1 .. SPDX-License-Identifier: BSD-3-Clause
2 Copyright(c) 2010-2014 Intel Corporation.
7 This chapter describes the packages required to compile the DPDK.
11 If the DPDK is being used on an Intel® Communications Chipset 89xx Series platform,
12 please consult the *Intel® Communications Chipset 89xx Series Software for Linux Getting Started Guide*.
14 BIOS Setting Prerequisite on x86
15 --------------------------------
17 For the majority of platforms, no special BIOS settings are needed to use basic DPDK functionality.
18 However, for additional HPET timer and power management functionality,
19 and high performance of small packets, BIOS setting changes may be needed.
20 Consult the section on :ref:`Enabling Additional Functionality <Enabling_Additional_Functionality>`
21 for more information on the required changes.
25 If UEFI secure boot is enabled, the Linux kernel may disallow the use of
26 UIO on the system. Therefore, devices for use by DPDK should be bound to the
27 ``vfio-pci`` kernel module rather than ``igb_uio`` or ``uio_pci_generic``.
28 For more details see :ref:`linux_gsg_binding_kernel`.
30 Compilation of the DPDK
31 -----------------------
33 **Required Tools and Libraries:**
37 The setup commands and installed packages needed on various systems may be different.
38 For details on Linux distributions and the versions tested, please consult the DPDK Release Notes.
40 * General development tools including a supported C compiler such as gcc (version 4.9+) or clang (version 3.4+).
42 * For RHEL/Fedora systems these can be installed using ``dnf groupinstall "Development Tools"``
43 * For Ubuntu/Debian systems these can be installed using ``apt install build-essential``
44 * For Alpine Linux, ``apk add gcc libc-dev bsd-compat-headers libexecinfo-dev``
46 * Python 3.5 or later.
48 * Meson (version 0.49.2+) and ninja
50 * ``meson`` & ``ninja-build`` packages in most Linux distributions
51 * If the packaged version is below the minimum version, the latest versions
52 can be installed from Python's "pip" repository: ``pip3 install meson ninja``
54 * ``pyelftools`` (version 0.22+)
56 * For Fedora systems it can be installed using ``dnf install python-pyelftools``
57 * For RHEL/CentOS systems it can be installed using ``pip3 install pyelftools``
58 * For Ubuntu/Debian it can be installed using ``apt install python3-pyelftools``
59 * For Alpine Linux, ``apk add py3-elftools``
61 * Library for handling NUMA (Non Uniform Memory Access).
63 * ``numactl-devel`` in RHEL/Fedora;
64 * ``libnuma-dev`` in Debian/Ubuntu;
65 * ``numactl-dev`` in Alpine Linux
69 Please ensure that the latest patches are applied to third party libraries
70 and software to avoid any known vulnerabilities.
75 * Intel® C++ Compiler (icc). For installation, additional libraries may be required.
76 See the icc Installation Guide found in the Documentation directory under the compiler installation.
78 * IBM® Advance ToolChain for Powerlinux. This is a set of open source development tools and runtime libraries
79 which allows users to take leading edge advantage of IBM's latest POWER hardware features on Linux. To install
80 it, see the IBM official installation document.
82 **Additional Libraries**
84 A number of DPDK components, such as libraries and poll-mode drivers (PMDs) have additional dependencies.
85 For DPDK builds, the presence or absence of these dependencies will be automatically detected
86 enabling or disabling the relevant components appropriately.
88 In each case, the relevant library development package (``-devel`` or ``-dev``) is needed to build the DPDK components.
90 For libraries the additional dependencies include:
92 * libarchive: for some unit tests using tar to get their resources.
94 * libelf: to compile and use the bpf library.
96 For poll-mode drivers, the additional dependencies for each driver can be
97 found in that driver's documentation in the relevant DPDK guide document,
98 e.g. :doc:`../nics/index`
101 Building DPDK Applications
102 --------------------------
104 The tool pkg-config or pkgconf, integrated in most build systems,
105 must be used to parse options and dependencies from libdpdk.pc.
109 pkg-config 0.27, supplied with RHEL-7,
110 does not process the Libs.private section correctly,
111 resulting in statically linked applications not being linked properly.
114 Running DPDK Applications
115 -------------------------
117 To run a DPDK application, some customization may be required on the target machine.
124 * Kernel version >= 3.16
126 The kernel version required is based on the oldest long term stable kernel available
127 at kernel.org when the DPDK version is in development.
128 Compatibility for recent distribution kernels will be kept, notably RHEL/CentOS 7.
130 The kernel version in use can be checked using the command::
134 * glibc >= 2.7 (for features related to cpuset)
136 The version can be checked using the ``ldd --version`` command.
138 * Kernel configuration
140 In the Fedora OS and other common distributions, such as Ubuntu, or Red Hat Enterprise Linux,
141 the vendor supplied kernel configurations can be used to run most DPDK applications.
