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
51 - Identifying your adapter using `Intel Support
52 <http://www.intel.com/support>`_ and get the latest NVM/FW images.
54 - Follow the DPDK :ref:`Getting Started Guide for Linux <linux_gsg>` to setup the basic DPDK environment.
56 - To get better performance on Intel platforms, please follow the "How to get best performance with NICs on Intel platforms"
57 section of the :ref:`Getting Started Guide for Linux <linux_gsg>`.
59 - Upgrade the NVM/FW version following the `Intel® Ethernet NVM Update Tool Quick Usage Guide for Linux
60 <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.
62 - For information about supported media, please refer to this document: `Intel® Ethernet Controller X710/XXV710/XL710 Feature Support Matrix
63 <http://www.intel.com/content/dam/www/public/us/en/documents/release-notes/xl710-ethernet-controller-feature-matrix.pdf>`_.
67 * Some adapters based on the Intel(R) Ethernet Controller 700 Series only
68 support Intel Ethernet Optics modules. On these adapters, other modules are not
69 supported and will not function.
71 * For connections based on Intel(R) Ethernet Controller 700 Series,
72 support is dependent on your system board. Please see your vendor for details.
74 * In all cases Intel recommends using Intel Ethernet Optics; other modules
75 may function but are not validated by Intel. Contact Intel for supported media types.
77 Recommended Matching List
78 -------------------------
80 It is highly recommended to upgrade the i40e kernel driver and firmware to
81 avoid the compatibility issues with i40e PMD. Here is the suggested matching
82 list which has been tested and verified. The detailed information can refer
83 to chapter Tested Platforms/Tested NICs in release notes.
