/rawrtc

WebRTC and ORTC with a little bit of RAWR!

Primary LanguageCBSD 2-Clause "Simplified" LicenseBSD-2-Clause

RAWRTC

Build Status

A WebRTC and ORTC library with a small footprint that runs everywhere.

Features

The following list represents all features that are planned for RAWRTC. Features with a check mark are already implemented.

Prerequisites

The following packages are required:

  • git
  • cmake >= 3.2
  • pkg-config (pkgconf for newer FreeBSD versions)
  • SSL development libraries (libssl-dev on Debian, openssl on OSX and FreeBSD)
  • GNU make (gmake on FreeBSD for re and rew dependencies)

Meson (Alternative Build System)

If you want to use Meson instead of CMake, you have to install both the Meson build system and Ninja. Use CMake for now. Meson will be updated later.

Build

The following instruction will use a custom prefix to avoid installing the necessary dependencies and this library system-wide.

Dependencies

cd <path-to-rawrtc>
./make-dependencies.sh

Package Configuration Path

The following environment variable is required for both Meson and CMake to find the previously built dependencies:

export PKG_CONFIG_PATH=${PWD}/build/prefix/lib/pkgconfig

Note that this command will need to be repeated once the terminal has been closed.

Compile

Meson

cd <path-to-rawrtc>
meson build --default-library=static --prefix=${PWD}/build/prefix
cd build
ninja install

CMake

cd <path-to-rawrtc>/build
cmake -DCMAKE_INSTALL_PREFIX=${PWD}/prefix ..
make install

Run

RAWRTC provides a lot of tools that can be used for quick testing purposes and to get started. Let's go through them one by one. If you just want to check out data channels and browser interoperation, skip to the data-channel-sctp tool chapter.

Because we have used a custom prefix, we need to add the prefix to the path to run the various binaries. To be able to find the shared library when running a binary, the library path has to be set as well. Note: We assume that you are in the build directory.

export LD_LIBRARY_PATH=${PWD}/prefix/lib:${LD_LIBRARY_PATH}
export PATH=${PWD}/prefix/bin:${PATH}

Most of the tools have required or optional arguments which are shared among tools. Below is a description for the various arguments:

ice-role

Determines the ICE role to be used by the ICE transport, where 0 means controlled and 1 means controlling.

redirect-ip

The IP address on which an SCTP stack is listening.

Used in conjunction with redirect-port. Only used by the SCTP redirect transport.

redirect-port

The port number on which an SCTP stack is listening.

Used in conjunction with redirect-ip. Only used by the SCTP redirect transport.

sctp-port

The port number the internal SCTP stack is supposed to use. Defaults to 5000.

Note: It doesn't matter which port you choose unless you want to be able to debug SCTP messages. In this case, it's easier to distinguish the peers by their port numbers.

maximum-message-size

The maximum message size of an SCTP message the external SCTP stack is able to handle. 0 indicates that messages of arbitrary size can be handled. Defaults to 0.

Only used by the SCTP redirect transport.

ice-candidate-type

If supplied, one or more specific ICE candidate types will be enabled and all other ICE candidate types will be disabled. Can be one of the following strings:

  • host
  • srflx
  • prflx
  • relay

Note that this has no effect on the gathering policy. The candidates will be gathered but they will simply be ignored by the tool.

If not supplied, all ICE candidate types are enabled.

ice-gatherer

The ICE gatherer tool gathers and prints ICE candidates. Once gathering is complete, the tool exits.

Usage:

ice-gatherer

ice-transport-loopback

The ICE transport loopback tool starts two ICE transport instances which establish an ICE connection. Once you see the following line for both clients A and B, the ICE connection has been established:

(<client>) ICE transport state: connected

Usage:

ice-transport-loopback [<ice-candidate-type> ...]

dtls-transport-loopback

The DTLS transport loopback tool starts two DTLS transport instances which work on top of an established ICE transport connection. As soon as the DTLS connection has been established, it uses an internal interface to send raw data on the DTLS transport to the other peer. There's currently no way to verify that the data has been received but you can trace the packets using Wireshark.

