1.. SPDX-License-Identifier: GPL-2.0+
2
3=================================================================
4Linux Base Driver for the Intel(R) Ethernet Controller 700 Series
5=================================================================
6
7Intel 40 Gigabit Linux driver.
8Copyright(c) 1999-2018 Intel Corporation.
9
10Contents
11========
12
13- Overview
14- Identifying Your Adapter
15- Intel(R) Ethernet Flow Director
16- Additional Configurations
17- Known Issues
18- Support
19
20
21Driver information can be obtained using ethtool, lspci, and ifconfig.
22Instructions on updating ethtool can be found in the section Additional
23Configurations later in this document.
24
25For questions related to hardware requirements, refer to the documentation
26supplied with your Intel adapter. All hardware requirements listed apply to use
27with Linux.
28
29
30Identifying Your Adapter
31========================
32The driver is compatible with devices based on the following:
33
34 * Intel(R) Ethernet Controller X710
35 * Intel(R) Ethernet Controller XL710
36 * Intel(R) Ethernet Network Connection X722
37 * Intel(R) Ethernet Controller XXV710
38
39For the best performance, make sure the latest NVM/FW is installed on your
40device.
41
42For information on how to identify your adapter, and for the latest NVM/FW
43images and Intel network drivers, refer to the Intel Support website:
44https://www.intel.com/support
45
46SFP+ and QSFP+ Devices
47----------------------
48For information about supported media, refer to this document:
49https://www.intel.com/content/dam/www/public/us/en/documents/release-notes/xl710-ethernet-controller-feature-matrix.pdf
50
51NOTE: Some adapters based on the Intel(R) Ethernet Controller 700 Series only
52support Intel Ethernet Optics modules. On these adapters, other modules are not
53supported and will not function.  In all cases Intel recommends using Intel
54Ethernet Optics; other modules may function but are not validated by Intel.
55Contact Intel for supported media types.
56
57NOTE: For connections based on Intel(R) Ethernet Controller 700 Series, support
58is dependent on your system board. Please see your vendor for details.
59
60NOTE: In systems that do not have adequate airflow to cool the adapter and
61optical modules, you must use high temperature optical modules.
62
63Virtual Functions (VFs)
64-----------------------
65Use sysfs to enable VFs. For example::
66
67  #echo $num_vf_enabled > /sys/class/net/$dev/device/sriov_numvfs #enable VFs
68  #echo 0 > /sys/class/net/$dev/device/sriov_numvfs #disable VFs
69
70For example, the following instructions will configure PF eth0 and the first VF
71on VLAN 10::
72
73  $ ip link set dev eth0 vf 0 vlan 10
74
75VLAN Tag Packet Steering
76------------------------
77Allows you to send all packets with a specific VLAN tag to a particular SR-IOV
78virtual function (VF). Further, this feature allows you to designate a
79particular VF as trusted, and allows that trusted VF to request selective
80promiscuous mode on the Physical Function (PF).
81
82To set a VF as trusted or untrusted, enter the following command in the
83Hypervisor::
84
85  # ip link set dev eth0 vf 1 trust [on|off]
86
87Once the VF is designated as trusted, use the following commands in the VM to
88set the VF to promiscuous mode.
89
90::
91
92  For promiscuous all:
93  #ip link set eth2 promisc on
94  Where eth2 is a VF interface in the VM
95
96  For promiscuous Multicast:
97  #ip link set eth2 allmulticast on
98  Where eth2 is a VF interface in the VM
99
100NOTE: By default, the ethtool priv-flag vf-true-promisc-support is set to
101"off",meaning that promiscuous mode for the VF will be limited. To set the
102promiscuous mode for the VF to true promiscuous and allow the VF to see all
103ingress traffic, use the following command::
104
105  #ethtool -set-priv-flags p261p1 vf-true-promisc-support on
106
107The vf-true-promisc-support priv-flag does not enable promiscuous mode; rather,
108it designates which type of promiscuous mode (limited or true) you will get
109when you enable promiscuous mode using the ip link commands above. Note that
110this is a global setting that affects the entire device. However,the
111vf-true-promisc-support priv-flag is only exposed to the first PF of the
112device. The PF remains in limited promiscuous mode (unless it is in MFP mode)
113regardless of the vf-true-promisc-support setting.