85 For X710/XL710/XXV710,
87 +--------------+-----------------------+------------------+
88 | DPDK version | Kernel driver version | Firmware version |
89 +==============+=======================+==================+
90 | 20.02 | 2.10.19 | 7.20 |
91 +--------------+-----------------------+------------------+
92 | 19.11 | 2.9.21 | 7.00 |
93 +--------------+-----------------------+------------------+
94 | 19.08 | 2.8.43 | 7.00 |
95 +--------------+-----------------------+------------------+
96 | 19.05 | 2.7.29 | 6.80 |
97 +--------------+-----------------------+------------------+
98 | 19.02 | 2.7.26 | 6.80 |
99 +--------------+-----------------------+------------------+
100 | 18.11 | 2.4.6 | 6.01 |
101 +--------------+-----------------------+------------------+
102 | 18.08 | 2.4.6 | 6.01 |
103 +--------------+-----------------------+------------------+
104 | 18.05 | 2.4.6 | 6.01 |
105 +--------------+-----------------------+------------------+
106 | 18.02 | 2.4.3 | 6.01 |
107 +--------------+-----------------------+------------------+
108 | 17.11 | 2.1.26 | 6.01 |
109 +--------------+-----------------------+------------------+
110 | 17.08 | 2.0.19 | 6.01 |
111 +--------------+-----------------------+------------------+
112 | 17.05 | 1.5.23 | 5.05 |
113 +--------------+-----------------------+------------------+
114 | 17.02 | 1.5.23 | 5.05 |
115 +--------------+-----------------------+------------------+
116 | 16.11 | 1.5.23 | 5.05 |
117 +--------------+-----------------------+------------------+
118 | 16.07 | 1.4.25 | 5.04 |
119 +--------------+-----------------------+------------------+
120 | 16.04 | 1.4.25 | 5.02 |
121 +--------------+-----------------------+------------------+
126 +--------------+-----------------------+------------------+
127 | DPDK version | Kernel driver version | Firmware version |
128 +==============+=======================+==================+
129 | 20.02 | 2.10.19 | 4.11 |
130 +--------------+-----------------------+------------------+
131 | 19.11 | 2.9.21 | 4.10 |
132 +--------------+-----------------------+------------------+
133 | 19.08 | 2.9.21 | 4.10 |
134 +--------------+-----------------------+------------------+
135 | 19.05 | 2.7.29 | 3.33 |
136 +--------------+-----------------------+------------------+
137 | 19.02 | 2.7.26 | 3.33 |
138 +--------------+-----------------------+------------------+
139 | 18.11 | 2.4.6 | 3.33 |
140 +--------------+-----------------------+------------------+
143 Pre-Installation Configuration
144 ------------------------------
149 The following options can be modified in the ``config`` file.
150 Please note that enabling debugging options may affect system performance.
152 - ``CONFIG_RTE_LIBRTE_I40E_PMD`` (default ``y``)
154 Toggle compilation of the ``librte_pmd_i40e`` driver.
156 - ``CONFIG_RTE_LIBRTE_I40E_DEBUG_*`` (default ``n``)
158 Toggle display of generic debugging messages.
160 - ``CONFIG_RTE_LIBRTE_I40E_RX_ALLOW_BULK_ALLOC`` (default ``y``)
162 Toggle bulk allocation for RX.
164 - ``CONFIG_RTE_LIBRTE_I40E_INC_VECTOR`` (default ``n``)
166 Toggle the use of Vector PMD instead of normal RX/TX path.
167 To enable vPMD for RX, bulk allocation for Rx must be allowed.
169 - ``CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC`` (default ``n``)
171 Toggle to use a 16-byte RX descriptor, by default the RX descriptor is 32 byte.
173 - ``CONFIG_RTE_LIBRTE_I40E_QUEUE_NUM_PER_PF`` (default ``64``)
175 Number of queues reserved for PF.
177 - ``CONFIG_RTE_LIBRTE_I40E_QUEUE_NUM_PER_VM`` (default ``4``)
179 Number of queues reserved for each VMDQ Pool.
181 Runtime Config Options
182 ~~~~~~~~~~~~~~~~~~~~~~
184 - ``Reserved number of Queues per VF`` (default ``4``)
186 The number of reserved queue per VF is determined by its host PF. If the
187 PCI address of an i40e PF is aaaa:bb.cc, the number of reserved queues per
188 VF can be configured with EAL parameter like -w aaaa:bb.cc,queue-num-per-vf=n.
189 The value n can be 1, 2, 4, 8 or 16. If no such parameter is configured, the
190 number of reserved queues per VF is 4 by default. If VF request more than
191 reserved queues per VF, PF will able to allocate max to 16 queues after a VF
195 - ``Support multiple driver`` (default ``disable``)
197 There was a multiple driver support issue during use of 700 series Ethernet
198 Adapter with both Linux kernel and DPDK PMD. To fix this issue, ``devargs``
199 parameter ``support-multi-driver`` is introduced, for example::
201 -w 84:00.0,support-multi-driver=1
203 With the above configuration, DPDK PMD will not change global registers, and
204 will switch PF interrupt from IntN to Int0 to avoid interrupt conflict between
205 DPDK and Linux Kernel.
207 - ``Support VF Port Representor`` (default ``not enabled``)
209 The i40e PF PMD supports the creation of VF port representors for the control
210 and monitoring of i40e virtual function devices. Each port representor
211 corresponds to a single virtual function of that device. Using the ``devargs``
212 option ``representor`` the user can specify which virtual functions to create
213 port representors for on initialization of the PF PMD by passing the VF IDs of
214 the VFs which are required.::
216 -w DBDF,representor=[0,1,4]
218 Currently hot-plugging of representor ports is not supported so all required
219 representors must be specified on the creation of the PF.
221 - ``Use latest supported vector`` (default ``disable``)
223 Latest supported vector path may not always get the best perf so vector path was
224 recommended to use only on later platform. But users may want the latest vector path
225 since it can get better perf in some real work loading cases. So ``devargs`` param
226 ``use-latest-supported-vec`` is introduced, for example::
228 -w 84:00.0,use-latest-supported-vec=1
230 - ``Enable validation for VF message`` (default ``not enabled``)
232 The PF counts messages from each VF. If in any period of seconds the message
233 statistic from a VF exceeds maximal limitation, the PF will ignore any new message
234 from that VF for some seconds.
