1 .. SPDX-License-Identifier: BSD-3-Clause
2 Copyright(c) 2016 Intel Corporation.
7 The i40e PMD (**librte_net_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 | 20.11 | 2.13.10 | 8.00 |
92 +--------------+-----------------------+------------------+
93 | 20.08 | 2.12.6 | 7.30 |
94 +--------------+-----------------------+------------------+
95 | 20.05 | 2.11.27 | 7.30 |
96 +--------------+-----------------------+------------------+
97 | 20.02 | 2.10.19 | 7.20 |
98 +--------------+-----------------------+------------------+
99 | 19.11 | 2.9.21 | 7.00 |
100 +--------------+-----------------------+------------------+
101 | 19.08 | 2.8.43 | 7.00 |
102 +--------------+-----------------------+------------------+
103 | 19.05 | 2.7.29 | 6.80 |
104 +--------------+-----------------------+------------------+
105 | 19.02 | 2.7.26 | 6.80 |
106 +--------------+-----------------------+------------------+
107 | 18.11 | 2.4.6 | 6.01 |
108 +--------------+-----------------------+------------------+
109 | 18.08 | 2.4.6 | 6.01 |
110 +--------------+-----------------------+------------------+
111 | 18.05 | 2.4.6 | 6.01 |
112 +--------------+-----------------------+------------------+
113 | 18.02 | 2.4.3 | 6.01 |
114 +--------------+-----------------------+------------------+
115 | 17.11 | 2.1.26 | 6.01 |
116 +--------------+-----------------------+------------------+
117 | 17.08 | 2.0.19 | 6.01 |
118 +--------------+-----------------------+------------------+
119 | 17.05 | 1.5.23 | 5.05 |
120 +--------------+-----------------------+------------------+
121 | 17.02 | 1.5.23 | 5.05 |
122 +--------------+-----------------------+------------------+
123 | 16.11 | 1.5.23 | 5.05 |
124 +--------------+-----------------------+------------------+
125 | 16.07 | 1.4.25 | 5.04 |
126 +--------------+-----------------------+------------------+
127 | 16.04 | 1.4.25 | 5.02 |
128 +--------------+-----------------------+------------------+
133 +--------------+-----------------------+------------------+
134 | DPDK version | Kernel driver version | Firmware version |
135 +==============+=======================+==================+
136 | 20.11 | 2.13.10 | 5.00 |
137 +--------------+-----------------------+------------------+
138 | 20.08 | 2.12.6 | 4.11 |
139 +--------------+-----------------------+------------------+
140 | 20.05 | 2.11.27 | 4.11 |
141 +--------------+-----------------------+------------------+
142 | 20.02 | 2.10.19 | 4.11 |
143 +--------------+-----------------------+------------------+
144 | 19.11 | 2.9.21 | 4.10 |
145 +--------------+-----------------------+------------------+
146 | 19.08 | 2.9.21 | 4.10 |
147 +--------------+-----------------------+------------------+
148 | 19.05 | 2.7.29 | 3.33 |
149 +--------------+-----------------------+------------------+
150 | 19.02 | 2.7.26 | 3.33 |
151 +--------------+-----------------------+------------------+
152 | 18.11 | 2.4.6 | 3.33 |
153 +--------------+-----------------------+------------------+
156 Pre-Installation Configuration
157 ------------------------------
162 The following options can be modified in the ``config/rte_config.h`` file.
164 - ``RTE_LIBRTE_I40E_QUEUE_NUM_PER_PF`` (default ``64``)
166 Number of queues reserved for PF.
168 - ``RTE_LIBRTE_I40E_QUEUE_NUM_PER_VM`` (default ``4``)
