1 .. SPDX-License-Identifier: BSD-3-Clause
2 Copyright(c) 2016 Intel Corporation.
7 The i40e PMD (librte_pmd_i40e) provides poll mode driver support for
8 10/25/40 Gbps Intel® Ethernet 700 Series Network Adapters based on
9 the Intel Ethernet Controller X710/XL710/XXV710 and Intel Ethernet
10 Connection X722 (only support part of features).
16 Features of the i40e PMD are:
18 - Multiple queues for TX and RX
19 - Receiver Side Scaling (RSS)
21 - Packet type information
25 - VLAN/QinQ stripping and inserting
29 - Port hardware statistics
31 - Link state information
33 - Mirror on port, VLAN and VSI
34 - Interrupt mode for RX
35 - Scattered and gather for TX and RX
36 - Vector Poll mode driver
41 - IEEE1588/802.1AS timestamping
42 - VF Daemon (VFD) - EXPERIMENTAL
43 - Dynamic Device Personalization (DDP)
44 - Queue region configuration
45 - Virtual Function Port Representors
46 - Malicious Device Drive event catch and notify
52 - Identifying your adapter using `Intel Support
53 <http://www.intel.com/support>`_ and get the latest NVM/FW images.
55 - Follow the DPDK :ref:`Getting Started Guide for Linux <linux_gsg>` to setup the basic DPDK environment.
57 - To get better performance on Intel platforms, please follow the "How to get best performance with NICs on Intel platforms"
58 section of the :ref:`Getting Started Guide for Linux <linux_gsg>`.
60 - Upgrade the NVM/FW version following the `Intel® Ethernet NVM Update Tool Quick Usage Guide for Linux
61 <https://www-ssl.intel.com/content/www/us/en/embedded/products/networking/nvm-update-tool-quick-linux-usage-guide.html>`_ and `Intel® Ethernet NVM Update Tool: Quick Usage Guide for EFI <https://www.intel.com/content/www/us/en/embedded/products/networking/nvm-update-tool-quick-efi-usage-guide.html>`_ if needed.
63 - For information about supported media, please refer to this document: `Intel® Ethernet Controller X710/XXV710/XL710 Feature Support Matrix
64 <http://www.intel.com/content/dam/www/public/us/en/documents/release-notes/xl710-ethernet-controller-feature-matrix.pdf>`_.
68 * Some adapters based on the Intel(R) Ethernet Controller 700 Series only
69 support Intel Ethernet Optics modules. On these adapters, other modules are not
70 supported and will not function.
72 * For connections based on Intel(R) Ethernet Controller 700 Series,
73 support is dependent on your system board. Please see your vendor for details.
75 * In all cases Intel recommends using Intel Ethernet Optics; other modules
76 may function but are not validated by Intel. Contact Intel for supported media types.
78 Recommended Matching List
79 -------------------------
81 It is highly recommended to upgrade the i40e kernel driver and firmware to
82 avoid the compatibility issues with i40e PMD. Here is the suggested matching
83 list which has been tested and verified. The detailed information can refer
84 to chapter Tested Platforms/Tested NICs in release notes.
