maki/openairinterface5g
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
2b796874ca974589a94cf98bcf63b63c8bfa8317
openairinterface5g/doc/ORAN_FHI7.2_Tutorial.md
T

28 KiB
Raw Blame History

OAI 7.2 Fronthaul Interface 5G SA Tutorial

Table of Contents

[[TOC]]

Prerequisites

The hardware on which we have tried this tutorial:

Hardware (CPU,RAM) Operating System (kernel) NIC (Vendor,Driver,Firmware)
Intel(R) Xeon(R) Gold 6354 36-Core, 128GB RHEL 9.2 (5.14.0-284.18.1.rt14.303.el9_2.x86_64) Intel X710, i40e, 9.20 0x8000d95e 22.0.9
Intel(R) Xeon(R) Gold 6354 36-Core, 128GB Ubuntu 22.04.3 LTS (5.15.0-1033-realtime) Intel X710, i40e, 9.00 0x8000cfeb 21.5.9
AMD EPYC 9374F 32-Core Processor, 128GB Ubuntu 22.04.2 LTS (5.15.0-1038-realtime) Intel E810 ,ice, 4.00 0x8001184e 1.3236.0

NOTE: These are not minimum hardware requirements. This is the configuration of our servers. The NIC card should support hardware PTP time stamping.

NICs we have tested so far:

Vendor Firmware Version
Intel X710 9.20 0x8000d95e 22.0.9
Intel E810 4.00 0x8001184e 1.3236.0
Intel XXV710 6.02 0x80003888

PTP enabled switches and grandmaster clock we have in are lab:

Vendor Software Version
CISCO C93180YC-FX3 10.2(4)
Fibrolan Falcon-RX/812/G 8.0.25.4
Qulsar Qg2 (Grandmaster) 12.1.27

Radio units we are testing/integrating:

Vendor Software Version
VVDN LPRU 03-v3.0.4
LiteON RU 01.00.08/02.00.03
Benetel 650 v0.8.1

Tested libxran releases:

Vendor
oran_e_maintenance_release_v1.0

Note: the libxran driver of OAI identifies the above version as "5.1.0" (E is fifth letter, then 1.0).

Configure your server

  1. Disable Hyperthreading (HT) in your BIOS. In all our servers HT is always disabled.
  2. We recommend you to start with a fresh installation of OS (either RHEL or Ubuntu). You have to install realtime kernel on your OS (Operating System). Based on your OS you can search how to install realtime kernel.
  3. Install realtime kernel for your OS
  4. Change the boot commands based on the below section. They can be performed either via tuned or via manually building the kernel

CPU allocation

This section is important to read, regardless of the operating system you are using.

Your server could be:

  • One NUMA Node (See one NUMA node example): all the processors are sharing a single memory system.
  • Two NUMA Node (see two NUMA node example): processors are grouped in 2 memory systems.
    • Usually the even (ie 0,2,4,...) CPUs are on the 1st socket
    • And the odd (ie (1,3,5,...) CPUs are on the 2nd socket

DPDK, OAI and kernel threads require to be properly allocated to extract maximum real-time performance for your use case.

  1. NOTE: Currently the default OAI 7.2 configuration file requires isolated CPUs 0,2,4 for DPDK/libXRAN, CPU 6 for ru_thread and CPU 8 for L1_rx_thread. It is preferrable to have all these threads on the same socket.
  2. Allocating CPUs to the OAI nr-softmodem is done using the --thread-pool option. Allocating 4 CPUs is the minimal configuration but we recommend to allocate at least 8 CPUs. And they can be on a different socket as the DPDK threads.
  3. And to avoid kernel preempting these allocated CPUs, it is better to force the kernel to use un-allocated CPUs.

Let summarize for example on a 32-CPU single NUMA node system, regardless of the number of sockets:

Applicative Threads Allocated CPUs
XRAN DPDK usage 0,2,4
OAI ru_thread 6
OAI L1_rx_thread 8
OAI nr-softmodem 1,3,5,7,9,11,13,15
kernel 16-31

In below example we have shown the output of /proc/cmdline for two different servers, each of them have different number of numa nodes. Be careful in isolating the CPUs in your environment. Apart from CPU allocation there are additional parameters which are important to be present in your boot command.