114
115Now add a VLAN interface on the VF interface::
116
117  #ip link add link eth2 name eth2.100 type vlan id 100
118
119Note that the order in which you set the VF to promiscuous mode and add the
120VLAN interface does not matter (you can do either first). The end result in
121this example is that the VF will get all traffic that is tagged with VLAN 100.
122
123Intel(R) Ethernet Flow Director
124-------------------------------
125The Intel Ethernet Flow Director performs the following tasks:
126
127- Directs receive packets according to their flows to different queues.
128- Enables tight control on routing a flow in the platform.
129- Matches flows and CPU cores for flow affinity.
130- Supports multiple parameters for flexible flow classification and load
131  balancing (in SFP mode only).
132
133NOTE: The Linux i40e driver supports the following flow types: IPv4, TCPv4, and
134UDPv4. For a given flow type, it supports valid combinations of IP addresses
135(source or destination) and UDP/TCP ports (source and destination). For
136example, you can supply only a source IP address, a source IP address and a
137destination port, or any combination of one or more of these four parameters.
138
139NOTE: The Linux i40e driver allows you to filter traffic based on a
140user-defined flexible two-byte pattern and offset by using the ethtool user-def
141and mask fields. Only L3 and L4 flow types are supported for user-defined
142flexible filters. For a given flow type, you must clear all Intel Ethernet Flow
143Director filters before changing the input set (for that flow type).
144
145To enable or disable the Intel Ethernet Flow Director::
146
147  # ethtool -K ethX ntuple <on|off>
148
149When disabling ntuple filters, all the user programmed filters are flushed from
150the driver cache and hardware. All needed filters must be re-added when ntuple
151is re-enabled.
152
153To add a filter that directs packet to queue 2, use -U or -N switch::
154
155  # ethtool -N ethX flow-type tcp4 src-ip 192.168.10.1 dst-ip \
156  192.168.10.2 src-port 2000 dst-port 2001 action 2 [loc 1]
157
158To set a filter using only the source and destination IP address::
159
160  # ethtool -N ethX flow-type tcp4 src-ip 192.168.10.1 dst-ip \
161  192.168.10.2 action 2 [loc 1]
162
163To see the list of filters currently present::
164
165  # ethtool <-u|-n> ethX
166
167Application Targeted Routing (ATR) Perfect Filters
168--------------------------------------------------
169ATR is enabled by default when the kernel is in multiple transmit queue mode.
170An ATR Intel Ethernet Flow Director filter rule is added when a TCP-IP flow
171starts and is deleted when the flow ends. When a TCP-IP Intel Ethernet Flow
172Director rule is added from ethtool (Sideband filter), ATR is turned off by the
173driver. To re-enable ATR, the sideband can be disabled with the ethtool -K
174option. For example::
175
176  ethtool -K [adapter] ntuple [off|on]
177
178If sideband is re-enabled after ATR is re-enabled, ATR remains enabled until a
179TCP-IP flow is added. When all TCP-IP sideband rules are deleted, ATR is
180automatically re-enabled.
181
182Packets that match the ATR rules are counted in fdir_atr_match stats in
183ethtool, which also can be used to verify whether ATR rules still exist.
184
185Sideband Perfect Filters
186------------------------
187Sideband Perfect Filters are used to direct traffic that matches specified
188characteristics. They are enabled through ethtool's ntuple interface. To add a
189new filter use the following command::
190
191  ethtool -U <device> flow-type <type> src-ip <ip> dst-ip <ip> src-port <port> \
192  dst-port <port> action <queue>
193
194Where:
195  <device> - the ethernet device to program
196  <type> - can be ip4, tcp4, udp4, or sctp4
197  <ip> - the ip address to match on
198  <port> - the port number to match on
199  <queue> - the queue to direct traffic towards (-1 discards matching traffic)
200
201Use the following command to display all of the active filters::
202
203  ethtool -u <device>
204
205Use the following command to delete a filter::
206
207  ethtool -U <device> delete <N>
208
209Where <N> is the filter id displayed when printing all the active filters, and
210may also have been specified using "loc <N>" when adding the filter.
