1 .. SPDX-License-Identifier: GPL-2.0+
3 =================================================================
4 Linux Base Driver for the Intel(R) Ethernet Controller 800 Series
5 =================================================================
7 Intel ice Linux driver.
8 Copyright(c) 2018-2021 Intel Corporation.
14 - Identifying Your Adapter
16 - Additional Features & Configurations
17 - Performance Optimization
20 The associated Virtual Function (VF) driver for this driver is iavf.
22 Driver information can be obtained using ethtool and lspci.
24 For questions related to hardware requirements, refer to the documentation
25 supplied with your Intel adapter. All hardware requirements listed apply to use
28 This driver supports XDP (Express Data Path) and AF_XDP zero-copy. Note that
29 XDP is blocked for frame sizes larger than 3KB.
32 Identifying Your Adapter
33 ========================
34 For information on how to identify your adapter, and for the latest Intel
35 network drivers, refer to the Intel Support website:
36 https://www.intel.com/support
42 Packet drops may occur under receive stress
43 -------------------------------------------
44 Devices based on the Intel(R) Ethernet Controller 800 Series are designed to
45 tolerate a limited amount of system latency during PCIe and DMA transactions.
46 If these transactions take longer than the tolerated latency, it can impact the
47 length of time the packets are buffered in the device and associated memory,
48 which may result in dropped packets. These packets drops typically do not have
49 a noticeable impact on throughput and performance under standard workloads.
51 If these packet drops appear to affect your workload, the following may improve
54 1) Make sure that your system's physical memory is in a high-performance
55 configuration, as recommended by the platform vendor. A common
56 recommendation is for all channels to be populated with a single DIMM
58 2) In your system's BIOS/UEFI settings, select the "Performance" profile.
59 3) Your distribution may provide tools like "tuned," which can help tweak
60 kernel settings to achieve better standard settings for different workloads.
63 Configuring SR-IOV for improved network security
64 ------------------------------------------------
65 In a virtualized environment, on Intel(R) Ethernet Network Adapters that
66 support SR-IOV, the virtual function (VF) may be subject to malicious behavior.
67 Software-generated layer two frames, like IEEE 802.3x (link flow control), IEEE
68 802.1Qbb (priority based flow-control), and others of this type, are not
69 expected and can throttle traffic between the host and the virtual switch,
70 reducing performance. To resolve this issue, and to ensure isolation from
71 unintended traffic streams, configure all SR-IOV enabled ports for VLAN tagging
72 from the administrative interface on the PF. This configuration allows
73 unexpected, and potentially malicious, frames to be dropped.
75 See "Configuring VLAN Tagging on SR-IOV Enabled Adapter Ports" later in this
76 README for configuration instructions.
79 Do not unload port driver if VF with active VM is bound to it
80 -------------------------------------------------------------
81 Do not unload a port's driver if a Virtual Function (VF) with an active Virtual
82 Machine (VM) is bound to it. Doing so will cause the port to appear to hang.
83 Once the VM shuts down, or otherwise releases the VF, the command will
87 Additional Features and Configurations
88 ======================================
92 The driver utilizes the ethtool interface for driver configuration and
93 diagnostics, as well as displaying statistical information. The latest ethtool
94 version is required for this functionality. Download it at:
95 https://kernel.org/pub/software/network/ethtool/
97 NOTE: The rx_bytes value of ethtool does not match the rx_bytes value of
98 Netdev, due to the 4-byte CRC being stripped by the device. The difference
99 between the two rx_bytes values will be 4 x the number of Rx packets. For
100 example, if Rx packets are 10 and Netdev (software statistics) displays
101 rx_bytes as "X", then ethtool (hardware statistics) will display rx_bytes as
102 "X+40" (4 bytes CRC x 10 packets).
105 Viewing Link Messages
106 ---------------------
107 Link messages will not be displayed to the console if the distribution is
108 restricting system messages. In order to see network driver link messages on
109 your console, set dmesg to eight by entering the following::
113 NOTE: This setting is not saved across reboots.
119 Intel(R) Ethernet Flow Director
120 -------------------------------
121 The Intel Ethernet Flow Director performs the following tasks:
123 - Directs receive packets according to their flows to different queues
124 - Enables tight control on routing a flow in the platform
125 - Matches flows and CPU cores for flow affinity
127 NOTE: This driver supports the following flow types:
138 Each flow type supports valid combinations of IP addresses (source or
139 destination) and UDP/TCP/SCTP ports (source and destination). You can supply
140 only a source IP address, a source IP address and a destination port, or any
141 combination of one or more of these four parameters.
