Structure
This section walks you through RIOT’s structure. Once you understand this structure, you will easily find your way around in RIOT’s code base.
RIOT’s code base is structured into five groups.
- The kernel (
core
) - Platform specific code (
cpu
;boards
) - Device drivers (
drivers
) - Libraries and network code (
sys
;pkg
) - Applications for demonstrating features and for testing (
examples
;tests
)
In addition RIOT includes a collection of scripts for various tasks (dist
) as
well as a predefined environment for generating this documentation (doc
).
The structural groups are projected onto the directory structure of RIOT, where each of these groups resides in one or two directories in the main RIOT directory.
The following list gives a more detailed description of each of RIOT’s top-level directories:
core
This directory contains the actual kernel. The kernel consists of the scheduler, inter-process-communication (messaging), threading, and thread synchronization, as well as supporting data-structures and type definitions.
See Kernel for further information and API documentations.
boards
The platform dependent code is split into two logic elements: CPUs and boards, while maintaining a strict 1-to-n relationship, a board has exactly one CPU, while a CPU can be part of n boards. The CPU part contains all generic, CPU specific code (see below).
The board part contains the specific configuration for the CPU it contains. This configuration mainly includes the peripheral configuration and pin-mapping, the configuration of on-board devices, and the CPU’s clock configuration.
On top of the source and header files needed for each board, this directory
additionally may include some script and configuration files needed for
interfacing with the board. These are typically custom flash/debug scripts or
e.g. OpenOCD configuration files. For most boards, these files are located in a
dist
sub-directory of the board.
See here Boards for further information.
cpu
For each supported CPU this directory contains a sub-directory with the name of
the CPU. These directories then contain all CPU specific configurations, such
as implementations of power management (LPM), interrupt handling and vectors,
startup code, clock initialization code and thread handling (e.g. context
switching) code. For most CPUs you will also find the linker scripts in the
ldscripts
sub-directory.
In the periph
sub-directory of each CPU you can find the implementations of
the CPU’s peripheral drivers like SPI, UART, GPIO, etc. See Peripheral Driver Interface
for their API documentation.
Many CPUs share a certain amount of their code (e.g. all ARM Cortex-M based
CPUs share the same code for task switching and interrupt handling). This
shared code is put in its own directories, following a xxxxx_common
naming
scheme. Examples for this is code shared across architectures (e.g.
cortexm_common
, msp430_comon
) or code shared among vendors (e.g.
stm32_common
).
See CPU for more detailed information.
drivers
This directory contains the drivers for external devices such as network interfaces, sensors and actuators. Each device driver is put into its own sub-directory with the name of that device.
All of RIOT’s device drivers are based on the peripheral driver API (e.g. SPI,
GPIO, etc.) and other RIOT modules like the xtimer
. This way the drivers are
completely platform agnostic and they don’t have any dependencies into the CPU
and board code.
See Drivers for more details.
sys
RIOT follows a modular design paradigm where everything is supposed to be a module. All of these modules that are not part of the hardware abstraction nor device drivers can be found in this directory. The libraries include data structures (e.g. bloom, color), crypto libraries (e.g. hashes, AES) , high-level APIs (e.g. Posix implementations), memory management (e.g. malloc), the RIOT shell and many more.
See System for a complete list of available libraries
sys/net
The sys/net
sub-directory needs to be explicitly mentioned, as this is where
all the networking code in RIOT resides. Here you can find the network stack
implementations (e.g. the “GNRC” stack) as well as network stack agnostic code as
header definitions or network types.
See Networking for more details on networking code.
pkg
RIOT comes with support for a number of external libraries (e.g.
ccn-lite,
microcoap). The way they are included is
that RIOT ships with a custom Makefile for each supported library that
downloads the library and optionally applies a number of patches to make it
work with RIOT. These Makefiles and patches can be found in the pkg
directory.
See Packages for a detailed description on how this works.
examples
Here you find a number of example applications that demonstrate certain features of RIOT. The default example found in this directory is a good starting point for anyone who is new to RIOT.
For more information best browse that directory and have a look at the
README.md
files that ship with each example.
To create your own application - here or anywhere else - see Setup Development Environment
tests
Many features/modules in RIOT come with their own test application, which are located in this directory. In contrary to the examples these tests are mostly focusing on a single aspect than on a set of features. Despite for testing, you might consider these tests also for insights on understanding RIOT.
dist & doc
All the tooling around RIOT can be found in these two folders.
doc
contains the doxygen configuration and also contains the compiled doxygen
output after running make doc
.
Lastly, the dist
directory contains tools to help you with RIOT. These
include
the serial terminal application pyterm
, generic scripts for flashing,
debugging, resetting (e.g. support for OpenOCD,
Jlink), as well as code enabling easy
integration to open testbeds such as the IoT-LAB.
Furthermore you can find here scripts to do all kind of code and style checks.