93. L3fwd Functional test plan

93.1. Description

The L3fwd application is a simple example of packet processing using DPDK to demonstrate usage of poll and event mode packet I/O mechanism. The application performs L3 forwarding. For more, please consult Data Plane Development Kit L3 Forwarding Sample Application.

93.2. Prerequisites

93.2.1. Topology

It requires at least 1 port connected traffic generator:

Port0 --- TG0

2 ports is better:

Port0 --- TG0
Port1 --- TG1

93.2.2. Hardware

This suite focus on l3fwd application, so any standard Ethernet Network Adapter is qualified.

93.2.3. Software

  • dpdk l3fwd application
  • scapy to send packets from traffic generator, which is usually another Ethernet Network Adapter.

93.2.4. General Set Up

Here assume that 0000:18:00.0 and 0000:18:00.1 are DUT ports, and ens785f0 and ens785f1 are tester interfaces.

  1. Build DPDK and l3fwd application:

    <dpdk dir># meson -Dexamples=l3fwd <dpdk build dir>
<dpdk dir># ninja -C <dpdk build dir>

  2. Get the DUT ports and tester interfaces:

    <dpdk dir># ./usertools/dpdk-devbind.py -s
0000:18:00.0 'Ethernet Controller E810-C for QSFP 1592' if=ens785f0 drv=ice unused=vfio-pci
0000:18:00.1 'Ethernet Controller E810-C for QSFP 1592' if=ens785f1 drv=ice unused=vfio-pci

  3. Bind the DUT ports to vfio-pci:

    <dpdk dir># ./usertools/dpdk-devbind.py -b vfio-pci 0000:18:00.0 0000:18:00.1
0000:18:00.0 'Ethernet Controller E810-C for QSFP 1592' drv=vfio-pci unused=ice
0000:18:00.1 'Ethernet Controller E810-C for QSFP 1592' drv=vfio-pci unused=ice

93.3. Test Case

The l3fwd application has a number of command line options, here list the key options will be tested:

./dpdk-l3fwd [EAL options] -- -p PORTMASK
                         --rule_ipv4=FILE
                         --rule_ipv6=FILE
                         [-P]
                         [--lookup LOOKUP_METHOD]
                         --config(port,queue,lcore)[,(port,queue,lcore)]
                         [--rx-queue-size NPKTS]
                         [--tx-queue-size NPKTS]
                         ...

Go through every options is not efficiency, therefore 3 combinations are created to make sure every key options are tested.

Port number Rx queues per port Rule source Lookup Promise Queue size
1 1 default default(lpm) default(on) default(1024)
1 4 file em enable 2048
2 4 file em enable 2048

The table shows:

  • No. 1 covers most default options with least port, it only requires 1 port.
  • No. 2 covers the other situations against to default, it also requires 1 port.
  • No. 3 is almost same as No.2 except it requires 2 ports.

Note

  • If there is only 1 port, choose test case 1 and 2 to test.
  • If there are 2 ports, choose test case 1 and 3 to test.

Packet pattern:

  • IPv4:

    [Ether(dst="<dtsmac>", src="<srcmac>")/IP(src="1.2.3.4",dst="<match>")/Raw("x"*80)]

  • IPv6:

    [Ether(dst="<dtsmac>", src="<srcmac>")/IPv6(src="fe80::b696:91ff:fe9f:64b9",dst="<match>")/Raw("x"*80)]

Note

20 packets with increased src_ip are used in every examination.

93.3.1. Test Case 1: 1 port 1 queue with default setting

  1. Launch l3fwd:

    ./<build_dir>/examples/dpdk-l3fwd -l <lcore> -n 4 -- -p 0x1 --config="(0,0,<lcore>)"
./build/examples/dpdk-l3fwd -l 1 -n 4 -- -p 0x1 --config="(0,0,1)" --eth-dest=0,b4:96:91:9f:64:b9

    Here list some output logs which helps you understand l3fwd.

