 ConnectX-5
1485 ============== ===== ===== ========= ===== ========== =============
1487 .. table:: Minimal SW/HW versions for rte_flow offloads
1489 +-----------------------+-----------------+-----------------+
1490 | Offload | with E-Switch | with NIC |
1491 +=======================+=================+=================+
1492 | Count | | DPDK 19.05 | | DPDK 19.02 |
1493 | | | OFED 4.6 | | OFED 4.6 |
1494 | | | rdma-core 24 | | rdma-core 23 |
1495 | | | ConnectX-5 | | ConnectX-5 |
1496 +-----------------------+-----------------+-----------------+
1497 | Drop | | DPDK 19.05 | | DPDK 18.11 |
1498 | | | OFED 4.6 | | OFED 4.5 |
1499 | | | rdma-core 24 | | rdma-core 23 |
1500 | | | ConnectX-5 | | ConnectX-4 |
1501 +-----------------------+-----------------+-----------------+
1502 | Queue / RSS | | | | DPDK 18.11 |
1503 | | | N/A | | OFED 4.5 |
1504 | | | | | rdma-core 23 |
1505 | | | | | ConnectX-4 |
1506 +-----------------------+-----------------+-----------------+
1507 | Shared action | | | | |
1508 | | | :numref:`sact`| | :numref:`sact`|
1511 +-----------------------+-----------------+-----------------+
1512 | | VLAN | | DPDK 19.11 | | DPDK 19.11 |
1513 | | (of_pop_vlan / | | OFED 4.7-1 | | OFED 4.7-1 |
1514 | | of_push_vlan / | | ConnectX-5 | | ConnectX-5 |
1515 | | of_set_vlan_pcp / | | | | |
1516 | | of_set_vlan_vid) | | | | |
1517 +-----------------------+-----------------+-----------------+
1518 | Encapsulation | | DPDK 19.05 | | DPDK 19.02 |
1519 | (VXLAN / NVGRE / RAW) | | OFED 4.7-1 | | OFED 4.6 |
1520 | | | rdma-core 24 | | rdma-core 23 |
1521 | | | ConnectX-5 | | ConnectX-5 |
1522 +-----------------------+-----------------+-----------------+
1523 | Encapsulation | | DPDK 19.11 | | DPDK 19.11 |
1524 | GENEVE | | OFED 4.7-3 | | OFED 4.7-3 |
1525 | | | rdma-core 27 | | rdma-core 27 |
1526 | | | ConnectX-5 | | ConnectX-5 |
1527 +-----------------------+-----------------+-----------------+
1528 | Tunnel Offload | | DPDK 20.11 | | DPDK 20.11 |
1529 | | | OFED 5.1-2 | | OFED 5.1-2 |
1530 | | | rdma-core 32 | | N/A |
1531 | | | ConnectX-5 | | ConnectX-5 |
1532 +-----------------------+-----------------+-----------------+
1533 | | Header rewrite | | DPDK 19.05 | | DPDK 19.02 |
1534 | | (set_ipv4_src / | | OFED 4.7-1 | | OFED 4.7-1 |
1535 | | set_ipv4_dst / | | rdma-core 24 | | rdma-core 24 |
1536 | | set_ipv6_src / | | ConnectX-5 | | ConnectX-5 |
1537 | | set_ipv6_dst / | | | | |
1538 | | set_tp_src / | | | | |
1539 | | set_tp_dst / | | | | |
1540 | | dec_ttl / | | | | |
1541 | | set_ttl / | | | | |
1542 | | set_mac_src / | | | | |
1543 | | set_mac_dst) | | | | |
1544 +-----------------------+-----------------+-----------------+
1545 | | Header rewrite | | DPDK 20.02 | | DPDK 20.02 |
1546 | | (set_dscp) | | OFED 5.0 | | OFED 5.0 |
1547 | | | | rdma-core 24 | | rdma-core 24 |
1548 | | | | ConnectX-5 | | ConnectX-5 |
1549 +-----------------------+-----------------+-----------------+
1550 | Jump | | DPDK 19.05 | | DPDK 19.02 |
1551 | | | OFED 4.7-1 | | OFED 4.7-1 |
1552 | | | rdma-core 24 | | N/A |
1553 | | | ConnectX-5 | | ConnectX-5 |
1554 +-----------------------+-----------------+-----------------+
1555 | Mark / Flag | | DPDK 19.05 | | DPDK 18.11 |
1556 | | | OFED 4.6 | | OFED 4.5 |
1557 | | | rdma-core 24 | | rdma-core 23 |
1558 | | | ConnectX-5 | | ConnectX-4 |
1559 +-----------------------+-----------------+-----------------+
1560 | Meta data | | DPDK 19.11 | | DPDK 19.11 |
1561 | | | OFED 4.7-3 | | OFED 4.7-3 |
1562 | | | rdma-core 26 | | rdma-core 26 |
1563 | | | ConnectX-5 | | ConnectX-5 |
1564 +-----------------------+-----------------+-----------------+
1565 | Port ID | | DPDK 19.05 | | N/A |
1566 | | | OFED 4.7-1 | | N/A |
1567 | | | rdma-core 24 | | N/A |
1568 | | | ConnectX-5 | | N/A |
1569 +-----------------------+-----------------+-----------------+
1570 | Hairpin | | | | DPDK 19.11 |
1571 | | | N/A | | OFED 4.7-3 |
1572 | | | | | rdma-core 26 |
1573 | | | | | ConnectX-5 |
1574 +-----------------------+-----------------+-----------------+
