Cross-development means developing and compiling programs on another platform then it intended to run upon. It is a common approach for Pi, since it’s also how Raspian is built. For quite some time I did compile Pi applications on the Pi itself. Which works quite well, but has one large drawback: performance. Being able to compile these programs on my 8 core 16GB desktop, improved compilation time enorms. And it is quite simple. I have used 2 different approaches to setup a cross build environment:
Let's have a look!
GCC is also used to cross compile Linux applications. Applications can be compiled for 32-bit or 64-bit Linux systems.
Check the current target architecture by logging into the Pi:
# gcc -dumpmachine ------------------ arm-linux-gnueabihf
Install all the required build tools:
# apt-get update # apt-get upgrade # apt-get install crossbuild-essential-armhf
Setup dpkg for armhf architecture:
dpkg --add-architecture armhf apt-get update
Installing the development libraries for new arhitecture:
apt-get install libssl-dev:armhf apt-get install gnutls-dev:armhf apt-get install libmicrohttpd-dev:armhf apt-get install libgpiod-dev:armhf
The first step is to install the development tools on the desktop, or host system. From the command line run the following:
That will involve s # apt-get update # apt-get upgrade # apt-get install build-essential # apt-get install gcc-arm-linux-gnueabihf # apt-get install g++-arm-linux-gnueabihf # apt-get install gcc-aarch64-linux-gnu # apt-get install g++-aarch64-linux-gnu
The first line makes sure that the system is up to date. The second instruction installs the general build tools. The third installs the C and C++ compiler and build tools for the Pi’s ARM processor. The ARM architecture designation, arm-linux-gnueabihf-, is used as a prefix to distinguish the ARM tools from the host system tools. Note the dashThat will involve s at the end of the string. Test the installation by entering the commant below. This reports the version of the G++ compiler installed and other information:
arm-linux-gnueabihf-g++ -v
Depending on the application and required libraries, it is possible that not all libraries are available. Easiest way is to copy these over from an existing Raspberry installation. On the Raspberry these are available in: /usr/lib/arm-linux-gnueabihf. On the target build environment the library path is: /usr/arm-linux-gnueabihf/lib/. We will be placing these retrieved libraries in a separate subdirectory (extra).
mkdir /usr/arm-linux-gnueabihf/lib/extra scp root@192.168.178.xx:/usr/lib/arm-linux-gnueabihf/libxxxx.* /usr/arm-linux-gnueabihf/lib/extra
Now we need to tell the loader (ld) to look for libraries in this extra directory
nano /etc/ld.so.conf.d/arm-linux-gnueabihf.conf Insert the path "/usr/arm-linux-gnueabihf/lib/extra" into this file. Then save and run: ldconfig
The host system should now be ready to build a Raspberry Pi program. Let’s test it with a minimal test application, just to see if everything is working. Create the following test.cpp file:
#include <iostream>
using namespace std;
int main()
{
cout << "This is our first test!" << endl;
return 0;
}
Build the program. The first line compiles the file test.cpp and the second links the compiler output to build the executable.
arm-linux-gnueabihf-g++ -O3 -g3 -Wall -c -fPIC -o test.o test.cpp arm-linux-gnueabihf-g++ -o test test.o
Now let's check the architecture of the executable with the 'file' command. If this shows 'ARM'it is compiled for an ARM processor and can transfer it to the Pi.
file test test: ELF 32-bit LSB executable, ARM, EABI5 version 1 (SYSV), dynamically linked, interpreter /lib/ld-linux-armhf.so.3, for GNU/Linux 3.2.0, BuildID[sha1]=3a73dacec4a89fa2c6ea9751d18743ba9bba33ac, not stripped scp test pi@192.168.178.xx:/tmp
That's all. I told you this was simple.
We will gdb on host and Pi to debug with the GCC toolset. First install the debug capability on our desktop system that works with target systems regardless of the processor architecture.
sudo apt-get install gdb-multiarch
Now install a debug server on the Pi system that can be connected to from our desktop and control execution on the target.
sudo apt-get install gdbserver
Start the server on the Pi. The command starts the server, runs it continuously. The third parameter is the computer name or IP address of the host computer. In my case it is “192.168.178.18”. 2001 is the port that the host and target will use for communication. Be aware that running the gdbserver is a potential security hole on the Pi. So make sure it can only be accessed from your local network.
gdbserver --multi 192.168.178.18:2001
The server uses stdout to display error messages and the output from the debugged program.
The program gdb-multiarch is an interactive program. On the host, enter the command:
gdb-multiarch
This prints a lot of output followed by the prompt (gdb). You might want to add the quiet option (-q) to the command line. At the prompt, using the name of your Pi, enter the following commands and you should see the responses, as shown in italics here:
target extended-remote 192.168.178.61:2001 Remote debugging using 192.168.178.61:2001 set remote exec-file remote/test [[no response]] file test Reading symbols from test...done.
The target command tells gdb the name of the target system, 192.168.178.61 with port, 2001. If it worked, you’ll get the response shown; otherwise a timeout message. The set remote line specifies the file location to use on the Pi target system. Remember that the executable needs to be copied onto the target system every time it is changed. We’ll see how to do that from within gdb, below. The last command, file, is the path and filename on the host computer. This can be a directory other than where the debugger is running. Now, type the following and get the response as shown:
run Starting program: /tmp/hello [[ a bunch more lines, some warnings and stuff to ignore for now ]] [Inferior 1 (process 234) exited normally]
If it works you’ll see that last line. Now switch to the SSH terminal connected to the Pi and you’ll see a couple of lines of status, the output from the program, a blank line, and a line reporting the exit status of the program. If all this doesn’t work check the error message from the server. It should explain what is happening. Often there is a mismatch between what you told gdb and the location of the executable.
The debugger provides a large number of commands for manipulating breakpoints, continuing execution, listing the program, etc. For instance, type in main and the program stops when main() is reached. A ‘c’ runs the program after a breakpoint. To help with debugging, there is an interesting mode, the Terminal User Interface, or tui. You can access and leave it by typing in Ctrl-x, Ctrl-a. This mode provides windows that show the source, registers, assembly language, and other information.