1.. SPDX-License-Identifier: BSD-3-Clause 2 Copyright(c) 2010-2015 Intel Corporation. 3 Copyright 2017 Mellanox Technologies, Ltd 4 All rights reserved. 5 6.. _linux_gsg_linux_drivers: 7 8Linux Drivers 9============= 10 11Different PMDs may require different kernel drivers in order to work properly. 12Depends on the PMD being used, a corresponding kernel driver should be load 13and bind to the network ports. 14 15UIO 16--- 17 18A small kernel module to set up the device, map device memory to user-space and register interrupts. 19In many cases, the standard ``uio_pci_generic`` module included in the Linux kernel 20can provide the uio capability. This module can be loaded using the command: 21 22.. code-block:: console 23 24 sudo modprobe uio_pci_generic 25 26.. note:: 27 28 ``uio_pci_generic`` module doesn't support the creation of virtual functions. 29 30As an alternative to the ``uio_pci_generic``, there is the ``igb_uio`` module 31which can be found in the repository `dpdk-kmods <http://git.dpdk.org/dpdk-kmods>`_. 32It can be loaded as shown below: 33 34.. code-block:: console 35 36 sudo modprobe uio 37 sudo insmod igb_uio.ko 38 39.. note:: 40 41 If UEFI secure boot is enabled, the Linux kernel may disallow the use of 42 UIO on the system. Therefore, devices for use by DPDK should be bound to the 43 ``vfio-pci`` kernel module rather than any UIO-based module. 44 For more details see :ref:`linux_gsg_binding_kernel` below. 45 46.. note:: 47 48 If the devices used for DPDK are bound to the ``uio_pci_generic`` kernel module, 49 please make sure that the IOMMU is disabled or passthrough. One can add 50 ``intel_iommu=off`` or ``amd_iommu=off`` or ``intel_iommu=on iommu=pt`` in GRUB 51 command line on x86_64 systems, or add ``iommu.passthrough=1`` on aarch64 system. 52 53Since DPDK release 1.7 onward provides VFIO support, use of UIO is optional 54for platforms that support using VFIO. 55 56VFIO 57---- 58 59A more robust and secure driver in compare to the ``UIO``, relying on IOMMU protection. 60To make use of VFIO, the ``vfio-pci`` module must be loaded: 61 62.. code-block:: console 63 64 sudo modprobe vfio-pci 65 66Note that in order to use VFIO, your kernel must support it. 67VFIO kernel modules have been included in the Linux kernel since version 3.6.0 and are usually present by default, 68however please consult your distributions documentation to make sure that is the case. 69 70The ``vfio-pci`` module since Linux version 5.7 supports the creation of virtual 71functions. After the PF is bound to vfio-pci module, the user can create the VFs 72by sysfs interface, and these VFs are bound to vfio-pci module automatically. 73 74When the PF is bound to vfio-pci, it has initial VF token generated by random. For 75security reason, this token is write only, the user can't read it from the kernel 76directly. To access the VF, the user needs to start the PF with token parameter to 77setup a VF token in UUID format, then the VF can be accessed with this new token. 78 79Since the ``vfio-pci`` module uses the VF token as internal data to provide the 80collaboration between SR-IOV PF and VFs, so DPDK can use the same VF token for all 81PF devices which bound to one application. This VF token can be specified by the EAL 82parameter ``--vfio-vf-token``. 83 84.. code-block:: console 85 86 1. Generate the VF token by uuid command 87 14d63f20-8445-11ea-8900-1f9ce7d5650d 88 89 2. sudo modprobe vfio-pci enable_sriov=1 90 91 2. ./usertools/dpdk-devbind.py -b vfio-pci 0000:86:00.0 92 93 3. echo 2 > /sys/bus/pci/devices/0000:86:00.0/sriov_numvfs 94 95 4. Start the PF: 96 <build_dir>/app/dpdk-testpmd -l 22-25 -n 4 -w 86:00.0 \ 97 --vfio-vf-token=14d63f20-8445-11ea-8900-1f9ce7d5650d --file-prefix=pf -- -i 98 99 5. Start the VF: 100 <build_dir>/app/dpdk-testpmd -l 26-29 -n 4 -w 86:02.0 \ 101 --vfio-vf-token=14d63f20-8445-11ea-8900-1f9ce7d5650d --file-prefix=vf0 -- -i 102 103Also, to use VFIO, both kernel and BIOS must support and be configured to use IO virtualization (such as Intel® VT-d). 104 105.. note:: 106 107 ``vfio-pci`` module doesn't support the creation of virtual functions before Linux version 5.7. 108 109For proper operation of VFIO when running DPDK applications as a non-privileged user, correct permissions should also be set up. 110This can be done by using the DPDK setup script (called dpdk-setup.sh and located in the usertools directory). 111 112.. note:: 113 114 VFIO can be used without IOMMU. While this is just as unsafe as using UIO, it does make it possible for the user to keep the degree of device access and programming that VFIO has, in situations where IOMMU is not available. 115 116.. _bifurcated_driver: 117 118Bifurcated Driver 119----------------- 120 121PMDs which use the bifurcated driver co-exists with the device kernel driver. 