Understanding Linking is crucial because large Rust projects can spend roughly 50% of build time in linking. If you have ever wondered how the rust compiler, linker, loader, and operating system work together to run your Rust program - this talk might spark your interest. In this talk, Shriram will share a high level overview of the rust compilation pipeline, how the compiler is changing to become demand driven, how rustc uses LLVM, what goes inside a linker, what’s inside the object files, what do the link-time optimization flags mean, why rust prefers static linking, and are there ways to do dynamic linking.

By the end of the talk, you should have a better understanding of how Rust programs are linked, and how the linker magically makes everything fit together like pieces of a puzzle.

Here’s what we’ll cover:

• Why understanding linking is important?
• Rust Compilation: A High-Level Overview
  • What is Demand Driven Compilation a.k.a Query Based Compilation?
• Object Files: What’s inside them? (ELF Format)
• Symbol Resolution: How the linker matches variable and function names to their memory addresses.
• A simple program to check if we can invoke the linker directly on object files instead of using cargo
• Link Time Optimization: How the linker can optimize your code at link time.
• Why Rust Prefers Static Linking?
  • Understanding ABI
  • Taking a “Stabby” - a Stable ABI for Rust?
• Experiments with Dynamic Linking