Oracle scans the smaller table(u need to verify this that it does that) , creates a
hash function based on join condition and places this table in hash_area_size
then the second is joined to the first retrieving all matching records using the hash function(table).
when this table does not fit into the hash_area_size , then the hash table is divided into partitions.
the larger table is also partitioned but is not one to one but based on the match to the join hash function
the rows which are matched are returned , rest is kept for next hash table partition.