Question

I was asked these questions in a interview.

1. Why exactly Binary code and Data was separated i.e why did they plan for Data segment why not everything inside code segment?

2.

class A
{
private :
    int i;
public:
    void show()
    {
        printf("hello");
    }
};
int main()
{
    A* a = NULL; (what happens in object table?)
        A* aa =  new A();  (what happens in object table?)

        a->show();
    aa->show();

    delete aa;
    return 0;
}

How exactly aa and a are different and how exactly object behaves inside memory.

Answer 1

code segment vs data segment

code segment is read-only, while data segment is read/write, if you mix these 2 sections together, update the data while keep the code secure becomes a challenge: one example is as Lol4t0 pointed out:

As to the OS memory management, the code segment will be swap out to
its original executable file on the file system, while the data
segment, which is deemed to change, are swapped out to paging file. if you mixed them
together, you may lose the advantage of reusing the executable as paging file.

and also, code segment are usually load into memory pages which are readonly (VirtualAlloc(PAGE_READONLY))

a(null) vs aa (non-null)

a and aa themselves are simply stack variable of type A*, but a points to NULL, while aa points to an object allocated in heap.

a->show() is translated into:

A_show(a)
//which is:
A_show(NULL)

because no member variable is referenced in show(), this should works fine.

aa->show() is translated into:

A_show(aa)

here aa is a valid address, so even if you reference member variable in show(), it would work.

Notice unlike virtual functions, which is resolved in runtime thus require a vptr in each object, member functions are just normal functions which take the this as first parameter, and resolved by compiler in compile time.