122On such model the NIC is controlled by the kernel, while the data 123path is performed by the PMD directly on top of the device. 124 125Such model has the following benefits: 126 127 - It is secure and robust, as the memory management and isolation 128 is done by the kernel. 129 - It enables the user to use legacy linux tools such as ``ethtool`` or 130 ``ifconfig`` while running DPDK application on the same network ports. 131 - It enables the DPDK application to filter only part of the traffic, 132 while the rest will be directed and handled by the kernel driver. 133 The flow bifurcation is performed by the NIC hardware. 134 As an example, using :ref:`flow_isolated_mode` allows to choose 135 strictly what is received in DPDK. 136 137More about the bifurcated driver can be found in 138`Mellanox Bifurcated DPDK PMD 139<https://www.dpdk.org/wp-content/uploads/sites/35/2016/10/Day02-Session04-RonyEfraim-Userspace2016.pdf>`__. 140 141.. _linux_gsg_binding_kernel: 142 143Binding and Unbinding Network Ports to/from the Kernel Modules 144-------------------------------------------------------------- 145 146.. note:: 147 148 PMDs Which use the bifurcated driver should not be unbind from their kernel drivers. this section is for PMDs which use the UIO or VFIO drivers. 149 150As of release 1.4, DPDK applications no longer automatically unbind all supported network ports from the kernel driver in use. 151Instead, in case the PMD being used use the UIO or VFIO drivers, all ports that are to be used by an DPDK application must be bound to the 152``uio_pci_generic``, ``igb_uio`` or ``vfio-pci`` module before the application is run. 153For such PMDs, any network ports under Linux* control will be ignored and cannot be used by the application. 154 155To bind ports to the ``uio_pci_generic``, ``igb_uio`` or ``vfio-pci`` module for DPDK use, 156and then subsequently return ports to Linux* control, 157a utility script called dpdk-devbind.py is provided in the usertools subdirectory. 158This utility can be used to provide a view of the current state of the network ports on the system, 159and to bind and unbind those ports from the different kernel modules, including the uio and vfio modules. 160The following are some examples of how the script can be used. 161A full description of the script and its parameters can be obtained by calling the script with the ``--help`` or ``--usage`` options. 162Note that the uio or vfio kernel modules to be used, should be loaded into the kernel before 163running the ``dpdk-devbind.py`` script. 164 165.. warning:: 166 167 Due to the way VFIO works, there are certain limitations to which devices can be used with VFIO. 168 Mainly it comes down to how IOMMU groups work. 169 Any Virtual Function device can be used with VFIO on its own, but physical devices will require either all ports bound to VFIO, 170 or some of them bound to VFIO while others not being bound to anything at all. 171 172 If your device is behind a PCI-to-PCI bridge, the bridge will then be part of the IOMMU group in which your device is in. 173 Therefore, the bridge driver should also be unbound from the bridge PCI device for VFIO to work with devices behind the bridge. 174 175.. warning:: 176 177 While any user can run the dpdk-devbind.py script to view the status of the network ports, 178 binding or unbinding network ports requires root privileges. 179 180To see the status of all network ports on the system: 181 182.. code-block:: console 183 184 ./usertools/dpdk-devbind.py --status 185 186 Network devices using DPDK-compatible driver 187 ============================================ 188 0000:82:00.0 '82599EB 10-GbE NIC' drv=uio_pci_generic unused=ixgbe 189 0000:82:00.1 '82599EB 10-GbE NIC' drv=uio_pci_generic unused=ixgbe 190 191 Network devices using kernel driver 192 =================================== 193 0000:04:00.0 'I350 1-GbE NIC' if=em0 drv=igb unused=uio_pci_generic *Active* 194 0000:04:00.1 'I350 1-GbE NIC' if=eth1 drv=igb unused=uio_pci_generic 195 0000:04:00.2 'I350 1-GbE NIC' if=eth2 drv=igb unused=uio_pci_generic 196 0000:04:00.3 'I350 1-GbE NIC' if=eth3 drv=igb unused=uio_pci_generic 197 198 Other network devices 199 ===================== 200 <none> 201 202To bind device ``eth1``,``04:00.1``, to the ``uio_pci_generic`` driver: 203 204.. code-block:: console 205 206 ./usertools/dpdk-devbind.py --bind=uio_pci_generic 04:00.1 207 208or, alternatively, 209 210.. code-block:: console 211 212 ./usertools/dpdk-devbind.py --bind=uio_pci_generic eth1 213 214To restore device ``82:00.0`` to its original kernel binding: 215 216.. code-block:: console 217 218 ./usertools/dpdk-devbind.py --bind=ixgbe 82:00.0 219