211
212The following example matches TCP traffic sent from 192.168.0.1, port 5300,
213directed to 192.168.0.5, port 80, and sends it to queue 7::
214
215  ethtool -U enp130s0 flow-type tcp4 src-ip 192.168.0.1 dst-ip 192.168.0.5 \
216  src-port 5300 dst-port 80 action 7
217
218For each flow-type, the programmed filters must all have the same matching
219input set. For example, issuing the following two commands is acceptable::
220
221  ethtool -U enp130s0 flow-type ip4 src-ip 192.168.0.1 src-port 5300 action 7
222  ethtool -U enp130s0 flow-type ip4 src-ip 192.168.0.5 src-port 55 action 10
223
224Issuing the next two commands, however, is not acceptable, since the first
225specifies src-ip and the second specifies dst-ip::
226
227  ethtool -U enp130s0 flow-type ip4 src-ip 192.168.0.1 src-port 5300 action 7
228  ethtool -U enp130s0 flow-type ip4 dst-ip 192.168.0.5 src-port 55 action 10
229
230The second command will fail with an error. You may program multiple filters
231with the same fields, using different values, but, on one device, you may not
232program two tcp4 filters with different matching fields.
233
234Matching on a sub-portion of a field is not supported by the i40e driver, thus
235partial mask fields are not supported.
236
237The driver also supports matching user-defined data within the packet payload.
238This flexible data is specified using the "user-def" field of the ethtool
239command in the following way:
240
241+----------------------------+--------------------------+
242| 31    28    24    20    16 | 15    12    8    4    0  |
243+----------------------------+--------------------------+
244| offset into packet payload | 2 bytes of flexible data |
245+----------------------------+--------------------------+
246
247For example,
248
249::
250
251  ... user-def 0x4FFFF ...
252
253tells the filter to look 4 bytes into the payload and match that value against
2540xFFFF. The offset is based on the beginning of the payload, and not the
255beginning of the packet. Thus
256
257::
258
259  flow-type tcp4 ... user-def 0x8BEAF ...
260
261would match TCP/IPv4 packets which have the value 0xBEAF 8 bytes into the
262TCP/IPv4 payload.
263
264Note that ICMP headers are parsed as 4 bytes of header and 4 bytes of payload.
265Thus to match the first byte of the payload, you must actually add 4 bytes to
266the offset. Also note that ip4 filters match both ICMP frames as well as raw
267(unknown) ip4 frames, where the payload will be the L3 payload of the IP4 frame.
268
269The maximum offset is 64. The hardware will only read up to 64 bytes of data
270from the payload. The offset must be even because the flexible data is 2 bytes
271long and must be aligned to byte 0 of the packet payload.
272
273The user-defined flexible offset is also considered part of the input set and
274cannot be programmed separately for multiple filters of the same type. However,
275the flexible data is not part of the input set and multiple filters may use the
276same offset but match against different data.
277
278To create filters that direct traffic to a specific Virtual Function, use the
279"action" parameter. Specify the action as a 64 bit value, where the lower 32
280bits represents the queue number, while the next 8 bits represent which VF.
281Note that 0 is the PF, so the VF identifier is offset by 1. For example::
282
283  ... action 0x800000002 ...
284
285specifies to direct traffic to Virtual Function 7 (8 minus 1) into queue 2 of
286that VF.
287
288Note that these filters will not break internal routing rules, and will not
289route traffic that otherwise would not have been sent to the specified Virtual
290Function.
291
292Setting the link-down-on-close Private Flag
293-------------------------------------------
294When the link-down-on-close private flag is set to "on", the port's link will
295go down when the interface is brought down using the ifconfig ethX down command.