Modifying the linux command line usually requires to edit the /etc/default/grub, run a grub command and reboot the server.

One NUMA NODE

Below is the output of /proc/cmdline of a single numa node server,

NUMA:
  NUMA node(s):          1
  NUMA node0 CPU(s):     0-31

isolcpus=0-15 nohz_full=0-15 rcu_nocbs=0-15 kthread_cpus=16-31 rcu_nocb_poll nosoftlockup default_hugepagesz=1GB hugepagesz=1G hugepages=20 amd_iommu=on iommu=pt mitigations=off skew_tick=1 selinux=0 enforcing=0 tsc=nowatchdog nmi_watchdog=0 softlockup_panic=0 audit=0 vt.handoff=7

Example taken for AMD EPYC 9374F 32-Core Processor

Two numa nodes

Below is the output of /proc/cmdline of a two numa node server,

NUMA:
  NUMA node(s):          2
  NUMA node0 CPU(s):     0,2,4,6,8,10,12,14,16,18,20,22,24,26,28,30,32,34
  NUMA node1 CPU(s):     1,3,5,7,9,11,13,15,17,19,21,23,25,27,29,31,33,35

mitigations=off usbcore.autosuspend=-1 intel_iommu=on intel_iommu=pt selinux=0 enforcing=0 nmi_watchdog=0 softlockup_panic=0 audit=0 skew_tick=1 isolcpus=managed_irq,domain,0,2,4,6,8,10,12,14,16 nohz_full=0,2,4,6,8,10,12,14,16 rcu_nocbs=0,2,4,6,8,10,12,14,16 rcu_nocb_poll intel_pstate=disable nosoftlockup tsc=nowatchdog cgroup_disable=memory mce=off hugepagesz=1G hugepages=40 hugepagesz=2M hugepages=0 default_hugepagesz=1G isolcpus=managed_irq,domain,0,2,4,6,8,10,12,14 kthread_cpus=18-35 intel_pstate=disable nosoftlockup tsc=reliable

Example taken for Intel(R) Xeon(R) Gold 6354 CPU @ 3.00GHz

Common

Configure your servers to maximum performance mode either via OS or in BIOS. If you want to disable CPU sleep state via OS then use the below command:

# to disable
sudo cpupower idle-set -D 0
#to enable
sudo cpupower idle-set -E

Set the tuned profile to realtime. If the tuned-adm command is not installed then you have to install it.

tuned-adm profile realtime

The above information we have gathered either from O-RAN documents or via our own experiments. In case you would like to read the O-RAN documents then here are the links:

  1. O-RAN-SC O-DU Setup Configuration
  2. O-RAN Cloud Platform Reference Designs 2.0,O-RAN.WG6.CLOUD-REF-v02.00,February 2021

PTP configuration

  1. You can install linuxptp rpm or debian package. It will install ptp4l and phc2sys.
#RHEL
sudo dnf install linuxptp -y
#Ubuntu
sudo apt install linuxptp -y

Once installed you can use this configuration file for ptp4l (/etc/ptp4l.conf). Here the clock domain is 24 so you can adjust it according to your PTP GM clock domain

[global]
domainNumber            24
slaveOnly               1
time_stamping           hardware
tx_timestamp_timeout    1
logging_level           6
summary_interval        0
#priority1               127

[your_PTP_ENABLED_NIC]
network_transport       L2
hybrid_e2e              0

Probably you need to increase tx_timestamp_timeout to 50 or 100 for Intel E-810. You will see that in the logs of ptp.

