# tapif-io Component This component implements a tap networking interface that provides connectivity to host network using `tuntap` interface in Linux. It could be used to route lwip traffic to host side network, typically when working with the **Linux target**. ## How to use this component ### Usage of the API 1) Add the path to this component to the `EXTRA_COMPONENT_DIRS` in your project makefile ```cmake list(APPEND EXTRA_COMPONENT_DIRS "$ENV{IDF_PATH}/examples/common_components/tapif_io") ``` 2) Include lwip and linux side of the configuration ```cpp #include "esp_netif.h" // esp-netif #include "tapio.h" // esp-netif's driver side #include "lwip/tapif.h" // esp-netif's network stack side ``` 3) Configure the esp-netif a) setup the linux tap I/O config ```cpp esp_netif_driver_ifconfig_t driver_cfg = { .handle = tapio_create(), .transmit = tapio_output, }; ``` b) configure the lwip netif for the tap interface ```cpp struct esp_netif_netstack_config stack_cfg = { .lwip = { .init_fn = lwip_tapif_init, .input_fn = lwip_tapif_input, } }; ``` c) configure the esp-netif basic parameters ```cpp esp_netif_inherent_config_t base_cfg = { .if_key = "TAP", // unique name of the interface .flags = ESP_NETIF_FLAG_AUTOUP, // no dhcp client, starts when it's set up .ip_info = &ip_info, // add static IP info .route_prio = 100 // priority for setting default gateway }; ``` 4) Initialize and attach the esp_netif to the I/O handle ```cpp esp_netif_t *tap_netif = esp_netif_new(&cfg); esp_netif_attach(tap_netif, driver_cfg.handle); ``` ### Host side networking 1) Create a new tun/tap interface type named `tap0` a) You can run the script `./make_tap_netif` b) Update the IP address of the interface to correspond to the configured static IP in previous step 2) Start the application and send/receive the packets via `tap0` interface * it is possible to create server or client test application listening or connecting to this interface. * it is also possible to route these packets to external network (using routing rules or simply by ip forwarding if using the same subnet) #### Common networking/routing examples ##### Isolated internal connection Is useful to experiment with one interface with no intention to connect to internet or external facilities. Typically, when we want to create a server listening on the `tap0` interface and run a client in lwip, e.g. the default `tcp_client` socket example in IDF. * Create the tap interface using `./make_tap_netif` and set the IP address **not to overlap** with any other IPv4 network range (e.g. `ip addr add 192.168.5.1/24 dev tap0`) * Configure the `tapif_io` component to use static address from that range (e.g. `192.168.5.x`) * Configure the `tcp_client` example to connect to the tap interface IP address (e.g. `192.168.5.1`) * Execute a tcp server listening on the tap interface and the configured port (e.g. `nc -l 3333`) * Build and run the `tcp_client` example to send and receive data between the server created in the previous step. ##### Connecting to the external network using IP forwarding This allows using full-featured network facilities of your host network, but a care must be taken to the selected IP addresses to avoid potential conflicts. * Set the IP address of the `tap0` interface from the range used by your host system's default gateway (e.g. `ip addr add 192.168.0.123/24 dev tap0`, assuming the default netif is `eth0` with IP range of `192.168.0.x` and this address doesn't overlap with any other IP address in this network) * Configure the `tapif_io` with another address from the same range, e.g. ```text CONFIG_EXAMPLE_CONNECT_TAPIF_IP_ADDR="192.168.0.100" CONFIG_EXAMPLE_CONNECT_TAPIF_NETMASK="255.255.255.0" CONFIG_EXAMPLE_CONNECT_TAPIF_GW="192.168.0.1" ``` assuming that the default gateway of your host network is configured to `192.168.0.1` * Build and run the lwip example to interact with the host network, e.g, to send an HTTP request to a publicly available http server (if the server is reachable from your host network) (Note, that the IP forwarding must be enabled in the host system: ```bash echo 1 > /proc/sys/net/ipv4/ip_forward ``` ) ##### Routing the internal interface to the host network with IP tables Uses an isolated interface with routing and NAT-ing between interfaces * Configure the `tap0` interface address **not to overlap** with any other IPv4 network range. * Setup `MASQUERADE` target to route network traffic between `tap0` and your default network interface (`eth0` in the below example). ```bash sudo iptables -t nat -A POSTROUTING -o eth0 -j MASQUERADE sudo iptables -A FORWARD -i eth0 -o tap0 -m state --state RELATED,ESTABLISHED -j ACCEPT sudo iptables -A FORWARD -i tap0 -o eth0 -j ACCEPT ``` ##### Using DHCP It's also possible to configure the lwip interface to use DHCP client (common setup for most default network interfaces, such as Ethernet or WiFi station) and set up a DHCP server on the host machine to assign the IP address dynamically. 1) **Configure and set the `esp-netif` up** * Same as in [API usage](#Usage-of-the-API), but update the base esp-netif config `3c)` to enable DHCP client ```cpp esp_netif_inherent_config_t base_cfg = { .if_key = "TAP", .flags = (esp_netif_flags_t)(ESP_NETIF_DHCP_CLIENT | ESP_NETIF_FLAG_EVENT_IP_MODIFIED | ESP_NETIF_FLAG_AUTOUP), .route_prio = 100 }; ``` * After starting the netif, tell the lwIP that we're connected ```cpp esp_netif_action_connected(tap_netif, 0, 0, 0); ``` * Wait for the IP address to be assigned. This could be implemented as a wait loop below, as the esp-event currently doesn't support IP events on Linux target. ```cpp esp_netif_ip_info_t ip_info = {}; while (ip_info.ip.addr == 0) { ESP_LOGI("tap-init", "No IP assigned, waiting..."); usleep(1000000); esp_netif_get_ip_info(tap_netif, &ip_info); } ESP_LOGI("tap-init", "Assigned IP address:"IPSTR ",", IP2STR(&ip_info.ip)); ``` 2) **Configure forwarding/routing** if needed based on the previous sections. 3) **Configure the DHCP server on the host machine** Example for `isc-dhcp-server` ```bash INTERFACES="tap0"; authoritative; subnet 192.168.5.0 netmask 255.255.255.0 { range 192.168.5.2 192.168.5.200; option routers 192.168.5.1; option domain-name-servers 8.8.8.8; } ```