296
297Use ethtool to view and set link-down-on-close, as follows::
298
299  ethtool --show-priv-flags ethX
300  ethtool --set-priv-flags ethX link-down-on-close [on|off]
301
302Viewing Link Messages
303---------------------
304Link messages will not be displayed to the console if the distribution is
305restricting system messages. In order to see network driver link messages on
306your console, set dmesg to eight by entering the following::
307
308  dmesg -n 8
309
310NOTE: This setting is not saved across reboots.
311
312Jumbo Frames
313------------
314Jumbo Frames support is enabled by changing the Maximum Transmission Unit (MTU)
315to a value larger than the default value of 1500.
316
317Use the ifconfig command to increase the MTU size. For example, enter the
318following where <x> is the interface number::
319
320  ifconfig eth<x> mtu 9000 up
321
322Alternatively, you can use the ip command as follows::
323
324  ip link set mtu 9000 dev eth<x>
325  ip link set up dev eth<x>
326
327This setting is not saved across reboots. The setting change can be made
328permanent by adding 'MTU=9000' to the file::
329
330  /etc/sysconfig/network-scripts/ifcfg-eth<x> // for RHEL
331  /etc/sysconfig/network/<config_file> // for SLES
332
333NOTE: The maximum MTU setting for Jumbo Frames is 9702. This value coincides
334with the maximum Jumbo Frames size of 9728 bytes.
335
336NOTE: This driver will attempt to use multiple page sized buffers to receive
337each jumbo packet. This should help to avoid buffer starvation issues when
338allocating receive packets.
339
340ethtool
341-------
342The driver utilizes the ethtool interface for driver configuration and
343diagnostics, as well as displaying statistical information. The latest ethtool
344version is required for this functionality. Download it at:
345https://www.kernel.org/pub/software/network/ethtool/
346
347Supported ethtool Commands and Options for Filtering
348----------------------------------------------------
349-n --show-nfc
350  Retrieves the receive network flow classification configurations.
351
352rx-flow-hash tcp4|udp4|ah4|esp4|sctp4|tcp6|udp6|ah6|esp6|sctp6
353  Retrieves the hash options for the specified network traffic type.
354
355-N --config-nfc
356  Configures the receive network flow classification.
357
358rx-flow-hash tcp4|udp4|ah4|esp4|sctp4|tcp6|udp6|ah6|esp6|sctp6 m|v|t|s|d|f|n|r...
359  Configures the hash options for the specified network traffic type.
360
361udp4 UDP over IPv4
362udp6 UDP over IPv6
363
364f Hash on bytes 0 and 1 of the Layer 4 header of the Rx packet.
365n Hash on bytes 2 and 3 of the Layer 4 header of the Rx packet.
366
367Speed and Duplex Configuration
368------------------------------
369In addressing speed and duplex configuration issues, you need to distinguish
370between copper-based adapters and fiber-based adapters.
371
372In the default mode, an Intel(R) Ethernet Network Adapter using copper
373connections will attempt to auto-negotiate with its link partner to determine
374the best setting. If the adapter cannot establish link with the link partner
375using auto-negotiation, you may need to manually configure the adapter and link
376partner to identical settings to establish link and pass packets. This should
377only be needed when attempting to link with an older switch that does not
378support auto-negotiation or one that has been forced to a specific speed or
379duplex mode. Your link partner must match the setting you choose. 1 Gbps speeds
380and higher cannot be forced. Use the autonegotiation advertising setting to
381manually set devices for 1 Gbps and higher.
382
383NOTE: You cannot set the speed for devices based on the Intel(R) Ethernet
384Network Adapter XXV710 based devices.
385
386Speed, duplex, and autonegotiation advertising are configured through the
387ethtool utility.
388
389Caution: Only experienced network administrators should force speed and duplex
390or change autonegotiation advertising manually. The settings at the switch must
391always match the adapter settings. Adapter performance may suffer or your
392adapter may not operate if you configure the adapter differently from your
393switch.