235 Format -- "maximal-message@period-seconds:ignore-seconds"
238 -w 84:00.0,vf_msg_cfg=80@120:180
240 Vector RX Pre-conditions
241 ~~~~~~~~~~~~~~~~~~~~~~~~
242 For Vector RX it is assumed that the number of descriptor rings will be a power
243 of 2. With this pre-condition, the ring pointer can easily scroll back to the
244 head after hitting the tail without a conditional check. In addition Vector RX
245 can use this assumption to do a bit mask using ``ring_size - 1``.
247 Driver compilation and testing
248 ------------------------------
250 Refer to the document :ref:`compiling and testing a PMD for a NIC <pmd_build_and_test>`
254 SR-IOV: Prerequisites and sample Application Notes
255 --------------------------------------------------
257 #. Load the kernel module:
259 .. code-block:: console
263 Check the output in dmesg:
265 .. code-block:: console
267 i40e 0000:83:00.1 ens802f0: renamed from eth0
269 #. Bring up the PF ports:
271 .. code-block:: console
275 #. Create VF device(s):
277 Echo the number of VFs to be created into the ``sriov_numvfs`` sysfs entry
282 .. code-block:: console
284 echo 2 > /sys/devices/pci0000:00/0000:00:03.0/0000:81:00.0/sriov_numvfs
287 #. Assign VF MAC address:
289 Assign MAC address to the VF using iproute2 utility. The syntax is:
291 .. code-block:: console
293 ip link set <PF netdev id> vf <VF id> mac <macaddr>
297 .. code-block:: console
299 ip link set ens802f0 vf 0 mac a0:b0:c0:d0:e0:f0
301 #. Assign VF to VM, and bring up the VM.
302 Please see the documentation for the *I40E/IXGBE/IGB Virtual Function Driver*.
306 Follow instructions available in the document
307 :ref:`compiling and testing a PMD for a NIC <pmd_build_and_test>`
312 .. code-block:: console
315 EAL: PCI device 0000:83:00.0 on NUMA socket 1
316 EAL: probe driver: 8086:1572 rte_i40e_pmd
317 EAL: PCI memory mapped at 0x7f7f80000000
318 EAL: PCI memory mapped at 0x7f7f80800000
319 PMD: eth_i40e_dev_init(): FW 5.0 API 1.5 NVM 05.00.02 eetrack 8000208a
320 Interactive-mode selected
321 Configuring Port 0 (socket 0)
324 PMD: i40e_dev_rx_queue_setup(): Rx Burst Bulk Alloc Preconditions are
325 satisfied.Rx Burst Bulk Alloc function will be used on port=0, queue=0.
328 Port 0: 68:05:CA:26:85:84
329 Checking link statuses...
330 Port 0 Link Up - speed 10000 Mbps - full-duplex
336 Sample Application Notes
337 ------------------------
342 Vlan filter only works when Promiscuous mode is off.
344 To start ``testpmd``, and add vlan 10 to port 0:
346 .. code-block:: console
348 ./app/testpmd -l 0-15 -n 4 -- -i --forward-mode=mac
351 testpmd> set promisc 0 off
352 testpmd> rx_vlan add 10 0
358 The Flow Director works in receive mode to identify specific flows or sets of flows and route them to specific queues.
359 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.
