Paging

Paging lets a program's memory be scattered all over physical RAM instead of sitting in one continuous block — and lets parts of it stay on disk until they're actually touched. The OS cuts a program's address space into fixed-size pages and cuts RAM into equal-size frames. A page can go in *any* free frame; a page table records which frame each page landed in.

A virtual address is two numbers

Programs use virtual addresses that say nothing about where data physically sits. The hardware splits each one into a page number and an offset into that page. If the page size is a power of two, this split is just slicing the bits: the low bits are the offset, the high bits are the page number. Translation keeps the offset and swaps the page number for its frame.

// Page size = 1024 bytes (2^10), so the low 10 bits are the offset.
// Page table: pageTable[pageNumber] = frameNumber
const pageSize = 1024;

function translate(virtualAddr, pageTable) {
  const page   = Math.floor(virtualAddr / pageSize);
  const offset = virtualAddr % pageSize;
  const frame  = pageTable[page];          // which frame holds this page
  return frame * pageSize + offset;        // physical address
}

Worked example

1. 1Page size is 1024 bytes, and the program reads virtual address 1500.
2. 2Page number = 1500 / 1024 = 1; offset = 1500 % 1024 = 476.
3. 3The page table says page 1 currently lives in frame 4.
4. 4Physical address = 4 * 1024 + 476 = 4572 — same offset, new frame.
5. 5If page 1 were marked *not loaded*, this access would trigger a page fault and the OS would fetch it from disk first.