170 Number of queues reserved for each VMDQ Pool.
172 Runtime Config Options
173 ~~~~~~~~~~~~~~~~~~~~~~
175 - ``Reserved number of Queues per VF`` (default ``4``)
177 The number of reserved queue per VF is determined by its host PF. If the
178 PCI address of an i40e PF is aaaa:bb.cc, the number of reserved queues per
179 VF can be configured with EAL parameter like -a aaaa:bb.cc,queue-num-per-vf=n.
180 The value n can be 1, 2, 4, 8 or 16. If no such parameter is configured, the
181 number of reserved queues per VF is 4 by default. If VF request more than
182 reserved queues per VF, PF will able to allocate max to 16 queues after a VF
186 - ``Support multiple driver`` (default ``disable``)
188 There was a multiple driver support issue during use of 700 series Ethernet
189 Adapter with both Linux kernel and DPDK PMD. To fix this issue, ``devargs``
190 parameter ``support-multi-driver`` is introduced, for example::
192 -a 84:00.0,support-multi-driver=1
194 With the above configuration, DPDK PMD will not change global registers, and
195 will switch PF interrupt from IntN to Int0 to avoid interrupt conflict between
196 DPDK and Linux Kernel.
198 - ``Support VF Port Representor`` (default ``not enabled``)
200 The i40e PF PMD supports the creation of VF port representors for the control
201 and monitoring of i40e virtual function devices. Each port representor
202 corresponds to a single virtual function of that device. Using the ``devargs``
203 option ``representor`` the user can specify which virtual functions to create
204 port representors for on initialization of the PF PMD by passing the VF IDs of
205 the VFs which are required.::
207 -a DBDF,representor=[0,1,4]
209 Currently hot-plugging of representor ports is not supported so all required
210 representors must be specified on the creation of the PF.
212 - ``Enable validation for VF message`` (default ``not enabled``)
214 The PF counts messages from each VF. If in any period of seconds the message
215 statistic from a VF exceeds maximal limitation, the PF will ignore any new message
216 from that VF for some seconds.
217 Format -- "maximal-message@period-seconds:ignore-seconds"
220 -a 84:00.0,vf_msg_cfg=80@120:180
222 Vector RX Pre-conditions
223 ~~~~~~~~~~~~~~~~~~~~~~~~
224 For Vector RX it is assumed that the number of descriptor rings will be a power
225 of 2. With this pre-condition, the ring pointer can easily scroll back to the
226 head after hitting the tail without a conditional check. In addition Vector RX
227 can use this assumption to do a bit mask using ``ring_size - 1``.
229 Driver compilation and testing
230 ------------------------------
232 Refer to the document :ref:`compiling and testing a PMD for a NIC <pmd_build_and_test>`
236 SR-IOV: Prerequisites and sample Application Notes
237 --------------------------------------------------
239 #. Load the kernel module:
241 .. code-block:: console
245 Check the output in dmesg:
247 .. code-block:: console
249 i40e 0000:83:00.1 ens802f0: renamed from eth0
251 #. Bring up the PF ports:
253 .. code-block:: console
257 #. Create VF device(s):
259 Echo the number of VFs to be created into the ``sriov_numvfs`` sysfs entry
264 .. code-block:: console
266 echo 2 > /sys/devices/pci0000:00/0000:00:03.0/0000:81:00.0/sriov_numvfs
269 #. Assign VF MAC address:
271 Assign MAC address to the VF using iproute2 utility. The syntax is:
273 .. code-block:: console
275 ip link set <PF netdev id> vf <VF id> mac <macaddr>
279 .. code-block:: console
281 ip link set ens802f0 vf 0 mac a0:b0:c0:d0:e0:f0
283 #. Assign VF to VM, and bring up the VM.
