[ENV]
OS : Ubuntu Server 20.04.4 LTS
Kernel Ver : 5.4.0-42-generic
HCA Model : MCX653105A-HDAT
HCA F/W : 20.33.1048
OFED Ver : MLNX_OFED_LINUX-5.6-2.0.9.0
I have same two HCA card.
mlx5_0: Link type is Infiniband
mlx5_1: Link type is Ethernet.
I had confirm HCA (mlx5_1) can be use bandwidth limit per VF in Ethernet mode.
[ENV] OS : Ubuntu Server 20.04.4 LTS Kernel Ver : 5.4.0-42-generic HCA Model : MCX653105A-HDAT HCA F/W : 20.33.1048 OFED Ver : MLNX_OFED_LINUX-5.6-2.0.9.0
I have same two HCA card. mlx5_0: Link type is Infiniband mlx5_1: Link type is Ethernet.
I had confirm HCA (mlx5_1) can be use bandwidth limit per VF in Ethernet mode.
[mlx5_1] : Ehternet mode root@test-svr:~# mlxconfig -d mlx5_1 q | egrep 'NUM_OF_VFS| SRIOV_EN |LINK_TYPE_P1' NUM_OF_VFS 16 SRIOV_EN True(1) LINK_TYPE_P1 ETH(2)
root@test-svr:~# ls -l /sys/class/net/eth4/device/sriov/0/ total 0 -rw-r--r-- 1 root root 4096 Oct 19 12:19 config -rw-r--r-- 1 root root 4096 Oct 19 12:19 group -rw-r--r-- 1 root root 4096 Oct 19 12:19 link_state -rw-r--r-- 1 root root 4096 Oct 19 12:19 mac -rw-r--r-- 1 root root 4096 Oct 19 12:19 max_tx_rate
-rw-r--r-- 1 root root 4096 Oct 19 12:19 min_tx_rate -rw-r--r-- 1 root root 4096 Oct 19 12:19 node drwxr-xr-x 2 root root 0 Oct 19 12:19 paging_control -rw-r--r-- 1 root root 4096 Oct 19 12:19 spoofcheck -rw-r--r-- 1 root root 4096 Oct 19 12:19 stats -rw-r--r-- 1 root root 4096 Oct 19 12:19 trunk -rw-r--r-- 1 root root 4096 Oct 19 12:19 trust -rw-r--r-- 1 root root 4096 Oct 19 12:19 vlan
But, I can't find SR-IOV VF Configuration files in Infiniband mode.
[mlx5_0] : Infiniband Mode root@test-svr:~# mlxconfig -d mlx5_0 q | egrep 'NUM_OF_VFS| SRIOV_EN |LINK_TYPE_P1' NUM_OF_VFS 16 SRIOV_EN True(1) LINK_TYPE_P1 IB(1)
root@test-svr:~# ls -l /sys/class/net/ib0/device/sriov/0/ total 0 -rw-r--r-- 1 root root 4096 Oct 19 12:19 node -rw-r--r-- 1 root root 4096 Oct 19 12:19 policy -rw-r--r-- 1 root root 4096 Oct 19 12:19 port
Q1. Is supports bandwidth limit per VF in Infiniband mode ?
If you need more information, please tell me.
Thanks.