2 Copyright (C) Cavium, Inc. 2016.
5 Redistribution and use in source and binary forms, with or without
6 modification, are permitted provided that the following conditions
9 * Redistributions of source code must retain the above copyright
10 notice, this list of conditions and the following disclaimer.
11 * Redistributions in binary form must reproduce the above copyright
12 notice, this list of conditions and the following disclaimer in
13 the documentation and/or other materials provided with the
15 * Neither the name of Cavium, Inc nor the names of its
16 contributors may be used to endorse or promote products derived
17 from this software without specific prior written permission.
19 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
20 "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
21 LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
22 A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
23 OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
24 SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
25 LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
26 DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
27 THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
28 (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
29 OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
31 ThunderX NICVF Poll Mode Driver
32 ===============================
34 The ThunderX NICVF PMD (**librte_pmd_thunderx_nicvf**) provides poll mode driver
35 support for the inbuilt NIC found in the **Cavium ThunderX** SoC family
36 as well as their virtual functions (VF) in SR-IOV context.
38 More information can be found at `Cavium, Inc Official Website
39 <http://www.cavium.com/ThunderX_ARM_Processors.html>`_.
44 Features of the ThunderX PMD are:
46 - Multiple queues for TX and RX
47 - Receive Side Scaling (RSS)
48 - Packet type information
52 - Port hardware statistics
54 - Link state information
55 - Scattered and gather for TX and RX
59 - Multi queue set support (up to 96 queues (12 queue sets)) per port
61 Supported ThunderX SoCs
62 -----------------------
69 - Follow the DPDK :ref:`Getting Started Guide for Linux <linux_gsg>` to setup the basic DPDK environment.
71 Pre-Installation Configuration
72 ------------------------------
77 The following options can be modified in the ``config`` file.
78 Please note that enabling debugging options may affect system performance.
80 - ``CONFIG_RTE_LIBRTE_THUNDERX_NICVF_PMD`` (default ``y``)
82 Toggle compilation of the ``librte_pmd_thunderx_nicvf`` driver.
84 - ``CONFIG_RTE_LIBRTE_THUNDERX_NICVF_DEBUG_INIT`` (default ``n``)
86 Toggle display of initialization related messages.
88 - ``CONFIG_RTE_LIBRTE_THUNDERX_NICVF_DEBUG_RX`` (default ``n``)
90 Toggle display of receive fast path run-time message
92 - ``CONFIG_RTE_LIBRTE_THUNDERX_NICVF_DEBUG_TX`` (default ``n``)
94 Toggle display of transmit fast path run-time message
96 - ``CONFIG_RTE_LIBRTE_THUNDERX_NICVF_DEBUG_DRIVER`` (default ``n``)
98 Toggle display of generic debugging messages
100 - ``CONFIG_RTE_LIBRTE_THUNDERX_NICVF_DEBUG_MBOX`` (default ``n``)
102 Toggle display of PF mailbox related run-time check messages
104 Driver compilation and testing
105 ------------------------------
107 Refer to the document :ref:`compiling and testing a PMD for a NIC <pmd_build_and_test>`
110 To compile the ThunderX NICVF PMD for Linux arm64 gcc,
111 use arm64-thunderx-linuxapp-gcc as target.
116 SR-IOV: Prerequisites and sample Application Notes
117 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
119 Current ThunderX NIC PF/VF kernel modules maps each physical Ethernet port
120 automatically to virtual function (VF) and presented them as PCIe-like SR-IOV device.
121 This section provides instructions to configure SR-IOV with Linux OS.
123 #. Verify PF devices capabilities using ``lspci``:
125 .. code-block:: console
131 .. code-block:: console
133 0002:01:00.0 Ethernet controller: Cavium Networks Device a01e (rev 01)
135 Capabilities: [100 v1] Alternative Routing-ID Interpretation (ARI)
137 Capabilities: [180 v1] Single Root I/O Virtualization (SR-IOV)
139 Kernel driver in use: thunder-nic
144 Unless ``thunder-nic`` driver is in use make sure your kernel config includes ``CONFIG_THUNDER_NIC_PF`` setting.
146 #. Verify VF devices capabilities and drivers using ``lspci``:
148 .. code-block:: console
154 .. code-block:: console
156 0002:01:00.1 Ethernet controller: Cavium Networks Device 0011 (rev 01)
158 Capabilities: [100 v1] Alternative Routing-ID Interpretation (ARI)
160 Kernel driver in use: thunder-nicvf
163 0002:01:00.2 Ethernet controller: Cavium Networks Device 0011 (rev 01)
165 Capabilities: [100 v1] Alternative Routing-ID Interpretation (ARI)
167 Kernel driver in use: thunder-nicvf
172 Unless ``thunder-nicvf`` driver is in use make sure your kernel config includes ``CONFIG_THUNDER_NIC_VF`` setting.
174 #. Pass VF device to VM context (PCIe Passthrough):
176 The VF devices may be passed through to the guest VM using qemu or
177 virt-manager or virsh etc.
