- It provides an abstraction layer for the application on top of the TCP/IP stack. This will allow applications to choose between IP stacks in the future.
- The APIs it provides are thread safe, even if the underlying TCP/IP stack APIs are not.
ESP-IDF currently implements ESP-NETIF for the lwIP TCP/IP stack only. However, the adapter itself is TCP/IP implementation agnostic and different implementations are possible.
Some ESP-NETIF API functions are intended to be called by application code, for example to get/set interface IP addresses, configure DHCP. Other functions are intended for internal ESP-IDF use by the network driver layer.
In many cases, applications do not need to call ESP-NETIF APIs directly as they are called from the default network event handlers.
Overall application interaction with a specific IO driver for communication media and configured TCP/IP network stack is abstracted using ESP-NETIF APIs and outlined as below:
ESP-NETIF is an intermediary between an IO driver and a network stack, connecting packet data path between these two. As that it provides a set of interfaces for attaching a driver to ESP-NETIF object (runtime) and configuring a network stack (compile time). In addition to that a set of API is provided to control network interface lifecycle and its TCP/IP properties. As an overview, the ESP-NETIF public interface could be divided into these 6 groups:
The ESP-NETIF L2 TAP interface is ESP-IDF mechanism utilized to access Data Link Layer (L2 per OSI/ISO) for frame reception and transmission from user application. Its typical usage in embedded world might be implementation of non-IP related protocols such as PTP, Wake on LAN and others. Note that only Ethernet (IEEE 802.3) is currently supported.
From user perspective, the ESP-NETIF L2 TAP interface is accessed using file descriptors of VFS which provides a file-like interfacing (using functions like ``open()``, ``read()``, ``write()``, etc). Refer to :doc:`/api-reference/storage/vfs` to learn more.
There is only one ESP-NETIF L2 TAP interface device (path name) available. However multiple file descriptors with different configuration can be opened at a time since the ESP-NETIF L2 TAP interface can be understood as generic entry point to Layer 2 infrastructure. Important is then specific configuration of particular file descriptor. It can be configured to give an access to specific Network Interface identified by ``if_key`` (e.g. `ETH_DEF`) and to filter only specific frames based on their type (e.g. Ethernet type in case of IEEE 802.3). Filtering only specific frames is crucial since the ESP-NETIF L2 TAP needs to exist along with IP stack and so the IP related traffic (IP, ARP, etc.) should not be passed directly to the user application. Even though such option is still configurable, it is not recommended in standard use cases. Filtering is also advantageous from a perspective the user’s application gets access only to frame types it is interested in and the remaining traffic is either passed to other L2 TAP file descriptors or to IP stack.
To be able to use the ESP-NETIF L2 TAP interface, it needs to be enabled in Kconfig by :ref:`CONFIG_ESP_NETIF_L2_TAP` first and then registered by :cpp:func:`esp_vfs_l2tap_intf_register()` prior usage of any VFS function.
Once the ESP-NETIF L2 TAP is registered, it can be opened at path name “/dev/net/tap”. The same path name can be opened multiple times up to :ref:`CONFIG_ESP_NETIF_L2_TAP_MAX_FDS` and multiple file descriptors with with different configuration may access the Data Link Layer frames.
The ESP-NETIF L2 TAP can be opened with ``O_NONBLOCK`` file status flag to the ``read()`` does not block. Note that the ``write()`` may block in current implementation when accessing a Network interface since it is a shared resource among multiple ESP-NETIF L2 TAP file descriptors and IP stack, and there is currently no queuing mechanism deployed. The file status flag can be retrieved and modified using ``fcntl()``.
On success, ``open()`` returns the new file descriptor (a nonnegative integer). On error, -1 is returned and ``errno`` is set to indicate the error.
ioctl()
^^^^^^^
The newly opened ESP-NETIF L2 TAP file descriptor needs to be configured prior its usage since it is not bounded to any specific Network Interface and no frame type filter is configured. The following configuration options are available to do so:
*``L2TAP_S_INTF_DEVICE`` - bounds the file descriptor to specific Network Interface which is identified by its ``if_key``. ESP-NETIF Network Interface ``if_key`` is passed to ``ioctl()`` as the third parameter. Note that default Network Interfaces ``if_key``'s used in ESP-IDF can be found in :component_file:`esp_netif/include/esp_netif_defaults.h`.
*``L2TAP_S_DEVICE_DRV_HNDL`` - is other way how to bound the file descriptor to specific Network Interface. In this case the Network interface is identified directly by IO Driver handle (e.g. :cpp:type:`esp_eth_handle_t` in case of Ethernet). The IO Driver handle is passed to ``ioctl()`` as the third parameter.
