[AMD Official Use Only - AMD Internal Distribution Only]
Hi Jasvinder,
I was facing similar issue with Cx-7 and it appears that the issue is related to the network namespace.
By configuring the Pod/container to use the "host's network namespace," the issue is effectively resolved.
----
spec:
nodeSelector:
"feature.node.kubernetes.io/test-node": "true"
hostNetwork: true
containers:
----
Thanks & Regards,
Sivaprasad
From: Singh, Jasvinder <jasvinder.singh@intel.com>
Sent: Tuesday, January 28, 2025 4:45 PM
To: users@dpdk.org; Raslan Darawsheh <rasland@nvidia.com>
Cc: Marjanovic, Nemanja <nemanja.marjanovic@intel.com>
Subject: mlx5_common: failed to load mlx5_eth driver issue
Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links,
or responding. |
Hi Folks,
I am trying to run dpdk-testpmd in a container with NVIDIA ConnectX-7 NIC card on the system and facing issue described below. Details on the system setup are as follow-
Host Configuration for OFED and CX7:
Nvidia CX7 is configured to use Ethernet as opposed to InfiniBand with the following steps:
OFED driver installation:
./mlnxofedinstall --without-dkms --add-kernel-support --kernel 6.8.0-51-generic --without-fw-update --force
Configuration:
mlxconfig -d /dev/mst/mt4129_pciconf0 set LINK_TYPE_P1=2
mlxconfig -d /dev/mst/mt4129_pciconf1 set LINK_TYPE_P1=2
SRIOV has been also enabled on the NIC.
The above configuration resulted in successfully deploying dpdk-testpmd on virtual functions on the host.
Container Configuration for OFED:
Ubuntu 22.04 used within container.
# Install mlnx OFED Drivers
&& tar -xvf MLNX_OFED_LINUX-24.07-0.6.1.0-ubuntu24.04-x86_64.tgz \
&& cd MLNX_OFED_LINUX-24.07-0.6.1.0-ubuntu24.04-x86_64 \
&& ./mlnxofedinstall --force --without-fw-update --dpdk --upstream-libs
Dpdk-testpmd error log:
root@tpmd-0:/tmp/dpdk-build/app# ./dpdk-testpmd -n 4 -a 60:00.1 -a 60:00.2 -- -i
EAL: Detected CPU lcores: 128
EAL: Detected NUMA nodes: 1
EAL: Detected static linkage of DPDK
EAL: Multi-process socket /var/run/dpdk/rte/mp_socket
EAL: Selected IOVA mode 'VA'
EAL: Probe PCI driver: mlx5_pci (15b3:101e) device: 0000:60:00.1 (socket 0)
mlx5_net: Unable to recognize master/representors on the multiple IB devices.
mlx5_common: Failed to load driver mlx5_eth
EAL: Requested device 0000:60:00.1 cannot be used
EAL: Probe PCI driver: mlx5_pci (15b3:101e) device: 0000:60:00.2 (socket 0)
mlx5_net: Unable to recognize master/representors on the multiple IB devices.
mlx5_common: Failed to load driver mlx5_eth
EAL: Requested device 0000:60:00.2 cannot be used
EAL: Bus (pci) probe failed.
TELEMETRY: No legacy callbacks, legacy socket not created
testpmd: No probed ethernet devices
Interactive-mode selected
testpmd: create a new mbuf pool <mb_pool_0>: n=1163456, size=2176, socket=0
testpmd: preferred mempool ops selected: ring_mp_mc
Done
testpmd> quit
Please provide help to resolve above.
Thanks,
Jasvinder