To verify that the DTLS connection establishes, wait for the following line for both clients A and B:

(<client>) DTLS transport state change: connected

Usage:

dtls-transport-loopback [<ice-candidate-type> ...]

sctp-transport-loopback

The SCTP transport loopback tool starts two SCTP transport instances which work on top of an established DTLS transport connection. As soon as the SCTP connection has been established, it uses an internal interface to send raw data on the SCTP transport to the other peer.

To verify that the SCTP connection establishes, wait for the following line for both clients A and B:

(<client>) SCTP transport state change: connected

The tool will output a warning (four times) in case the data has been transmitted successfully:

Ignored incoming DCEP control message with unknown type: 72

This warning is entirely valid as this tool sends invalid DCEP messages for testing purposes.

Usage:

sctp-transport-loopback [<ice-candidate-type> ...]

sctp-redirect-transport

The SCTP redirect transport tool starts an SCTP redirect transport on top of an established DTLS transport to relay SCTP messages from and to a third party. This tool has been developed to be able to test data channel implementations without having to write the required DTLS and ICE stacks. An example of such a testing tool is dctt which uses the kernel SCTP stack of FreeBSD.

Usage:

sctp-redirect-transport <0|1 (ice-role)> <redirect-ip> <redirect-port>
                        [<sctp-port>] [<maximum-message-size>]
                        [<ice-candidate-type> ...]

data-channel-sctp-loopback

The data channel SCTP loopback tool creates several data channels on top of an abstracted SCTP data transport. As soon as a data channel is open, a message will be sent to the other peer. Furthermore, another message will be sent on a specific channel after a brief timeout.

To verify that a data channels opens, wait for the following line:

(<client>) Data channel open: <channel-label>

The tool will send some large (16 MiB) test data to the other peer depending on the ICE role. We are able to do this because RAWRTC handles data channel messages correctly and does not have a maximum message size limitation compared to most other implementations (check out this article for a detailed explanation).

Usage:

data-channel-sctp-loopback [<ice-candidate-type> ...]

data-channel-sctp

The data channel SCTP tool creates several data channels on top of an abstracted SCTP data transport:

  1. A pre-negotiated data channel with the label cat-noises and the id 0 that is reliable and ordered. In the WebRTC JS API, the channel would be created by invoking:

    peerConnection.createDataChannel('cat-noises', {
    ordered: true,
    id: 0
});

  2. A data channel with the label bear-noises that is reliable but unordered. In the WebRTC JS API, the channel would be created by invoking:

    peerConnection.createDataChannel('bear-noises', {
    ordered: true,
    maxRetransmits: 0
});

To establish a connection with another peer, the following procecure must be followed:

  1. The JSON blob after Local Parameters: must be pasted into the other peer you want to establish a connection with. This can be either a browser instance that uses the WebRTC-RAWRTC browser tool or another instance of this tool.

  2. The other peer's local parameters in form of a JSON blob must be pasted into this tool's instance.

  3. Once you've pasted the local parameters into each other's instance, the peer connection can be established by pressing Enter in both instances (click the Start button in the browser).

The tool will send some test data to the other peer depending on the ICE role. However, the browser tool behaves a bit differently. Check the log output of the tool instances (console output in the browser) to see what data has been sent and whether it has been received successfully.

In the browser, you can use the created data channels by accessing peer.dc['<channel-name>'], for example:

peer.dc['example-channel'].send('RAWR!')

Usage:

data-channel-sctp <0|1 (ice-role)> [<sctp-port>] [<ice-candidate-type> ...]

data-channel-sctp-echo

The data channel SCTP echo tool behaves just like any other echo server: It echoes received data on any data channel back to the sender.

The necessary peer connection establishment steps are identical to the ones described for the data-channel-sctp tool.

Usage:

data-channel-sctp-echo <0|1 (ice-role)> [<sctp-port>] [<ice-candidate-type> ...]