86 For X710/XL710/XXV710,
88 +--------------+-----------------------+------------------+
89 | DPDK version | Kernel driver version | Firmware version |
90 +==============+=======================+==================+
91 | 20.02 | 2.10.19 | 7.20 |
92 +--------------+-----------------------+------------------+
93 | 19.11 | 2.9.21 | 7.00 |
94 +--------------+-----------------------+------------------+
95 | 19.08 | 2.8.43 | 7.00 |
96 +--------------+-----------------------+------------------+
97 | 19.05 | 2.7.29 | 6.80 |
98 +--------------+-----------------------+------------------+
99 | 19.02 | 2.7.26 | 6.80 |
100 +--------------+-----------------------+------------------+
101 | 18.11 | 2.4.6 | 6.01 |
102 +--------------+-----------------------+------------------+
103 | 18.08 | 2.4.6 | 6.01 |
104 +--------------+-----------------------+------------------+
105 | 18.05 | 2.4.6 | 6.01 |
106 +--------------+-----------------------+------------------+
107 | 18.02 | 2.4.3 | 6.01 |
108 +--------------+-----------------------+------------------+
109 | 17.11 | 2.1.26 | 6.01 |
110 +--------------+-----------------------+------------------+
111 | 17.08 | 2.0.19 | 6.01 |
112 +--------------+-----------------------+------------------+
113 | 17.05 | 1.5.23 | 5.05 |
114 +--------------+-----------------------+------------------+
115 | 17.02 | 1.5.23 | 5.05 |
116 +--------------+-----------------------+------------------+
117 | 16.11 | 1.5.23 | 5.05 |
118 +--------------+-----------------------+------------------+
119 | 16.07 | 1.4.25 | 5.04 |
120 +--------------+-----------------------+------------------+
121 | 16.04 | 1.4.25 | 5.02 |
122 +--------------+-----------------------+------------------+
127 +--------------+-----------------------+------------------+
128 | DPDK version | Kernel driver version | Firmware version |
129 +==============+=======================+==================+
130 | 20.02 | 2.10.19 | 4.11 |
131 +--------------+-----------------------+------------------+
132 | 19.11 | 2.9.21 | 4.10 |
133 +--------------+-----------------------+------------------+
134 | 19.08 | 2.9.21 | 4.10 |
135 +--------------+-----------------------+------------------+
136 | 19.05 | 2.7.29 | 3.33 |
137 +--------------+-----------------------+------------------+
138 | 19.02 | 2.7.26 | 3.33 |
139 +--------------+-----------------------+------------------+
140 | 18.11 | 2.4.6 | 3.33 |
141 +--------------+-----------------------+------------------+
144 Pre-Installation Configuration
145 ------------------------------
150 The following options can be modified in the ``config`` file.
151 Please note that enabling debugging options may affect system performance.
153 - ``CONFIG_RTE_LIBRTE_I40E_PMD`` (default ``y``)
155 Toggle compilation of the ``librte_pmd_i40e`` driver.
157 - ``CONFIG_RTE_LIBRTE_I40E_DEBUG_*`` (default ``n``)
159 Toggle display of generic debugging messages.
161 - ``CONFIG_RTE_LIBRTE_I40E_RX_ALLOW_BULK_ALLOC`` (default ``y``)
163 Toggle bulk allocation for RX.
165 - ``CONFIG_RTE_LIBRTE_I40E_INC_VECTOR`` (default ``n``)
167 Toggle the use of Vector PMD instead of normal RX/TX path.
168 To enable vPMD for RX, bulk allocation for Rx must be allowed.
170 - ``CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC`` (default ``n``)
172 Toggle to use a 16-byte RX descriptor, by default the RX descriptor is 32 byte.
174 - ``CONFIG_RTE_LIBRTE_I40E_QUEUE_NUM_PER_PF`` (default ``64``)
176 Number of queues reserved for PF.
178 - ``CONFIG_RTE_LIBRTE_I40E_QUEUE_NUM_PER_VM`` (default ``4``)
180 Number of queues reserved for each VMDQ Pool.
182 Runtime Config Options
183 ~~~~~~~~~~~~~~~~~~~~~~
185 - ``Reserved number of Queues per VF`` (default ``4``)
187 The number of reserved queue per VF is determined by its host PF. If the
188 PCI address of an i40e PF is aaaa:bb.cc, the number of reserved queues per
189 VF can be configured with EAL parameter like -w aaaa:bb.cc,queue-num-per-vf=n.
190 The value n can be 1, 2, 4, 8 or 16. If no such parameter is configured, the
191 number of reserved queues per VF is 4 by default. If VF request more than
192 reserved queues per VF, PF will able to allocate max to 16 queues after a VF
196 - ``Support multiple driver`` (default ``disable``)
198 There was a multiple driver support issue during use of 700 series Ethernet
199 Adapter with both Linux kernel and DPDK PMD. To fix this issue, ``devargs``
200 parameter ``support-multi-driver`` is introduced, for example::
202 -w 84:00.0,support-multi-driver=1
204 With the above configuration, DPDK PMD will not change global registers, and
205 will switch PF interrupt from IntN to Int0 to avoid interrupt conflict between
206 DPDK and Linux Kernel.
208 - ``Support VF Port Representor`` (default ``not enabled``)
210 The i40e PF PMD supports the creation of VF port representors for the control
211 and monitoring of i40e virtual function devices. Each port representor
212 corresponds to a single virtual function of that device. Using the ``devargs``
213 option ``representor`` the user can specify which virtual functions to create
214 port representors for on initialization of the PF PMD by passing the VF IDs of
215 the VFs which are required.::
217 -w DBDF,representor=[0,1,4]
219 Currently hot-plugging of representor ports is not supported so all required
220 representors must be specified on the creation of the PF.
