DPDK Connection Error with Liteon FlexFi

[Sig-thd]

Hello,

I need some help with dpdk connection to Liteon FlexFi:

System:

• Dell Poweredge R740
• Intel(R) Xeon(R) Gold 6212U (24 CPUs)
• 256 GB RAM
• Intel XVV710 DA2 Card
• Intel E810 DA2 Card (testing)
• RU: Liteon FlexFi FF-RFI078I4
• PTP Switch: Fibrolan Falcon RX

Software:

• Ubuntu 24.04
• Kernel 6.8.1-1015-realtime
• srsRAN release 25.10
• DPDK 23.11.5
• LinuxPTP v4.2
• Intel Driver i40e - 2.28.13
• Firmware XVV710 - 9.55
• Intel Driver ice - 2.3.15
• Firmware E810 - 4.90
• FlexFi FW - 02.00.10

I have this setup working without dpdk using the following config files. To establish the connection to the RU I had to add a VLAN Interface (VLAN564) which is used for Data transmission to the RU. (enp179 is the XVV710 card)

enp179s0f0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9600 qdisc mq state UP mode DEFAULT group default qlen 1000
link/ether 40:a6:b7:2a:xx:xx brd ff:ff:ff:ff:ff:ff

vlan564@enp179s0f0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9000 qdisc noqueue state UP mode DEFAULT group default qlen 1000
link/ether 40:a6:b7:2a:xx:xx brd ff:ff:ff:ff:ff:ff

Config Files:
gnb.yml.txt
srsran.cfg.txt

When I want to use DPDK I delete interface vlan564@enp179s0f0 and start the following script, what differs from the srsran manual is that I have to set VLAN for interface vf 0 to vlan 564 otherwise I don't get a connection to the RU.
https://docs.srsran.com/projects/project/en/latest/tutorials/source/oranRU/source/index.html

#!/bin/bash

sudo modprobe vfio-pci
sleep 2
ethtool -G enp179s0f0 tx 4096 rx 4096
sleep 2
sudo sh -c 'echo 1 > /sys/class/net/enp179s0f0/device/sriov_numvfs'
sleep 2
sudo ip link set enp179s0f0 vf 0 mac 00:33:22:33:00:12 vlan 564 mtu 9600
sleep 2
sudo ip link set enp179s0f0 vf 0 spoofchk off
sleep 2
sudo dpdk-devbind.py --bind=vfio-pci 0000:b3:02.0

dpdk-devbind.py -s

Network devices using DPDK-compatible driver
============================================
0000:b3:02.0 'Ethernet Virtual Function 700 Series 154c' drv=vfio-pci unused=iavf

I changed the grub cmdline to the following and added huge pages:

BOOT_IMAGE=/vmlinuz-6.8.1-1015-realtime root=/dev/mapper/ubuntu--vg-ubuntu--lv ro quiet splash intel_iommu=on iommu=pt hugepagesz=1G hugepages=20 default_hugepagesz=1G nosoftlockup mitigations=off skew_tick=1 nmi_watchdog=0 softlockup_panic=0 enforcing=0 audit=0 intel_pstate=disable vt.handoff=7

HugePages_Total:      20
HugePages_Free:       20
HugePages_Rsvd:        0
HugePages_Surp:        0
Hugepagesize:    1048576 kB
Hugetlb:        20971520 kB

FlexFi is configured as following:

# show running-config 
Band Width = 40000000
Center Frequency = 3779040000
Compression Bit = 9
Control and User Plane vlan = 564
M Plane vlan = 0
default gateway = 192.168.1.1
dpd mode : Enable
DU MAC Address = 003322330012
phase compensation mode : Enable
RX attenuation = 11
TX attenuation = 17
subcarrier spacing = 1
rj45_vlan_ip = 10.101.131.61
SFP_vlan_ip = 10.101.131.62
SFP_non_vlan_static_ip = 192.168.88.12
prach eAxC-id port 0, 1, 2, 3 = 0x0004, 0x0005, 0x0006, 0x0007
slotid = 0x00000001
jumboframe = 0x00000001
sync source : PTP

With these settings and the following gnb_dpdk.yml config I can establish a connection to the RU and the RU gives the status DU Connected.

gnb_dpdk.yml.txt

# show oru-status 
Sync State  : SYNCHRONIZED
RF State    : Ready
DPD         : Ready
DuConnected : Ready

--== srsRAN gNB (commit d2f4b70dda) ==--

EAL: Detected CPU lcores: 24
EAL: Detected NUMA nodes: 1
EAL: Detected shared linkage of DPDK
EAL: Multi-process socket /var/run/dpdk/rte/mp_socket
EAL: Selected IOVA mode 'VA'
EAL: VFIO support initialized
EAL: Using IOMMU type 1 (Type 1)
EAL: Probe PCI driver: net_iavf (8086:154c) device: 0000:b3:02.0 (socket 0)
TELEMETRY: No legacy callbacks, legacy socket not created
iavf_set_rx_function(): request RXDID[1] in Queue[0] is legacy, set rx_pkt_burst as legacy for all queues
Initializing the Open Fronthaul Interface for sector#0: ul_compr=[BFP,9], dl_compr=[BFP,9], prach_compr=[BFP,9], prach_cp_enabled=true
Cell pci=3, bw=40 MHz, 4T2R, dl_arfcn=651936 (n78), dl_freq=3779.04 MHz, dl_ssb_arfcn=650976, ul_freq=3779.04 MHz