Create the configuration file for phc2sys (/etc/sysconfig/phc2sys)

OPTIONS="-a -r -r -n 24"

The service of ptp4l (/usr/lib/systemd/system/ptp4l.service) should be configured as below:

[Unit]
Description=Precision Time Protocol (PTP) service
After=network-online.target
Wants=network-online.target

[Service]
Type=simple
EnvironmentFile=-/etc/sysconfig/ptp4l
ExecStart=/usr/sbin/ptp4l $OPTIONS

[Install]
WantedBy=multi-user.target

and service of phc2sys (/usr/lib/systemd/system/phc2sys.service) should be configured as below:

[Unit]
Description=Synchronize system clock or PTP hardware clock (PHC)
After=ntpdate.service ptp4l.service

[Service]
Type=simple
EnvironmentFile=-/etc/sysconfig/phc2sys
ExecStart=/usr/sbin/phc2sys $OPTIONS

[Install]
WantedBy=multi-user.target

Debugging PTP issues

You can see these steps in case your ptp logs have erorrs or rms reported in ptp4l logs is more than 100ms.

  1. Make sure that you have skew_tick=1 in /proc/cmdline
  2. For Intel E-810 cards set tx_timestamp_timeout to 50 or 100 if there are errors in ptp4l logs
  3. Other time sources than PTP, such as NTP or chrony timesources, should be disabled. Make sure they are enabled as further below.
  4. If rms or delay remain high, you can try pinning the PTP process to an isolated CPU.
#to check there is NTP enabled or not
timedatectl | grep NTP
#to disable
timedatectl set-ntp false

DPDK (Data Plane Development Kit)

Download DPDK version 20.11.7.

# on debian
sudo apt install wget xz-utils
# on Fedora/RHEL
sudo dnf install wget xz
cd
wget http://fast.dpdk.org/rel/dpdk-20.11.7.tar.xz

DPDK Compilation and Installation

# Installing meson : it should pull ninja-build and compiler packages
# on debian
sudo apt install meson
# on Fedora/RHEL
sudo dnf install meson
tar xvf dpdk-20.11.7.tar.xz && cd dpdk-stable-20.11.7

meson build
ninja -C build
sudo ninja install -C build

Verify the installation is complete

Check if the LD cache contains the DPDK Shared Objects after update:

sudo ldconfig -v | grep rte_
	librte_fib.so.0.200.2 -> librte_fib.so.0.200.2
	librte_telemetry.so.0.200.2 -> librte_telemetry.so.0.200.2
	librte_compressdev.so.0.200.2 -> librte_compressdev.so.0.200.2
	librte_gro.so.20.0 -> librte_gro.so.20.0.2
	librte_mempool_dpaa.so.20.0 -> librte_mempool_dpaa.so.20.0.2
	librte_distributor.so.20.0 -> librte_distributor.so.20.0.2
	librte_rawdev_dpaa2_cmdif.so.20.0 -> librte_rawdev_dpaa2_cmdif.so.20.0.2
	librte_mempool.so.20.0 -> librte_mempool.so.20.0.2
	librte_pmd_octeontx2_crypto.so.20.0 -> librte_pmd_octeontx2_crypto.so.20.0.2
	librte_common_cpt.so.20.0 -> librte_common_cpt.so.20.0.2
....

You may not have the /usr/local/lib, /usr/local/lib64, or custom DPDK installation paths in the LD_LIBRARY_PATH. In this case, add it as below; if you installed into a custom

sudo echo "/usr/local/lib" > /etc/ld.so.conf.d/local-lib.conf
sudo echo "/usr/local/lib64" >> /etc/ld.so.conf.d/local-lib.conf
sudo ldconfig
sudo ldconfig -v | grep rte_