361 The default input set of each flow type is::
363 ipv4-other : src_ip_address, dst_ip_address
364 ipv4-frag : src_ip_address, dst_ip_address
365 ipv4-tcp : src_ip_address, dst_ip_address, src_port, dst_port
366 ipv4-udp : src_ip_address, dst_ip_address, src_port, dst_port
367 ipv4-sctp : src_ip_address, dst_ip_address, src_port, dst_port,
369 ipv6-other : src_ip_address, dst_ip_address
370 ipv6-frag : src_ip_address, dst_ip_address
371 ipv6-tcp : src_ip_address, dst_ip_address, src_port, dst_port
372 ipv6-udp : src_ip_address, dst_ip_address, src_port, dst_port
373 ipv6-sctp : src_ip_address, dst_ip_address, src_port, dst_port,
375 l2_payload : ether_type
377 The flex payload is selected from offset 0 to 15 of packet's payload by default, while it is masked out from matching.
379 Start ``testpmd`` with ``--disable-rss`` and ``--pkt-filter-mode=perfect``:
381 .. code-block:: console
383 ./app/testpmd -l 0-15 -n 4 -- -i --disable-rss --pkt-filter-mode=perfect \
384 --rxq=8 --txq=8 --nb-cores=8 --nb-ports=1
386 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:
388 .. code-block:: console
390 testpmd> flow_director_filter 0 mode IP add flow ipv4-udp \
391 src 2.2.2.3 32 dst 2.2.2.5 32 vlan 0 flexbytes () \
392 fwd pf queue 1 fd_id 1
394 Check the flow director status:
396 .. code-block:: console
398 testpmd> show port fdir 0
400 ######################## FDIR infos for port 0 ####################
402 SUPPORTED FLOW TYPE: ipv4-frag ipv4-tcp ipv4-udp ipv4-sctp ipv4-other
403 ipv6-frag ipv6-tcp ipv6-udp ipv6-sctp ipv6-other
406 max_len: 16 payload_limit: 480
407 payload_unit: 2 payload_seg: 3
408 bitmask_unit: 2 bitmask_num: 2
411 src_ipv4: 0x00000000,
412 dst_ipv4: 0x00000000,
415 src_ipv6: 0x00000000,0x00000000,0x00000000,0x00000000,
416 dst_ipv6: 0x00000000,0x00000000,0x00000000,0x00000000
417 FLEX PAYLOAD SRC OFFSET:
418 L2_PAYLOAD: 0 1 2 3 4 5 6 ...
419 L3_PAYLOAD: 0 1 2 3 4 5 6 ...
420 L4_PAYLOAD: 0 1 2 3 4 5 6 ...
422 ipv4-udp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
423 ipv4-tcp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
424 ipv4-sctp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
425 ipv4-other: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
426 ipv4-frag: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
427 ipv6-udp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
428 ipv6-tcp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
429 ipv6-sctp: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
430 ipv6-other: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
431 ipv6-frag: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
432 l2_payload: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
433 guarant_count: 1 best_count: 0
434 guarant_space: 512 best_space: 7168
441 Delete all flow director rules on a port:
443 .. code-block:: console
445 testpmd> flush_flow_director 0
450 The Intel® Ethernet 700 Series support a feature called
453 A Virtual Ethernet Bridge (VEB) is an IEEE Edge Virtual Bridging (EVB) term
454 for functionality that allows local switching between virtual endpoints within
455 a physical endpoint and also with an external bridge/network.
457 A "Floating" VEB doesn't have an uplink connection to the outside world so all
458 switching is done internally and remains within the host. As such, this
459 feature provides security benefits.
461 In addition, a Floating VEB overcomes a limitation of normal VEBs where they
462 cannot forward packets when the physical link is down. Floating VEBs don't need
463 to connect to the NIC port so they can still forward traffic from VF to VF
464 even when the physical link is down.
466 Therefore, with this feature enabled VFs can be limited to communicating with
467 each other but not an outside network, and they can do so even when there is
468 no physical uplink on the associated NIC port.
470 To enable this feature, the user should pass a ``devargs`` parameter to the
473 -w 84:00.0,enable_floating_veb=1
475 In this configuration the PMD will use the floating VEB feature for all the
476 VFs created by this PF device.
478 Alternatively, the user can specify which VFs need to connect to this floating
479 VEB using the ``floating_veb_list`` argument::
481 -w 84:00.0,enable_floating_veb=1,floating_veb_list=1;3-4
483 In this example ``VF1``, ``VF3`` and ``VF4`` connect to the floating VEB,
484 while other VFs connect to the normal VEB.
