Suppose you want to make a nested function call (i.e. a call to a function from inside of another function) using a jal rather than a call for performance reasons. how would the push and pop pseudo-op

Suppose you want to make a nested function call (i.e. a call to a function from inside of another function) using a jal rather than a call for performance reasons. how would the push and pop pseudo-ops be proprely ordered along with the jal so that the previous return address isn't lost?