Copyright © 2008-2011 Tim Blechmann
Distributed under the Boost Software License, Version 1.0. (See accompanying file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
Table of Contents
The term non-blocking denotes concurrent data structures, which do not use traditional synchronization primitives like guards to ensure thread-safety. Maurice Herlihy and Nir Shavit (compare "The Art of Multiprocessor Programming") distinguish between 3 types of non-blocking data structures, each having different properties:
Some data structures can only be implemented in a lock-free manner, if they
are used under certain restrictions. The relevant aspects for the implementation
boost.lockfree are the number of producer and consumer
threads. Single-producer (sp)
or multiple producer (mp)
means that only a single thread or multiple concurrent threads are allowed
to add data to a data structure. Single-consumer
(sc) or Multiple-consumer
(mc) denote the equivalent for the removal
of data from the data structure.
Non-blocking data structures do not rely on locks and mutexes to ensure thread-safety. The synchronization is done completely in user-space without any direct interaction with the operating system . This implies that they are not prone to issues like priority inversion (a low-priority thread needs to wait for a high-priority thread).
Instead of relying on guards, non-blocking data structures require atomic operations (specific CPU instructions executed without interruption). This means that any thread either sees the state before or after the operation, but no intermediate state can be observed. Not all hardware supports the same set of atomic instructions. If it is not available in hardware, it can be emulated in software using guards. However this has the obvious drawback of losing the lock-free property.
When discussing the performance of non-blocking data structures, one has to distinguish between amortized and worst-case costs. The definition of 'lock-free' and 'wait-free' only mention the upper bound of an operation. Therefore lock-free data structures are not necessarily the best choice for every use case. In order to maximise the throughput of an application one should consider high-performance concurrent data structures .
Lock-free data structures will be a better choice in order to optimize the latency of a system or to avoid priority inversion, which may be necessary in real-time applications. In general we advise to consider if lock-free data structures are necessary or if concurrent data structures are sufficient. In any case we advice to perform benchmarks with different data structures for a specific workload.
Apart from locks and mutexes (which we are not using in
anyway), there are three other aspects, that could violate lock-freedom:
Some architectures do not provide the necessary atomic operations in natively in hardware. If this is not the case, they are emulated in software using spinlocks, which by itself is blocking.
Allocating memory from the operating system is not lock-free. This makes
it impossible to implement true dynamically-sized non-blocking data structures.
The node-based data structures of
a memory pool to allocate the internal nodes. If this memory pool is
exhausted, memory for new nodes has to be allocated from the operating
system. However all data structures of
can be configured to avoid memory allocations (instead the specific calls
will fail). This is especially useful for real-time systems that require
lock-free memory allocations.
The C++ exception handling does not give any guarantees about its real-time behavior. We therefore do not encourage the use of exceptions and exception handling in lock-free code.
boost.lockfree implements three lock-free data structures:
The data structures can be configured with Boost.Parameter-style templates:
Configures the data structure as fixed sized. The internal nodes are stored inside an array and they are addressed by array indexing. This limits the possible size of the queue to the number of elements that can be addressed by the index type (usually 2**16-2), but on platforms that lack double-width compare-and-exchange instructions, this is the best way to achieve lock-freedom.
Sets the capacity of a data structure at compile-time. This implies that a data structure is fixed-sized.
Defines the allocator.
boost.lockfree supports stateful
allocator and is compatible with Boost.Interprocess
 Spinlocks do not directly interact with the operating system either. However it is possible that the owning thread is preempted by the operating system, which violates the lock-free property.
Last revised: December 08, 2022 at 01:12:23 GMT