Linearizability is a well-established consistency and correctness criterion for concurrent data types. An important feature of linearizability is Herlihy and Wing's locality principle, which says that a concurrent system is linearizable if and only if all of its constituent parts (so-called objects) are linearizable. This paper presents $P$-compositionality, which generalizes the idea behind the locality principle to operations on the same concurrent data type. We implement $P$-compositionality in a novel linearizability checker. Our experiments with over nine implementations of concurrent sets, including Intel's TBB library, show that our linearizability checker is one order of magnitude faster and/or more space efficient than the state-of-the-art algorithm.