    • The DUT port is 0000:18:00.0:

      EAL: Probe PCI driver: net_ice (8086:1592) device: 0000:18:00.0 (socket 0)

    • The lookup method is lpm and use default table. DUT mac address is 40:A6:B7:7B:3F:00, the egress packets dst mac is 02:00:00:00:00:00:

      Neither ACL, LPM, EM, or FIB selected, defaulting to LPM
L3FWD: Missing 1 or more rule files, using default instead

    • Every port creates 1 rx queue and 1 tx queue:

      Creating queues: nb_rxq=1 nb_txq=1
Address:40:A6:B7:7B:3F:00, Destination:02:00:00:00:00:00

    • Route rules:

      LPM: Adding route 198.18.0.0 / 24 (0) [0000:18:00.0]
LPM: Adding route 2001:200:: / 64 (0) [0000:18:00.0]

    • RX PATH “AVX2 OFFLOAD Vector Rx” is used:

      ice_set_rx_function(): Using AVX2 OFFLOAD Vector Rx (port 0)

    • lcore 1 is used to polling port 0 rx queue 0:

      L3FWD:  -- lcoreid=1 portid=0 rxqueueid=0

    • Link status, Packets sending to DUT have to wait port link up:

      Port 0 Link up at 100 Gbps FDX Autoneg

  2. run tcpdump to capture packets on tester interface:

    tcpdump -i <TG interface> -vvv -Q in -e
tcpdump -i ens2f0 -vvv -Q in -e

  3. TG send both 20 ipv4 and ipv6 packets which match the route table:

    >>> sendp([Ether(dst="<matched mac>", src="<src mac>")/IP(src="<src ip>",dst="<198.168.0.x>")/Raw("x"*80)], iface="<tester tx port interface>")
>>> sendp([Ether(dst="<matched mac>", src="<src mac>")/IPv6(src="<src ip>",dst="<2001:200::x>")/Raw("x"*80)], iface="<tester tx port interface>")

    >>> sendp([Ether(dst="40:A6:B7:7B:3F:00", src="b4:96:91:9f:64:b9")/IP(src="1.2.3.4",dst="198.168.0.1")/Raw("x"*80)], iface="ens2f0")
>>> sendp([Ether(dst="40:A6:B7:7B:3F:00", src="b4:96:91:9f:64:b9")/IPv6(src="fe80::b696:91ff:fe9f:64b9",dst="2001:200::")/Raw("x"*80)], iface="ens2f0")

  4. Check if the packets forwarded to TG, get the packets informartion from tcpdump output:

    07:44:32.770005 40:a6:b7:7b:3f:00 (oui Unknown) > 02:00:00:00:00:00 (oui Unknown), ethertype IPv4 (0x0800), length 114: (tos 0x0, ttl 63, id 1, offset 0, flags [none], proto Options (0), length 100)
    1.2.3.4 > 198.168.0.1:  hopopt 80
07:53:08.206002 40:a6:b7:7b:3f:00 (oui Unknown) > 02:00:00:00:00:00 (oui Unknown), ethertype IPv6 (0x86dd), length 134: (hlim 64, next-header unknown (59) payload length: 80) test > 2001:200::: no next header

    Pass Criteria: Both the IPv4 and IPv6 packets are matched with the the transmitted packets.

93.3.2. Test Case 2: 1 port 4 queue with non-default setting

  1. Launch l3fwd:

    ./<build_dir>/examples/dpdk-l3fwd -l <lcore0,lcore1> -n 4 -- -p 0x1 --config="(0,0,<lcore0>),(0,1,<lcore0>),(0,2,<lcore1>),(0,3,<lcore1>)" -P --rule_ipv4="./examples/l3fwd/em_default_v4.cfg" --rule_ipv6="./examples/l3fwd/em_default_v6.cfg" --lookup=em --rx-queue-size=2048 --tx-queue-size=2048
./build/examples/dpdk-l3fwd -l 1,2 -n 4 -- -p 0x1 --config="(0,0,1),(0,1,1),(0,2,2),(0,3,2)" -P --rule_ipv4="./examples/l3fwd/em_default_v4.cfg" --rule_ipv6="./examples/l3fwd/em_default_v6.cfg" --lookup=em --rx-queue-size=2048 --tx-queue-size=2048 --parse-ptype

    “–parse-ptype” is optional, add it if DUT do not support to parse RTE_PTYPE_L3_IPV4_EXT and RTE_PTYPE_L3_IPV6_EXT.