284 Please see the documentation for the *I40E/IXGBE/IGB Virtual Function Driver*.
288 Follow instructions available in the document
289 :ref:`compiling and testing a PMD for a NIC <pmd_build_and_test>`
294 .. code-block:: console
297 EAL: PCI device 0000:83:00.0 on NUMA socket 1
298 EAL: probe driver: 8086:1572 rte_i40e_pmd
299 EAL: PCI memory mapped at 0x7f7f80000000
300 EAL: PCI memory mapped at 0x7f7f80800000
301 PMD: eth_i40e_dev_init(): FW 5.0 API 1.5 NVM 05.00.02 eetrack 8000208a
302 Interactive-mode selected
303 Configuring Port 0 (socket 0)
306 PMD: i40e_dev_rx_queue_setup(): Rx Burst Bulk Alloc Preconditions are
307 satisfied.Rx Burst Bulk Alloc function will be used on port=0, queue=0.
310 Port 0: 68:05:CA:26:85:84
311 Checking link statuses...
312 Port 0 Link Up - speed 10000 Mbps - full-duplex
318 Sample Application Notes
319 ------------------------
324 Vlan filter only works when Promiscuous mode is off.
326 To start ``testpmd``, and add vlan 10 to port 0:
328 .. code-block:: console
330 ./<build_dir>/app/dpdk-testpmd -l 0-15 -n 4 -- -i --forward-mode=mac
333 testpmd> set promisc 0 off
334 testpmd> rx_vlan add 10 0
340 The Flow Director works in receive mode to identify specific flows or sets of flows and route them to specific queues.
341 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.
343 The default input set of each flow type is::
345 ipv4-other : src_ip_address, dst_ip_address
346 ipv4-frag : src_ip_address, dst_ip_address
347 ipv4-tcp : src_ip_address, dst_ip_address, src_port, dst_port
348 ipv4-udp : src_ip_address, dst_ip_address, src_port, dst_port
349 ipv4-sctp : src_ip_address, dst_ip_address, src_port, dst_port,
351 ipv6-other : src_ip_address, dst_ip_address
352 ipv6-frag : src_ip_address, dst_ip_address
353 ipv6-tcp : src_ip_address, dst_ip_address, src_port, dst_port
354 ipv6-udp : src_ip_address, dst_ip_address, src_port, dst_port
355 ipv6-sctp : src_ip_address, dst_ip_address, src_port, dst_port,
357 l2_payload : ether_type
359 The flex payload is selected from offset 0 to 15 of packet's payload by default, while it is masked out from matching.
361 Start ``testpmd`` with ``--disable-rss`` and ``--pkt-filter-mode=perfect``:
363 .. code-block:: console
365 ./<build_dir>/app/dpdk-testpmd -l 0-15 -n 4 -- -i --disable-rss \
366 --pkt-filter-mode=perfect --rxq=8 --txq=8 --nb-cores=8 \
369 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:
