A) A Background

1. UNIX, POSIX, Linux, and the ABI/API stack

In the 1970s, UNIX was created at Bell Labs as a small, elegant OS meant to be built upon, but licensing led to forks (BSD, System V, etc.) and fragmentation. This drove IEEE's POSIX effort in the 1980s to standardize a common API across UNIX-like systems. The GNU project aimed to build a free UNIX replacement but lacked a kernel — Linus Torvalds' 1991 Linux kernel filled that gap, and because it closely follows POSIX, it supports portable UNIX applications and became the heart of modern open-source OSes.

                                               Linus Torvalds : The developer of LINUX

The Linux kernel manages hardware and processes and exposes system calls (open, read, fork, exec, ...) which form the ABI (Application Binary Interface — binary-level contract). The POSIX API defines a consistent source-level interface (fopen, pthread_create, ...). libc (glibc/musl) sits between applications and the kernel, translating POSIX calls into actual syscalls via the correct ABI — enabling portability across hardware and kernel versions as long as the ABI is stable.

2. Linux Filesystem Principles

• Inodes: every file has an inode storing metadata (size, permissions, timestamps, owner, pointers to data blocks) — not the filename.
• Filename ↔ inode mapping: directories map names to inode numbers, so multiple names (links) can point to one inode.
• Blocks: storage is divided into blocks (512B–4KB); inodes point to these blocks (direct, indirect, double indirect, etc.).
• Hard links: a second directory entry pointing to the same inode; data persists until the last link is deleted.
• Symlinks: a special file containing a path to another file, with its own inode; can cross filesystems or point to nothing (broken link).
• Directories: special files listing filenames → inode numbers, making rename/move fast (just a directory-entry change).
• Everything is a file: devices, sockets, pipes, directories all share the file interface.
• Mounting: filesystems (ext4, vfat, ntfs, ...) mount into the unified tree at /, or subdirectories like /mnt, /home.
• Permissions: inode permission bits + ownership control access via the kernel.
• Filenames are independent of inodes (up to ~255 chars, reusable across dirs).
• Keeping filenames out of inodes enables linking, fast lookups, and inode reuse.
• Daemon: an application that runs in the background.