There is need to parallelize iteration of CHT iteration, i.e. multiple threads being allowed to work on a single (large) CHT during safepoint with no concurrent modifications of any kind.

In GC we often parallelize HT iteration on a set of buckets basis in such situations.