In print_pointer(), the 'ptr' variable is on the stack, so the address
of it can change with every call, as it can with any stack variable.
The same thing can happen in C.

e.g. if you combined your C code with the following test harness, it
would also print out different values for &ptr.

void indirect_print(void* ptr) { print_pointer(ptr); }

int main(int argc, char **argv) {
  void* ptr = (void*) (intptr_t) 0x1234;
  print_pointer(ptr);
  indirect_print(ptr);
}


The value of the pointer, which is the thing you are interested in,
does not change.  Neither does it with FFI (as your output showed).


2009/12/10 Simon Chiang <simon.a.chiang@gmail.com>:

> I've noticed the address of pointers can change when passed back into
> an attached function multiple times.  How can I get the address to
> stay the same?
>
> For example, say I have these functions:
>
>  #include <stdio.h>
>
>  void *get_pointer()
>  {
>    void *ptr;
>    return(ptr);
>  }
>
>  int print_pointer( void *ptr )
>  {
>    printf("pointer: %p\naddress: %p\n", ptr, &ptr);
>    return(0);
>  }
>
> And then I do this:
>
>  module ExampleCode
>    attach_function :get_pointer, [], :pointer
>    attach_function :print_pointer, [:pointer], :int
>  end
>
>  ptr = ExampleCode.get_pointer()
>
>  puts
>  puts "First Time:"
>  ExampleCode.print_pointer(ptr)
>
>  puts
>  puts "Second Time:"
>  ExampleCode.print_pointer(ptr)
>
> I get this for an output:
>
>  First Time:
>  pointer: 0x6249
>  address: 0x7fff5fbfef38
>
>  Second Time:
>  pointer: 0x6249
>  address: 0x7fff5fbfeef8
>
> I'm on OS X 10.6.2, using FFI 0.5.4.  Thanks!
>