143 NOTE: This driver allows you to filter traffic based on a user-defined flexible
144 two-byte pattern and offset by using the ethtool user-def and mask fields. Only
145 L3 and L4 flow types are supported for user-defined flexible filters. For a
146 given flow type, you must clear all Intel Ethernet Flow Director filters before
147 changing the input set (for that flow type).
150 Flow Director Filters
151 ---------------------
152 Flow Director filters are used to direct traffic that matches specified
153 characteristics. They are enabled through ethtool's ntuple interface. To enable
154 or disable the Intel Ethernet Flow Director and these filters::
156 # ethtool -K <ethX> ntuple <off|on>
158 NOTE: When you disable ntuple filters, all the user programmed filters are
159 flushed from the driver cache and hardware. All needed filters must be re-added
160 when ntuple is re-enabled.
162 To display all of the active filters::
166 To add a new filter::
168 # ethtool -U <ethX> flow-type <type> src-ip <ip> [m <ip_mask>] dst-ip <ip>
169 [m <ip_mask>] src-port <port> [m <port_mask>] dst-port <port> [m <port_mask>]
173 <ethX> - the Ethernet device to program
174 <type> - can be ip4, tcp4, udp4, sctp4, ip6, tcp6, udp6, sctp6
175 <ip> - the IP address to match on
176 <ip_mask> - the IPv4 address to mask on
177 NOTE: These filters use inverted masks.
178 <port> - the port number to match on
179 <port_mask> - the 16-bit integer for masking
180 NOTE: These filters use inverted masks.
181 <queue> - the queue to direct traffic toward (-1 discards the
186 # ethtool -U <ethX> delete <N>
188 Where <N> is the filter ID displayed when printing all the active filters,
189 and may also have been specified using "loc <N>" when adding the filter.
193 To add a filter that directs packet to queue 2::
195 # ethtool -U <ethX> flow-type tcp4 src-ip 192.168.10.1 dst-ip \
196 192.168.10.2 src-port 2000 dst-port 2001 action 2 [loc 1]
198 To set a filter using only the source and destination IP address::
200 # ethtool -U <ethX> flow-type tcp4 src-ip 192.168.10.1 dst-ip \
201 192.168.10.2 action 2 [loc 1]
203 To set a filter based on a user-defined pattern and offset::
205 # ethtool -U <ethX> flow-type tcp4 src-ip 192.168.10.1 dst-ip \
206 192.168.10.2 user-def 0x4FFFF action 2 [loc 1]
208 where the value of the user-def field contains the offset (4 bytes) and
209 the pattern (0xffff).
211 To match TCP traffic sent from 192.168.0.1, port 5300, directed to 192.168.0.5,
212 port 80, and then send it to queue 7::
214 # ethtool -U enp130s0 flow-type tcp4 src-ip 192.168.0.1 dst-ip 192.168.0.5
215 src-port 5300 dst-port 80 action 7
217 To add a TCPv4 filter with a partial mask for a source IP subnet::
219 # ethtool -U <ethX> flow-type tcp4 src-ip 192.168.0.0 m 0.255.255.255 dst-ip
220 192.168.5.12 src-port 12600 dst-port 31 action 12
224 For each flow-type, the programmed filters must all have the same matching
225 input set. For example, issuing the following two commands is acceptable::
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 src-ip 192.168.0.5 src-port 55 action 10
230 Issuing the next two commands, however, is not acceptable, since the first
231 specifies src-ip and the second specifies dst-ip::
233 # ethtool -U enp130s0 flow-type ip4 src-ip 192.168.0.1 src-port 5300 action 7
234 # ethtool -U enp130s0 flow-type ip4 dst-ip 192.168.0.5 src-port 55 action 10
236 The second command will fail with an error. You may program multiple filters
237 with the same fields, using different values, but, on one device, you may not
238 program two tcp4 filters with different matching fields.
