1 .. SPDX-License-Identifier: BSD-3-Clause
2 Copyright(c) 2018 Intel Corporation.
7 The ice PMD (**librte_net_ice**) provides poll mode driver support for
8 10/25/50/100 Gbps Intel® Ethernet 800 Series Network Adapters based on
9 the Intel Ethernet Controller E810 and Intel Ethernet Connection E822/E823.
14 - Follow the DPDK :ref:`Getting Started Guide for Linux <linux_gsg>` to setup the basic DPDK environment.
16 - To get better performance on Intel platforms, please follow the "How to get best performance with NICs on Intel platforms"
17 section of the :ref:`Getting Started Guide for Linux <linux_gsg>`.
19 - Please follow the matching list to download specific kernel driver, firmware and DDP package from
20 `https://www.intel.com/content/www/us/en/search.html?ws=text#q=e810&t=Downloads&layout=table`.
22 - To understand what is DDP package and how it works, please review `Intel® Ethernet Controller E810 Dynamic
23 Device Personalization (DDP) for Telecommunications Technology Guide <https://cdrdv2.intel.com/v1/dl/getContent/617015>`_.
25 - To understand DDP for COMMs usage with DPDK, please review `Intel® Ethernet 800 Series Telecommunication (Comms)
26 Dynamic Device Personalization (DDP) Package <https://cdrdv2.intel.com/v1/dl/getContent/618651>`_.
31 - Follow the :doc:`guide for Windows <../windows_gsg/run_apps>`
32 to setup the basic DPDK environment.
34 - Identify the Intel® Ethernet adapter and get the latest NVM/FW version.
36 - To access any Intel® Ethernet hardware, load the NetUIO driver in place of existing built-in (inbox) driver.
38 - To load NetUIO driver, follow the steps mentioned in `dpdk-kmods repository
39 <https://git.dpdk.org/dpdk-kmods/tree/windows/netuio/README.rst>`_.
41 - Loading of private Dynamic Device Personalization (DDP) package is not supported on Windows.
44 Recommended Matching List
45 -------------------------
47 It is highly recommended to upgrade the ice kernel driver, firmware and DDP package
48 to avoid the compatibility issues with ice PMD.
49 Here is the suggested matching list which has been tested and verified.
50 The detailed information can refer to chapter Tested Platforms/Tested NICs in release notes.
52 +-----------+---------------+-----------------+-----------+--------------+-----------+
53 | DPDK | Kernel Driver | OS Default DDP | COMMS DDP | Wireless DDP | Firmware |
54 +===========+===============+=================+===========+==============+===========+
55 | 20.11 | 1.3.2 | 1.3.20 | 1.3.24 | N/A | 2.3 |
56 +-----------+---------------+-----------------+-----------+--------------+-----------+
57 | 21.02 | 1.4.11 | 1.3.24 | 1.3.28 | 1.3.4 | 2.4 |
58 +-----------+---------------+-----------------+-----------+--------------+-----------+
59 | 21.05 | 1.6.5 | 1.3.26 | 1.3.30 | 1.3.6 | 3.0 |
60 +-----------+---------------+-----------------+-----------+--------------+-----------+
61 | 21.08 | 1.7.16 | 1.3.27 | 1.3.31 | 1.3.7 | 3.1 |
62 +-----------+---------------+-----------------+-----------+--------------+-----------+
63 | 21.11 | 1.7.16 | 1.3.27 | 1.3.31 | 1.3.7 | 3.1 |
64 +-----------+---------------+-----------------+-----------+--------------+-----------+
65 | 22.03 | 1.8.3 | 1.3.28 | 1.3.35 | 1.3.8 | 3.2 |
66 +-----------+---------------+-----------------+-----------+--------------+-----------+
68 Pre-Installation Configuration
69 ------------------------------
72 Runtime Config Options
73 ~~~~~~~~~~~~~~~~~~~~~~
75 - ``Safe Mode Support`` (default ``0``)
77 If driver failed to load OS package, by default driver's initialization failed.
78 But if user intend to use the device without OS package, user can take ``devargs``
79 parameter ``safe-mode-support``, for example::
81 -a 80:00.0,safe-mode-support=1
83 Then the driver will be initialized successfully and the device will enter Safe Mode.
84 NOTE: In Safe mode, only very limited features are available, features like RSS,
85 checksum, fdir, tunneling ... are all disabled.