179 Example qemu guest launch command:
181 .. code-block:: console
183 sudo qemu-system-aarch64 -name vm1 \
184 -machine virt,gic_version=3,accel=kvm,usb=off \
186 -smp 4,sockets=1,cores=8,threads=1 \
187 -nographic -nodefaults \
188 -kernel <kernel image> \
189 -append "root=/dev/vda console=ttyAMA0 rw hugepagesz=512M hugepages=3" \
190 -device vfio-pci,host=0002:01:00.1 \
191 -drive file=<rootfs.ext3>,if=none,id=disk1,format=raw \
192 -device virtio-blk-device,scsi=off,drive=disk1,id=virtio-disk1,bootindex=1 \
193 -netdev tap,id=net0,ifname=tap0,script=/etc/qemu-ifup_thunder \
194 -device virtio-net-device,netdev=net0 \
198 #. Enable **VFIO-NOIOMMU** mode (optional):
200 .. code-block:: console
202 echo 1 > /sys/module/vfio/parameters/enable_unsafe_noiommu_mode
206 **VFIO-NOIOMMU** is required only when running in VM context and should not be enabled otherwise.
210 Follow instructions available in the document
211 :ref:`compiling and testing a PMD for a NIC <pmd_build_and_test>`
216 .. code-block:: console
218 ./arm64-thunderx-linuxapp-gcc/app/testpmd -l 0-3 -n 4 -w 0002:01:00.2 \
219 -- -i --disable-hw-vlan-filter --disable-crc-strip --no-flush-rx \
224 PMD: rte_nicvf_pmd_init(): librte_pmd_thunderx nicvf version 1.0
227 EAL: probe driver: 177d:11 rte_nicvf_pmd
228 EAL: using IOMMU type 1 (Type 1)
229 EAL: PCI memory mapped at 0x3ffade50000
230 EAL: Trying to map BAR 4 that contains the MSI-X table.
231 Trying offsets: 0x40000000000:0x0000, 0x10000:0x1f0000
232 EAL: PCI memory mapped at 0x3ffadc60000
233 PMD: nicvf_eth_dev_init(): nicvf: device (177d:11) 2:1:0:2
234 PMD: nicvf_eth_dev_init(): node=0 vf=1 mode=tns-bypass sqs=false
235 loopback_supported=true
236 PMD: nicvf_eth_dev_init(): Port 0 (177d:11) mac=a6:c6:d9:17:78:01
237 Interactive-mode selected
238 Configuring Port 0 (socket 0)
241 PMD: nicvf_dev_configure(): Configured ethdev port0 hwcap=0x0
242 Port 0: A6:C6:D9:17:78:01
243 Checking link statuses...
244 Port 0 Link Up - speed 10000 Mbps - full-duplex
248 Multiple Queue Set per DPDK port configuration
249 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
251 There are two types of VFs:
256 Each port consists of a primary VF and n secondary VF(s). Each VF provides 8 Tx/Rx queues to a port.
257 When a given port is configured to use more than 8 queues, it requires one (or more) secondary VF.
258 Each secondary VF adds 8 additional queues to the queue set.
260 During PMD driver initialization, the primary VF's are enumerated by checking the
261 specific flag (see sqs message in DPDK boot log - sqs indicates secondary queue set).
262 They are at the beginning of VF list (the remain ones are secondary VF's).
264 The primary VFs are used as master queue sets. Secondary VFs provide
265 additional queue sets for primary ones. If a port is configured for more then
266 8 queues than it will request for additional queues from secondary VFs.
268 Secondary VFs cannot be shared between primary VFs.
270 Primary VFs are present on the beginning of the 'Network devices using kernel
271 driver' list, secondary VFs are on the remaining on the remaining part of the list.
275 The VNIC driver in the multiqueue setup works differently than other drivers like `ixgbe`.
276 We need to bind separately each specific queue set device with the ``usertools/dpdk-devbind.py`` utility.
280 Depending on the hardware used, the kernel driver sets a threshold ``vf_id``. VFs that try to attached with an id below or equal to
281 this boundary are considered primary VFs. VFs that try to attach with an id above this boundary are considered secondary VFs.
284 Example device binding
285 ~~~~~~~~~~~~~~~~~~~~~~
287 If a system has three interfaces, a total of 18 VF devices will be created
288 on a non-NUMA machine.