371 .. code-block:: console
373 testpmd> flow create 0 ingress pattern eth / ipv4 src is 2.2.2.3 \
374 dst is 2.2.2.5 / udp src is 32 dst is 32 / end \
375 actions mark id 1 / queue index 1 / end
377 Check the flow director status:
379 .. code-block:: console
381 testpmd> show port fdir 0
383 ######################## FDIR infos for port 0 ####################
385 SUPPORTED FLOW TYPE: ipv4-frag ipv4-tcp ipv4-udp ipv4-sctp ipv4-other
386 ipv6-frag ipv6-tcp ipv6-udp ipv6-sctp ipv6-other
389 max_len: 16 payload_limit: 480
390 payload_unit: 2 payload_seg: 3
391 bitmask_unit: 2 bitmask_num: 2
394 src_ipv4: 0x00000000,
395 dst_ipv4: 0x00000000,
398 src_ipv6: 0x00000000,0x00000000,0x00000000,0x00000000,
399 dst_ipv6: 0x00000000,0x00000000,0x00000000,0x00000000
400 FLEX PAYLOAD SRC OFFSET:
401 L2_PAYLOAD: 0 1 2 3 4 5 6 ...
402 L3_PAYLOAD: 0 1 2 3 4 5 6 ...
403 L4_PAYLOAD: 0 1 2 3 4 5 6 ...
405 ipv4-udp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
406 ipv4-tcp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
407 ipv4-sctp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
408 ipv4-other: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
409 ipv4-frag: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
410 ipv6-udp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
411 ipv6-tcp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
412 ipv6-sctp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
413 ipv6-other: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
414 ipv6-frag: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
415 l2_payload: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
416 guarant_count: 1 best_count: 0
417 guarant_space: 512 best_space: 7168
427 The Intel® Ethernet 700 Series support a feature called
430 A Virtual Ethernet Bridge (VEB) is an IEEE Edge Virtual Bridging (EVB) term
431 for functionality that allows local switching between virtual endpoints within
432 a physical endpoint and also with an external bridge/network.
434 A "Floating" VEB doesn't have an uplink connection to the outside world so all
435 switching is done internally and remains within the host. As such, this
436 feature provides security benefits.
438 In addition, a Floating VEB overcomes a limitation of normal VEBs where they
439 cannot forward packets when the physical link is down. Floating VEBs don't need
440 to connect to the NIC port so they can still forward traffic from VF to VF
441 even when the physical link is down.
443 Therefore, with this feature enabled VFs can be limited to communicating with
444 each other but not an outside network, and they can do so even when there is
445 no physical uplink on the associated NIC port.
447 To enable this feature, the user should pass a ``devargs`` parameter to the
450 -a 84:00.0,enable_floating_veb=1
452 In this configuration the PMD will use the floating VEB feature for all the
453 VFs created by this PF device.
455 Alternatively, the user can specify which VFs need to connect to this floating
456 VEB using the ``floating_veb_list`` argument::
458 -a 84:00.0,enable_floating_veb=1,floating_veb_list=1;3-4
460 In this example ``VF1``, ``VF3`` and ``VF4`` connect to the floating VEB,
461 while other VFs connect to the normal VEB.
463 The current implementation only supports one floating VEB and one regular
464 VEB. VFs can connect to a floating VEB or a regular VEB according to the
465 configuration passed on the EAL command line.
467 The floating VEB functionality requires a NIC firmware version of 5.0
470 Dynamic Device Personalization (DDP)
471 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
473 The Intel® Ethernet 700 Series except for the Intel Ethernet Connection
474 X722 support a feature called "Dynamic Device Personalization (DDP)",
475 which is used to configure hardware by downloading a profile to support
476 protocols/filters which are not supported by default. The DDP
477 functionality requires a NIC firmware version of 6.0 or greater.
479 Current implementation supports GTP-C/GTP-U/PPPoE/PPPoL2TP/ESP,
480 steering can be used with rte_flow API.
482 GTPv1 package is released, and it can be downloaded from
483 https://downloadcenter.intel.com/download/27587.
485 PPPoE package is released, and it can be downloaded from
486 https://downloadcenter.intel.com/download/28040.
488 ESP-AH package is released, and it can be downloaded from
489 https://downloadcenter.intel.com/download/29446.
491 Load a profile which supports GTP and store backup profile:
493 .. code-block:: console
495 testpmd> ddp add 0 ./gtp.pkgo,./backup.pkgo
497 Delete a GTP profile and restore backup profile:
499 .. code-block:: console
501 testpmd> ddp del 0 ./backup.pkgo
503 Get loaded DDP package info list:
505 .. code-block:: console
507 testpmd> ddp get list 0
509 Display information about a GTP profile:
511 .. code-block:: console
513 testpmd> ddp get info ./gtp.pkgo
515 Input set configuration
516 ~~~~~~~~~~~~~~~~~~~~~~~
517 Input set for any PCTYPE can be configured with user defined configuration,
518 For example, to use only 48bit prefix for IPv6 src address for IPv6 TCP RSS:
520 .. code-block:: console
522 testpmd> port config 0 pctype 43 hash_inset clear all
523 testpmd> port config 0 pctype 43 hash_inset set field 13
524 testpmd> port config 0 pctype 43 hash_inset set field 14
525 testpmd> port config 0 pctype 43 hash_inset set field 15
527 Queue region configuration
528 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
529 The Intel® Ethernet 700 Series supports a feature of queue regions
530 configuration for RSS in the PF, so that different traffic classes or
531 different packet classification types can be separated to different
532 queues in different queue regions. There is an API for configuration
533 of queue regions in RSS with a command line. It can parse the parameters
534 of the region index, queue number, queue start index, user priority, traffic
535 classes and so on. Depending on commands from the command line, it will call
536 i40e private APIs and start the process of setting or flushing the queue
537 region configuration. As this feature is specific for i40e only private
538 APIs are used. These new ``test_pmd`` commands are as shown below. For
539 details please refer to :doc:`../testpmd_app_ug/index`.