240 The ice driver does not support matching on a subportion of a field, thus
241 partial mask fields are not supported.
244 Flex Byte Flow Director Filters
245 -------------------------------
246 The driver also supports matching user-defined data within the packet payload.
247 This flexible data is specified using the "user-def" field of the ethtool
248 command in the following way:
252 ============================== ============================
253 ``31 28 24 20 16`` ``15 12 8 4 0``
254 ``offset into packet payload`` ``2 bytes of flexible data``
255 ============================== ============================
261 ... user-def 0x4FFFF ...
263 tells the filter to look 4 bytes into the payload and match that value against
264 0xFFFF. The offset is based on the beginning of the payload, and not the
265 beginning of the packet. Thus
269 flow-type tcp4 ... user-def 0x8BEAF ...
271 would match TCP/IPv4 packets which have the value 0xBEAF 8 bytes into the
274 Note that ICMP headers are parsed as 4 bytes of header and 4 bytes of payload.
275 Thus to match the first byte of the payload, you must actually add 4 bytes to
276 the offset. Also note that ip4 filters match both ICMP frames as well as raw
277 (unknown) ip4 frames, where the payload will be the L3 payload of the IP4
280 The maximum offset is 64. The hardware will only read up to 64 bytes of data
281 from the payload. The offset must be even because the flexible data is 2 bytes
282 long and must be aligned to byte 0 of the packet payload.
284 The user-defined flexible offset is also considered part of the input set and
285 cannot be programmed separately for multiple filters of the same type. However,
286 the flexible data is not part of the input set and multiple filters may use the
287 same offset but match against different data.
292 Allows you to set the hash bytes per flow type and any combination of one or
293 more options for Receive Side Scaling (RSS) hash byte configuration.
297 # ethtool -N <ethX> rx-flow-hash <type> <option>
300 tcp4 signifying TCP over IPv4
301 udp4 signifying UDP over IPv4
302 tcp6 signifying TCP over IPv6
303 udp6 signifying UDP over IPv6
304 And <option> is one or more of:
305 s Hash on the IP source address of the Rx packet.
306 d Hash on the IP destination address of the Rx packet.
307 f Hash on bytes 0 and 1 of the Layer 4 header of the Rx packet.
308 n Hash on bytes 2 and 3 of the Layer 4 header of the Rx packet.
311 Accelerated Receive Flow Steering (aRFS)
312 ----------------------------------------
313 Devices based on the Intel(R) Ethernet Controller 800 Series support
314 Accelerated Receive Flow Steering (aRFS) on the PF. aRFS is a load-balancing
315 mechanism that allows you to direct packets to the same CPU where an
316 application is running or consuming the packets in that flow.
320 - aRFS requires that ntuple filtering is enabled via ethtool.
321 - aRFS support is limited to the following packet types:
323 - TCP over IPv4 and IPv6
324 - UDP over IPv4 and IPv6
325 - Nonfragmented packets
327 - aRFS only supports Flow Director filters, which consist of the
328 source/destination IP addresses and source/destination ports.
329 - aRFS and ethtool's ntuple interface both use the device's Flow Director. aRFS
330 and ntuple features can coexist, but you may encounter unexpected results if
331 there's a conflict between aRFS and ntuple requests. See "Intel(R) Ethernet
332 Flow Director" for additional information.
336 1. Enable the Intel Ethernet Flow Director and ntuple filters using ethtool.
340 # ethtool -K <ethX> ntuple on
342 2. Set up the number of entries in the global flow table. For example:
346 # NUM_RPS_ENTRIES=16384
347 # echo $NUM_RPS_ENTRIES > /proc/sys/net/core/rps_sock_flow_entries
349 3. Set up the number of entries in the per-queue flow table. For example:
354 # for file in /sys/class/net/$IFACE/queues/rx-*/rps_flow_cnt; do
355 # echo $(($NUM_RPS_ENTRIES/$NUM_RX_QUEUES)) > $file;
358 4. Disable the IRQ balance daemon (this is only a temporary stop of the service
359 until the next reboot).
363 # systemctl stop irqbalance
365 5. Configure the interrupt affinity.
