LFTHREADS

Abstract

This extended abstract presents LFTHREADS, a thread library entirely based on lock-free methods, i.e. no spinlocks or similar synchronization mechanisms are employed in the implementation of the multithreading. Since lockfreedom is highly desirable in multiprocessors/multicores due to its advantages in parallelism, fault-tolerance, convoy-avoidance and more, there is an increased demand in lock-free methods in parallel applications, hence also in multiprocessor/multicore system services. LFTHREADS is the first thread library that provides a lock-free implementation of blocking synchronization primitives for application threads; although the latter may sound like a contradicting goal, such objects have several benefits: e.g. library operations that block and unblock threads on the same synchronization object can make progress in parallel while maintaining the desired thread-level semantics and without having to wait for any "low" operations among them. Besides, as no spin-locks or similar synchronization mechanisms are employed, memory contention can be reduced and processors/cores are able to do useful work. As a consequence, applications, too, can enjoy enhanced parallelism and fault-tolerance. For the synchronization in LFTHREADS we have introduced a new method, which we call responsibility hand-off (RHO), that does not need any special kernel support. The RHO method is also of independent interest, as it can also serve as a tool for lock-free token passing, management of contention and interaction between scheduling and synchronization. This paper gives an outline and the context of LFTHREADS. For more details the reader is refered to [7] and [8].