486 The current implementation only supports one floating VEB and one regular
487 VEB. VFs can connect to a floating VEB or a regular VEB according to the
488 configuration passed on the EAL command line.
490 The floating VEB functionality requires a NIC firmware version of 5.0
493 Dynamic Device Personalization (DDP)
494 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
496 The Intel® Ethernet 700 Series except for the Intel Ethernet Connection
497 X722 support a feature called "Dynamic Device Personalization (DDP)",
498 which is used to configure hardware by downloading a profile to support
499 protocols/filters which are not supported by default. The DDP
500 functionality requires a NIC firmware version of 6.0 or greater.
502 Current implementation supports GTP-C/GTP-U/PPPoE/PPPoL2TP/ESP,
503 steering can be used with rte_flow API.
505 GTPv1 package is released, and it can be downloaded from
506 https://downloadcenter.intel.com/download/27587.
508 PPPoE package is released, and it can be downloaded from
509 https://downloadcenter.intel.com/download/28040.
511 ESP-AH package is not released yet.
513 Load a profile which supports GTP and store backup profile:
515 .. code-block:: console
517 testpmd> ddp add 0 ./gtp.pkgo,./backup.pkgo
519 Delete a GTP profile and restore backup profile:
521 .. code-block:: console
523 testpmd> ddp del 0 ./backup.pkgo
525 Get loaded DDP package info list:
527 .. code-block:: console
529 testpmd> ddp get list 0
531 Display information about a GTP profile:
533 .. code-block:: console
535 testpmd> ddp get info ./gtp.pkgo
537 Input set configuration
538 ~~~~~~~~~~~~~~~~~~~~~~~
539 Input set for any PCTYPE can be configured with user defined configuration,
540 For example, to use only 48bit prefix for IPv6 src address for IPv6 TCP RSS:
542 .. code-block:: console
544 testpmd> port config 0 pctype 43 hash_inset clear all
545 testpmd> port config 0 pctype 43 hash_inset set field 13
546 testpmd> port config 0 pctype 43 hash_inset set field 14
547 testpmd> port config 0 pctype 43 hash_inset set field 15
549 Queue region configuration
550 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
551 The Intel® Ethernet 700 Series supports a feature of queue regions
552 configuration for RSS in the PF, so that different traffic classes or
553 different packet classification types can be separated to different
554 queues in different queue regions. There is an API for configuration
555 of queue regions in RSS with a command line. It can parse the parameters
556 of the region index, queue number, queue start index, user priority, traffic
557 classes and so on. Depending on commands from the command line, it will call
558 i40e private APIs and start the process of setting or flushing the queue
559 region configuration. As this feature is specific for i40e only private
560 APIs are used. These new ``test_pmd`` commands are as shown below. For
561 details please refer to :doc:`../testpmd_app_ug/index`.
563 .. code-block:: console
565 testpmd> set port (port_id) queue-region region_id (value) \
566 queue_start_index (value) queue_num (value)
567 testpmd> set port (port_id) queue-region region_id (value) flowtype (value)
568 testpmd> set port (port_id) queue-region UP (value) region_id (value)
569 testpmd> set port (port_id) queue-region flush (on|off)