1575 | 2-port Hairpin | | | | DPDK 20.11 |
1576 | | | N/A | | OFED 5.1-2 |
1578 | | | | | ConnectX-5 |
1579 +-----------------------+-----------------+-----------------+
1580 | Metering | | DPDK 19.11 | | DPDK 19.11 |
1581 | | | OFED 4.7-3 | | OFED 4.7-3 |
1582 | | | rdma-core 26 | | rdma-core 26 |
1583 | | | ConnectX-5 | | ConnectX-5 |
1584 +-----------------------+-----------------+-----------------+
1585 | Sampling | | DPDK 20.11 | | DPDK 20.11 |
1586 | | | OFED 5.1-2 | | OFED 5.1-2 |
1587 | | | rdma-core 32 | | N/A |
1588 | | | ConnectX-5 | | ConnectX-5 |
1589 +-----------------------+-----------------+-----------------+
1590 | Encapsulation | | DPDK 21.02 | | DPDK 21.02 |
1591 | GTP PSC | | OFED 5.2 | | OFED 5.2 |
1592 | | | rdma-core 35 | | rdma-core 35 |
1593 | | | ConnectX-6 Dx| | ConnectX-6 Dx |
1594 +-----------------------+-----------------+-----------------+
1595 | Encapsulation | | DPDK 21.02 | | DPDK 21.02 |
1596 | GENEVE TLV option | | OFED 5.2 | | OFED 5.2 |
1597 | | | rdma-core 34 | | rdma-core 34 |
1598 | | | ConnectX-6 Dx | | ConnectX-6 Dx |
1599 +-----------------------+-----------------+-----------------+
1600 | Modify Field | | DPDK 21.02 | | DPDK 21.02 |
1601 | | | OFED 5.2 | | OFED 5.2 |
1602 | | | rdma-core 35 | | rdma-core 35 |
1603 | | | ConnectX-5 | | ConnectX-5 |
1604 +-----------------------+-----------------+-----------------+
1606 .. table:: Minimal SW/HW versions for shared action offload
1609 +-----------------------+-----------------+-----------------+
1610 | Shared Action | with E-Switch | with NIC |
1611 +=======================+=================+=================+
1612 | RSS | | | | DPDK 20.11 |
1613 | | | N/A | | OFED 5.2 |
1614 | | | | | rdma-core 33 |
1615 | | | | | ConnectX-5 |
1616 +-----------------------+-----------------+-----------------+
1617 | Age | | DPDK 20.11 | | DPDK 20.11 |
1618 | | | OFED 5.2 | | OFED 5.2 |
1619 | | | rdma-core 32 | | rdma-core 32 |
1620 | | | ConnectX-6 Dx| | ConnectX-6 Dx |
1621 +-----------------------+-----------------+-----------------+
1626 MARK and META items are interrelated with datapath - they might move from/to
1627 the applications in mbuf fields. Hence, zero value for these items has the
1628 special meaning - it means "no metadata are provided", not zero values are
1629 treated by applications and PMD as valid ones.
1631 Moreover in the flow engine domain the value zero is acceptable to match and
1632 set, and we should allow to specify zero values as rte_flow parameters for the
1633 META and MARK items and actions. In the same time zero mask has no meaning and
1634 should be rejected on validation stage.
1639 Flows are not cached in the driver.
1640 When stopping a device port, all the flows created on this port from the
1641 application will be flushed automatically in the background.
1642 After stopping the device port, all flows on this port become invalid and
1643 not represented in the system.
1644 All references to these flows held by the application should be discarded
1645 directly but neither destroyed nor flushed.
1647 The application should re-create the flows as required after the port restart.
1652 Compared to librte_net_mlx4 that implements a single RSS configuration per
1653 port, librte_net_mlx5 supports per-protocol RSS configuration.
1655 Since ``testpmd`` defaults to IP RSS mode and there is currently no
1656 command-line parameter to enable additional protocols (UDP and TCP as well
1657 as IP), the following commands must be entered from its CLI to get the same
1658 behavior as librte_net_mlx4::
1661 > port config all rss all
1667 This section demonstrates how to launch **testpmd** with Mellanox
1668 ConnectX-4/ConnectX-5/ConnectX-6/BlueField devices managed by librte_net_mlx5.
1670 #. Load the kernel modules::
1672 modprobe -a ib_uverbs mlx5_core mlx5_ib
1674 Alternatively if MLNX_OFED/MLNX_EN is fully installed, the following script
1677 /etc/init.d/openibd restart
1681 User space I/O kernel modules (uio and igb_uio) are not used and do
1682 not have to be loaded.