87 - ``Generic Flow Pipeline Mode Support`` (default ``0``)
89 In pipeline mode, a flow can be set at one specific stage by setting parameter
90 ``priority``. Currently, we support two stages: priority = 0 or !0. Flows with
91 priority 0 located at the first pipeline stage which typically be used as a firewall
92 to drop the packet on a blocklist(we called it permission stage). At this stage,
93 flow rules are created for the device's exact match engine: switch. Flows with priority
94 !0 located at the second stage, typically packets are classified here and be steered to
95 specific queue or queue group (we called it distribution stage), At this stage, flow
96 rules are created for device's flow director engine.
97 For none-pipeline mode, ``priority`` is ignored, a flow rule can be created as a flow director
98 rule or a switch rule depends on its pattern/action and the resource allocation situation,
99 all flows are virtually at the same pipeline stage.
100 By default, generic flow API is enabled in none-pipeline mode, user can choose to
101 use pipeline mode by setting ``devargs`` parameter ``pipeline-mode-support``,
104 -a 80:00.0,pipeline-mode-support=1
106 - ``Protocol extraction for per queue``
108 Configure the RX queues to do protocol extraction into mbuf for protocol
109 handling acceleration, like checking the TCP SYN packets quickly.
111 The argument format is::
113 -a 18:00.0,proto_xtr=<queues:protocol>[<queues:protocol>...]
114 -a 18:00.0,proto_xtr=<protocol>
116 Queues are grouped by ``(`` and ``)`` within the group. The ``-`` character
117 is used as a range separator and ``,`` is used as a single number separator.
118 The grouping ``()`` can be omitted for single element group. If no queues are
119 specified, PMD will use this protocol extraction type for all queues.
121 Protocol is : ``vlan, ipv4, ipv6, ipv6_flow, tcp, ip_offset``.
123 .. code-block:: console
125 dpdk-testpmd -a 18:00.0,proto_xtr='[(1,2-3,8-9):tcp,10-13:vlan]'
127 This setting means queues 1, 2-3, 8-9 are TCP extraction, queues 10-13 are
128 VLAN extraction, other queues run with no protocol extraction.
130 .. code-block:: console
132 dpdk-testpmd -a 18:00.0,proto_xtr=vlan,proto_xtr='[(1,2-3,8-9):tcp,10-23:ipv6]'
134 This setting means queues 1, 2-3, 8-9 are TCP extraction, queues 10-23 are
135 IPv6 extraction, other queues use the default VLAN extraction.
137 The extraction metadata is copied into the registered dynamic mbuf field, and
138 the related dynamic mbuf flags is set.
