29. LiquidIO VF Poll Mode Driver

The LiquidIO VF PMD library (librte_pmd_lio) provides poll mode driver support for Cavium LiquidIO® II server adapter VFs. PF management and VF creation can be done using kernel driver.

More information can be found at Cavium Official Website.

29.1. Supported LiquidIO Adapters

  • LiquidIO II CN2350 210SV/225SV
  • LiquidIO II CN2350 210SVPT
  • LiquidIO II CN2360 210SV/225SV
  • LiquidIO II CN2360 210SVPT

29.2. Pre-Installation Configuration

The following options can be modified in the config file. Please note that enabling debugging options may affect system performance.

  • CONFIG_RTE_LIBRTE_LIO_PMD (default y)

    Toggle compilation of LiquidIO PMD.

  • CONFIG_RTE_LIBRTE_LIO_DEBUG_RX (default n)

    Toggle display of receive fast path run-time messages.

  • CONFIG_RTE_LIBRTE_LIO_DEBUG_TX (default n)

    Toggle display of transmit fast path run-time messages.

  • CONFIG_RTE_LIBRTE_LIO_DEBUG_MBOX (default n)

    Toggle display of mailbox messages.

  • CONFIG_RTE_LIBRTE_LIO_DEBUG_REGS (default n)

    Toggle display of register reads and writes.

29.3. SR-IOV: Prerequisites and Sample Application Notes

This section provides instructions to configure SR-IOV with Linux OS.

  1. Verify SR-IOV and ARI capabilities are enabled on the adapter using lspci:

    lspci -s <slot> -vvv
    

    Example output:

    [...]
    Capabilities: [148 v1] Alternative Routing-ID Interpretation (ARI)
    [...]
    Capabilities: [178 v1] Single Root I/O Virtualization (SR-IOV)
    [...]
    Kernel driver in use: LiquidIO
    
  2. Load the kernel module:

    modprobe liquidio
    
  3. Bring up the PF ports:

    ifconfig p4p1 up
    ifconfig p4p2 up
    
  4. Change PF MTU if required:

    ifconfig p4p1 mtu 9000
    ifconfig p4p2 mtu 9000
    
  5. Create VF device(s):

    Echo number of VFs to be created into "sriov_numvfs" sysfs entry of the parent PF.

    echo 1 > /sys/bus/pci/devices/0000:03:00.0/sriov_numvfs
    echo 1 > /sys/bus/pci/devices/0000:03:00.1/sriov_numvfs
    
  6. Assign VF MAC address:

    Assign MAC address to the VF using iproute2 utility. The syntax is:

    ip link set <PF iface> vf <VF id> mac <macaddr>
    

    Example output:

    ip link set p4p1 vf 0 mac F2:A8:1B:5E:B4:66
    
  7. Assign VF(s) to VM.

    The VF devices may be passed through to the guest VM using qemu or virt-manager or virsh etc.

    Example qemu guest launch command:

    ./qemu-system-x86_64 -name lio-vm -machine accel=kvm \
    -cpu host -m 4096 -smp 4 \
    -drive file=<disk_file>,if=none,id=disk1,format=<type> \
    -device virtio-blk-pci,scsi=off,drive=disk1,id=virtio-disk1,bootindex=1 \
    -device vfio-pci,host=03:00.3 -device vfio-pci,host=03:08.3
    
  8. Running testpmd

    Refer to the document compiling and testing a PMD for a NIC to run testpmd application.

    Note

    Use igb_uio instead of vfio-pci in VM.

    Example output:

    [...]
    EAL: PCI device 0000:03:00.3 on NUMA socket 0
    EAL:   probe driver: 177d:9712 net_liovf
    EAL:   using IOMMU type 1 (Type 1)
    PMD: net_liovf[03:00.3]INFO: DEVICE : CN23XX VF
    EAL: PCI device 0000:03:08.3 on NUMA socket 0
    EAL:   probe driver: 177d:9712 net_liovf
    PMD: net_liovf[03:08.3]INFO: DEVICE : CN23XX VF
    Interactive-mode selected
    USER1: create a new mbuf pool <mbuf_pool_socket_0>: n=171456, size=2176, socket=0
    Configuring Port 0 (socket 0)
    PMD: net_liovf[03:00.3]INFO: Starting port 0
    Port 0: F2:A8:1B:5E:B4:66
    Configuring Port 1 (socket 0)
    PMD: net_liovf[03:08.3]INFO: Starting port 1
    Port 1: 32:76:CC:EE:56:D7
    Checking link statuses...
    Port 0 Link Up - speed 10000 Mbps - full-duplex
    Port 1 Link Up - speed 10000 Mbps - full-duplex
    Done
    testpmd>
    
  9. Enabling VF promiscuous mode

    One VF per PF can be marked as trusted for promiscuous mode.

    ip link set dev <PF iface> vf <VF id> trust on
    

29.4. Limitations

29.4.1. VF MTU

VF MTU is limited by PF MTU. Raise PF value before configuring VF for larger packet size.

29.4.2. VLAN offload

Tx VLAN insertion is not supported and consequently VLAN offload feature is marked partial.

29.4.3. Ring size

Number of descriptors for Rx/Tx ring should be in the range 128 to 512.

29.4.4. CRC striping

LiquidIO adapters strip ethernet FCS of every packet coming to the host interface.