La chaîne d’outils d’assemblage, d’édition de liens et d’exécution

You write assembly in mnemonics like MOV and ADD, but the CPU only speaks binary machine code. The toolchain bridges that gap.

Step 1: the assembler translates each mnemonic into its machine-instruction bytes, producing an object file (.o or .obj).

An object file holds machine code plus metadata — symbols, sections, relocations. It isn't runnable yet because addresses aren't final.

Code and data live in named sections: .text for instructions, .data for initialized data, .bss for uninitialized data.

Step 2: the linker combines object files, resolves the symbol references between them, and assigns final addresses to build the executable.

When code calls a label defined elsewhere, the assembler leaves a placeholder. The linker fills in the real address during relocation.

Step 3: when you run it, the OS loader maps the sections into memory, sets up the stack, and jumps to the entry point.

Execution begins at a start symbol — typically _start for raw Linux assembly, or main when you link with the C runtime.

Static linking copies library code into the binary; dynamic linking loads shared libraries at run time, keeping the executable smaller.

Inspect each stage: objdump disassembles, readelf shows headers and sections, and nm lists the symbols.

The full flow: assembler turns .asm into objects, the linker joins objects into an executable, and the loader runs it as a process.

You learned the toolchain: assembler makes objects, the linker resolves symbols into an executable, and the loader turns it into a running process.