140 .. table:: Protocol extraction : ``vlan``
142 +----------------------------+----------------------------+
144 +======+===+=================+======+===+=================+
145 | PCP | D | VID | PCP | D | VID |
146 +------+---+-----------------+------+---+-----------------+
148 VLAN1 - single or EVLAN (first for QinQ).
150 VLAN2 - C-VLAN (second for QinQ).
152 .. table:: Protocol extraction : ``ipv4``
154 +----------------------------+----------------------------+
156 +======+=======+=============+==============+=============+
157 | Ver |Hdr Len| ToS | TTL | Protocol |
158 +------+-------+-------------+--------------+-------------+
160 IPHDR1 - IPv4 header word 4, "TTL" and "Protocol" fields.
162 IPHDR2 - IPv4 header word 0, "Ver", "Hdr Len" and "Type of Service" fields.
164 .. table:: Protocol extraction : ``ipv6``
166 +----------------------------+----------------------------+
168 +=====+=============+========+=============+==============+
169 | Ver |Traffic class| Flow | Next Header | Hop Limit |
170 +-----+-------------+--------+-------------+--------------+
172 IPHDR1 - IPv6 header word 3, "Next Header" and "Hop Limit" fields.
174 IPHDR2 - IPv6 header word 0, "Ver", "Traffic class" and high 4 bits of
177 .. table:: Protocol extraction : ``ipv6_flow``
179 +----------------------------+----------------------------+
181 +=====+=============+========+============================+
182 | Ver |Traffic class| Flow Label |
183 +-----+-------------+-------------------------------------+
185 IPHDR1 - IPv6 header word 1, 16 low bits of the "Flow Label" field.
187 IPHDR2 - IPv6 header word 0, "Ver", "Traffic class" and high 4 bits of
190 .. table:: Protocol extraction : ``tcp``
192 +----------------------------+----------------------------+
193 | TCPHDR2 | TCPHDR1 |
194 +============================+======+======+==============+
195 | Reserved |Offset| RSV | Flags |
196 +----------------------------+------+------+--------------+
198 TCPHDR1 - TCP header word 6, "Data Offset" and "Flags" fields.
202 .. table:: Protocol extraction : ``ip_offset``
204 +----------------------------+----------------------------+
206 +============================+============================+
207 | IPv6 HDR Offset | IPv4 HDR Offset |
208 +----------------------------+----------------------------+
210 IPHDR1 - Outer/Single IPv4 Header offset.
212 IPHDR2 - Outer/Single IPv6 Header offset.
214 Use ``rte_net_ice_dynf_proto_xtr_metadata_get`` to access the protocol
215 extraction metadata, and use ``RTE_PKT_RX_DYNF_PROTO_XTR_*`` to get the
216 metadata type of ``struct rte_mbuf::ol_flags``.
218 The ``rte_net_ice_dump_proto_xtr_metadata`` routine shows how to
219 access the protocol extraction result in ``struct rte_mbuf``.
221 - ``Hardware debug mask log support`` (default ``0``)
223 User can enable the related hardware debug mask such as ICE_DBG_NVM::
225 -a 0000:88:00.0,hw_debug_mask=0x80 --log-level=pmd.net.ice.driver:8
227 These ICE_DBG_XXX are defined in ``drivers/net/ice/base/ice_type.h``.
229 - ``1PPS out support``
231 The E810 supports four single-ended GPIO signals (SDP[20:23]). The 1PPS
232 signal outputs via SDP[20:23]. User can select GPIO pin index flexibly.
233 Pin index 0 means SDP20, 1 means SDP21 and so on. For example::
235 -a af:00.0,pps_out='[pin:0]'
237 - ``Low Rx latency`` (default ``0``)
239 vRAN workloads require low latency DPDK interface for the front haul
240 interface connection to Radio. By specifying ``1`` for parameter
241 ``rx_low_latency``, each completed Rx descriptor can be written immediately
242 to host memory and the Rx interrupt latency can be reduced to 2us::
244 -a 0000:88:00.0,rx_low_latency=1
246 As a trade-off, this configuration may cause the packet processing performance
247 degradation due to the PCI bandwidth limitation.
249 Driver compilation and testing
250 ------------------------------
252 Refer to the document :ref:`compiling and testing a PMD for a NIC <pmd_build_and_test>`
261 Vector PMD for RX and TX path are selected automatically. The paths
262 are chosen based on 2 conditions.
265 On the X86 platform, the driver checks if the CPU supports AVX2.
266 If it's supported, AVX2 paths will be chosen. If not, SSE is chosen.
267 If the CPU supports AVX512 and EAL argument ``--force-max-simd-bitwidth``
268 is set to 512, AVX512 paths will be chosen.