222 - ``Use latest supported vector`` (default ``disable``)
224 Latest supported vector path may not always get the best perf so vector path was
225 recommended to use only on later platform. But users may want the latest vector path
226 since it can get better perf in some real work loading cases. So ``devargs`` param
227 ``use-latest-supported-vec`` is introduced, for example::
229 -w 84:00.0,use-latest-supported-vec=1
231 - ``Enable validation for VF message`` (default ``not enabled``)
233 The PF counts messages from each VF. If in any period of seconds the message
234 statistic from a VF exceeds maximal limitation, the PF will ignore any new message
235 from that VF for some seconds.
236 Format -- "maximal-message@period-seconds:ignore-seconds"
239 -w 84:00.0,vf_msg_cfg=80@120:180
241 Vector RX Pre-conditions
242 ~~~~~~~~~~~~~~~~~~~~~~~~
243 For Vector RX it is assumed that the number of descriptor rings will be a power
244 of 2. With this pre-condition, the ring pointer can easily scroll back to the
245 head after hitting the tail without a conditional check. In addition Vector RX
246 can use this assumption to do a bit mask using ``ring_size - 1``.
248 Driver compilation and testing
249 ------------------------------
251 Refer to the document :ref:`compiling and testing a PMD for a NIC <pmd_build_and_test>`
255 SR-IOV: Prerequisites and sample Application Notes
256 --------------------------------------------------
258 #. Load the kernel module:
260 .. code-block:: console
264 Check the output in dmesg:
266 .. code-block:: console
268 i40e 0000:83:00.1 ens802f0: renamed from eth0
270 #. Bring up the PF ports:
272 .. code-block:: console
276 #. Create VF device(s):
278 Echo the number of VFs to be created into the ``sriov_numvfs`` sysfs entry
283 .. code-block:: console
285 echo 2 > /sys/devices/pci0000:00/0000:00:03.0/0000:81:00.0/sriov_numvfs
288 #. Assign VF MAC address:
290 Assign MAC address to the VF using iproute2 utility. The syntax is:
292 .. code-block:: console
294 ip link set <PF netdev id> vf <VF id> mac <macaddr>
298 .. code-block:: console
300 ip link set ens802f0 vf 0 mac a0:b0:c0:d0:e0:f0
302 #. Assign VF to VM, and bring up the VM.
303 Please see the documentation for the *I40E/IXGBE/IGB Virtual Function Driver*.
307 Follow instructions available in the document
308 :ref:`compiling and testing a PMD for a NIC <pmd_build_and_test>`
313 .. code-block:: console
316 EAL: PCI device 0000:83:00.0 on NUMA socket 1
317 EAL: probe driver: 8086:1572 rte_i40e_pmd
318 EAL: PCI memory mapped at 0x7f7f80000000
319 EAL: PCI memory mapped at 0x7f7f80800000
320 PMD: eth_i40e_dev_init(): FW 5.0 API 1.5 NVM 05.00.02 eetrack 8000208a
321 Interactive-mode selected
322 Configuring Port 0 (socket 0)
325 PMD: i40e_dev_rx_queue_setup(): Rx Burst Bulk Alloc Preconditions are
326 satisfied.Rx Burst Bulk Alloc function will be used on port=0, queue=0.
329 Port 0: 68:05:CA:26:85:84
330 Checking link statuses...
331 Port 0 Link Up - speed 10000 Mbps - full-duplex
337 Sample Application Notes
338 ------------------------
343 Vlan filter only works when Promiscuous mode is off.
345 To start ``testpmd``, and add vlan 10 to port 0:
347 .. code-block:: console
349 ./app/testpmd -l 0-15 -n 4 -- -i --forward-mode=mac
352 testpmd> set promisc 0 off
353 testpmd> rx_vlan add 10 0
359 The Flow Director works in receive mode to identify specific flows or sets of flows and route them to specific queues.
360 The Flow Director filters can match the different fields for different type of packet: flow type, specific input set per flow type and the flexible payload.