367 See ``/Documentation/core-api/irq/irq-affinity.rst``
370 To disable aRFS using ethtool::
372 # ethtool -K <ethX> ntuple off
374 NOTE: This command will disable ntuple filters and clear any aRFS filters in
375 software and hardware.
379 1. Set the server application on the desired CPU (e.g., CPU 4).
383 # taskset -c 4 netserver
385 2. Use netperf to route traffic from the client to CPU 4 on the server with
386 aRFS configured. This example uses TCP over IPv4.
390 # netperf -H <Host IPv4 Address> -t TCP_STREAM
393 Enabling Virtual Functions (VFs)
394 --------------------------------
395 Use sysfs to enable virtual functions (VF).
397 For example, you can create 4 VFs as follows::
399 # echo 4 > /sys/class/net/<ethX>/device/sriov_numvfs
401 To disable VFs, write 0 to the same file::
403 # echo 0 > /sys/class/net/<ethX>/device/sriov_numvfs
405 The maximum number of VFs for the ice driver is 256 total (all ports). To check
406 how many VFs each PF supports, use the following command::
408 # cat /sys/class/net/<ethX>/device/sriov_totalvfs
410 Note: You cannot use SR-IOV when link aggregation (LAG)/bonding is active, and
411 vice versa. To enforce this, the driver checks for this mutual exclusion.
414 Displaying VF Statistics on the PF
415 ----------------------------------
416 Use the following command to display the statistics for the PF and its VFs::
418 # ip -s link show dev <ethX>
420 NOTE: The output of this command can be very large due to the maximum number of
423 The PF driver will display a subset of the statistics for the PF and for all
424 VFs that are configured. The PF will always print a statistics block for each
425 of the possible VFs, and it will show zero for all unconfigured VFs.
428 Configuring VLAN Tagging on SR-IOV Enabled Adapter Ports
429 --------------------------------------------------------
430 To configure VLAN tagging for the ports on an SR-IOV enabled adapter, use the
431 following command. The VLAN configuration should be done before the VF driver
432 is loaded or the VM is booted. The VF is not aware of the VLAN tag being
433 inserted on transmit and removed on received frames (sometimes called "port
438 # ip link set dev <ethX> vf <id> vlan <vlan id>
440 For example, the following will configure PF eth0 and the first VF on VLAN 10::
442 # ip link set dev eth0 vf 0 vlan 10
445 Enabling a VF link if the port is disconnected
446 ----------------------------------------------
447 If the physical function (PF) link is down, you can force link up (from the
448 host PF) on any virtual functions (VF) bound to the PF.
450 For example, to force link up on VF 0 bound to PF eth0::
452 # ip link set eth0 vf 0 state enable
454 Note: If the command does not work, it may not be supported by your system.
457 Setting the MAC Address for a VF
458 --------------------------------
459 To change the MAC address for the specified VF::
461 # ip link set <ethX> vf 0 mac <address>
465 # ip link set <ethX> vf 0 mac 00:01:02:03:04:05
467 This setting lasts until the PF is reloaded.
469 NOTE: Assigning a MAC address for a VF from the host will disable any
470 subsequent requests to change the MAC address from within the VM. This is a
471 security feature. The VM is not aware of this restriction, so if this is
472 attempted in the VM, it will trigger MDD events.
475 Trusted VFs and VF Promiscuous Mode
476 -----------------------------------
477 This feature allows you to designate a particular VF as trusted and allows that
478 trusted VF to request selective promiscuous mode on the Physical Function (PF).
480 To set a VF as trusted or untrusted, enter the following command in the
483 # ip link set dev <ethX> vf 1 trust [on|off]
485 NOTE: It's important to set the VF to trusted before setting promiscuous mode.
486 If the VM is not trusted, the PF will ignore promiscuous mode requests from the
487 VF. If the VM becomes trusted after the VF driver is loaded, you must make a
488 new request to set the VF to promiscuous.
490 Once the VF is designated as trusted, use the following commands in the VM to
491 set the VF to promiscuous mode.
