Cross compiling toolchains in Docker images.
- Pre-built and configured toolchains for cross compiling.
- Most images also contain an emulator for the target system.
- Clean separation of build tools, source code, and build artifacts.
- Commands in the container are run as the calling user, so that any created files have the expected ownership, (i.e. not root).
- Make variables (CC, LD etc) are set to point to the appropriate tools in the container.
- Recent CMake and ninja are precompiled.
- Conan.io can be used as a package manager.
- Toolchain files configured for CMake.
- Current directory is mounted as the container's workdir,
/work
. - Works with the Docker for Mac and Docker for Windows.
dockcross make
: Build the Makefile in the current directory.dockcross cmake -Bbuild -H. -GNinja
: Run CMake with a build directory./build
for a CMakeLists.txt file in the current directory and generateninja
build configuration files.dockcross ninja -Cbuild
: Run ninja in the./build
directory.dockcross bash -c '$CC test/C/hello.c -o hello'
: Build the hello.c file with the compiler identified with theCC
environmental variable in the build environment.dockcross bash
: Run an interactive shell in the build environment.
Note that commands are executed verbatim. If any shell processing for environment variable expansion or redirection is required, please use bash -c 'command args...'.
This image does not need to be run manually. Instead, there is a helper script to execute build commands on source code existing on the local host filesystem. This script is bundled with the image.
To install the helper script, run one of the images with no arguments, and redirect the output to a file:
docker run --rm CROSS_COMPILER_IMAGE_NAME > ./dockcross chmod +x ./dockcross mv ./dockcross ~/bin/
Where CROSS_COMPILER_IMAGE_NAME is the name of the cross-compiler toolchain Docker instance, e.g. dockcross/linux-armv7.
Only 64-bit x86_64 images are provided; a 64-bit x86_64 host system is required.
For the impatient, here's how to compile a hello world for armv7:
cd ~/src/dockcross docker run --rm dockcross/linux-armv7 > ./dockcross-linux-armv7 chmod +x ./dockcross-linux-armv7 ./dockcross-linux-armv7 bash -c '$CC test/C/hello.c -o hello_arm'
Note how invoking any toolchain command (make, gcc, etc.) is just a matter of prepending the dockcross script on the commandline:
./dockcross-linux-armv7 [command] [args...]
The dockcross script will execute the given command-line inside the container, along with all arguments passed after the command. Commands that evaluate environmental variables in the image, like $CC above, should be executed in bash -c. The present working directory is mounted within the image, which can be used to make source code available in the Docker container.
- dockcross/base
- Base image for other toolchain images. From Debian Jessie with GCC, make, autotools, CMake, Ninja, Git, and Python.
- dockcross/linux-armv5
- Linux armv5 cross compiler toolchain for legacy devices like the Parrot AR Drone.
- dockcross/linux-mips
- Linux mips cross compiler toolchain for big endian 32-bit hard float MIPS GNU systems.
- dockcross/linux-ppc64le
- Linux PowerPC 64 little endian cross compiler toolchain for the POWER8, etc.
- dockcross/linux-x64
- Linux x86_64 / amd64 compiler. Since the Docker image is natively x86_64, this is not actually a cross compiler.
- dockcross/manylinux-x64
- Docker manylinux image for building Linux x86_64 / amd64 Python wheel packages. Also has support for the dockcross script, and it has installations of CMake, Ninja, and scikit-build
- dockcross/manylinux-x86
- Docker manylinux image for building Linux i686 Python wheel packages. Also has support for the dockcross script, and it has installations of CMake, Ninja, and scikit-build
- dockcross: C++ Write Once, Run Anywhere
- Cross-compiling binaries for multiple architectures with Docker
A special update command can be executed that will update the source cross-compiler Docker image or the dockcross script itself.
dockcross [--] command [args...]
: Forces a command to run inside the container (in case of a name clash with a built-in command), use--
before the command.dockcross update-image
: Fetch the latest version of the docker image.dockcross update-script
: Update the installed dockcross script with the one bundled in the image.dockcross update
: Update both the docker image, and the dockcross script.
To easily download all images, the convenience target display_images
could be used:
curl https://raw.githubusercontent.com/dockcross/dockcross/master/Makefile -o dockcross-Makefile for image in $(make -f dockcross-Makefile display_images); do echo "Pulling dockcross/$image" docker pull dockcross/$image done
To automatically install in ~/bin
the dockcross scripts for each images already downloaded, the
convenience target display_images
could be used:
curl https://raw.githubusercontent.com/dockcross/dockcross/master/Makefile -o dockcross-Makefile for image in $(make -f dockcross-Makefile display_images); do if [[ $(docker images -q dockcross/$image) == "" ]]; then echo "~/bin/dockcross-$image skipping: image not found locally" continue fi echo "~/bin/dockcross-$image ok" docker run dockcross/$image > ~/bin/dockcross-$image && \ chmod u+x ~/bin/dockcross-$image done
The following environmental variables and command-line options are used. In all cases, the command-line option overrides the environment variable.
This file is sourced, if it exists, before executing the rest of the dockcross script.
Default: ~/.dockcross
The Docker cross-compiler image to run.
Default: Image with which the script was created.
Extra arguments to pass to the docker run
command. Quote the entire set of
args if they contain spaces.
If a shell script named .dockcross
is found in the current directory where
the dockcross script is started, it is executed before the dockcross script
command
argument. The shell script is expected to have a shebang like
#!/bin/bash
.
For example, commands like git config --global advice.detachedHead false
can
be added to this script.
In order to extend Dockcross images with your own commands, one must:
- Use
FROM dockcross/<name_of_image>
. - Set
DEFAULT_DOCKCROSS_IMAGE
to a name of the tag you're planning to use for the image. This tag must then be used during the build phase, unless you mean to pass the resulting helper script theDOCKCROSS_IMAGE
argument.
An example Dockerfile would be:
FROM dockcross/linux-armv7 ENV DEFAULT_DOCKCROSS_IMAGE my_cool_image RUN apt-get install nano
And then in the shell:
docker build -t my_cool_image . # Builds the dockcross image. docker run my_cool_image > linux-armv7 # Creates a helper script named linux-armv7. chmod +x linux-armv7 # Gives the script execution permission. ./linux-armv7 bash # Runs the helper script with the argument "bash", which starts an interactive container using your extended image.
The key difference is that dockbuild images use the same method to conveniently isolate the build environment as dockcross but they do NOT provide a toolchain file.
---
Credits go to sdt/docker-raspberry-pi-cross-compiler, who invented the base of the dockcross script.