Master Linux Driver, Kernel, and Application Compilation in 5 Minutes

When compiling the kernel, the method of specifying the architecture and cross-compilation tool is the same as for compiling modules. After the above operations, Little Wang has become quite familiar. He also learned to set environment variables first before compiling.

Before the first configuration compilation, previous generated files need to be cleared. This can be done with the following command.

Next, configure the kernel. The kernel configuration can use one of the following commands.

Among these four methods, it is recommended to use make menuconfig, which does not depend on QT or GTK+ and is very intuitive. After using the make menuconfig command, you will enter the interface shown in Figure 1, where you can trim the options you need to compile into the kernel according to your needs. For boolean options, press Y or N to decide whether to compile into the kernel. [*] indicates selected, and [] indicates unselected. For tristate options, you can choose Y, M, or N based on your needs, <* > indicates static compilation, indicates compilation as a module, and < > indicates not compiling into the kernel.

Figure 1 Kernel Configuration Interface

After configuration, select save and then exit. After configuration, a .config file will be generated, which saves the configuration information for the Makefile to use when compiling the kernel.

In the configuration interface, we configure the necessary functions and support for the corresponding device drivers as needed, and then compile the kernel and device tree. The commands for compiling the kernel and device tree are as follows.

To speed up compilation, you can add the “-j” option after the compilation command. For example, if the PC’s CPU has 4 cores, the compilation command would be as follows.

If you want to compile certain drivers as modules, use the following command.

For the corresponding development board, manufacturers generally have default configurations, which are located in the kernel source arch/arm/configs directory. For example, use the following default configuration.

If there is an existing .config file, you can load the existing .config configuration file using the following command.

After the kernel and device tree compilation is complete, zImage and the corresponding dtb files will be generated. Burn the image to the target board according to the manufacturer’s provided method.

For example, the application file to be compiled is test.c, and it needs to be compiled to run on the ARM development board. The compilation command is as follows.

This is a very simple program and has only one file, so it can be compiled directly by entering the command. If the project is larger, this method is not advisable. Usually, a Makefile needs to be written for project management. A sample Makefile for the application is as follows.

After writing the Makefile, execute the make command to generate the target file test.

The executable file generated by the compilation can be downloaded to the development board via TFTP, NFS, etc. For setting up TFTP and NFS servers on the PC, you can refer to related articles online, which will not be elaborated here.

1. NFS File Transfer

Connect the target board to the local area network or directly connect it to the PC host with a crossover cable, set the target board’s IP to be in the same subnet as the host, and then perform the remote mount operation. Place the files to be transferred in the NFS server directory on the PC. Execute the following operations on the development board.

Before performing remote mounting, it is best to first check network communication using the ping command. Only when the ping is successful can normal mounting be performed; otherwise, check the network. If the remote mount fails even when ping is successful, check the other settings on the host and target machine.

2. TFTP File Transfer

Place the files to be transferred in the TFTP server directory on the PC. The TFTP download command format is as follows.

For example: download the hellodrv.ko file from the TFTP server with IP 192.168.1.168, execute the following command in the current directory of the development board.

Using any of the above methods will copy the compiled driver module hellodrv.ko and the test application to the target board.

Execute the following command to dynamically load the driver module into the kernel.

Use the following command to view the loaded modules.

If you need to remove the module, use the following command.

IoT9000A-LI is an ARM industrial control board developed by Guangzhou Zhiyuan Electronics Co., Ltd., based on the M6708-T series core board. The standard processor on the core board is i.MX 6U, Cortex™-A9 dual-core. The board integrates a large number of peripheral interfaces, including Gigabit Ethernet, audio, USB, HDMI, LVDS, LCD, miniPCIE, camera, CAN signal, UART signal, etc. The integrated multifunctional HD video engine can provide 1080P 60fps video decoding, 1080P 30fps video encoding, and includes a 2D, 3D graphics engine, meeting the rich graphics and high responsiveness requirements of next-generation applications such as consumer electronics, industrial, and automotive in-vehicle entertainment systems, as well as medical applications.

Figure 2 IoT9000A_LI Product Image

For related learning kits, click “Read Original” to learn more~