*``L2TAP_S_RCV_FILTER`` - sets the filter to frames with this type to be passed to the file descriptor. In case of Ethernet frames, the frames are to be filtered based on Length/Ethernet type field. In case the filter value is set less than or equal to 0x05DC, the Ethernet type field is considered to represent IEEE802.3 Length Field and all frames with values in interval <0, 0x05DC> at that field are to be passed to the file descriptor. The IEEE802.2 logical link control (LLC) resolution is then expected to be performed by user’s application. In case the filter value is set greater than 0x05DC, the Ethernet type field is considered to represent protocol identification and only frames which are equal to the set value are to be passed to the file descriptor.
All above set configuration options have getter counterpart option to read the current settings.
..warning::
The file descriptor needs to be firstly bounded to specific Network Interface by ``L2TAP_S_INTF_DEVICE`` or ``L2TAP_S_DEVICE_DRV_HNDL`` to be ``L2TAP_S_RCV_FILTER`` option available.
VLAN tagged frames are currently not recognized. If user needs to process VLAN tagged frames, they need set filter to be equal to VLAN tag (i.e. 0x8100 or 0x88A8) and process the VLAN tagged frames in user application.
``L2TAP_S_DEVICE_DRV_HNDL`` is particularly useful when user's application does not require usage of IP stack and so ESP-NETIF is not required to be initialized too. As a result, Network Interface cannot be identified by its ``if_key`` and hence it needs to be identified directly by its IO Driver handle.
Opened and configured ESP-NETIF L2 TAP file descriptor can be accessed by ``read()`` to get inbound frames. The read operation can be either blocking or non-blocking based on actual state of ``O_NONBLOCK`` file status flag. When the file status flag is set blocking, the read operation waits until a frame is received and context is switched to other task. When the file status flag is set non-blocking, the read operation returns immediately. In such case, either a frame is returned if it was already queued or the function indicates the queue is empty. The number of queued frames associated with one file descriptor is limited by :ref:`CONFIG_ESP_NETIF_L2_TAP_RX_QUEUE_SIZE` Kconfig option. Once the number of queued frames reach configured threshold, the newly arriving frames are dropped until the queue has enough room to accept incoming traffic (Tail Drop queue management).
| On success, ``read()`` returns the number of bytes read. Zero is returned when size of the destination buffer is 0. On error, -1 is returned, and ``errno`` is set to indicate the error.
|**EBADF** - not a valid file descriptor.
|**EAGAIN** - the file descriptor has been marked non-blocking (``O_NONBLOCK``), and the read would block.
write()
^^^^^^^
A raw Data Link Layer frame can be sent to Network Interface via opened and configured ESP-NETIF L2 TAP file descriptor. User’s application is responsible to construct the whole frame except for fields which are added automatically by the physical interface device. The following fields need to be constructed by the user's application in case of Ethernet link: source/destination MAC addresses, Ethernet type, actual protocol header and user data. See below for more information about Ethernet frame structure.
In other words, there is no additional frame processing performed by the ESP-NETIF L2 TAP interface. It only checks the Ethernet type of the frame is the same as the filter configured in the file descriptor. If the Ethernet type is different, an error is returned and the frame is not sent. Note that the ``write()`` may block in current implementation when accessing a Network interface since it is a shared resource among multiple ESP-NETIF L2 TAP file descriptors and IP stack, and there is currently no queuing mechanism deployed.
| On success, ``write()`` returns the number of bytes written. Zero is returned when size of the input buffer is 0. On error, -1 is returned, and ``errno`` is set to indicate the error.
|**EBADF** - not a valid file descriptor.
|**EBADMSG** - Ethernet type of the frame is different then file descriptor configured filter.
|**EIO** - Network interface not available or busy.
Opened ESP-NETIF L2 TAP file descriptor can be closed by the ``close()`` to free its allocated resources. The ESP-NETIF L2 TAP implementation of ``close()`` may block. On the other hand, it is thread safe and can be called from different task than the file descriptor is actually used. If such situation occurs and one task is blocked in I/O operation and another task tries to close the file descriptor, the first task is unblocked. The first's task read operation then ends with error.
The initialization code as well as registering event handlers for default interfaces, such as softAP and station, are provided in separate APIs to facilitate simple startup code for most applications:
Please note that these functions return the ``esp_netif`` handle, i.e. a pointer to a network interface object allocated and configured with default settings, which as a consequence, means that:
* The created object has to be destroyed if a network de-initialization is provided by an application using :cpp:func:`esp_netif_destroy_default_wifi()`.