493 For promiscuous all::
495 # ip link set <ethX> promisc on
496 Where <ethX> is a VF interface in the VM
498 For promiscuous Multicast::
500 # ip link set <ethX> allmulticast on
501 Where <ethX> is a VF interface in the VM
503 NOTE: By default, the ethtool private flag vf-true-promisc-support is set to
504 "off," meaning that promiscuous mode for the VF will be limited. To set the
505 promiscuous mode for the VF to true promiscuous and allow the VF to see all
506 ingress traffic, use the following command::
508 # ethtool --set-priv-flags <ethX> vf-true-promisc-support on
510 The vf-true-promisc-support private flag does not enable promiscuous mode;
511 rather, it designates which type of promiscuous mode (limited or true) you will
512 get when you enable promiscuous mode using the ip link commands above. Note
513 that this is a global setting that affects the entire device. However, the
514 vf-true-promisc-support private flag is only exposed to the first PF of the
515 device. The PF remains in limited promiscuous mode regardless of the
516 vf-true-promisc-support setting.
518 Next, add a VLAN interface on the VF interface. For example::
520 # ip link add link eth2 name eth2.100 type vlan id 100
522 Note that the order in which you set the VF to promiscuous mode and add the
523 VLAN interface does not matter (you can do either first). The result in this
524 example is that the VF will get all traffic that is tagged with VLAN 100.
527 Malicious Driver Detection (MDD) for VFs
528 ----------------------------------------
529 Some Intel Ethernet devices use Malicious Driver Detection (MDD) to detect
530 malicious traffic from the VF and disable Tx/Rx queues or drop the offending
531 packet until a VF driver reset occurs. You can view MDD messages in the PF's
532 system log using the dmesg command.
534 - If the PF driver logs MDD events from the VF, confirm that the correct VF
536 - To restore functionality, you can manually reload the VF or VM or enable
538 - When automatic VF resets are enabled, the PF driver will immediately reset
539 the VF and reenable queues when it detects MDD events on the receive path.
540 - If automatic VF resets are disabled, the PF will not automatically reset the
541 VF when it detects MDD events.
543 To enable or disable automatic VF resets, use the following command::
545 # ethtool --set-priv-flags <ethX> mdd-auto-reset-vf on|off
548 MAC and VLAN Anti-Spoofing Feature for VFs
549 ------------------------------------------
550 When a malicious driver on a Virtual Function (VF) interface attempts to send a
551 spoofed packet, it is dropped by the hardware and not transmitted.
553 NOTE: This feature can be disabled for a specific VF::
555 # ip link set <ethX> vf <vf id> spoofchk {off|on}
560 Jumbo Frames support is enabled by changing the Maximum Transmission Unit (MTU)
561 to a value larger than the default value of 1500.
563 Use the ifconfig command to increase the MTU size. For example, enter the
564 following where <ethX> is the interface number::
566 # ifconfig <ethX> mtu 9000 up
568 Alternatively, you can use the ip command as follows::
570 # ip link set mtu 9000 dev <ethX>
571 # ip link set up dev <ethX>
573 This setting is not saved across reboots.
576 NOTE: The maximum MTU setting for jumbo frames is 9702. This corresponds to the
577 maximum jumbo frame size of 9728 bytes.
579 NOTE: This driver will attempt to use multiple page sized buffers to receive
580 each jumbo packet. This should help to avoid buffer starvation issues when
581 allocating receive packets.
583 NOTE: Packet loss may have a greater impact on throughput when you use jumbo
584 frames. If you observe a drop in performance after enabling jumbo frames,
585 enabling flow control may mitigate the issue.
588 Speed and Duplex Configuration
589 ------------------------------
590 In addressing speed and duplex configuration issues, you need to distinguish
591 between copper-based adapters and fiber-based adapters.
593 In the default mode, an Intel(R) Ethernet Network Adapter using copper
594 connections will attempt to auto-negotiate with its link partner to determine
595 the best setting. If the adapter cannot establish link with the link partner
596 using auto-negotiation, you may need to manually configure the adapter and link
597 partner to identical settings to establish link and pass packets. This should
598 only be needed when attempting to link with an older switch that does not
599 support auto-negotiation or one that has been forced to a specific speed or
600 duplex mode. Your link partner must match the setting you choose. 1 Gbps speeds
601 and higher cannot be forced. Use the autonegotiation advertising setting to
602 manually set devices for 1 Gbps and higher.
