cpointerscomplement

What is the purpose of this C code?

I was looking over the source code for the "tfind" function from the "search.h" C library and I stumbled upon this line:

#define DEREFNODEPTR(NP) (node)((uintptr_t)(*(NP)) & ~((uintptr_t) 0x1))

This is how it's used:

    /* Find datum in search tree.
   KEY is the key to be located, ROOTP is the address of tree root,
   COMPAR the ordering function.  */

void *
__tfind (const void *key, void *const *vrootp, __compar_fn_t compar)
{
  node root;
  node *rootp = (node *) vrootp;
  if (rootp == NULL)
    return NULL;
  root = DEREFNODEPTR(rootp);
  CHECK_TREE (root);
  while (DEREFNODEPTR(rootp) != NULL)
    {
      root = DEREFNODEPTR(rootp);
      int r;
      r = (*compar) (key, root->key);
      if (r == 0)
        return root;
      rootp = r < 0 ? LEFTPTR(root) : RIGHTPTR(root);
    }
  return NULL;
}

So, why is one's complement needed in this case?

Thank you for your time and consideration.

Solution

  • It is likely the case that the implementation assumes the node pointers are to nodes that are always at least 2 byte aligned. This means that the least significant bit will always be 0 for a valid node pointer. This allows the bit to be "used" to store state (such as whether or not it has been visited before, or whether it is a red or black node, or some other thing...).

    The macro clears the LSB before accessing the value as a pointer. There is no one's complement requirement.