541 .. code-block:: console
543 testpmd> set port (port_id) queue-region region_id (value) \
544 queue_start_index (value) queue_num (value)
545 testpmd> set port (port_id) queue-region region_id (value) flowtype (value)
546 testpmd> set port (port_id) queue-region UP (value) region_id (value)
547 testpmd> set port (port_id) queue-region flush (on|off)
548 testpmd> show port (port_id) queue-region
555 RSS Flow supports to set hash input set, hash function, enable hash
556 and configure queues.
558 Configure queues as queue 0, 1, 2, 3.
560 .. code-block:: console
562 testpmd> flow create 0 ingress pattern end actions rss types end \
563 queues 0 1 2 3 end / end
565 Enable hash and set input set for ipv4-tcp.
567 .. code-block:: console
569 testpmd> flow create 0 ingress pattern eth / ipv4 / tcp / end \
570 actions rss types ipv4-tcp l3-src-only end queues end / end
572 Set symmetric hash enable for flow type ipv4-tcp.
574 .. code-block:: console
576 testpmd> flow create 0 ingress pattern eth / ipv4 / tcp / end \
577 actions rss types ipv4-tcp end queues end func symmetric_toeplitz / end
579 Set hash function as simple xor.
581 .. code-block:: console
583 testpmd> flow create 0 ingress pattern end actions rss types end \
584 queues end func simple_xor / end
586 Limitations or Known issues
587 ---------------------------
589 MPLS packet classification
590 ~~~~~~~~~~~~~~~~~~~~~~~~~~
592 For firmware versions prior to 5.0, MPLS packets are not recognized by the NIC.
593 The L2 Payload flow type in flow director can be used to classify MPLS packet
594 by using a command in testpmd like:
596 testpmd> flow_director_filter 0 mode IP add flow l2_payload ether \
597 0x8847 flexbytes () fwd pf queue <N> fd_id <M>
599 With the NIC firmware version 5.0 or greater, some limited MPLS support
600 is added: Native MPLS (MPLS in Ethernet) skip is implemented, while no
601 new packet type, no classification or offload are possible. With this change,
602 L2 Payload flow type in flow director cannot be used to classify MPLS packet
603 as with previous firmware versions. Meanwhile, the Ethertype filter can be
604 used to classify MPLS packet by using a command in testpmd like:
606 testpmd> flow create 0 ingress pattern eth type is 0x8847 / end \
607 actions queue index <M> / end
609 16 Byte RX Descriptor setting on DPDK VF
610 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
612 Currently the VF's RX descriptor mode is decided by PF. There's no PF-VF
613 interface for VF to request the RX descriptor mode, also no interface to notify
614 VF its own RX descriptor mode.
615 For all available versions of the i40e driver, these drivers don't support 16
616 byte RX descriptor. If the Linux i40e kernel driver is used as host driver,
617 while DPDK i40e PMD is used as the VF driver, DPDK cannot choose 16 byte receive
618 descriptor. The reason is that the RX descriptor is already set to 32 byte by
619 the i40e kernel driver.