604 Speed, duplex, and autonegotiation advertising are configured through the
605 ethtool utility. For the latest version, download and install ethtool from the
608 https://kernel.org/pub/software/network/ethtool/
610 To see the speed configurations your device supports, run the following::
614 Caution: Only experienced network administrators should force speed and duplex
615 or change autonegotiation advertising manually. The settings at the switch must
616 always match the adapter settings. Adapter performance may suffer or your
617 adapter may not operate if you configure the adapter differently from your
621 Data Center Bridging (DCB)
622 --------------------------
623 NOTE: The kernel assumes that TC0 is available, and will disable Priority Flow
624 Control (PFC) on the device if TC0 is not available. To fix this, ensure TC0 is
625 enabled when setting up DCB on your switch.
627 DCB is a configuration Quality of Service implementation in hardware. It uses
628 the VLAN priority tag (802.1p) to filter traffic. That means that there are 8
629 different priorities that traffic can be filtered into. It also enables
630 priority flow control (802.1Qbb) which can limit or eliminate the number of
631 dropped packets during network stress. Bandwidth can be allocated to each of
632 these priorities, which is enforced at the hardware level (802.1Qaz).
634 DCB is normally configured on the network using the DCBX protocol (802.1Qaz), a
635 specialization of LLDP (802.1AB). The ice driver supports the following
636 mutually exclusive variants of DCBX support:
638 1) Firmware-based LLDP Agent
639 2) Software-based LLDP Agent
641 In firmware-based mode, firmware intercepts all LLDP traffic and handles DCBX
642 negotiation transparently for the user. In this mode, the adapter operates in
643 "willing" DCBX mode, receiving DCB settings from the link partner (typically a
644 switch). The local user can only query the negotiated DCB configuration. For
645 information on configuring DCBX parameters on a switch, please consult the
646 switch manufacturer's documentation.
648 In software-based mode, LLDP traffic is forwarded to the network stack and user
649 space, where a software agent can handle it. In this mode, the adapter can
650 operate in either "willing" or "nonwilling" DCBX mode and DCB configuration can
651 be both queried and set locally. This mode requires the FW-based LLDP Agent to
656 - You can enable and disable the firmware-based LLDP Agent using an ethtool
657 private flag. Refer to the "FW-LLDP (Firmware Link Layer Discovery Protocol)"
658 section in this README for more information.
659 - In software-based DCBX mode, you can configure DCB parameters using software
660 LLDP/DCBX agents that interface with the Linux kernel's DCB Netlink API. We
661 recommend using OpenLLDP as the DCBX agent when running in software mode. For
662 more information, see the OpenLLDP man pages and
663 https://github.com/intel/openlldp.
664 - The driver implements the DCB netlink interface layer to allow the user space
665 to communicate with the driver and query DCB configuration for the port.
666 - iSCSI with DCB is not supported.
669 FW-LLDP (Firmware Link Layer Discovery Protocol)
670 ------------------------------------------------
671 Use ethtool to change FW-LLDP settings. The FW-LLDP setting is per port and
672 persists across boots.
676 # ethtool --set-priv-flags <ethX> fw-lldp-agent on
680 # ethtool --set-priv-flags <ethX> fw-lldp-agent off
682 To check the current LLDP setting::
684 # ethtool --show-priv-flags <ethX>
686 NOTE: You must enable the UEFI HII "LLDP Agent" attribute for this setting to
687 take effect. If "LLDP AGENT" is set to disabled, you cannot enable it from the
693 Ethernet Flow Control (IEEE 802.3x) can be configured with ethtool to enable
694 receiving and transmitting pause frames for ice. When transmit is enabled,
695 pause frames are generated when the receive packet buffer crosses a predefined
696 threshold. When receive is enabled, the transmit unit will halt for the time
697 delay specified when a pause frame is received.
699 NOTE: You must have a flow control capable link partner.
701 Flow Control is disabled by default.
703 Use ethtool to change the flow control settings.
705 To enable or disable Rx or Tx Flow Control::
707 # ethtool -A <ethX> rx <on|off> tx <on|off>
709 Note: This command only enables or disables Flow Control if auto-negotiation is
710 disabled. If auto-negotiation is enabled, this command changes the parameters
711 used for auto-negotiation with the link partner.