292 NUMA systems have 12 VFs per port and non-NUMA 6 VFs per port.
294 .. code-block:: console
296 # usertools/dpdk-devbind.py --status
298 Network devices using DPDK-compatible driver
299 ============================================
302 Network devices using kernel driver
303 ===================================
304 0000:01:10.0 'Device a026' if= drv=thunder-BGX unused=vfio-pci,uio_pci_generic
305 0000:01:10.1 'Device a026' if= drv=thunder-BGX unused=vfio-pci,uio_pci_generic
306 0002:01:00.0 'Device a01e' if= drv=thunder-nic unused=vfio-pci,uio_pci_generic
307 0002:01:00.1 'Device 0011' if=eth0 drv=thunder-nicvf unused=vfio-pci,uio_pci_generic
308 0002:01:00.2 'Device 0011' if=eth1 drv=thunder-nicvf unused=vfio-pci,uio_pci_generic
309 0002:01:00.3 'Device 0011' if=eth2 drv=thunder-nicvf unused=vfio-pci,uio_pci_generic
310 0002:01:00.4 'Device 0011' if= drv=thunder-nicvf unused=vfio-pci,uio_pci_generic
311 0002:01:00.5 'Device 0011' if= drv=thunder-nicvf unused=vfio-pci,uio_pci_generic
312 0002:01:00.6 'Device 0011' if= drv=thunder-nicvf unused=vfio-pci,uio_pci_generic
313 0002:01:00.7 'Device 0011' if= drv=thunder-nicvf unused=vfio-pci,uio_pci_generic
314 0002:01:01.0 'Device 0011' if= drv=thunder-nicvf unused=vfio-pci,uio_pci_generic
315 0002:01:01.1 'Device 0011' if= drv=thunder-nicvf unused=vfio-pci,uio_pci_generic
316 0002:01:01.2 'Device 0011' if= drv=thunder-nicvf unused=vfio-pci,uio_pci_generic
317 0002:01:01.3 'Device 0011' if= drv=thunder-nicvf unused=vfio-pci,uio_pci_generic
318 0002:01:01.4 'Device 0011' if= drv=thunder-nicvf unused=vfio-pci,uio_pci_generic
319 0002:01:01.5 'Device 0011' if= drv=thunder-nicvf unused=vfio-pci,uio_pci_generic
320 0002:01:01.6 'Device 0011' if= drv=thunder-nicvf unused=vfio-pci,uio_pci_generic
321 0002:01:01.7 'Device 0011' if= drv=thunder-nicvf unused=vfio-pci,uio_pci_generic
322 0002:01:02.0 'Device 0011' if= drv=thunder-nicvf unused=vfio-pci,uio_pci_generic
323 0002:01:02.1 'Device 0011' if= drv=thunder-nicvf unused=vfio-pci,uio_pci_generic
324 0002:01:02.2 'Device 0011' if= drv=thunder-nicvf unused=vfio-pci,uio_pci_generic
326 Other network devices
327 =====================
328 0002:00:03.0 'Device a01f' unused=vfio-pci,uio_pci_generic
331 We want to bind two physical interfaces with 24 queues each device, we attach two primary VFs
332 and four secondary queues. In our example we choose two 10G interfaces eth1 (0002:01:00.2) and eth2 (0002:01:00.3).
333 We will choose four secondary queue sets from the ending of the list (0002:01:01.7-0002:01:02.2).
336 #. Bind two primary VFs to the ``vfio-pci`` driver:
338 .. code-block:: console
340 usertools/dpdk-devbind.py -b vfio-pci 0002:01:00.2
341 usertools/dpdk-devbind.py -b vfio-pci 0002:01:00.3
343 #. Bind four primary VFs to the ``vfio-pci`` driver:
345 .. code-block:: console
347 usertools/dpdk-devbind.py -b vfio-pci 0002:01:01.7
348 usertools/dpdk-devbind.py -b vfio-pci 0002:01:02.0
349 usertools/dpdk-devbind.py -b vfio-pci 0002:01:02.1
350 usertools/dpdk-devbind.py -b vfio-pci 0002:01:02.2
352 The nicvf thunderx driver will make use of attached secondary VFs automatically during the interface configuration stage.
360 The ThunderX SoC family NICs strip the CRC for every packets coming into the
361 host interface. So, CRC will be stripped even when the
362 ``rxmode.hw_strip_crc`` member is set to 0 in ``struct rte_eth_conf``.
364 Maximum packet length
365 ~~~~~~~~~~~~~~~~~~~~~
367 The ThunderX SoC family NICs support a maximum of a 9K jumbo frame. The value
368 is fixed and cannot be changed. So, even when the ``rxmode.max_rx_pkt_len``
369 member of ``struct rte_eth_conf`` is set to a value lower than 9200, frames
370 up to 9200 bytes can still reach the host interface.
372 Maximum packet segments
373 ~~~~~~~~~~~~~~~~~~~~~~~
375 The ThunderX SoC family NICs support up to 12 segments per packet when working
376 in scatter/gather mode. So, setting MTU will result with ``EINVAL`` when the
377 frame size does not fit in the maximum number of segments.