Pushing the Return Address

To return to the caller a subroutine must have the correct return address in $ra when the jr instruction is performed.

But this address does not have to remain in $ra all the time the subroutine is running. It works fine to save the value in $ra on the stack. The value can be restored, later on, when it is time to return to the caller.

In the picture, QTSPIM starts main (step 1) then main calls subA with a jal instruction (step 2). This puts the return address to main in $ra.

When it gets control, subA pushes the return address on the stack (step 3). The return address that subA should use when it returns to main is now on the top of the stack.

The "push" and "pop" instructions in the picture are conceptual. Actual code does these operations in the usual way.

After pushing the return address, subA runs for a while. Then subA calls subB using a jal instruction (step 4). This puts the return address to subA in $ra.

subB receives control and pushes $ra on the stack (step 5). The return address that subB must use is now safely on the top of the stack. subB now runs for a while. When subB returns to subA it pops the return address from the stack into $ra. It then uses a jr $ra instruction (step 6).

The top of stack is now the return address to main. Control returns to subA, which runs for a while. subA returns to main by popping the stack into $ra and executing a jr $ra instruction (step 7).

Control returns to main. The stack is now logically empty. The return addresses have been popped in the reverse order that they were pushed.

The program ends when main uses exception handler service 10 (step 8). The most recent version of QtSPIM requires main to end by using exception handler service 10.