Check if the PKG-CONFIG tool discovers the libraries:

pkg-config --libs libdpdk --static
Possible output 
-lrte_node -lrte_graph -lrte_bpf -lrte_flow_classify -lrte_pipeline -lrte_table -lrte_port -lrte_fib -lrte_ipsec -lrte_vhost -lrte_stack -lrte_security -lrte_sched -lrte_reorder -lrte_rib -lrte_rawdev -lrte_pdump -lrte_power -lrte_member -lrte_lpm -lrte_latencystats -lrte_kni -lrte_jobstats -lrte_ip_frag -lrte_gso -lrte_gro -lrte_eventdev -lrte_efd -lrte_distributor -lrte_cryptodev -lrte_compressdev -lrte_cfgfile -lrte_bitratestats -lrte_bbdev -lrte_acl -lrte_timer -lrte_hash -lrte_metrics -lrte_cmdline -lrte_pci -lrte_ethdev -lrte_meter -lrte_net -lrte_mbuf -lrte_mempool -lrte_rcu -lrte_ring -lrte_eal -lrte_telemetry -lrte_kvargs -Wl,--whole-archive -lrte_common_cpt -lrte_common_dpaax -lrte_common_iavf -lrte_common_octeontx -lrte_common_octeontx2 -lrte_bus_dpaa -lrte_bus_fslmc -lrte_bus_ifpga -lrte_bus_pci -lrte_bus_vdev -lrte_bus_vmbus -lrte_mempool_bucket -lrte_mempool_dpaa -lrte_mempool_dpaa2 -lrte_mempool_octeontx -lrte_mempool_octeontx2 -lrte_mempool_ring -lrte_mempool_stack -lrte_pmd_af_packet -lrte_pmd_ark -lrte_pmd_atlantic -lrte_pmd_avp -lrte_pmd_axgbe -lrte_pmd_bond -lrte_pmd_bnxt -lrte_pmd_cxgbe -lrte_pmd_dpaa -lrte_pmd_dpaa2 -lrte_pmd_e1000 -lrte_pmd_ena -lrte_pmd_enetc -lrte_pmd_enic -lrte_pmd_failsafe -lrte_pmd_fm10k -lrte_pmd_i40e -lrte_pmd_hinic -lrte_pmd_hns3 -lrte_pmd_iavf -lrte_pmd_ice -lrte_pmd_igc -lrte_pmd_ixgbe -lrte_pmd_kni -lrte_pmd_liquidio -lrte_pmd_memif -lrte_pmd_netvsc -lrte_pmd_nfp -lrte_pmd_null -lrte_pmd_octeontx -lrte_pmd_octeontx2 -lrte_pmd_pfe -lrte_pmd_qede -lrte_pmd_ring -lrte_pmd_sfc -lrte_pmd_softnic -lrte_pmd_tap -lrte_pmd_thunderx -lrte_pmd_vdev_netvsc -lrte_pmd_vhost -lrte_pmd_virtio -lrte_pmd_vmxnet3 -lrte_rawdev_dpaa2_cmdif -lrte_rawdev_dpaa2_qdma -lrte_rawdev_ioat -lrte_rawdev_ntb -lrte_rawdev_octeontx2_dma -lrte_rawdev_octeontx2_ep -lrte_rawdev_skeleton -lrte_pmd_caam_jr -lrte_pmd_dpaa_sec -lrte_pmd_dpaa2_sec -lrte_pmd_nitrox -lrte_pmd_null_crypto -lrte_pmd_octeontx_crypto -lrte_pmd_octeontx2_crypto -lrte_pmd_crypto_scheduler -lrte_pmd_virtio_crypto -lrte_pmd_octeontx_compress -lrte_pmd_qat -lrte_pmd_ifc -lrte_pmd_dpaa_event -lrte_pmd_dpaa2_event -lrte_pmd_octeontx2_event -lrte_pmd_opdl_event -lrte_pmd_skeleton_event -lrte_pmd_sw_event -lrte_pmd_dsw_event -lrte_pmd_octeontx_event -lrte_pmd_bbdev_null -lrte_pmd_bbdev_turbo_sw -lrte_pmd_bbdev_fpga_lte_fec -lrte_pmd_bbdev_fpga_5gnr_fec -Wl,--no-whole-archive -lrte_node -lrte_graph -lrte_bpf -lrte_flow_classify -lrte_pipeline -lrte_table -lrte_port -lrte_fib -lrte_ipsec -lrte_vhost -lrte_stack -lrte_security -lrte_sched -lrte_reorder -lrte_rib -lrte_rawdev -lrte_pdump -lrte_power -lrte_member -lrte_lpm -lrte_latencystats -lrte_kni -lrte_jobstats -lrte_ip_frag -lrte_gso -lrte_gro -lrte_eventdev -lrte_efd -lrte_distributor -lrte_cryptodev -lrte_compressdev -lrte_cfgfile -lrte_bitratestats -lrte_bbdev -lrte_acl -lrte_timer -lrte_hash -lrte_metrics -lrte_cmdline -lrte_pci -lrte_ethdev -lrte_meter -lrte_net -lrte_mbuf -lrte_mempool -lrte_rcu -lrte_ring -lrte_eal -lrte_telemetry -lrte_kvargs -Wl,-Bdynamic -pthread -lm -ldl