713 Note: Flow Control auto-negotiation is part of link auto-negotiation. Depending
714 on your device, you may not be able to change the auto-negotiation setting.
718 - The ice driver requires flow control on both the port and link partner. If
719 flow control is disabled on one of the sides, the port may appear to hang on
721 - You may encounter issues with link-level flow control (LFC) after disabling
722 DCB. The LFC status may show as enabled but traffic is not paused. To resolve
723 this issue, disable and reenable LFC using ethtool::
725 # ethtool -A <ethX> rx off tx off
726 # ethtool -A <ethX> rx on tx on
732 This driver supports NAPI (Rx polling mode).
734 See :ref:`Documentation/networking/napi.rst <napi>` for more information.
738 This driver supports MACVLAN. Kernel support for MACVLAN can be tested by
739 checking if the MACVLAN driver is loaded. You can run 'lsmod | grep macvlan' to
740 see if the MACVLAN driver is loaded or run 'modprobe macvlan' to try to load
745 - In passthru mode, you can only set up one MACVLAN device. It will inherit the
746 MAC address of the underlying PF (Physical Function) device.
749 IEEE 802.1ad (QinQ) Support
750 ---------------------------
751 The IEEE 802.1ad standard, informally known as QinQ, allows for multiple VLAN
752 IDs within a single Ethernet frame. VLAN IDs are sometimes referred to as
753 "tags," and multiple VLAN IDs are thus referred to as a "tag stack." Tag stacks
754 allow L2 tunneling and the ability to segregate traffic within a particular
755 VLAN ID, among other uses.
759 - Receive checksum offloads and VLAN acceleration are not supported for 802.1ad
762 - 0x88A8 traffic will not be received unless VLAN stripping is disabled with
763 the following command::
765 # ethtool -K <ethX> rxvlan off
767 - 0x88A8/0x8100 double VLANs cannot be used with 0x8100 or 0x8100/0x8100 VLANS
768 configured on the same port. 0x88a8/0x8100 traffic will not be received if
769 0x8100 VLANs are configured.
771 - The VF can only transmit 0x88A8/0x8100 (i.e., 802.1ad/802.1Q) traffic if:
773 1) The VF is not assigned a port VLAN.
774 2) spoofchk is disabled from the PF. If you enable spoofchk, the VF will
775 not transmit 0x88A8/0x8100 traffic.
777 - The VF may not receive all network traffic based on the Inner VLAN header
778 when VF true promiscuous mode (vf-true-promisc-support) and double VLANs are
779 enabled in SR-IOV mode.
781 The following are examples of how to configure 802.1ad (QinQ)::
783 # ip link add link eth0 eth0.24 type vlan proto 802.1ad id 24
784 # ip link add link eth0.24 eth0.24.371 type vlan proto 802.1Q id 371
786 Where "24" and "371" are example VLAN IDs.
789 Tunnel/Overlay Stateless Offloads
790 ---------------------------------
791 Supported tunnels and overlays include VXLAN, GENEVE, and others depending on
792 hardware and software configuration. Stateless offloads are enabled by default.
794 To view the current state of all offloads::
799 UDP Segmentation Offload
800 ------------------------
801 Allows the adapter to offload transmit segmentation of UDP packets with
802 payloads up to 64K into valid Ethernet frames. Because the adapter hardware is
803 able to complete data segmentation much faster than operating system software,
804 this feature may improve transmission performance.
805 In addition, the adapter may use fewer CPU resources.
809 - The application sending UDP packets must support UDP segmentation offload.
811 To enable/disable UDP Segmentation Offload, issue the following command::
813 # ethtool -K <ethX> tx-udp-segmentation [off|on]
818 Requires kernel compiled with CONFIG_GNSS=y or CONFIG_GNSS=m.
819 Allows user to read messages from the GNSS hardware module and write supported
820 commands. If the module is physically present, a GNSS device is spawned:
822 The protocol of write command is dependent on the GNSS hardware module as the
823 driver writes raw bytes by the GNSS object to the receiver through i2c. Please
824 refer to the hardware GNSS module documentation for configuration details.
