Embox is a configurable RTOS designed for resource constrained and embedded systems. Embox's main idea is to use Linux software without Linux.

• VoIP phone on STM32F7Discovery based on PJSIP project
• Qt on STM32F7Discovery
• OpenCV on STM32F769i board
• A lot of programming languages are available: Python, Lisp, Java (phoneme), TCL, Ruby, Lua, JS, Scheme
• SSHD based on Dropbear project
• Mesa3d
• Quake3
• zeromq, qpid
• ...

• POSIX-compliant
• C++ support
• File systems: FAT, ext2/3/4, ...
• TCP/IP: BSD sockets; supported protocols: UDP, HTTP, ARP, NTP, ICMP, ...
• Cross-platform: ARM, MIPS, x86, RISC-V, Microblaze, SPARC, PowerPC, E2K
• Popular platforms: STM32, i.MX6, RaPi, ...
• Provides popular desktop software on MCU (e.g. STM32): Qt, OpenCV, PJSIP, dropbear, ...
• Unix-like shell utilities: ls, cat, mount, ...
• Popular programming languages: java, python, lua, tcl, lisp, ruby, js, scheme

Mailing lists:

• main (english): [email protected]
• russian: [email protected]

Telegram chats:

• main (english): https://t.me/embox_chat_en
• news: https://t.me/embox_news
• russian: https://t.me/embox_chat

• Wiki
• PDF
• GitHub repository with source for docs

Here's a quick overview on how to build and run Embox.

Required environment:

• gcc and make
• cross compiler for the target platform

For Debian-based systems (most packages are installed out of box though):

$ sudo apt-get install build-essential gcc-multilib curl libmpc-dev python

For Arch Linux:

$ sudo pacman -S make gcc-multilib cpio qemu
$ yay -S python2-bin

For MAC OS X (requires MacPorts installed):

$ sudo port install i386-elf-gcc i386-elf-binutils cpio gawk qemu

For any system with Docker (more info on wiki Emdocker):

$ ./scripts/docker/docker_start.sh
$ . ./scripts/docker/docker_rc.sh

First of all:

$ git clone https://github.com/embox/embox.git embox
$ cd embox

Since Embox is highly configurable project, it is necessary to specify modules to be built and params for them. To build the OS make command is used. All commands described below are called from the embox directory, which includes src/, templates/, ...

For configuring it is necessary to specify parameters and modules supposed to be included into the system. There are two ways to set up a current config:

• make confload-<template> - build artefacts generated in a fixed place
• ./confset <template> - build artefacts maintained in project dependent working paths

While make confload-<template> is used, files from the template's folder are copied into './conf' folder. You can modify them. While confset <template> is used, files from the template's folder are copied into ./work/<template>/conf folder and a link './conf' is created to ./work/<template>/conf. This usage is suited for quick switching between your templates, because full rebuilds are avoided. Moreover, working conf-settings are retained as well, when you switch build e.g. between real target and qemu and back again. Only downside of using the confset method: more SSD/HDD usage, dependent on number of parallel projects.

Embox has several templates prepared, to list them use the following command:

$ make confload

The simplest way to load a template is to specify its name in the command:

$ ./confset <template>

or

$ make confload-<template>

For the quick overview you can use one of qemu templates that exist for most architectures, that is, x86/qemu for x86:

$ ./confset x86/qemu   (yet no autocomplete/-suggest when only a part is typed: to Do)

or

$ make confload-x86/qemu

After configuring the project just run make to build:

$ make

The resulting image can now be run on QEMU. The simplest way is to execute ./scripts/qemu/auto_qemu script:

$ sudo ./scripts/qemu/auto_qemu

sudo is requried to setup a TUN/TAP device necessary to emulate networking.

After the system is loaded, you’ll see the embox> prompt, now you are able to run commands. For example, help lists all existing commands.

To test the connection:

ping 10.0.2.16

If everything's fine, you can connect to the Embox terminal via telnet.

To exit Qemu type ctrl + A and X after that.

You can use the same script with -s -S -machine accel=tcg flags for debugging:

$ sudo ./scripts/qemu/auto_qemu -s -S -machine accel=tcg

After running that QEMU waits for a connection from a gdb-client. Run gdb in the other terminal:

$ gdb ./build/base/bin/embox
...
(gdb) target extended-remote :1234
(gdb) continue

The system starts to load.

At any moment in gdb terminal you can type ctrl + C and see the stack of the current thread (backtrace) or set breakpoints (break <function name>, break <file name>:<line number>).

Embox supports the following CPU architectures: x86, ARM, Microblaze, SPARC, PPC, MIPS.

In order to work with architectures other than x86 you'll need a cross compiler. After installing the cross compiler just repeat the step above, starting with configuring:

make confload-<arch>/qemu
make
sudo ./scripts/qemu/auto_qemu

The output is quite the same as in the case of the x86 architecture.

Embox supports networking on qemu x86, ARM, MIPS and Microblaze.