If DPDK was installed into /usr/local/lib, /usr/local/lib64, or another custom path, you have to point to the right directory with PKG_CONFIG_PATH, for instance:

export PKG_CONFIG_PATH=$PKG_CONFIG_PATH:/usr/local/lib64/pkgconfig/
pkg-config --libs libdpdk --static

If you want to de-install this version of DPDK

Go back to the version folder you used to build and install

cd ~/dpdk-stable-20.11.7
sudo ninja deinstall -C build

Build OAI-FHI gNB

Clone OAI code base in a suitable repository, here we are cloning in ~/openairinterface5g directory,

git clone https://gitlab.eurecom.fr/oai/openairinterface5g.git ~/openairinterface5g
cd ~/openairinterface5g/

Build ORAN Fronthaul Interface Library

Download ORAN FHI DU library and checkout the correct version.

git clone https://gerrit.o-ran-sc.org/r/o-du/phy.git ~/phy
cd ~/phy
git checkout oran_e_maintenance_release_v1.0

Apply the patch (available in oai_folder/cmake_targets/tools/oran_fhi_integration_patches/E):

git apply ~/openairinterface5g/cmake_targets/tools/oran_fhi_integration_patches/E/oaioran_E.patch

Compile the fronthaul interface library by calling make and the option XRAN_LIB_SO=1 to have it build a shared object. Note that we provide two environment variables RTE_SDK for the path to the source tree of DPDK, and XRAN_DIR to set the path to the fronthaul library.

cd ~/phy/fhi_lib/lib
make clean
RTE_SDK=~/dpdk-stable-20.11.7/ XRAN_DIR=~/phy/fhi_lib make XRAN_LIB_SO=1
...
[AR] build/libxran.so
./build/libxran.so

The shared library object ~/phy/fhi_lib/lib/build/libxran.so must be present before proceeding.

Build OAI gNB

You can now proceed building OAI. You build it the same way as for other radios, providing option -t oran_fhlib_5g. Additionally, you need to provide it the location of the FH library: --cmake-opt -Dxran_LOCATION=PATH. Note that since we cannot use ~ here, we resort to $HOME, which is equivalent. Finally, if you needed to define PKG_CONFIG_PATH previously, you need to do so now, too.

# You should have already cloned above
cd ~/openairinterface5g/cmake_targets
# on debian
sudo apt install -y libnuma-dev
# on RHEL
sudo dnf install -y numactl-devel
export PKG_CONFIG_PATH=/opt/dpdk/lib64/pkgconfig/
./build_oai -I  # if you never installed OAI, use this command once before the next line
./build_oai --gNB --ninja -t oran_fhlib_5g --cmake-opt -Dxran_LOCATION=$HOME/phy/fhi_lib/lib