362 The default input set of each flow type is::
364 ipv4-other : src_ip_address, dst_ip_address
365 ipv4-frag : src_ip_address, dst_ip_address
366 ipv4-tcp : src_ip_address, dst_ip_address, src_port, dst_port
367 ipv4-udp : src_ip_address, dst_ip_address, src_port, dst_port
368 ipv4-sctp : src_ip_address, dst_ip_address, src_port, dst_port,
370 ipv6-other : src_ip_address, dst_ip_address
371 ipv6-frag : src_ip_address, dst_ip_address
372 ipv6-tcp : src_ip_address, dst_ip_address, src_port, dst_port
373 ipv6-udp : src_ip_address, dst_ip_address, src_port, dst_port
374 ipv6-sctp : src_ip_address, dst_ip_address, src_port, dst_port,
376 l2_payload : ether_type
378 The flex payload is selected from offset 0 to 15 of packet's payload by default, while it is masked out from matching.
380 Start ``testpmd`` with ``--disable-rss`` and ``--pkt-filter-mode=perfect``:
382 .. code-block:: console
384 ./app/testpmd -l 0-15 -n 4 -- -i --disable-rss --pkt-filter-mode=perfect \
385 --rxq=8 --txq=8 --nb-cores=8 --nb-ports=1
387 Add a rule to direct ``ipv4-udp`` packet whose ``dst_ip=2.2.2.5, src_ip=2.2.2.3, src_port=32, dst_port=32`` to queue 1:
389 .. code-block:: console
391 testpmd> flow_director_filter 0 mode IP add flow ipv4-udp \
392 src 2.2.2.3 32 dst 2.2.2.5 32 vlan 0 flexbytes () \
393 fwd pf queue 1 fd_id 1
395 Check the flow director status:
397 .. code-block:: console
399 testpmd> show port fdir 0
401 ######################## FDIR infos for port 0 ####################
403 SUPPORTED FLOW TYPE: ipv4-frag ipv4-tcp ipv4-udp ipv4-sctp ipv4-other
404 ipv6-frag ipv6-tcp ipv6-udp ipv6-sctp ipv6-other
407 max_len: 16 payload_limit: 480
408 payload_unit: 2 payload_seg: 3
409 bitmask_unit: 2 bitmask_num: 2
412 src_ipv4: 0x00000000,
413 dst_ipv4: 0x00000000,
416 src_ipv6: 0x00000000,0x00000000,0x00000000,0x00000000,
417 dst_ipv6: 0x00000000,0x00000000,0x00000000,0x00000000
418 FLEX PAYLOAD SRC OFFSET:
419 L2_PAYLOAD: 0 1 2 3 4 5 6 ...
420 L3_PAYLOAD: 0 1 2 3 4 5 6 ...
421 L4_PAYLOAD: 0 1 2 3 4 5 6 ...
423 ipv4-udp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
424 ipv4-tcp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
425 ipv4-sctp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
426 ipv4-other: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
427 ipv4-frag: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
428 ipv6-udp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
429 ipv6-tcp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
430 ipv6-sctp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
431 ipv6-other: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
432 ipv6-frag: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
433 l2_payload: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
434 guarant_count: 1 best_count: 0
435 guarant_space: 512 best_space: 7168
442 Delete all flow director rules on a port:
444 .. code-block:: console
446 testpmd> flush_flow_director 0
451 The Intel® Ethernet 700 Series support a feature called
454 A Virtual Ethernet Bridge (VEB) is an IEEE Edge Virtual Bridging (EVB) term
455 for functionality that allows local switching between virtual endpoints within
456 a physical endpoint and also with an external bridge/network.
458 A "Floating" VEB doesn't have an uplink connection to the outside world so all
459 switching is done internally and remains within the host. As such, this
460 feature provides security benefits.
462 In addition, a Floating VEB overcomes a limitation of normal VEBs where they
463 cannot forward packets when the physical link is down. Floating VEBs don't need
464 to connect to the NIC port so they can still forward traffic from VF to VF
465 even when the physical link is down.
467 Therefore, with this feature enabled VFs can be limited to communicating with
468 each other but not an outside network, and they can do so even when there is
469 no physical uplink on the associated NIC port.
471 To enable this feature, the user should pass a ``devargs`` parameter to the
474 -w 84:00.0,enable_floating_veb=1
476 In this configuration the PMD will use the floating VEB feature for all the
477 VFs created by this PF device.
479 Alternatively, the user can specify which VFs need to connect to this floating
480 VEB using the ``floating_veb_list`` argument::
482 -w 84:00.0,enable_floating_veb=1,floating_veb_list=1;3-4
484 In this example ``VF1``, ``VF3`` and ``VF4`` connect to the floating VEB,
485 while other VFs connect to the normal VEB.
