Virtual Memory in One Diagram

One diagram, one concept at a time. We build the virtual memory system from the ground up: what the hardware does, what the kernel does, and why the illusion of contiguous private memory works so well that most engineers never think about it.

Topics covered:

• Virtual vs. physical addresses: why the translation exists
• Page tables: the four-level hierarchy that maps virtual to physical
• TLB: the hardware cache that makes translation fast
• Demand paging: why malloc is a promise, not an allocation
• Page faults: what happens when you first touch a page
• Copy-on-write: why fork() is fast and how it avoids copying
• Large pages: reducing TLB misses for large data structures
• Swap: when the kernel evicts pages to disk and why it's catastrophic