394
395An Intel(R) Ethernet Network Adapter using fiber-based connections, however,
396will not attempt to auto-negotiate with its link partner since those adapters
397operate only in full duplex and only at their native speed.
398
399NAPI
400----
401NAPI (Rx polling mode) is supported in the i40e driver.
402
403See :ref:`Documentation/networking/napi.rst <napi>` for more information.
404
405Flow Control
406------------
407Ethernet Flow Control (IEEE 802.3x) can be configured with ethtool to enable
408receiving and transmitting pause frames for i40e. When transmit is enabled,
409pause frames are generated when the receive packet buffer crosses a predefined
410threshold. When receive is enabled, the transmit unit will halt for the time
411delay specified when a pause frame is received.
412
413NOTE: You must have a flow control capable link partner.
414
415Flow Control is on by default.
416
417Use ethtool to change the flow control settings.
418
419To enable or disable Rx or Tx Flow Control::
420
421  ethtool -A eth? rx <on|off> tx <on|off>
422
423Note: This command only enables or disables Flow Control if auto-negotiation is
424disabled. If auto-negotiation is enabled, this command changes the parameters
425used for auto-negotiation with the link partner.
426
427To enable or disable auto-negotiation::
428
429  ethtool -s eth? autoneg <on|off>
430
431Note: Flow Control auto-negotiation is part of link auto-negotiation. Depending
432on your device, you may not be able to change the auto-negotiation setting.
433
434RSS Hash Flow
435-------------
436Allows you to set the hash bytes per flow type and any combination of one or
437more options for Receive Side Scaling (RSS) hash byte configuration.
438
439::
440
441  # ethtool -N <dev> rx-flow-hash <type> <option>
442
443Where <type> is:
444  tcp4	signifying TCP over IPv4
445  udp4	signifying UDP over IPv4
446  tcp6	signifying TCP over IPv6
447  udp6	signifying UDP over IPv6
448And <option> is one or more of:
449  s	Hash on the IP source address of the Rx packet.
450  d	Hash on the IP destination address of the Rx packet.
451  f	Hash on bytes 0 and 1 of the Layer 4 header of the Rx packet.
452  n	Hash on bytes 2 and 3 of the Layer 4 header of the Rx packet.
453
454MAC and VLAN anti-spoofing feature
455----------------------------------
456When a malicious driver attempts to send a spoofed packet, it is dropped by the
457hardware and not transmitted.
458NOTE: This feature can be disabled for a specific Virtual Function (VF)::
459
460  ip link set <pf dev> vf <vf id> spoofchk {off|on}
461
462IEEE 1588 Precision Time Protocol (PTP) Hardware Clock (PHC)
463------------------------------------------------------------
464Precision Time Protocol (PTP) is used to synchronize clocks in a computer
465network. PTP support varies among Intel devices that support this driver. Use
466"ethtool -T <netdev name>" to get a definitive list of PTP capabilities
467supported by the device.
468
469IEEE 802.1ad (QinQ) Support
470---------------------------
471The IEEE 802.1ad standard, informally known as QinQ, allows for multiple VLAN
472IDs within a single Ethernet frame. VLAN IDs are sometimes referred to as
473"tags," and multiple VLAN IDs are thus referred to as a "tag stack." Tag stacks
474allow L2 tunneling and the ability to segregate traffic within a particular
475VLAN ID, among other uses.
476
477The following are examples of how to configure 802.1ad (QinQ)::
478
479  ip link add link eth0 eth0.24 type vlan proto 802.1ad id 24
480  ip link add link eth0.24 eth0.24.371 type vlan proto 802.1Q id 371
481
482Where "24" and "371" are example VLAN IDs.
483
484NOTES:
485  Receive checksum offloads, cloud filters, and VLAN acceleration are not
486  supported for 802.1ad (QinQ) packets.
487
488VXLAN and GENEVE Overlay HW Offloading
489--------------------------------------
490Virtual Extensible LAN (VXLAN) allows you to extend an L2 network over an L3
491network, which may be useful in a virtualized or cloud environment. Some
492Intel(R) Ethernet Network devices perform VXLAN processing, offloading it from
493the operating system. This reduces CPU utilization.