487 The current implementation only supports one floating VEB and one regular
488 VEB. VFs can connect to a floating VEB or a regular VEB according to the
489 configuration passed on the EAL command line.
491 The floating VEB functionality requires a NIC firmware version of 5.0
494 Dynamic Device Personalization (DDP)
495 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
497 The Intel® Ethernet 700 Series except for the Intel Ethernet Connection
498 X722 support a feature called "Dynamic Device Personalization (DDP)",
499 which is used to configure hardware by downloading a profile to support
500 protocols/filters which are not supported by default. The DDP
501 functionality requires a NIC firmware version of 6.0 or greater.
503 Current implementation supports GTP-C/GTP-U/PPPoE/PPPoL2TP/ESP,
504 steering can be used with rte_flow API.
506 GTPv1 package is released, and it can be downloaded from
507 https://downloadcenter.intel.com/download/27587.
509 PPPoE package is released, and it can be downloaded from
510 https://downloadcenter.intel.com/download/28040.
512 ESP-AH package is not released yet.
514 Load a profile which supports GTP and store backup profile:
516 .. code-block:: console
518 testpmd> ddp add 0 ./gtp.pkgo,./backup.pkgo
520 Delete a GTP profile and restore backup profile:
522 .. code-block:: console
524 testpmd> ddp del 0 ./backup.pkgo
526 Get loaded DDP package info list:
528 .. code-block:: console
530 testpmd> ddp get list 0
532 Display information about a GTP profile:
534 .. code-block:: console
536 testpmd> ddp get info ./gtp.pkgo
538 Input set configuration
539 ~~~~~~~~~~~~~~~~~~~~~~~
540 Input set for any PCTYPE can be configured with user defined configuration,
541 For example, to use only 48bit prefix for IPv6 src address for IPv6 TCP RSS:
543 .. code-block:: console
545 testpmd> port config 0 pctype 43 hash_inset clear all
546 testpmd> port config 0 pctype 43 hash_inset set field 13
547 testpmd> port config 0 pctype 43 hash_inset set field 14
548 testpmd> port config 0 pctype 43 hash_inset set field 15
550 Queue region configuration
551 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
552 The Intel® Ethernet 700 Series supports a feature of queue regions
553 configuration for RSS in the PF, so that different traffic classes or
554 different packet classification types can be separated to different
555 queues in different queue regions. There is an API for configuration
556 of queue regions in RSS with a command line. It can parse the parameters
557 of the region index, queue number, queue start index, user priority, traffic
558 classes and so on. Depending on commands from the command line, it will call
559 i40e private APIs and start the process of setting or flushing the queue
560 region configuration. As this feature is specific for i40e only private
561 APIs are used. These new ``test_pmd`` commands are as shown below. For
562 details please refer to :doc:`../testpmd_app_ug/index`.
564 .. code-block:: console
566 testpmd> set port (port_id) queue-region region_id (value) \
567 queue_start_index (value) queue_num (value)
568 testpmd> set port (port_id) queue-region region_id (value) flowtype (value)
569 testpmd> set port (port_id) queue-region UP (value) region_id (value)
570 testpmd> set port (port_id) queue-region flush (on|off)
571 testpmd> show port (port_id) queue-region
578 RSS Flow supports to set hash input set, hash function, enable hash
579 and configure queue region.
