Death of a process

It’s nature’s rule that no one can live forever. So, unfortunately, processes also have to die. A process dies by itself by calling exit() system call. This is how it selfdestructs. Processes can also die from a received signal or exception, which they cannot handle. Exit() then calls do_ exit(), which calls exit_mm(), exit_sem(), exit_files() and exit_ fs(), to release mm_struct, to de-queue processes waiting for the semaphore, to decrement count of usage of file descriptors and file system data, respectively. It also calls schedule() so that the next process can be scheduled.

With this, all resources are released except the kernel stack, task_struct and thread_info. That means the process is in the zombie state. Now it’s the responsibility of the parent process to inform the kernel that its child is not required; then the kernel removes the process completely by calling release_task(), which again calls put_task_struct() to free the pages containing the process’ kernel stack and thread_info structure and deallocate the slab cache containing the task_struct.