Mmm... so when looking for a non-existent key you have to go through the whole table. Hardly O(1)!
A better strategy would be to first search the given range for the first empty slot, and exclude everything after that point. Then proceed looking for tophash in the thus-limited interval. The return value has 3 possible values: 1) key found, 2) reached end of interval without finding but it might be in the other half, 3) found an empty slot so the key is not present, not even in the second half.
A better strategy would be to first search the given range for the first empty slot, and exclude everything after that point. Then proceed looking for tophash in the thus-limited interval. The return value has 3 possible values: 1) key found, 2) reached end of interval without finding but it might be in the other half, 3) found an empty slot so the key is not present, not even in the second half.