620 In the future, if the Linux i40e driver supports 16 byte RX descriptor, user
621 should make sure the DPDK VF uses the same RX descriptor mode, 16 byte or 32
622 byte, as the PF driver.
624 The same rule for DPDK PF + DPDK VF. The PF and VF should use the same RX
625 descriptor mode. Or the VF RX will not work.
627 Receive packets with Ethertype 0x88A8
628 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
630 Due to the FW limitation, PF can receive packets with Ethertype 0x88A8
631 only when floating VEB is disabled.
633 Incorrect Rx statistics when packet is oversize
634 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
636 When a packet is over maximum frame size, the packet is dropped.
637 However, the Rx statistics, when calling `rte_eth_stats_get` incorrectly
638 shows it as received.
640 RX/TX statistics may be incorrect when register overflowed
641 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
643 The rx_bytes/tx_bytes statistics register is 48 bit length.
644 Although this limitation is enlarged to 64 bit length on the software side,
645 but there is no way to detect if the overflow occurred more than once.
646 So rx_bytes/tx_bytes statistics data is correct when statistics are
647 updated at least once between two overflows.
649 VF & TC max bandwidth setting
650 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
652 The per VF max bandwidth and per TC max bandwidth cannot be enabled in parallel.
653 The behavior is different when handling per VF and per TC max bandwidth setting.
654 When enabling per VF max bandwidth, SW will check if per TC max bandwidth is
655 enabled. If so, return failure.
656 When enabling per TC max bandwidth, SW will check if per VF max bandwidth
657 is enabled. If so, disable per VF max bandwidth and continue with per TC max
660 TC TX scheduling mode setting
661 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
663 There are 2 TX scheduling modes for TCs, round robin and strict priority mode.
664 If a TC is set to strict priority mode, it can consume unlimited bandwidth.
665 It means if APP has set the max bandwidth for that TC, it comes to no
667 It's suggested to set the strict priority mode for a TC that is latency
668 sensitive but no consuming much bandwidth.
670 VF performance is impacted by PCI extended tag setting
671 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
673 To reach maximum NIC performance in the VF the PCI extended tag must be
674 enabled. The DPDK i40e PF driver will set this feature during initialization,
675 but the kernel PF driver does not. So when running traffic on a VF which is
676 managed by the kernel PF driver, a significant NIC performance downgrade has
677 been observed (for 64 byte packets, there is about 25% line-rate downgrade for
678 a 25GbE device and about 35% for a 40GbE device).
680 For kernel version >= 4.11, the kernel's PCI driver will enable the extended
681 tag if it detects that the device supports it. So by default, this is not an
682 issue. For kernels <= 4.11 or when the PCI extended tag is disabled it can be
683 enabled using the steps below.
685 #. Get the current value of the PCI configure register::
687 setpci -s <XX:XX.X> a8.w
691 value = value | 0x100
693 #. Set the PCI configure register with new value::
695 setpci -s <XX:XX.X> a8.w=<value>
700 The VF vlan strip function is only supported in the i40e kernel driver >= 2.1.26.
705 DCB works only when RSS is enabled.
707 Global configuration warning
708 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
710 I40E PMD will set some global registers to enable some function or set some
711 configure. Then when using different ports of the same NIC with Linux kernel
712 and DPDK, the port with Linux kernel will be impacted by the port with DPDK.
713 For example, register I40E_GL_SWT_L2TAGCTRL is used to control L2 tag, i40e
714 PMD uses I40E_GL_SWT_L2TAGCTRL to set vlan TPID. If setting TPID in port A
715 with DPDK, then the configuration will also impact port B in the NIC with
716 kernel driver, which don't want to use the TPID.
717 So PMD reports warning to clarify what is changed by writing global register.
722 When programming cloud filters for IPv4/6_UDP/TCP/SCTP with SRC port only or DST port only,
723 it will make any cloud filter using inner_vlan or tunnel key invalid. Default configuration will be
724 recovered only by NIC core reset.