494
495VXLAN offloading is controlled by the Tx and Rx checksum offload options
496provided by ethtool. That is, if Tx checksum offload is enabled, and the
497adapter has the capability, VXLAN offloading is also enabled.
498
499Support for VXLAN and GENEVE HW offloading is dependent on kernel support of
500the HW offloading features.
501
502Multiple Functions per Port
503---------------------------
504Some adapters based on the Intel Ethernet Controller X710/XL710 support
505multiple functions on a single physical port. Configure these functions through
506the System Setup/BIOS.
507
508Minimum TX Bandwidth is the guaranteed minimum data transmission bandwidth, as
509a percentage of the full physical port link speed, that the partition will
510receive. The bandwidth the partition is awarded will never fall below the level
511you specify.
512
513The range for the minimum bandwidth values is:
5141 to ((100 minus # of partitions on the physical port) plus 1)
515For example, if a physical port has 4 partitions, the range would be:
5161 to ((100 - 4) + 1 = 97)
517
518The Maximum Bandwidth percentage represents the maximum transmit bandwidth
519allocated to the partition as a percentage of the full physical port link
520speed. The accepted range of values is 1-100. The value is used as a limiter,
521should you chose that any one particular function not be able to consume 100%
522of a port's bandwidth (should it be available). The sum of all the values for
523Maximum Bandwidth is not restricted, because no more than 100% of a port's
524bandwidth can ever be used.
525
526NOTE: X710/XXV710 devices fail to enable Max VFs (64) when Multiple Functions
527per Port (MFP) and SR-IOV are enabled. An error from i40e is logged that says
528"add vsi failed for VF N, aq_err 16". To workaround the issue, enable less than
52964 virtual functions (VFs).
530
531Data Center Bridging (DCB)
532--------------------------
533DCB is a configuration Quality of Service implementation in hardware. It uses
534the VLAN priority tag (802.1p) to filter traffic. That means that there are 8
535different priorities that traffic can be filtered into. It also enables
536priority flow control (802.1Qbb) which can limit or eliminate the number of
537dropped packets during network stress. Bandwidth can be allocated to each of
538these priorities, which is enforced at the hardware level (802.1Qaz).
539
540Adapter firmware implements LLDP and DCBX protocol agents as per 802.1AB and
541802.1Qaz respectively. The firmware based DCBX agent runs in willing mode only
542and can accept settings from a DCBX capable peer. Software configuration of
543DCBX parameters via dcbtool/lldptool are not supported.
544
545NOTE: Firmware LLDP can be disabled by setting the private flag disable-fw-lldp.
546
547The i40e driver implements the DCB netlink interface layer to allow user-space
548to communicate with the driver and query DCB configuration for the port.
549
550NOTE:
551The kernel assumes that TC0 is available, and will disable Priority Flow
552Control (PFC) on the device if TC0 is not available. To fix this, ensure TC0 is
553enabled when setting up DCB on your switch.
554
555Interrupt Rate Limiting
556-----------------------
557:Valid Range: 0-235 (0=no limit)
558
559The Intel(R) Ethernet Controller XL710 family supports an interrupt rate
560limiting mechanism. The user can control, via ethtool, the number of
561microseconds between interrupts.
562
563Syntax::
564
565  # ethtool -C ethX rx-usecs-high N
566
567The range of 0-235 microseconds provides an effective range of 4,310 to 250,000
568interrupts per second. The value of rx-usecs-high can be set independently of
569rx-usecs and tx-usecs in the same ethtool command, and is also independent of
570the adaptive interrupt moderation algorithm. The underlying hardware supports
571granularity in 4-microsecond intervals, so adjacent values may result in the
572same interrupt rate.
573
574One possible use case is the following::
575
576  # ethtool -C ethX adaptive-rx off adaptive-tx off rx-usecs-high 20 rx-usecs \
577    5 tx-usecs 5
578
579The above command would disable adaptive interrupt moderation, and allow a
580maximum of 5 microseconds before indicating a receive or transmit was complete.