581 Configure queue region as queue 0, 1, 2, 3.
583 .. code-block:: console
585 testpmd> flow create 0 ingress pattern end actions rss types end \
586 queues 0 1 2 3 end / end
588 Enable hash and set input set for ipv4-tcp.
590 .. code-block:: console
592 testpmd> flow create 0 ingress pattern eth / ipv4 / tcp / end \
593 actions rss types ipv4-tcp l3-src-only end queues end / end
595 Set symmetric hash enable for flow type ipv4-tcp.
597 .. code-block:: console
599 testpmd> flow create 0 ingress pattern eth / ipv4 / tcp / end \
600 actions rss types ipv4-tcp end queues end func symmetric_toeplitz / end
602 Set hash function as simple xor.
604 .. code-block:: console
606 testpmd> flow create 0 ingress pattern end actions rss types end \
607 queues end func simple_xor / end
609 Limitations or Known issues
610 ---------------------------
612 MPLS packet classification
613 ~~~~~~~~~~~~~~~~~~~~~~~~~~
615 For firmware versions prior to 5.0, MPLS packets are not recognized by the NIC.
616 The L2 Payload flow type in flow director can be used to classify MPLS packet
617 by using a command in testpmd like:
619 testpmd> flow_director_filter 0 mode IP add flow l2_payload ether \
620 0x8847 flexbytes () fwd pf queue <N> fd_id <M>
622 With the NIC firmware version 5.0 or greater, some limited MPLS support
623 is added: Native MPLS (MPLS in Ethernet) skip is implemented, while no
624 new packet type, no classification or offload are possible. With this change,
625 L2 Payload flow type in flow director cannot be used to classify MPLS packet
626 as with previous firmware versions. Meanwhile, the Ethertype filter can be
627 used to classify MPLS packet by using a command in testpmd like:
629 testpmd> ethertype_filter 0 add mac_ignr 00:00:00:00:00:00 ethertype \
632 16 Byte RX Descriptor setting on DPDK VF
633 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
635 Currently the VF's RX descriptor mode is decided by PF. There's no PF-VF
636 interface for VF to request the RX descriptor mode, also no interface to notify
637 VF its own RX descriptor mode.
638 For all available versions of the i40e driver, these drivers don't support 16
639 byte RX descriptor. If the Linux i40e kernel driver is used as host driver,
640 while DPDK i40e PMD is used as the VF driver, DPDK cannot choose 16 byte receive
641 descriptor. The reason is that the RX descriptor is already set to 32 byte by
642 the i40e kernel driver. That is to say, user should keep
643 ``CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC=n`` in config file.
644 In the future, if the Linux i40e driver supports 16 byte RX descriptor, user
645 should make sure the DPDK VF uses the same RX descriptor mode, 16 byte or 32
646 byte, as the PF driver.
648 The same rule for DPDK PF + DPDK VF. The PF and VF should use the same RX
649 descriptor mode. Or the VF RX will not work.
651 Receive packets with Ethertype 0x88A8
652 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
654 Due to the FW limitation, PF can receive packets with Ethertype 0x88A8
655 only when floating VEB is disabled.
657 Incorrect Rx statistics when packet is oversize
658 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
660 When a packet is over maximum frame size, the packet is dropped.
661 However, the Rx statistics, when calling `rte_eth_stats_get` incorrectly
662 shows it as received.
664 VF & TC max bandwidth setting
665 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
667 The per VF max bandwidth and per TC max bandwidth cannot be enabled in parallel.
668 The behavior is different when handling per VF and per TC max bandwidth setting.
669 When enabling per VF max bandwidth, SW will check if per TC max bandwidth is
670 enabled. If so, return failure.
671 When enabling per TC max bandwidth, SW will check if per VF max bandwidth
672 is enabled. If so, disable per VF max bandwidth and continue with per TC max
675 TC TX scheduling mode setting
676 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
678 There are 2 TX scheduling modes for TCs, round robin and strict priority mode.
679 If a TC is set to strict priority mode, it can consume unlimited bandwidth.
680 It means if APP has set the max bandwidth for that TC, it comes to no
682 It's suggested to set the strict priority mode for a TC that is latency
683 sensitive but no consuming much bandwidth.
685 VF performance is impacted by PCI extended tag setting
686 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
688 To reach maximum NIC performance in the VF the PCI extended tag must be
689 enabled. The DPDK i40e PF driver will set this feature during initialization,
690 but the kernel PF driver does not. So when running traffic on a VF which is
691 managed by the kernel PF driver, a significant NIC performance downgrade has
692 been observed (for 64 byte packets, there is about 25% line-rate downgrade for
693 a 25GbE device and about 35% for a 40GbE device).
695 For kernel version >= 4.11, the kernel's PCI driver will enable the extended
696 tag if it detects that the device supports it. So by default, this is not an
697 issue. For kernels <= 4.11 or when the PCI extended tag is disabled it can be
698 enabled using the steps below.
700 #. Get the current value of the PCI configure register::
702 setpci -s <XX:XX.X> a8.w
706 value = value | 0x100
708 #. Set the PCI configure register with new value::
710 setpci -s <XX:XX.X> a8.w=<value>
715 The VF vlan strip function is only supported in the i40e kernel driver >= 2.1.26.
720 DCB works only when RSS is enabled.
722 Global configuration warning
723 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
725 I40E PMD will set some global registers to enable some function or set some
726 configure. Then when using different ports of the same NIC with Linux kernel
727 and DPDK, the port with Linux kernel will be impacted by the port with DPDK.