570 testpmd> show port (port_id) queue-region
572 Limitations or Known issues
573 ---------------------------
575 MPLS packet classification
576 ~~~~~~~~~~~~~~~~~~~~~~~~~~
578 For firmware versions prior to 5.0, MPLS packets are not recognized by the NIC.
579 The L2 Payload flow type in flow director can be used to classify MPLS packet
580 by using a command in testpmd like:
582 testpmd> flow_director_filter 0 mode IP add flow l2_payload ether \
583 0x8847 flexbytes () fwd pf queue <N> fd_id <M>
585 With the NIC firmware version 5.0 or greater, some limited MPLS support
586 is added: Native MPLS (MPLS in Ethernet) skip is implemented, while no
587 new packet type, no classification or offload are possible. With this change,
588 L2 Payload flow type in flow director cannot be used to classify MPLS packet
589 as with previous firmware versions. Meanwhile, the Ethertype filter can be
590 used to classify MPLS packet by using a command in testpmd like:
592 testpmd> ethertype_filter 0 add mac_ignr 00:00:00:00:00:00 ethertype \
595 16 Byte RX Descriptor setting on DPDK VF
596 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
598 Currently the VF's RX descriptor mode is decided by PF. There's no PF-VF
599 interface for VF to request the RX descriptor mode, also no interface to notify
600 VF its own RX descriptor mode.
601 For all available versions of the i40e driver, these drivers don't support 16
602 byte RX descriptor. If the Linux i40e kernel driver is used as host driver,
603 while DPDK i40e PMD is used as the VF driver, DPDK cannot choose 16 byte receive
604 descriptor. The reason is that the RX descriptor is already set to 32 byte by
605 the i40e kernel driver. That is to say, user should keep
606 ``CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC=n`` in config file.
607 In the future, if the Linux i40e driver supports 16 byte RX descriptor, user
608 should make sure the DPDK VF uses the same RX descriptor mode, 16 byte or 32
609 byte, as the PF driver.
611 The same rule for DPDK PF + DPDK VF. The PF and VF should use the same RX
612 descriptor mode. Or the VF RX will not work.
614 Receive packets with Ethertype 0x88A8
615 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
617 Due to the FW limitation, PF can receive packets with Ethertype 0x88A8
618 only when floating VEB is disabled.
620 Incorrect Rx statistics when packet is oversize
621 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
623 When a packet is over maximum frame size, the packet is dropped.
624 However, the Rx statistics, when calling `rte_eth_stats_get` incorrectly
625 shows it as received.
627 VF & TC max bandwidth setting
628 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
630 The per VF max bandwidth and per TC max bandwidth cannot be enabled in parallel.
631 The behavior is different when handling per VF and per TC max bandwidth setting.
632 When enabling per VF max bandwidth, SW will check if per TC max bandwidth is
633 enabled. If so, return failure.
634 When enabling per TC max bandwidth, SW will check if per VF max bandwidth
635 is enabled. If so, disable per VF max bandwidth and continue with per TC max
638 TC TX scheduling mode setting
639 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
641 There are 2 TX scheduling modes for TCs, round robin and strict priority mode.
642 If a TC is set to strict priority mode, it can consume unlimited bandwidth.
643 It means if APP has set the max bandwidth for that TC, it comes to no
645 It's suggested to set the strict priority mode for a TC that is latency
646 sensitive but no consuming much bandwidth.
648 VF performance is impacted by PCI extended tag setting
649 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
651 To reach maximum NIC performance in the VF the PCI extended tag must be
652 enabled. The DPDK i40e PF driver will set this feature during initialization,
653 but the kernel PF driver does not. So when running traffic on a VF which is
654 managed by the kernel PF driver, a significant NIC performance downgrade has
655 been observed (for 64 byte packets, there is about 25% line-rate downgrade for
656 a 25GbE device and about 35% for a 40GbE device).
658 For kernel version >= 4.11, the kernel's PCI driver will enable the extended
659 tag if it detects that the device supports it. So by default, this is not an
660 issue. For kernels <= 4.11 or when the PCI extended tag is disabled it can be
661 enabled using the steps below.
663 #. Get the current value of the PCI configure register::
665 setpci -s <XX:XX.X> a8.w
669 value = value | 0x100
671 #. Set the PCI configure register with new value::
673 setpci -s <XX:XX.X> a8.w=<value>
678 The VF vlan strip function is only supported in the i40e kernel driver >= 2.1.26.
683 DCB works only when RSS is enabled.
685 Global configuration warning
686 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
688 I40E PMD will set some global registers to enable some function or set some
689 configure. Then when using different ports of the same NIC with Linux kernel
690 and DPDK, the port with Linux kernel will be impacted by the port with DPDK.
