Bringing practical lock-free synchronization to 64-bit applications

Abstract

Many lock-free data structures in the literature exploit techniques that are possible only because state-of-the-art 64-bit processors are still running 32-bit operating systems and applications. As software catches up to hardware, "64-bitclean" lock-free data structures, which cannot use such techniques, are needed.We present several 64-bit-clean lock-free implementations: load-linked/store-conditional variables of arbitrary size, a FIFO queue, and a freelist. In addition to being portable to 64-bit software, our implementations also improve on previous ones in that they are space-adaptive and do not require knowledge of the number of threads that will access them.

Practical, Fast and Simple Concurrent FIFO Queues Using Single Word Synchronization Primitives

“…Current and emerging 64-bit architectures do not provide atomic access to more than 64-bit quantities, thus it is no longer possible to pack a large reference counter along with pointer-wide values in 64-bit applications. When application software exploiting these 64-bit capabilities becomes widespread [2], their algorithm will be of limited use. Tsigas and Zhang [15] proposed the first practical non-blocking FIFO queue based on a circular array using single word synchronization primitives found on all modern architectures and suitable for 64-bit applications.…”

“…Doherty et al [2] present the first link-based FIFO queue that is population-oblivious and has a space consumption depending only on the number of queued items and the number of threads currently accessing the queue. However, their algorithm introduces significant overheads and trades memory space for computational time.…”

“…To succeed, the following 4 statements must be true: (1) Tail < Head + Q_LENGTH, i.e., the queue is not full, (2) Tail does not change value since it was last read on line E5, (3) Q[Tail mod Q_LENGTH] = null, and (4) Q[Tail mod Q_LENGTH] has not changed since it was last read on line E9. Provided the queue is not full, statements (2) and (4) Therefore ENQ(y) can occur if and only if the previous enqueue operation is completely done.…”

Efficient Wait-Free Implementation of Multiword LL/SC Variables

“…In the following we explain (i) what it means for SC to succeed or fail, and (ii) the rule for determining the SC's success or failure. [1,2,6,11,14,16,17]. Most of this research is focused on implementing small LL/SC objects, i.e., LL/SC objects whose value fits in a single machine word (which is 64-bits in the case of most machines) [2,6,11,14,16,17].…”

“…[1,2,6,11,14,16,17]. Most of this research is focused on implementing small LL/SC objects, i.e., LL/SC objects whose value fits in a single machine word (which is 64-bits in the case of most machines) [2,6,11,14,16,17]. However, many existing applications [1,4,12,13] need large LL/SC objects, i.e., LL/SC objects whose value does not fit in a single machine word.…”