Usage¶

In this section, it is assumed that Virtual Accelerator has been properly installed and configured. See Getting Started for more details.

You can manage VLANs either under Linux or in the fast path.

#  modprobe vlan

Linux¶

You can manage VLAN devices via the Linux commands below.

If you do not receive netlink notifications from the lower interface, set the lower interface in promiscuous and allmulticast modes:

$ sudo ip link set eth2 promisc on
$ sudo ip link set eth2 allmulticast on

Interface eth2 should now have the flags ALL_MULTI and PROMISC:

$ fpcmd iface
...
6:eth2 [VR-0] ifid=6 (port 1) <UP|RUNNING|PROMISC|FWD4|FWD6> (0x73)
        type=ether mac=00:02:02:00:00:21 mtu=1500 tcp4mss=0 tcp6mss=0
        IPv4 routes=0  IPv6 routes=0
        if_ops: rx_dev=none tx_dev=none ip_output=none
...

Create a VLAN device:
```
$ sudo ip link add link eth2 name eth2.1000 type vlan id 1000
```
This adds the VLAN (802.1Q) device number 1000 to the link interface eth2. The new VLAN interface name is eth2.1000. A VLAN tag is appended to frames traversing this VLAN device.
Display information about the VLAN device:
```
$ ip -d link show eth2.1000
```
Delete the VLAN device:
```
$ sudo ip link delete eth2.1000
```

Egress QoS mapping¶

This feature allows setting the VLAN priority field of outgoing packets based on the priority of input packets.

In Linux, it depends on both the TOS to skb priority mapping table and the skb priority to VLAN priority mapping table. The former is hardcoded and the later can be configured via iproute2. Besides that, skb priority can be forced by an iptables rule with CLASSIFY target(–set-class).

The TOS octet in the IP header has an interesting history having been redefined several times. The TOS to skb priority table is based on RFC 1349:

3.  Specification of the Type of Service Octet

                0     1     2     3     4     5     6     7
             +-----+-----+-----+-----+-----+-----+-----+-----+
             |                 |                       |     |
             |   PRECEDENCE    |          TOS          | MBZ |
             |                 |                       |     |
             +-----+-----+-----+-----+-----+-----+-----+-----+

4.  Specification of the TOS Field

                    1000   --   minimize delay
                    0100   --   maximize throughput
                    0010   --   maximize reliability
                    0001   --   minimize monetary cost
                    0000   --   normal service

In fact, Linux uses only the bit 3 and 4. Here is the mapping:

bit #3	bit# 4	skb priority
0	0	0
0	1	2
1	0	6
1	1	4

Here is an example on how to configure the skb priority to VLAN priority mapping table.

$ sudo ip link add link eth2 name eth2.1000 type vlan id 1000 egress-qos-map 2:6 4:5

The resulting mapping for the previous example is described in the following table:

bit #3	bit# 4	skb priority	vlan priority
0	0	0	0
0	1	2	6
1	0	6	0
1	1	4	5

As a result, incoming packets with tos 0x18 will get the vlan priority field set to 5.

Note

$ sudo ip link add link eth2 name eth2.1000 type vlan help
Usage: ... vlan [ protocol VLANPROTO ] id VLANID                [ FLAG-LIST ]
                [ ingress-qos-map QOS-MAP ] [ egress-qos-map QOS-MAP ]
VLANPROTO: [ 802.1Q / 802.1ad ]
VLANID := 0-4095
FLAG-LIST := [ FLAG-LIST ] FLAG
FLAG := [ reorder_hdr { on | off } ] [ gvrp { on | off } ] [ mvrp { on | off } ]
        [ loose_binding { on | off } ]
QOS-MAP := [ QOS-MAP ] QOS-MAPPING
QOS-MAPPING := FROM:TO

Linux packet priority can also be set by iptables rules. For example:

$ iptables -t mangle -A POSTROUTING -p udp -o eth2.1000 -j CLASSIFY --set-class 0:2

By the rule, the skb priority of an udp packet to be sent on eth2.1000 will be set to 2. For the previous egress qos mapping example, these udp packets will get the vlan priority field set to 6.

Ingress QoS mapping¶

This feature allows mapping VLAN header prio field to the Linux internal packet priority(skb->priority) on incoming VLAN packets.

The ingress-qos-map parameters can be configured by the above ip link add command.

Fast path¶

You can manage VLAN devices via the fp-cli commands below.

Displaying VLAN devices¶

Synopsis

vlan

Example

<fp-0> vlan
eth2-vr0:
        eth2.1000-vr0: vlanid: 1000 vlanproto: 0x8100
                Ingress QOS mappings:
                       0:2 1:6 2:4 3:0 4:0 5:0 6:0 7:0
                Egress QOS mappings:
                       2:6 4:5

Note

The fp-cli vlan command displays the egress mapping between mbuf’s priority value and the three priority bit values set in the VLAN tag.

Displaying VLAN information¶

Synopsis

iface

Example

<fp-0> iface
...
6:eth2.1000 [VR-0] ifid=6 (virtual) <FWD4|FWD6> (0x60)
        type=vlan mac=00:02:02:00:00:21 mtu=1500 tcp4mss=0 tcp6mss=0
        IPv4 routes=0  IPv6 routes=0
        if_ops: rx_dev=none tx_dev=vlan ip_output=none
        vlan id 1000 vlan proto 0x8100 link eth2 <REORDER_HDR>
...

Providing options¶

Some capabilities can be tuned for this module.

--ifaces¶

Maximum number of VLAN interfaces

Default value: 127
Memory footprint per VLAN interfaces: 150 B
Range: 0 .. 50000

Example

FP_OPTIONS="--mod-opt=vlan:--ifaces=512"

Then fast path can manage up to 512 VLAN interfaces.

--hash-order¶

Size order of VLAN interfaces hash table. Value automatically updated if --ifaces is changed.

Default value: 8
Range: 1 .. 31

Example

FP_OPTIONS="--mod-opt=vlan:--hash-order=10"

Note

See Fast Path Capabilities documentation for impact of the available memory on the default value of configurable capabilities

Tip

To get optimal performance, apply the following ratios to the two parameters:

Parameter	Value
–hash-order	`N`
–ifaces	`2 ** N`