Note that if you're just experimenting you can also use LLDB's breakpoint commands + "thread ret"/reg write to control logic.

But I've had some weird issues using "thread ret" (even when using it as the first instruction in a function when it shouldn't corrupt the stack), so patching memory is probably cleaner.