N2: Connection to AMF on 10.45.1.34:38412 completed
==== gNB started ===
Type <h> to view help

UEs can see the network if I trigger a network search but they cannot connect to the network. What I found is that there is probably an issue with the network forwarding. If I use eththool -S I can see that data is transmitted over Interface VF 0 but no traffic is received by VF 0. Instead the traffic only goes to to physical interface:

NIC statistics:
     rx_packets: 849956
     tx_packets: 5990
     rx_bytes: 1875803408
     tx_bytes: 358190
     rx_errors: 0
     tx_errors: 0
     rx_dropped: 0
     tx_dropped: 0
     collisions: 0
     vf000.rx_bytes: 0
     vf000.rx_unicast: 0
     vf000.rx_multicast: 0
     vf000.rx_broadcast: 0
     vf000.rx_discards: 0
     vf000.rx_unknown_protocol: 0
     vf000.tx_bytes: 2347002168
     vf000.tx_unicast: 888118
     vf000.tx_multicast: 0
     vf000.tx_broadcast: 0
     vf000.tx_discards: 0
     vf000.tx_errors: 0

I tried the same with E810 Card to see if this is a Hardware error, but no same behavior. I also tried with setting vlan 0 on vf 0 and without the vlan settings inside the gnb.yml file, but with these changes a connection to the RU was not possible.

Any ideas what I am doing wrong here?

[TanishqCh07]

Hi,

From your description it looks like the RU link comes up correctly (DPD, sync, DU Connected all good), but the UL user-plane traffic never reaches the VF. The key detail is in your ethtool -S output:

vf000.tx_* increasing
vf000.rx_* all zero
pf rx_packets increasing

This means:

The PF is receiving packets,

but the packets are never forwarded down to the VF,

so the DPDK process bound to VF0 never sees UL traffic.

This usually points to one of three issues:
(1) VLAN steering for the VF is not configured correctly,
(2) PF-side switching is blocking packets to the VF,
(3) SR-IOV offload/stripping on the NIC is not matching the FlexFi VLAN tags.

Below are the most likely causes and fixes.

1️⃣ VLAN steering mismatch between PF/VF and FlexFi

You said the RU traffic is on VLAN 564, and it only works when you manually assign:

ip link set enp179s0f0 vf 0 vlan 564

This setting forces the NIC to:

strip VLAN 564 on RX before delivering frames to the VF

forward only VLAN 564 traffic to that VF

The problem is:
FlexFi sends VLAN-tagged UL frames, but your VF expects untagged frames after VLAN stripping.

If the NIC is not stripping correctly, packets stay on the PF and never get forwarded to the VF.

Try disabling VLAN insertion/stripping:
ip link set enp179s0f0 vf 0 vlan 0
ip link set enp179s0f0 vf 0 spoofchk off

Then in DPDK config, add the VLAN ID explicitly:

ru_ofh:
vlan: 564

Many O-RU vendors expect the DU to accept tagged packets, not stripped ones.

2️⃣ PF switching disabled (default on newer Intel drivers)

Newer i40e/ice drivers often disable the internal switch when SR-IOV is enabled.
When this happens the PF will happily receive packets, but it will not forward them to any VF unless you explicitly enable VEPA/bridge mode.

Try:

echo 2 > /sys/class/net/enp179s0f0/bridge_mode

Modes:

0 = VEB (default)

2 = VEPA (forces forwarding even in some SR-IOV setups)

This fixes RX forwarding problems on VFs in a lot of ORAN setups.

3️⃣ Intel VFs drop packets when MTU or VLAN doesn’t match exactly

FlexFi is using MTU 9000 (jumbo) and VLAN 564.
Make sure the VF MTU is also set and remains applied after binding to VFIO:

ip link set enp179s0f0 mtu 9600
ip link set enp179s0f0 vf 0 mtu 9000

AND in DPDK config:

dpdk:
mbuf:
data_room_size: 9000

If the VF MTU is smaller than FlexFi's UL packets → the NIC silently drops them.

4️⃣ Try disabling hardware VLAN filtering on the PF

VFs sometimes do not inherit the VLAN filter properly:

ip link set enp179s0f0 vf 0 vlan 564
ip link set enp179s0f0 rx-vlan-filter off
ip link set enp179s0f0 rx-vlan-stag-filter off

5️⃣ Confirm that the RU actually sends data toward the DU

FlexFi should send UL U-plane to the DU MAC you configured:

DU MAC Address = 003322330012

Check that this matches the MAC assigned to VF0 after binding to vfio-pci:

dpdk-devbind.py -s

DPDK changes the interface name and sometimes resets the MAC.
If FlexFi sends UL data to a wrong MAC, it will arrive at PF but not at VF.

✔️ What I would test first (ordered)

Disable VLAN stripping on the VF

ip link set enp179s0f0 vf 0 vlan 0

Enable VEPA mode

echo 2 > /sys/class/net/enp179s0f0/bridge_mode

Set MTU on VF and DPDK

ip link set enp179s0f0 vf 0 mtu 9000

Verify MAC address FlexFi → VF

Enable promiscuous mode in DPDK
Sometimes VFs won’t pass VLAN frames unless in promisc mode:

dpdk:
promiscuous: true