Efficient and Reliable Lock-Free Memory Reclamation Based on Reference Counting

Abstract

We present an efficient and practical lock-free method for semiautomatic (application-guided) memory reclamation based on reference counting, aimed for use with arbitrary lock-free dynamic data structures. The method guarantees the safety of local as well as global references, supports arbitrary memory reuse, uses atomic primitives that are available in modern computer systems, and provides an upper bound on the amount of memory waiting to be reclaimed. To the best of our knowledge, this is the first lock-free method that provides all of these properties. We provide analytical and experimental study of the method. The experiments conducted have shown that the method can also provide significant performance improvements for lock-free algorithms of dynamic data structures that require strong memory management.

Cache-Aware Lock-Free Queues for Multiple Producers/Consumers and Weak Memory Consistency

“…All dynamic queues (including the new algorithm) have been implemented to support queue sizes only limited by the system's memory, i.e., using lock-free management schemes [7] or [2] and lock-free free-lists where appropriate. For the new implementation, the size of the array block (BLOCK SIZE) is chosen to fit within one cache line.…”

“…For our implementation of the new lock-free queue algorithm, we have selected the lock-free memory management scheme proposed by Gidenstam et al [2] which makes use of the CAS and FAA atomic synchronization primitives. The interface defined by the memory management scheme is listed in Program 1 and are fully described in [2].…”

“…The interface defined by the memory management scheme is listed in Program 1 and are fully described in [2]. Using this scheme we can assure that an array block can only be reclaimed when there is no next pointer in the linked list pointing to it and that there are no local references to it from pending concurrent operations or from pointers in thread-local storage.…”

“…Note also that the additional fields head and tail in the array block are only used for indicating either fullness or emptiness of the whole array, and not any intermediate status. In order to simplify the description of our new algorithm, we have omitted some of the details of applying the operations of the memory management [2]. In actual implementations, ReleaseRef calls should be inserted at appropriate places whenever a variable holding a safe pointer goes out of scope or is reassigned.…”

Lock-free deques and doubly linked lists

“…The underlying data structure is fully consistent when it is idle, and the algorithm supports bidirectional traversals also through deleted nodes. The improved algorithm is compatible 5 with the lock-free memory reclamation scheme by Gidenstam et al [20,21], and thus the data structure is fully dynamic in size and there is an upper bound of the amount of memory occupied at any time. The algorithm, which also allows disjoint accesses on the data structure to perform in parallel, is described in detail later in this paper together with the aspects concerning the underlying lockfree memory management.…”