270 - ``Offload features``
271 The supported HW offload features are described in the document ice.ini,
272 A value "P" means the offload feature is not supported by vector path.
273 If any not supported features are used, ICE vector PMD is disabled and the
274 normal paths are chosen.
276 Malicious driver detection (MDD)
277 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
279 It's not appropriate to send a packet, if this packet's destination MAC address
280 is just this port's MAC address. If SW tries to send such packets, HW will
281 report a MDD event and drop the packets.
283 The APPs based on DPDK should avoid providing such packets.
285 Device Config Function (DCF)
286 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
288 This section demonstrates ICE DCF PMD, which shares the core module with ICE
291 A DCF (Device Config Function) PMD bounds to the device's trusted VF with ID 0,
292 it can act as a sole controlling entity to exercise advance functionality (such
293 as switch, ACL) for the rest VFs.
295 The DCF PMD needs to advertise and acquire DCF capability which allows DCF to
296 send AdminQ commands that it would like to execute over to the PF and receive
297 responses for the same from PF.
301 .. figure:: img/ice_dcf.*
303 DCF Communication flow.
307 echo 4 > /sys/bus/pci/devices/0000\:18\:00.0/sriov_numvfs
309 #. Enable the VF0 trust on::
311 ip link set dev enp24s0f0 vf 0 trust on
313 #. Bind the VF0, and run testpmd with 'cap=dcf' devarg::
315 dpdk-testpmd -l 22-25 -n 4 -a 18:01.0,cap=dcf -- -i
317 #. Monitor the VF2 interface network traffic::
319 tcpdump -e -nn -i enp24s1f2
321 #. Create one flow to redirect the traffic to VF2 by DCF::
323 flow create 0 priority 0 ingress pattern eth / ipv4 src is 192.168.0.2 \
324 dst is 192.168.0.3 / end actions vf id 2 / end
326 #. Send the packet, and it should be displayed on tcpdump::
328 sendp(Ether(src='3c:fd:fe:aa:bb:78', dst='00:00:00:01:02:03')/IP(src=' \
329 192.168.0.2', dst="192.168.0.3")/TCP(flags='S')/Raw(load='XXXXXXXXXX'), \
330 iface="enp24s0f0", count=10)
332 Sample Application Notes
333 ------------------------
338 Vlan filter only works when Promiscuous mode is off.
340 To start ``testpmd``, and add vlan 10 to port 0:
342 .. code-block:: console
344 ./app/dpdk-testpmd -l 0-15 -n 4 -- -i
347 testpmd> rx_vlan add 10 0
349 Limitations or Known issues
350 ---------------------------
352 The Intel E810 requires a programmable pipeline package be downloaded
353 by the driver to support normal operations. The E810 has a limited
354 functionality built in to allow PXE boot and other use cases, but the
355 driver must download a package file during the driver initialization
358 The default DDP package file name is ice.pkg. For a specific NIC, the
359 DDP package supposed to be loaded can have a filename: ice-xxxxxx.pkg,
360 where 'xxxxxx' is the 64-bit PCIe Device Serial Number of the NIC. For
361 example, if the NIC's device serial number is 00-CC-BB-FF-FF-AA-05-68,
362 the device-specific DDP package filename is ice-00ccbbffffaa0568.pkg
363 (in hex and all low case). During initialization, the driver searches
364 in the following paths in order: /lib/firmware/updates/intel/ice/ddp
365 and /lib/firmware/intel/ice/ddp. The corresponding device-specific DDP
366 package will be downloaded first if the file exists. If not, then the
367 driver tries to load the default package. The type of loaded package
368 is stored in ``ice_adapter->active_pkg_type``.
370 A symbolic link to the DDP package file is also ok. The same package
371 file is used by both the kernel driver and the DPDK PMD.
375 Windows support: The DDP package is not supported on Windows so,
376 loading of the package is disabled on Windows.