726 High Performance of Small Packets on 40GbE NIC
727 ----------------------------------------------
729 As there might be firmware fixes for performance enhancement in latest version
730 of firmware image, the firmware update might be needed for getting high performance.
731 Check the Intel support website for the latest firmware updates.
732 Users should consult the release notes specific to a DPDK release to identify
733 the validated firmware version for a NIC using the i40e driver.
735 Use 16 Bytes RX Descriptor Size
736 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
738 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.
739 In ``config/rte_config.h`` set the following to use 16 bytes size RX descriptors::
741 #define RTE_LIBRTE_I40E_16BYTE_RX_DESC 1
743 Input set requirement of each pctype for FDIR
744 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
746 Each PCTYPE can only have one specific FDIR input set at one time.
747 For example, if creating 2 rte_flow rules with different input set for one PCTYPE,
748 it will fail and return the info "Conflict with the first rule's input set",
749 which means the current rule's input set conflicts with the first rule's.
750 Remove the first rule if want to change the input set of the PCTYPE.
752 Example of getting best performance with l3fwd example
753 ------------------------------------------------------
755 The following is an example of running the DPDK ``l3fwd`` sample application to get high performance with a
756 server with Intel Xeon processors and Intel Ethernet CNA XL710.
758 The example scenario is to get best performance with two Intel Ethernet CNA XL710 40GbE ports.
759 See :numref:`figure_intel_perf_test_setup` for the performance test setup.
761 .. _figure_intel_perf_test_setup:
763 .. figure:: img/intel_perf_test_setup.*
765 Performance Test Setup
768 1. Add two Intel Ethernet CNA XL710 to the platform, and use one port per card to get best performance.
769 The reason for using two NICs is to overcome a PCIe v3.0 limitation since it cannot provide 80GbE bandwidth
770 for two 40GbE ports, but two different PCIe v3.0 x8 slot can.
771 Refer to the sample NICs output above, then we can select ``82:00.0`` and ``85:00.0`` as test ports::
773 82:00.0 Ethernet [0200]: Intel XL710 for 40GbE QSFP+ [8086:1583]
774 85:00.0 Ethernet [0200]: Intel XL710 for 40GbE QSFP+ [8086:1583]
776 2. Connect the ports to the traffic generator. For high speed testing, it's best to use a hardware traffic generator.
778 3. Check the PCI devices numa node (socket id) and get the cores number on the exact socket id.
779 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
781 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
782 cores from different cores (e.g core18 and core19).
784 4. Bind these two ports to igb_uio.
786 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
787 will be required, and each queue pair will need a dedicated CPU core for receiving/transmitting packets.
789 6. The DPDK sample application ``l3fwd`` will be used for performance testing, with using two ports for bi-directional forwarding.
790 Compile the ``l3fwd sample`` with the default lpm mode.
792 7. The command line of running l3fwd would be something like the following::
794 ./dpdk-l3fwd -l 18-21 -n 4 -a 82:00.0 -a 85:00.0 \
795 -- -p 0x3 --config '(0,0,18),(0,1,19),(1,0,20),(1,1,21)'
797 This means that the application uses core 18 for port 0, queue pair 0 forwarding, core 19 for port 0, queue pair 1 forwarding,
798 core 20 for port 1, queue pair 0 forwarding, and core 21 for port 1, queue pair 1 forwarding.
800 8. Configure the traffic at a traffic generator.
802 * Start creating a stream on packet generator.
804 * Set the Ethernet II type to 0x0800.
806 Tx bytes affected by the link status change
807 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
809 For firmware versions prior to 6.01 for X710 series and 3.33 for X722 series, the tx_bytes statistics data is affected by
810 the link down event. Each time the link status changes to down, the tx_bytes decreases 110 bytes.