Re: Unable to initialize ConnectX-4 ports in DPDK 24.11 and 25.03

[Dariusz Sosnowski <dsosnowski@nvidia.com>]

Hi,

On Sat, May 24, 2025 at 01:50:45AM +0100, Charles Oluwaseun Oluboyo wrote:
> Hello,
> 
> I have a ConnectX-4 dual-100Gbps NIC that works with DPDK versions
> 24.03 and 24.07 but not with 24.11 or 25.03.
> 
> 0000:04:00.0 'MT27700 Family [ConnectX-4] 1013' numa_node=0
> if=enp4s0f0np0 drv=mlx5_core unused=vfio-pci
> 0000:04:00.1 'MT27700 Family [ConnectX-4] 1013' numa_node=0
> if=enp4s0f1np1 drv=mlx5_core unused=vfio-pci
> 
> Setting the log levels to debug, the newer versions are unable to
> configure more than one (1) description on any queue:
> 
> ETHDEV: Invalid value for nb_tx_desc(=1024), should be: <= 1, >= 0,
> and a product of 1
> 
> Whereas using the older version, I am able to set a larger number of
> descriptors. Tried 512, 1024, 2048 and 4096.
> 
> ********************* Infos for port 0  *********************
> MAC address: EC:0D:9A:35:1A:16
> Device name: 04:00.0
> Driver name: mlx5_pci
> Firmware-version: 12.24.1000
> Devargs:
> Connect to socket: 0
> memory allocation on the socket: 0
> Link status: up
> Link speed: 100 Gbps
> Link duplex: full-duplex
> Autoneg status: On
> MTU: 2030
> Promiscuous mode: enabled
> Allmulticast mode: disabled
> Maximum number of MAC addresses: 128
> Maximum number of MAC addresses of hash filtering: 0
> VLAN offload:
>   strip off, filter off, extend off, qinq strip off
> Hash key size in bytes: 40
> Redirection table size: 4
> Supported RSS offload flow types:
>   ipv4  ipv4-frag  ipv4-tcp  ipv4-udp  ipv4-other  ipv6
>   ipv6-frag  ipv6-tcp  ipv6-udp  ipv6-other  ipv6-ex
>   ipv6-tcp-ex  ipv6-udp-ex  esp  l4-dst-only  l4-src-only
>   l3-dst-only  l3-src-only
> Minimum size of RX buffer: 32
> Maximum configurable length of RX packet: 65536
> Maximum configurable size of LRO aggregated packet: 65280
> Current number of RX queues: 4
> Max possible RX queues: 1024
> Max possible number of RXDs per queue: 65535
> Min possible number of RXDs per queue: 0
> RXDs number alignment: 1
> Current number of TX queues: 4
> Max possible TX queues: 1024
> Max possible number of TXDs per queue: 65535
> Min possible number of TXDs per queue: 0
> TXDs number alignment: 1
> Max segment number per packet: 40
> Max segment number per MTU/TSO: 40
> Device capabilities: 0x10( FLOW_SHARED_OBJECT_KEEP )
> Switch name: 04:00.0
> Switch domain Id: 0
> Switch Port Id: 65535
> Device error handling mode: none
> Device private info:
>   none
> 
> Is this a known issue? Please, what should I do next?
> 
> Oluwaseun

One of the patches merged to 24.11 introduced an issue with queues
descriptor limits on older NICs.

A patch with a fix for that is pending for merge to 25.07.

Would you be able apply the following patches and test on your side?

- https://patches.dpdk.org/project/dpdk/patch/20250424133128.133900-1-viacheslavo@nvidia.com/
- https://patches.dpdk.org/project/dpdk/patch/20250514075530.419423-1-viacheslavo@nvidia.com/

Best regards,
Dariusz Sosnowski