You can optionally check that everything has been linked properly with

ldd ran_build/build/liboran_fhlib_5g.so
Possible output 
#check if all the libraries are properly linked to liboai_transpro.so
ldd ran_build/build/liboran_fhlib_5g.so
    linux-vdso.so.1 (0x00007fffb459e000)
    librte_node.so.21 => /usr/local/lib64/librte_node.so.21 (0x00007fd358690000)
    librte_graph.so.21 => /usr/local/lib64/librte_graph.so.21 (0x00007fd358685000)
    librte_bpf.so.21 => /usr/local/lib64/librte_bpf.so.21 (0x00007fd358672000)
    librte_flow_classify.so.21 => /usr/local/lib64/librte_flow_classify.so.21 (0x00007fd35866c000)
    librte_pipeline.so.21 => /usr/local/lib64/librte_pipeline.so.21 (0x00007fd35862f000)
    librte_table.so.21 => /usr/local/lib64/librte_table.so.21 (0x00007fd358612000)
    librte_port.so.21 => /usr/local/lib64/librte_port.so.21 (0x00007fd3585f8000)
    librte_fib.so.21 => /usr/local/lib64/librte_fib.so.21 (0x00007fd3585e9000)
...
    libm.so.6 => /lib64/libm.so.6 (0x00007fd357eb1000)
    libnuma.so.1 => /lib64/libnuma.so.1 (0x00007fd357ea1000)
    libc.so.6 => /lib64/libc.so.6 (0x00007fd357c98000)
    /lib64/ld-linux-x86-64.so.2 (0x00007fd3587c7000)
    libelf.so.1 => /lib64/libelf.so.1 (0x00007fd357c7d000)
    libz.so.1 => /lib64/libz.so.1 (0x00007fd357c61000)

Note that you might also call cmake directly instead of using build_oai:

cd ~/openairinterface5g
mkdir build && cd build
cmake .. -GNinja -DOAI_FHI72=ON -Dxran_LOCATION=$HOME/phy/fhi_lib/lib
ninja nr-softmodem oran_fhlib_5g params_libconfig

Configure the RU

Contact the RU vendor to get the configuration manual, and configure the RU appropriately. You can orient on the OAI configuration files mentioned further below.

We are evaluating if we can share RU configuration steps.

Configure OAI gNB

Sample configuration files for OAI gNB, specific to the manufacturer of the radio unit, are available at:

  1. LITE-ON RU: gnb.sa.band78.273prb.fhi72.4x4-liteon.conf (band n78, 273 PRBs, 3.5GHz center freq, 4x4 antenna configuration with 9 bit I/Q samples (compressed) for PUSCH/PDSCH/PRACH, 2-layer DL MIMO, UL SISO)
  2. Benetel 650 RU: gnb.sa.band78.273prb.fhi72.4x2-benetel650.conf (band n78, 273 PRBs, 3.5GHz center freq, 4x2 antenna configuration with 9 bit I/Q samples (compressed) for PUSCH/PDSCH/PRACH, 2-layer DL MIMO, UL SISO)
  3. VVDN RU: gnb.sa.band77.273prb.fhi72.4x4-vvdn.conf (band n77, 273 PRBs, 4.0GHz center freq, 4x4 antenna configuration with 9 bit I/Q samples (compressed) for PUSCH/PDSCH/PRACH, 2-layer DL MIMO, UL SISO)

Adapt the OAI-gNB configuration file to your system/workspace

Edit the sample OAI gNB configuration file and check following parameters:

  • gNBs section

    • The PLMN section shall match the one defined in the AMF
    • amf_ip_address shall be the correct AMF IP address in your system
    • GNB_INTERFACE_NAME_FOR_NG_AMF and GNB_IPV4_ADDRESS_FOR_NG_AMF shall match your DU N2 interface name and IP address
    • GNB_INTERFACE_NAME_FOR_NGU and GNB_IPV4_ADDRESS_FOR_NGU shall match your DU N3 interface name and IP address
    • prach_ConfigurationIndex
    • prach_msg1_FrequencyStart
    • Adjust the frequency, bandwidth and SSB position

  • L1s section

    • Set an isolated core for L1 thread L1_rx_thread_core, in our environment we are using CPU 8
    • phase_compensation should be set to 0 to disable when it is performed in the RU and set to 1 when it should be performed on the DU side

  • RUs section

    • Set an isolated core for RU thread ru_thread_core, in our environment we are using CPU 6

  • fhi_72 (FrontHaul Interface) section: this config follows the structure that is employed by the xRAN library (xran_fh_init and xran_fh_config structs in the code):

