How to cross compile C++ for ARM from an X86/X64 computer

Today we are going to see how to compile a C++ application for an ARM processor from a computer with x86 or x64 architecture.

To do this, first of all, we must install the necessary dependencies.

sudo apt-get install libc6-armel-cross libc6-dev-armel-cross binutils-arm-linux-gnueabi libncurses5-dev build-essential bison flex libssl-dev bc

Next, we install the appropriate version of G++ for our ARM architecture. One of the biggest complications is the designation of ARM processors, which determines the necessary dependencies.

If in doubt, this post can be helpful. But, in summary, ARMv7 and earlier processors are 32-bit, while ARMv8 and later are 64-bit (compatible with 32-bit).

On the other hand, there are cross compilers for floating point operations executed by software (such as gnueabi) and floating point operations implemented in hardware (such as gnueabihf). If possible to use the latter, it is preferred over the former.

So, depending on the type of processor we have, we do the following,

For ARM 32bits

sudo apt-get install gcc-arm-linux-gnueabihf g++-arm-linux-gnueabihf

For ARM 64bits

sudo apt-get install gcc-aarch64-linux-gnu g++-aarch64-linux-gnu

Now, to test, let’s create a simple “hello world” program. To do this, we create a ‘main.cpp’ file with the following content.

int main(int argc, char *argv[]) {
    cout << "Hello world!" << endl;
    return 0;
}

To compile it, we run the following command.

For ARM 32bits

arm-linux-gnueabihf-g++ main.cpp -o hello

For ARM 64bits

aarch64-linux-gnu-gcc main.cpp -o hello

We move the ‘hello’ file to our target machine, and change the execution permissions (for example, to 777).

We run our test program with the following command.

./hello