The Linux task_struct

Whether you're opening a browser, launching your favorite IDE, or watching a video, you're simply running a process. We know a process has a PID, a status, and consumes resources. But have you ever wondered how this is represented at the lowest level, right inside the kernel?

In Linux, a process isn't just a magic box; it's a specific C object called task_struct.

This structure is the heart of the Linux kernel. It represents any schedulable entity. Whether it's a single-threaded process or a thread within a multithreaded app, it's all just a task_struct to the kernel. The difference depends only on how it was created, via system calls like clone(), fork(), or vfork(). These system calls belong to Unix-like kernels, like Linux, and not the Windows kernel.

It currently occupies approximately 7-8 KiB of memory and lives in include/linux/sched.h.

It's fascinating to think that the entire concept of "multitasking" on your computer is just the CPU hopping between instances of this one massive struct.

Fun fact: processes are just nodes of a massive tree, because every process has a parent, except the systemd process, which is the process with PID 1.