581However, instead of resulting in as many as 200,000 interrupts per second, it
582limits total interrupts per second to 50,000 via the rx-usecs-high parameter.
583
584Performance Optimization
585========================
586Driver defaults are meant to fit a wide variety of workloads, but if further
587optimization is required we recommend experimenting with the following settings.
588
589NOTE: For better performance when processing small (64B) frame sizes, try
590enabling Hyper threading in the BIOS in order to increase the number of logical
591cores in the system and subsequently increase the number of queues available to
592the adapter.
593
594Virtualized Environments
595------------------------
5961. Disable XPS on both ends by using the included virt_perf_default script
597or by running the following command as root::
598
599  for file in `ls /sys/class/net/<ethX>/queues/tx-*/xps_cpus`;
600  do echo 0 > $file; done
601
6022. Using the appropriate mechanism (vcpupin) in the vm, pin the cpu's to
603individual lcpu's, making sure to use a set of cpu's included in the
604device's local_cpulist: /sys/class/net/<ethX>/device/local_cpulist.
605
6063. Configure as many Rx/Tx queues in the VM as available. Do not rely on
607the default setting of 1.
608
609
610Non-virtualized Environments
611----------------------------
612Pin the adapter's IRQs to specific cores by disabling the irqbalance service
613and using the included set_irq_affinity script. Please see the script's help
614text for further options.
615
616- The following settings will distribute the IRQs across all the cores evenly::
617
618  # scripts/set_irq_affinity -x all <interface1> , [ <interface2>, ... ]
619
620- The following settings will distribute the IRQs across all the cores that are
621  local to the adapter (same NUMA node)::
622
623  # scripts/set_irq_affinity -x local <interface1> ,[ <interface2>, ... ]
624
625For very CPU intensive workloads, we recommend pinning the IRQs to all cores.
626
627For IP Forwarding: Disable Adaptive ITR and lower Rx and Tx interrupts per
628queue using ethtool.
629
630- Setting rx-usecs and tx-usecs to 125 will limit interrupts to about 8000
631  interrupts per second per queue.
632
633::
634
635  # ethtool -C <interface> adaptive-rx off adaptive-tx off rx-usecs 125 \
636    tx-usecs 125
637
638For lower CPU utilization: Disable Adaptive ITR and lower Rx and Tx interrupts
639per queue using ethtool.
640
641- Setting rx-usecs and tx-usecs to 250 will limit interrupts to about 4000
642  interrupts per second per queue.
643
644::
645
646  # ethtool -C <interface> adaptive-rx off adaptive-tx off rx-usecs 250 \
647    tx-usecs 250
648
649For lower latency: Disable Adaptive ITR and ITR by setting Rx and Tx to 0 using
650ethtool.
651
652::
653
654  # ethtool -C <interface> adaptive-rx off adaptive-tx off rx-usecs 0 \
655    tx-usecs 0
656
657Application Device Queues (ADq)
658-------------------------------
659Application Device Queues (ADq) allows you to dedicate one or more queues to a
660specific application. This can reduce latency for the specified application,
661and allow Tx traffic to be rate limited per application. Follow the steps below
662to set ADq.
663
6641. Create traffic classes (TCs). Maximum of 8 TCs can be created per interface.
665The shaper bw_rlimit parameter is optional.
666
667Example: Sets up two tcs, tc0 and tc1, with 16 queues each and max tx rate set
668to 1Gbit for tc0 and 3Gbit for tc1.
669
670::
671
672  # tc qdisc add dev <interface> root mqprio num_tc 2 map 0 0 0 0 1 1 1 1
673  queues 16@0 16@16 hw 1 mode channel shaper bw_rlimit min_rate 1Gbit 2Gbit
674  max_rate 1Gbit 3Gbit
675
676map: priority mapping for up to 16 priorities to tcs (e.g. map 0 0 0 0 1 1 1 1
677sets priorities 0-3 to use tc0 and 4-7 to use tc1)
678
679queues: for each tc, <num queues>@<offset> (e.g. queues 16@0 16@16 assigns
68016 queues to tc0 at offset 0 and 16 queues to tc1 at offset 16. Max total
681number of queues for all tcs is 64 or number of cores, whichever is lower.)