    • dpdk_devices: PCI addresses of NIC VFs binded to the DPDK
    • io_core: absolute CPU core ID for XRAN library
    • worker_cores: array of absolute CPU core IDs for XRAN library
    • du_addr: DU C- and U-plane MAC-addresses (format UU:VV:WW:XX:YY:ZZ, hexadecimal numbers)
    • ru_addr: RU C- and U-plane MAC-addresses (format UU:VV:WW:XX:YY:ZZ, hexadecimal numbers)
    • mtu: Maximum Transmission Unit for the RU, specified by RU vendor
    • fh_config: parameters that need to match RU parameters
      • timing parameters (starting with T) depend on the RU: Tadv_cp_dl is a single number, the rest pairs of numbers (x, y) specifying minimum and maximum delays
      • ru_config: RU-specific configuration:
        • iq_width: Width of DL/UL IQ samples: if 16, no compression, if <16, applies compression
        • iq_width_prach: Width of PRACH IQ samples: if 16, no compression, if <16, applies compression
        • fft_size: size of FFT performed by RU, set to 12 by default
      • prach_config: PRACH-specific configuration
        • eAxC_offset: PRACH antenna offset
        • kbar: the PRACH guard interval, provided in RU

xRAN SRS reception is not supported. The eAxC offsets for DL/UL are determined from the antenna configuration.

Configure Network Interfaces and DPDK VFs

The 7.2 fronthaul uses the xran library, which requires DPDK. In this step, we need to configure network interfaces to send data to the RU, and configure DPDK to bind to the corresponding PCI interfaces. More specifically, in the following we use SR-IOV to create multiple virtual functions (VFs) through which Control plane (C plane) and User plane (U plane) traffic will flow. The following commands are not persistant, and have to be repeated after reboot.

In the following, we will use these short hands:

  • physical-interface: Physical network interface through which you can access the RU
  • vlan: VLAN tags as defined in the RU configuration
  • mtu: the MTU as specified by the RU vendor, and supported by the NIC
  • du-c-plane-mac-addr: DU C plane MAC address
  • lspci-address-c-plane-vf: PCI bus address of the VF for C plane
  • du-u-plane-mac-addr: DU U plane MAC address
  • lspci-address-u-plane-vf: PCI bus address of the VF for U plane

For both the MAC addresses, you might use the MAC addresses which are pre-configured in the RUs (typically 00:11:22:33:44:66, but that is not always the case). Note that if your system has Intel E-810 NIC cards/ICE driver, you have to choose different MAC addresses (valid for above-mentioned kernels). If the RU vendor requires untagged traffic, remove the VLAN tagging in the below command and configure VLAN on the switch as "access VLAN". In case the MTU is different than 1500, you have to update the MTU on the switch interface as well.

First, set maximum ring buffers:

sudo ethtool -g <physical-interface>
sudo ethtool -G <physical-interface> rx 4096     # assumes 4096 is max
sudo ethtool -G <physical-interface> tx 4096     # assumes 4096 is max

Set the maximum MTU in the physical interface:

sudo ifconfig <physical-interface> mtu <mtu>

(Re-)create two VFs, load the Linux "Base Driver for Intel Ethernet Adaptive Virtual Function" (in case you use Intel ethernet card), and set up the VFs.

sudo modprobe iavf
sudo sh -c echo "0" > /sys/class/net/<physical-interface>/device/sriov_numvfs
sudo sh -c echo "2" > /sys/class/net/<physical-interface>/device/sriov_numvfs
sudo ip link set <physical-interface> vf 0 mac <du-c-plane-mac-addr> vlan <vlan> mtu <mtu> spoofchk off
sudo ip link set <physical-interface> vf 1 mac <du-u-plane-mac-addr> vlan <vlan> mtu <mtu> spoofchk off

After running the above commands, the kernel created virtual functions that have been assigned a PCI address under the same device and vendor ID. For instance, use sudo lshw -c network -businfo to get a list of PCI addresses and interface names, locate the PCI address of <physical-interface>, then use lspci | grep Virtual to get all virtual interfaces and use the ones with the same Device/Vendor ID parts (first two numbers).