143 For other kernel builds, options which should be enabled for DPDK include:
147 * PROC_PAGE_MONITOR support
149 * HPET and HPET_MMAP configuration options should also be enabled if HPET support is required.
150 See the section on :ref:`High Precision Event Timer (HPET) Functionality <High_Precision_Event_Timer>` for more details.
152 .. _linux_gsg_hugepages:
154 Use of Hugepages in the Linux Environment
155 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
157 Hugepage support is required for the large memory pool allocation used for packet buffers
158 (the HUGETLBFS option must be enabled in the running kernel as indicated the previous section).
159 By using hugepage allocations, performance is increased since fewer pages are needed,
160 and therefore less Translation Lookaside Buffers (TLBs, high speed translation caches),
161 which reduce the time it takes to translate a virtual page address to a physical page address.
162 Without hugepages, high TLB miss rates would occur with the standard 4k page size, slowing performance.
164 Reserving Hugepages for DPDK Use
165 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
167 The reservation of hugepages can be performed at run time.
168 This is done by echoing the number of hugepages required
169 to a ``nr_hugepages`` file in the ``/sys/kernel/`` directory
170 corresponding to a specific page size (in Kilobytes).
171 For a single-node system, the command to use is as follows
172 (assuming that 1024 of 2MB pages are required)::
174 echo 1024 > /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages
176 On a NUMA machine, the above command will usually divide the number of hugepages
177 equally across all NUMA nodes (assuming there is enough memory on all NUMA nodes).
178 However, pages can also be reserved explicitly on individual NUMA nodes
179 using a ``nr_hugepages`` file in the ``/sys/devices/`` directory::
181 echo 1024 > /sys/devices/system/node/node0/hugepages/hugepages-2048kB/nr_hugepages
182 echo 1024 > /sys/devices/system/node/node1/hugepages/hugepages-2048kB/nr_hugepages
184 The tool ``dpdk-hugepages.py`` can be used to manage hugepages.
188 Some kernel versions may not allow reserving 1 GB hugepages at run time,
189 so reserving them at boot time may be the only option.
190 Please see below for instructions.
194 In the general case, reserving hugepages at run time is perfectly fine,
195 but in use cases where having lots of physically contiguous memory is required,
196 it is preferable to reserve hugepages at boot time,
197 as that will help in preventing physical memory from becoming heavily fragmented.
199 To reserve hugepages at boot time, a parameter is passed to the Linux kernel on the kernel command line.
201 For 2 MB pages, just pass the hugepages option to the kernel. For example, to reserve 1024 pages of 2 MB, use::
205 For other hugepage sizes, for example 1G pages, the size must be specified explicitly and
206 can also be optionally set as the default hugepage size for the system.
207 For example, to reserve 4G of hugepage memory in the form of four 1G pages, the following options should be passed to the kernel::
209 default_hugepagesz=1G hugepagesz=1G hugepages=4
213 The hugepage sizes that a CPU supports can be determined from the CPU flags on Intel architecture.
214 If pse exists, 2M hugepages are supported; if pdpe1gb exists, 1G hugepages are supported.
215 On IBM Power architecture, the supported hugepage sizes are 16MB and 16GB.
219 For 64-bit applications, it is recommended to use 1 GB hugepages if the platform supports them.
221 In the case of a dual-socket NUMA system,
222 the number of hugepages reserved at boot time is generally divided equally between the two sockets
223 (on the assumption that sufficient memory is present on both sockets).
225 See the Documentation/admin-guide/kernel-parameters.txt file in your Linux source tree for further details of these and other kernel options.
227 Using Hugepages with the DPDK
228 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
230 If secondary process support is not required, DPDK is able to use hugepages
231 without any configuration by using "in-memory" mode.
232 Please see :doc:`linux_eal_parameters` for more details.
234 If secondary process support is required,
235 mount points for hugepages need to be created.
236 On modern Linux distributions, a default mount point for hugepages
237 is provided by the system and is located at ``/dev/hugepages``.
238 This mount point will use the default hugepage size
239 set by the kernel parameters as described above.
241 However, in order to use hugepage sizes other than the default, it is necessary
242 to manually create mount points for those hugepage sizes (e.g. 1GB pages).
244 To make the hugepages of size 1GB available for DPDK use,
245 following steps must be performed::
248 mount -t hugetlbfs pagesize=1GB /mnt/huge
250 The mount point can be made permanent across reboots, by adding the following line to the ``/etc/fstab`` file::
252 nodev /mnt/huge hugetlbfs pagesize=1GB 0 0