827 Performance Optimization
828 ========================
829 Driver defaults are meant to fit a wide variety of workloads, but if further
830 optimization is required, we recommend experimenting with the following
834 Rx Descriptor Ring Size
835 -----------------------
836 To reduce the number of Rx packet discards, increase the number of Rx
837 descriptors for each Rx ring using ethtool.
839 Check if the interface is dropping Rx packets due to buffers being full
840 (rx_dropped.nic can mean that there is no PCIe bandwidth)::
842 # ethtool -S <ethX> | grep "rx_dropped"
844 If the previous command shows drops on queues, it may help to increase
845 the number of descriptors using 'ethtool -G'::
847 # ethtool -G <ethX> rx <N>
848 Where <N> is the desired number of ring entries/descriptors
850 This can provide temporary buffering for issues that create latency while
851 the CPUs process descriptors.
854 Interrupt Rate Limiting
855 -----------------------
856 This driver supports an adaptive interrupt throttle rate (ITR) mechanism that
857 is tuned for general workloads. The user can customize the interrupt rate
858 control for specific workloads, via ethtool, adjusting the number of
859 microseconds between interrupts.
861 To set the interrupt rate manually, you must disable adaptive mode::
863 # ethtool -C <ethX> adaptive-rx off adaptive-tx off
865 For lower CPU utilization:
867 Disable adaptive ITR and lower Rx and Tx interrupts. The examples below
868 affect every queue of the specified interface.
870 Setting rx-usecs and tx-usecs to 80 will limit interrupts to about
871 12,500 interrupts per second per queue::
873 # ethtool -C <ethX> adaptive-rx off adaptive-tx off rx-usecs 80 tx-usecs 80
877 Disable adaptive ITR and ITR by setting rx-usecs and tx-usecs to 0
880 # ethtool -C <ethX> adaptive-rx off adaptive-tx off rx-usecs 0 tx-usecs 0
882 Per-queue interrupt rate settings:
884 The following examples are for queues 1 and 3, but you can adjust other
887 To disable Rx adaptive ITR and set static Rx ITR to 10 microseconds or
888 about 100,000 interrupts/second, for queues 1 and 3::
890 # ethtool --per-queue <ethX> queue_mask 0xa --coalesce adaptive-rx off
893 To show the current coalesce settings for queues 1 and 3::
895 # ethtool --per-queue <ethX> queue_mask 0xa --show-coalesce
897 Bounding interrupt rates using rx-usecs-high:
899 :Valid Range: 0-236 (0=no limit)
901 The range of 0-236 microseconds provides an effective range of 4,237 to
902 250,000 interrupts per second. The value of rx-usecs-high can be set
903 independently of rx-usecs and tx-usecs in the same ethtool command, and is
904 also independent of the adaptive interrupt moderation algorithm. The
905 underlying hardware supports granularity in 4-microsecond intervals, so
906 adjacent values may result in the same interrupt rate.
908 The following command would disable adaptive interrupt moderation, and allow
909 a maximum of 5 microseconds before indicating a receive or transmit was
910 complete. However, instead of resulting in as many as 200,000 interrupts per
911 second, it limits total interrupts per second to 50,000 via the rx-usecs-high
916 # ethtool -C <ethX> adaptive-rx off adaptive-tx off rx-usecs-high 20
917 rx-usecs 5 tx-usecs 5
920 Virtualized Environments
921 ------------------------
922 In addition to the other suggestions in this section, the following may be
923 helpful to optimize performance in VMs.
925 Using the appropriate mechanism (vcpupin) in the VM, pin the CPUs to
926 individual LCPUs, making sure to use a set of CPUs included in the
927 device's local_cpulist: ``/sys/class/net/<ethX>/device/local_cpulist``.
929 Configure as many Rx/Tx queues in the VM as available. (See the iavf driver
930 documentation for the number of queues supported.) For example::
932 # ethtool -L <virt_interface> rx <max> tx <max>
937 For general information, go to the Intel support website at:
938 https://www.intel.com/support/
940 If an issue is identified with the released source code on a supported kernel
941 with a supported adapter, email the specific information related to the issue
942 to intel-wired-lan@lists.osuosl.org.
947 Intel is a trademark or registered trademark of Intel Corporation or its
948 subsidiaries in the United States and/or other countries.
950 * Other names and brands may be claimed as the property of others.