Example

The machine in this example has an Intel X710 card. The interface in question is eno12409. Running lshw gives:

$ sudo lshw -c network -businfo
Bus info          Device     Class          Description
=======================================================
[...]
pci@0000:31:00.1  eno12409   network        Ethernet Controller X710 for 10GbE SFP+
[...]

We see the PCI address 31:00.1. Listing the virtual interfaces through lspci, we get

$ lspci | grep Virtual
31:06.0 Ethernet controller: Intel Corporation Ethernet Virtual Function 700 Series (rev 02)
31:06.1 Ethernet controller: Intel Corporation Ethernet Virtual Function 700 Series (rev 02)
98:11.0 Ethernet controller: Intel Corporation Ethernet Adaptive Virtual Function (rev 02)
98:11.1 Ethernet controller: Intel Corporation Ethernet Adaptive Virtual Function (rev 02)
98:11.2 Ethernet controller: Intel Corporation Ethernet Adaptive Virtual Function (rev 02)
98:11.3 Ethernet controller: Intel Corporation Ethernet Adaptive Virtual Function (rev 02)
98:11.4 Ethernet controller: Intel Corporation Ethernet Adaptive Virtual Function (rev 02)
98:11.5 Ethernet controller: Intel Corporation Ethernet Adaptive Virtual Function (rev 02)

The hardware card 31:00.1 has two associated virtual functions 31:06.0 and 31:06.1.

Now, unbind any pre-existing DPDK devices, load the "Virtual Function I/O" driver vfio_pci, and bind DPDK to these devices:

sudo /usr/local/bin/dpdk-devbind.py --unbind <lspci-address-c-plane-vf>
sudo /usr/local/bin/dpdk-devbind.py --unbind <lspci-address-u-plane-vf>
sudo modprobe vfio_pci
sudo /usr/local/bin/dpdk-devbind.py --bind vfio-pci <lspci-address-c-plane-vf>
sudo /usr/local/bin/dpdk-devbind.py --bind vfio-pci <lspci-address-u-plane-vf>

We recommand to put the above commands into a script file to quickly repeat them.

Example script for Benetel 650 with Intel X710 on host 
set -x

sudo ethtool -G eno12409 rx 4096
sudo ethtool -G eno12409 tx 4096
sudo ifconfig eno12409 mtu 9216

sudo modprobe iavf
sudo sh -c 'echo 0 > /sys/class/net/eno12409/device/sriov_numvfs'
sudo sh -c 'echo 2 > /sys/class/net/eno12409/device/sriov_numvfs'
sudo ip link set eno12409 vf 0 mac 00:11:22:33:44:67 vlan 3 qos 0 spoofchk off mtu 9216
sudo ip link set eno12409 vf 1 mac 00:11:22:33:44:66 vlan 3 qos 0 spoofchk off mtu 9216

sudo /usr/local/bin/dpdk-devbind.py --unbind 31:06.0
sudo /usr/local/bin/dpdk-devbind.py --unbind 31:06.1
sudo modprobe vfio-pci
sudo /usr/local/bin/dpdk-devbind.py --bind vfio-pci 31:06.0
sudo /usr/local/bin/dpdk-devbind.py --bind vfio-pci 31:06.1

Start OAI gNB

cd ran_build/build

sudo ./nr-softmodem -O ../../../targets/PROJECTS/GENERIC-NR-5GC/CONF/oran.fh.band78.fr1.273PRB.conf --sa --reorder-thread-disable 1 --thread-pool <list of non isolated cpus>

For example if you have two numa nodes (for example 18 CPU per socket) in your system and odd cores are non isolated then you can put the thread-pool on 1,3,5,7,9,11,13,15. Else if you have 1 numa node either you can use isolated cores or non isolated. Just make sure that isolated cores are not the ones defined earlier.

Reference in New Issue View Git Blame Copy Permalink