mlx5 with inbox driver 100G is not detecting

Hi,

I am trying to up 100G Ethernet driver. I am using inbox driver, I am not using MLNX OFED stack.

I am inserting drivers manually in the sequence of ib_cm…mlx5_core,mlx5_ib

mkdir /sys/kernel/config/rdma_cm/mlx5_0

echo “RoCE v2” > /sys/kernel/config/rdma_cm/$mlx5_0/ports/1/default_roce_mode

But with ethtool command is not showing 100G, Am I missing some thing?

root@xhdipsspdk1 mailto:root@xhdipsspdk1 ~]# ethtool enp5s0

Settings for enp5s0:

Supported ports: [ FIBRE ]

Supported link modes: 1000baseT/Full

1000baseKX/Full

40000baseKR4/Full

40000baseCR4/Full

40000baseSR4/Full

40000baseLR4/Full

Supported pause frame use: Symmetric Receive-only

Supports auto-negotiation: No

Advertised link modes: 1000baseT/Full

1000baseKX/Full

40000baseKR4/Full

40000baseCR4/Full

40000baseSR4/Full

40000baseLR4/Full

Advertised pause frame use: No

Advertised auto-negotiation: No

Speed: Unknown!

Duplex: Unknown! (255)

Port: FIBRE

PHYAD: 0

Transceiver: internal

Auto-negotiation: off

Supports Wake-on: d

Wake-on: d

Supports Wake-on: d

Wake-on: d

Link detected: no

Hi,

Please, perform the same test with MOFED in order to check if the issue related to inbox driver. I would also suggest to check the cable P/N and try connect ports in loopback ( one to another, using same cable).

Now I can able to work with Inbox and MLNX OFED stack. RHEL 7.0 and kernel 4.2

OS is Centos 7.3 at the moment, kernel is 3.10.0. Everything works as expected with MOFED (well, sort of).

I’ve encountered the same problem with the latest centos7 inbox rpms. Back to mofed it seems.

What OS and kernel are you use ?

Keep in mind that support for 25G/50G/100G was added to ethtool v4.8.

You may want to check which version your are running.

https://git.kernel.org/pub/scm/network/ethtool/ethtool.git/commit/?id=766a1ea227353fa3c14f4333fc616754ab3ef06d https://git.kernel.org/pub/scm/network/ethtool/ethtool.git/commit/?id=766a1ea227353fa3c14f4333fc616754ab3ef06d