    • Route rules:

      EM: Adding route 198.18.0.0, 198.18.0.1, 9, 9, 17 (0) [0000:18:00.0]

  2. run tcpdump to capture packets on tester interface:

    tcpdump -i <TG interface> -vvv -Q in -e
tcpdump -i ens2f0 -vvv -Q in -e

  3. TG send both ipv4 and ipv6 packets which match the route table and distributed to all queues:

    >>> sendp([Ether(dst="<unmatched mac>", src="<src mac>")/IP(src="<src ip>",dst="<198.168.0.x>")/Raw("x"*80)], iface="<tester tx port interface>")
>>> sendp([Ether(dst="<unmatched mac>", src="<src mac>")/IPv6(src="<src ip>",dst="<match table>")/Raw("x"*80)], iface="<tester tx port interface>")

  4. Check if the packets forwarded to TG, get the packets informartion from tcpdump output:

    07:44:32.770005 40:a6:b7:7b:3f:00 (oui Unknown) > 02:00:00:00:00:00 (oui Unknown), ethertype IPv4 (0x0800), length 114: (tos 0x0, ttl 63, id 1, offset 0, flags [none], proto Options (0), length 100)
    1.2.3.4 > 198.168.0.1:  hopopt 80
07:53:08.206002 40:a6:b7:7b:3f:00 (oui Unknown) > 02:00:00:00:00:00 (oui Unknown), ethertype IPv6 (0x86dd), length 134: (hlim 64, next-header unknown (59) payload length: 80) test > 2001:200::: no next header

    Pass Criteria: Both the IPv4 and IPv6 packets are matched with the the transmitted packets.

93.3.3. Test Case 3: 2 ports 4 queues with non-default setting

  1. Launch l3fwd:

    ./<build_dir>/examples/dpdk-l3fwd -l <lcore0,lcore1> -n 4 -- -p 0x3 --config="(0,0,<lcore0>),(0,1,<lcore0>),(0,2,<lcore1>, ,(0,3,<lcore1>),(1,0,<lcore0>),(1,1,<lcore0>),(1,2,<lcore1>, ,(1,3,<lcore1>)" -P --rule_ipv4="rule_ipv4.cfg" --rule_ipv6="rule_ipv6.cfg" --lookup=em --rx-queue-size=2048 --tx-queue-size=2048

  2. run tcpdump to capture packets on tester interfaces:

    tcpdump -i <tester tx Port0 interface> -vvv -Q in -e
tcpdump -i <tester tx Port1 interface> -vvv -Q in -e

  3. All TG 2 ports send both ipv4 and ipv6 packets which match the route table and distributed to all queues:

    >>> sendp([Ether(dst="<unmatched mac>", src="<src mac>")/IP(src="<src ip>",dst="<198.168.0.x>")/Raw("x"*80)], iface="<tester tx Port0 interface>")
>>> sendp([Ether(dst="<unmatched mac>", src="<src mac>")/IPv6(src="<src ip>",dst="<match table>")/Raw("x"*80)], iface="<tester tx port0 interface>")
>>> sendp([Ether(dst="<unmatched mac>", src="<src mac>")/IP(src="<src ip>",dst="<198.168.0.x>")/Raw("x"*80)], iface="<tester tx Port1 interface>")
>>> sendp([Ether(dst="<unmatched mac>", src="<src mac>")/IPv6(src="<src ip>",dst="<match table>")/Raw("x"*80)], iface="<tester tx port1 interface>")

  4. Check if the packets forwarded to TG, run tcpdump to capture packets on tester interface:

    07:44:32.770005 40:a6:b7:7b:3f:00 (oui Unknown) > 02:00:00:00:00:00 (oui Unknown), ethertype IPv4 (0x0800), length 114: (tos 0x0, ttl 63, id 1, offset 0, flags [none], proto Options (0), length 100)
    1.2.3.4 > 198.168.0.1:  hopopt 80
07:53:08.206002 40:a6:b7:7b:3f:00 (oui Unknown) > 02:00:00:00:00:00 (oui Unknown), ethertype IPv6 (0x86dd), length 134: (hlim 64, next-header unknown (59) payload length: 80) test > 2001:200::: no next header

    Pass Criteria: Both the IPv4 and IPv6 packets are matched with the the transmitted packets.