728 For example, register I40E_GL_SWT_L2TAGCTRL is used to control L2 tag, i40e
729 PMD uses I40E_GL_SWT_L2TAGCTRL to set vlan TPID. If setting TPID in port A
730 with DPDK, then the configuration will also impact port B in the NIC with
731 kernel driver, which don't want to use the TPID.
732 So PMD reports warning to clarify what is changed by writing global register.
734 High Performance of Small Packets on 40GbE NIC
735 ----------------------------------------------
737 As there might be firmware fixes for performance enhancement in latest version
738 of firmware image, the firmware update might be needed for getting high performance.
739 Check the Intel support website for the latest firmware updates.
740 Users should consult the release notes specific to a DPDK release to identify
741 the validated firmware version for a NIC using the i40e driver.
743 Use 16 Bytes RX Descriptor Size
744 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
746 As i40e PMD supports both 16 and 32 bytes RX descriptor sizes, and 16 bytes size can provide helps to high performance of small packets.
747 Configuration of ``CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC`` in config files can be changed to use 16 bytes size RX descriptors.
749 Example of getting best performance with l3fwd example
750 ------------------------------------------------------
752 The following is an example of running the DPDK ``l3fwd`` sample application to get high performance with a
753 server with Intel Xeon processors and Intel Ethernet CNA XL710.
755 The example scenario is to get best performance with two Intel Ethernet CNA XL710 40GbE ports.
756 See :numref:`figure_intel_perf_test_setup` for the performance test setup.
758 .. _figure_intel_perf_test_setup:
760 .. figure:: img/intel_perf_test_setup.*
762 Performance Test Setup
765 1. Add two Intel Ethernet CNA XL710 to the platform, and use one port per card to get best performance.
766 The reason for using two NICs is to overcome a PCIe v3.0 limitation since it cannot provide 80GbE bandwidth
767 for two 40GbE ports, but two different PCIe v3.0 x8 slot can.
768 Refer to the sample NICs output above, then we can select ``82:00.0`` and ``85:00.0`` as test ports::
770 82:00.0 Ethernet [0200]: Intel XL710 for 40GbE QSFP+ [8086:1583]
771 85:00.0 Ethernet [0200]: Intel XL710 for 40GbE QSFP+ [8086:1583]
773 2. Connect the ports to the traffic generator. For high speed testing, it's best to use a hardware traffic generator.
775 3. Check the PCI devices numa node (socket id) and get the cores number on the exact socket id.
776 In this case, ``82:00.0`` and ``85:00.0`` are both in socket 1, and the cores on socket 1 in the referenced platform
778 Note: Don't use 2 logical cores on the same core (e.g core18 has 2 logical cores, core18 and core54), instead, use 2 logical
779 cores from different cores (e.g core18 and core19).
781 4. Bind these two ports to igb_uio.
783 5. As to Intel Ethernet CNA XL710 40GbE port, we need at least two queue pairs to achieve best performance, then two queues per port
784 will be required, and each queue pair will need a dedicated CPU core for receiving/transmitting packets.
786 6. The DPDK sample application ``l3fwd`` will be used for performance testing, with using two ports for bi-directional forwarding.
787 Compile the ``l3fwd sample`` with the default lpm mode.
789 7. The command line of running l3fwd would be something like the following::
791 ./l3fwd -l 18-21 -n 4 -w 82:00.0 -w 85:00.0 \
792 -- -p 0x3 --config '(0,0,18),(0,1,19),(1,0,20),(1,1,21)'
794 This means that the application uses core 18 for port 0, queue pair 0 forwarding, core 19 for port 0, queue pair 1 forwarding,
795 core 20 for port 1, queue pair 0 forwarding, and core 21 for port 1, queue pair 1 forwarding.
797 8. Configure the traffic at a traffic generator.
799 * Start creating a stream on packet generator.
801 * Set the Ethernet II type to 0x0800.
803 Tx bytes affected by the link status change
804 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
806 For firmware versions prior to 6.01 for X710 series and 3.33 for X722 series, the tx_bytes statistics data is affected by
807 the link down event. Each time the link status changes to down, the tx_bytes decreases 110 bytes.