ret returns from a function — it sends control back to whoever called it. It is the closing bracket that every call expects. When a function was called, the caller left a return address in the ra register (see call and jal); ret simply jumps to that address, picking up right after the original call. It is a pseudo-instruction for jumping to ra.

The whole correctness of ret rests on one condition: ra must still hold the right address when it runs. For a function that calls nothing else, that is automatic — nothing touched ra between entry and exit. But the moment a function makes a call of its own, that call overwrites ra with a new return address, erasing the one the function needs. This is why a function that calls others must save its ra on entry and restore it just before ret. Forget the restore and the function returns to the wrong place — a classic crash where the program jumps into nonsense or loops forever.

By convention, a function leaves its result in the a0 register before returning, and the caller looks for it there.

The example is the simplest complete function: square multiplies its argument by itself and then ret hands the result back. Because it calls nothing else, it needs no saving or restoring — just compute and return. Stepping through it in the simulator and watching control jump back to the caller is the clearest way to see calling and returning work.