682
683hw 1 mode channel: ‘channel’ with ‘hw’ set to 1 is a new new hardware
684offload mode in mqprio that makes full use of the mqprio options, the
685TCs, the queue configurations, and the QoS parameters.
686
687shaper bw_rlimit: for each tc, sets minimum and maximum bandwidth rates.
688Totals must be equal or less than port speed.
689
690For example: min_rate 1Gbit 3Gbit: Verify bandwidth limit using network
691monitoring tools such as `ifstat` or `sar -n DEV [interval] [number of samples]`
692
6932. Enable HW TC offload on interface::
694
695    # ethtool -K <interface> hw-tc-offload on
696
6973. Apply TCs to ingress (RX) flow of interface::
698
699    # tc qdisc add dev <interface> ingress
700
701NOTES:
702 - Run all tc commands from the iproute2 <pathtoiproute2>/tc/ directory.
703 - ADq is not compatible with cloud filters.
704 - Setting up channels via ethtool (ethtool -L) is not supported when the
705   TCs are configured using mqprio.
706 - You must have iproute2 latest version
707 - NVM version 6.01 or later is required.
708 - ADq cannot be enabled when any the following features are enabled: Data
709   Center Bridging (DCB), Multiple Functions per Port (MFP), or Sideband
710   Filters.
711 - If another driver (for example, DPDK) has set cloud filters, you cannot
712   enable ADq.
713 - Tunnel filters are not supported in ADq. If encapsulated packets do
714   arrive in non-tunnel mode, filtering will be done on the inner headers.
715   For example, for VXLAN traffic in non-tunnel mode, PCTYPE is identified
716   as a VXLAN encapsulated packet, outer headers are ignored. Therefore,
717   inner headers are matched.
718 - If a TC filter on a PF matches traffic over a VF (on the PF), that
719   traffic will be routed to the appropriate queue of the PF, and will
720   not be passed on the VF. Such traffic will end up getting dropped higher
721   up in the TCP/IP stack as it does not match PF address data.
722 - If traffic matches multiple TC filters that point to different TCs,
723   that traffic will be duplicated and sent to all matching TC queues.
724   The hardware switch mirrors the packet to a VSI list when multiple
725   filters are matched.
726
727
728Known Issues/Troubleshooting
729============================
730
731NOTE: 1 Gb devices based on the Intel(R) Ethernet Network Connection X722 do
732not support the following features:
733
734  * Data Center Bridging (DCB)
735  * QOS
736  * VMQ
737  * SR-IOV
738  * Task Encapsulation offload (VXLAN, NVGRE)
739  * Energy Efficient Ethernet (EEE)
740  * Auto-media detect
741
742Unexpected Issues when the device driver and DPDK share a device
743----------------------------------------------------------------
744Unexpected issues may result when an i40e device is in multi driver mode and
745the kernel driver and DPDK driver are sharing the device. This is because
746access to the global NIC resources is not synchronized between multiple
747drivers. Any change to the global NIC configuration (writing to a global
748register, setting global configuration by AQ, or changing switch modes) will
749affect all ports and drivers on the device. Loading DPDK with the
750"multi-driver" module parameter may mitigate some of the issues.
751
752TC0 must be enabled when setting up DCB on a switch
753---------------------------------------------------
754The kernel assumes that TC0 is available, and will disable Priority Flow
755Control (PFC) on the device if TC0 is not available. To fix this, ensure TC0 is
756enabled when setting up DCB on your switch.
757
758
759Support
760=======
761For general information, go to the Intel support website at:
762https://www.intel.com/support/
763
764If an issue is identified with the released source code on a supported kernel
765with a supported adapter, email the specific information related to the issue
766to intel-wired-lan@lists.osuosl.org.
767