Linux virtual network devices

Linux allows us to create virtual network devices and control them programmaticaly. We can read and produce raw IP or Ethernet packets. Such devices are called TUN or TAP and often referred to as TUN/TAP. TUN device is used to manipulate IP packets, TAP - Ethernet [1].

TUN/TAP has a lot of uses: we can inspect, modify, generate, etc. network packets. OpenVPN uses TUN/TAP to route all packets through proxy server. Thus we can use TUN/TAP to create misc VPN services, e.g. IP over DNS [2] [3].

Overview

+------+              +------------+          +------+
| eth0 |  <-------->  | Networking | <------> | tun0 |
+------+              |    stack   |          +------+
                      +------------+             ^
                                                 |
                                                 |
                                                 V
                                          +-------------+
                                          | Application |
                                          +-------------+

tun0 is virtual network device interface. It acts just like a regular interface. Except we can hook to it and control it from userspace application.

When application writes packets to tun0, they will be put to networking stack and treated as if they came from a regular NIC. When packets arrive to networking stack with destination address that is routed to tun0, they will be forwarded to userspace application.

Create TUN device from CLI

We can use ip CLI command to setup the TUN device:

# ip tuntap add mode tun tun0
# ip addr add 10.0.0.1/24 dev tun0
# ip link set tun0 up

Then the resulting routing table looks like:

# route -n
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
0.0.0.0         192.168.0.1     0.0.0.0         UG    600    0        0 wlp3s0
10.0.0.0        0.0.0.0         255.255.255.0   U     0      0        0 tun0

Which means that the packets with destination 10.0.0.X will be forwarded to tun0 interface.

Create TUN device with Python

The syscalls used by ip tuntap command might be called programmatically from any language including Python. To create TUN device we need to open /dev/net/tun and call ioctl() with specific parameters:

import os
from fcntl import ioctl
import struct

TUNSETIFF = 0x400454ca
IFF_TUN   = 0x0001
IFF_NO_PI = 0x1000

ftun = os.open("/dev/net/tun", os.O_RDWR)
ioctl(ftun, TUNSETIFF, struct.pack("16sH", b"tun0", IFF_TUN | IFF_NO_PI))

However, this is pretty low level and suits my learning needs very well. Although, we can definitely find some python libraries for this [5] [6].

Also, we can programmatically set up routes using Netlink based protocols. Fortunately there's a python package pyroute2 implementing Netlink. For example we can assign an IP address to TUN interface and bring up:

# pip install pyroute2==0.4.19
from pyroute2 import IPRoute

ip = IPRoute()
idx = ip.link_lookup(ifname='tun0')[0]
ip.addr('add', index=idx, address='10.0.0.1', prefixlen=24)
ip.link('set', index=idx, state='up')
ip.close()

Controlling TUN/TAP device

Once we have created and enabled tun0 interface we can receive and send raw IP or Ethernet packets, respectively.

Receiving packets

Let's start sending ICMP requests to 10.0.0.4 which gets routed to our TUN device:

$ route -n
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
0.0.0.0         192.168.0.1     0.0.0.0         UG    600    0        0 wlp3s0
10.0.0.0        0.0.0.0         255.255.255.0   U     0      0        0 tun0
192.168.0.0     0.0.0.0         255.255.255.0   U     600    0        0 wlp3s0

$ ping 10.0.0.4

Then we can receive those packets with a simple read:

import os

while True:
    raw_packet = os.read(ftun, 1500) # we get ftun descriptor by opening /dev/net/tun
    print(raw_packet)

The output is something like:

b'E\x00\x00T\xef]@\x00@\x017G\n\x00\x00\x01\n\x00\x00\x04\x08\x00M\xef%\xc2\x00\x05As\x9dY\x00\x00\x00\x00\xe1\xa9\x05\x00\x00\x00\x00\x00\x10\x11\x12\x13\x14\x15\x16\x17\x18\x19\x1a\x1b\x1c\x1d\x1e\x1f !"#$%&\'()*+,-./01234567

which is a raw IP packet with ICMP packet as data.

By the way, seems like Linux kernel is sending SSDP packets to the TUN interface. So don't get suprised to see some unexpected traffic.

Sending packets

To send raw IP packets we write them to TUN interface:

import os

icmp_req = b'E\x00\x00(\x00\x00\x00\x00@\x01`\xc2\n\x00\x00\x04\x08\x08'\
    '\x08\x08\x08\x00\x0f\xaa\x00{\x00\x01\x00\x00\x00\x00\x00\x00\x00\x00test'
os.write(ftun, icmp_req)

By the way, we can use pypacker to construct and parse raw packets.

Prerequisites

  • To use TUN/TAP devices python scripts must be run with root permissions.

  • To forward packets from TUN/TAP to other interfaces (eth0), packet forwarding must be enabled:

    # echo 1 > /proc/sys/net/ipv4/ip_forward
    # iptables -P FORWARD ACCEPT
  • To properly route outgoing packets NAT must be enabled:

    # iptables -t nat -A POSTROUTING -s 10.0.0.0/24 -j MASQUERADE

References

Comments