I am trying to understand the process of declaration and assignment of a primitive type at the back stage.
int i;
i = 3;
There are not always addresses involved. The compiler can put variables into registers if it finds that their address is never taken by the programmer. So you wouldn't need any access to the main memory. For example in your code above, what the compiler could generate could be as simple as
add $2, $0, 3
to put value 3 into register 2. As soon as you create a pointer and make it point to that variable, then you have an address, actually. And then the variable cannot be in a register only anymore.
Really stack is special area of proccess virtual memory, so everything in stack has memory addres. Head of the stack holded by ESP (SP) registry (x86 architecture). Stack addresses usualy lower than memory adresses, because stack located closer to begining of proccess vitrual memory then heap.
How else do stack buffer overflows occur? :) someone's got to be writing to a pointer to the stack.
Assuming you're talking about C or C++ (I can't tell), yes. You can access the address like so:
int i = 3;
int *k = &i; // k now is a pointer to i
*k = 4; // assigns the value k points to (i) to 4, i is now 4