#mutex #condvar #rwlock #once #thread

usync

fast, drop-in, synchronization primitives

4 releases (2 breaking)

Uses new Rust 2021

0.2.0 May 8, 2022
0.1.1 Mar 29, 2022
0.1.0 Mar 29, 2022
0.0.0 Oct 8, 2020

#78 in Concurrency

Download history 1/week @ 2022-02-20 1/week @ 2022-03-06 3/week @ 2022-03-13 3/week @ 2022-03-20 95/week @ 2022-03-27 14/week @ 2022-04-03 13/week @ 2022-04-24 19/week @ 2022-05-01 48/week @ 2022-05-08

80 downloads per month

MIT license

155KB
2.5K SLoC

usync

Crates.io Documentation

This library provides implementations of Mutex, RwLock, Condvar, Barrier and Once that are word-sized and generally fast as those in parking_lot. It also provides a ReentrantMutex type which supports recursive locking.

Features

The primitives provided by this library have several advantages over those in the Rust standard library:

  1. All types require only require 1 word of storage space. On the other hand the standard library primitives require a dynamically allocated Box to hold OS-specific synchronization primitives. The small size of Mutex in particular encourages the use of fine-grained locks to increase parallelism.
  2. Since they consist of just a single atomic variable, have constant initializers and don't need destructors, these primitives can be used as static global variables. The standard library primitives require dynamic initialization and thus need to be lazily initialized with lazy_static!.
  3. Uncontended lock acquisition and release is done through fast inline paths which only require a single atomic operation.
  4. Microcontention (a contended lock with a short critical section) is efficiently handled by spinning a few times while trying to acquire a lock.
  5. The locks are adaptive and will suspend a thread after a few failed spin attempts. This makes the locks suitable for both long and short critical sections.
  6. Condvar::notify_all will generally only wake up a single thread and requeue the rest to wait on the associated Mutex. This avoids a thundering herd problem where all threads try to acquire the lock at the same time.
  7. Mutex and RwLock allow raw unlocking without a RAII guard object.
  8. Mutex<()> and RwLock<()> allow raw locking without a RAII guard object.
  9. A ReentrantMutex type which supports recursive locking.
  10. Lock guards can be sent to other threads when the send_guard feature is enabled.

Userspace queues

To keep these primitives word sized, their state is multiplexed between counters, queues of threads, and combinations of both. This draws similarities to Windows' Slim Synchronization Primitives. No external locking of global queues as seen in Linux futex or parking_lot is employed. The queues are all embedded in each primitive and interacted with lock-free operations to decrease worst-case contention latency.

Having to juggle around queues with the synchronization state unfortunately means that "no spurious wakeups" cannot be guaranteed for Condvar and that extreme read-only workflows for RwLock can't use optimized atomics to improve throughput. These perf limits shouldn't matter in practice though, even more so when other cache effects come into play. On the bright side, writer/exclusive heavy workloads scale much better than existing solutions and are heavily optimized for micro-contention.

Nightly vs stable

There are a few restrictions when using this library on stable Rust:

  • You will have to use the const_* functions (e.g. const_mutex(val)) to statically initialize the locking primitives. Using e.g. Mutex::new(val) does not work on stable Rust yet.

To enable nightly-only functionality, you need to enable the nightly feature in Cargo (see below).

Usage

Add this to your Cargo.toml:

[dependencies]
usync = "0.1.0"

To enable nightly-only features, add this to your Cargo.toml instead:

[dependencies]
usync = { version = "0.1.0", features = ["nightly"] }

To allow sending MutexGuards and RwLock*Guards to other threads, enable the send_guard option.

License

Licensed under MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT).

Dependencies

~160KB