trex_stack_linux_based.cpp- Need help in understanding on how arp and ping works when linux_based namespaces are configured ?

[chris...@gmail.com]

Hi All , 

I am creating linux_based network namespaces using trex. 

I understand one end of veth pair is within namespace and the other end is in default linux namespace .

My trex_cfg used interface named test .

- port_limit: 2
  version   : 2
  #c         : 3
  interfaces: ['test', 'dummy']
  stack     : linux_based
  low_end   : true
  low_end_core: 15

DPDK args
 xx  --lcores  (0,1,2)@15  -n  4  --log-level  8  --main-lcore  0  --vdev=net_af_packet1000000,iface=test,blocksz=593920,framesz=9280,framecnt=256  --no-pci  --no-huge  -m  250  --legacy-mem

trex(service)>ns show-node -p 0 --mac 00:01:02:03:04:05

Setting port 0 in with namespace configuration               [SUCCESS]


wait_for_async_results                                       [SUCCESS]

{'nodes': [{'bpf': '',
            'ether': {'src': '00:01:02:03:04:05'},
            'ipv4': {'dst': '2.2.2.11', 'src': '2.2.2.110'},
            'ipv6': {'enabled': False},
            'is-shared-ns': False,
            'is_bird': False,
            'linux-ns': 'trex-a-0-1',
            'linux-veth-external': 'trex-a-0-1-T',
            'linux-veth-internal': 'trex-a-0-1-L',
            'vlan': {'tags': [622], 'tpids': []},
            'vlan_filter': 'no_tcp_udp'}]}
trex(service)>

On host :

38: trex-a-0-1-T@if37: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9280 qdisc noqueue state UP group default qlen 1000
    link/ether fa:eb:bb:53:97:8d brd ff:ff:ff:ff:ff:ff link-netns trex-a-0-1
53: test: <BROADCAST,MULTICAST,PROMISC,UP,LOWER_UP> mtu 9000 qdisc mq state UP group default qlen 1000
    link/ether 0e:05:9b:7f:72:89 brd ff:ff:ff:ff:ff:ff
    altname enp108s0f0v20
    inet6 fe80::c05:9bff:fe7f:7289/64 scope link
       valid_lft forever preferred_lft forever

Inside namespace :

ip netns exec trex-a-0-1 ip addr show

37: trex-a-0-1-L@if38: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9280 qdisc noqueue state UP group default qlen 1000
    link/ether 00:01:02:03:04:05 brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet 2.2.2.110/32 scope global trex-a-0-1-L
       valid_lft forever preferred_lft forever

ip netns exec trex-a-0-1 ip route show
default via 2.2.2.11 dev trex-a-0-1-L
2.2.2.11 dev trex-a-0-1-L scope link

Can someone help me understand , how arp and ping works , when i try pinging 2.2.2.11 which is outside host and resides on DUT ?

How does that namespace know that the packets have to be routed through linux interface test ?

When i check trex_stack_linux_based.cpp, i see the veth being bound to a linux socket sockfd = socket(AF_PACKET, SOCK_RAW, htons(ETH_P_ALL));.Further i am unable to understand how routing takes place. Can someone please help me understand this?