691 For example, register I40E_GL_SWT_L2TAGCTRL is used to control L2 tag, i40e
692 PMD uses I40E_GL_SWT_L2TAGCTRL to set vlan TPID. If setting TPID in port A
693 with DPDK, then the configuration will also impact port B in the NIC with
694 kernel driver, which don't want to use the TPID.
695 So PMD reports warning to clarify what is changed by writing global register.
697 High Performance of Small Packets on 40GbE NIC
698 ----------------------------------------------
700 As there might be firmware fixes for performance enhancement in latest version
701 of firmware image, the firmware update might be needed for getting high performance.
702 Check the Intel support website for the latest firmware updates.
703 Users should consult the release notes specific to a DPDK release to identify
704 the validated firmware version for a NIC using the i40e driver.
706 Use 16 Bytes RX Descriptor Size
707 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
709 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.
710 Configuration of ``CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC`` in config files can be changed to use 16 bytes size RX descriptors.
712 Example of getting best performance with l3fwd example
713 ------------------------------------------------------
715 The following is an example of running the DPDK ``l3fwd`` sample application to get high performance with a
716 server with Intel Xeon processors and Intel Ethernet CNA XL710.
718 The example scenario is to get best performance with two Intel Ethernet CNA XL710 40GbE ports.
719 See :numref:`figure_intel_perf_test_setup` for the performance test setup.
721 .. _figure_intel_perf_test_setup:
723 .. figure:: img/intel_perf_test_setup.*
725 Performance Test Setup
728 1. Add two Intel Ethernet CNA XL710 to the platform, and use one port per card to get best performance.
729 The reason for using two NICs is to overcome a PCIe v3.0 limitation since it cannot provide 80GbE bandwidth
730 for two 40GbE ports, but two different PCIe v3.0 x8 slot can.
731 Refer to the sample NICs output above, then we can select ``82:00.0`` and ``85:00.0`` as test ports::
733 82:00.0 Ethernet [0200]: Intel XL710 for 40GbE QSFP+ [8086:1583]
734 85:00.0 Ethernet [0200]: Intel XL710 for 40GbE QSFP+ [8086:1583]
736 2. Connect the ports to the traffic generator. For high speed testing, it's best to use a hardware traffic generator.
738 3. Check the PCI devices numa node (socket id) and get the cores number on the exact socket id.
739 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
741 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
742 cores from different cores (e.g core18 and core19).
744 4. Bind these two ports to igb_uio.
746 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
747 will be required, and each queue pair will need a dedicated CPU core for receiving/transmitting packets.
749 6. The DPDK sample application ``l3fwd`` will be used for performance testing, with using two ports for bi-directional forwarding.
750 Compile the ``l3fwd sample`` with the default lpm mode.
752 7. The command line of running l3fwd would be something like the following::
754 ./l3fwd -l 18-21 -n 4 -w 82:00.0 -w 85:00.0 \
755 -- -p 0x3 --config '(0,0,18),(0,1,19),(1,0,20),(1,1,21)'
757 This means that the application uses core 18 for port 0, queue pair 0 forwarding, core 19 for port 0, queue pair 1 forwarding,
758 core 20 for port 1, queue pair 0 forwarding, and core 21 for port 1, queue pair 1 forwarding.
760 8. Configure the traffic at a traffic generator.
762 * Start creating a stream on packet generator.
764 * Set the Ethernet II type to 0x0800.
766 Tx bytes affected by the link status change
767 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
769 For firmware versions prior to 6.01 for X710 series and 3.33 for X722 series, the tx_bytes statistics data is affected by
770 the link down event. Each time the link status changes to down, the tx_bytes decreases 110 bytes.