1684 #. Make sure Ethernet interfaces are in working order and linked to kernel
1685 verbs. Related sysfs entries should be present::
1687 ls -d /sys/class/net/*/device/infiniband_verbs/uverbs* | cut -d / -f 5
1696 #. Optionally, retrieve their PCI bus addresses for to be used with the allow list::
1699 for intf in eth2 eth3 eth4 eth5;
1701 (cd "/sys/class/net/${intf}/device/" && pwd -P);
1704 sed -n 's,.*/\(.*\),-a \1,p'
1713 #. Request huge pages::
1715 dpdk-hugepages.py --setup 2G
1717 #. Start testpmd with basic parameters::
1719 dpdk-testpmd -l 8-15 -n 4 -a 05:00.0 -a 05:00.1 -a 06:00.0 -a 06:00.1 -- --rxq=2 --txq=2 -i
1724 EAL: PCI device 0000:05:00.0 on NUMA socket 0
1725 EAL: probe driver: 15b3:1013 librte_net_mlx5
1726 PMD: librte_net_mlx5: PCI information matches, using device "mlx5_0" (VF: false)
1727 PMD: librte_net_mlx5: 1 port(s) detected
1728 PMD: librte_net_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:fe
1729 EAL: PCI device 0000:05:00.1 on NUMA socket 0
1730 EAL: probe driver: 15b3:1013 librte_net_mlx5
1731 PMD: librte_net_mlx5: PCI information matches, using device "mlx5_1" (VF: false)
1732 PMD: librte_net_mlx5: 1 port(s) detected
1733 PMD: librte_net_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:ff
1734 EAL: PCI device 0000:06:00.0 on NUMA socket 0
1735 EAL: probe driver: 15b3:1013 librte_net_mlx5
1736 PMD: librte_net_mlx5: PCI information matches, using device "mlx5_2" (VF: false)
1737 PMD: librte_net_mlx5: 1 port(s) detected
1738 PMD: librte_net_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:fa
1739 EAL: PCI device 0000:06:00.1 on NUMA socket 0
1740 EAL: probe driver: 15b3:1013 librte_net_mlx5
1741 PMD: librte_net_mlx5: PCI information matches, using device "mlx5_3" (VF: false)
1742 PMD: librte_net_mlx5: 1 port(s) detected
1743 PMD: librte_net_mlx5: port 1 MAC address is e4:1d:2d:e7:0c:fb
1744 Interactive-mode selected
1745 Configuring Port 0 (socket 0)
1746 PMD: librte_net_mlx5: 0x8cba80: TX queues number update: 0 -> 2
1747 PMD: librte_net_mlx5: 0x8cba80: RX queues number update: 0 -> 2
1748 Port 0: E4:1D:2D:E7:0C:FE
1749 Configuring Port 1 (socket 0)
1750 PMD: librte_net_mlx5: 0x8ccac8: TX queues number update: 0 -> 2
1751 PMD: librte_net_mlx5: 0x8ccac8: RX queues number update: 0 -> 2
1752 Port 1: E4:1D:2D:E7:0C:FF
1753 Configuring Port 2 (socket 0)
1754 PMD: librte_net_mlx5: 0x8cdb10: TX queues number update: 0 -> 2
1755 PMD: librte_net_mlx5: 0x8cdb10: RX queues number update: 0 -> 2
1756 Port 2: E4:1D:2D:E7:0C:FA
1757 Configuring Port 3 (socket 0)
1758 PMD: librte_net_mlx5: 0x8ceb58: TX queues number update: 0 -> 2
1759 PMD: librte_net_mlx5: 0x8ceb58: RX queues number update: 0 -> 2
1760 Port 3: E4:1D:2D:E7:0C:FB
1761 Checking link statuses...
1762 Port 0 Link Up - speed 40000 Mbps - full-duplex
1763 Port 1 Link Up - speed 40000 Mbps - full-duplex
1764 Port 2 Link Up - speed 10000 Mbps - full-duplex
1765 Port 3 Link Up - speed 10000 Mbps - full-duplex
1772 This section demonstrates how to dump flows. Currently, it's possible to dump
1773 all flows with assistance of external tools.
1775 #. 2 ways to get flow raw file:
1777 - Using testpmd CLI:
1779 .. code-block:: console
1781 testpmd> flow dump <port> <output_file>
1783 - call rte_flow_dev_dump api:
1785 .. code-block:: console
1787 rte_flow_dev_dump(port, file, NULL);
1789 #. Dump human-readable flows from raw file:
1791 Get flow parsing tool from: https://github.com/Mellanox/mlx_steering_dump
1793 .. code-block:: console
1795 mlx_steering_dump.py -f <output_file>