Sync

This header-only library is a small wrapper around std::mutex and std::shared_timed_mutex in order to better express which data they protect. The design of the library is heavily influenced by the Rust std::sync in particular std::sync::Mutex and std::sync::RwLock.

Requirements

• C++14
• CMake 3.14 (only when using CPM for installation)

Installation

Either,

• copy the content of the include folder somewhere and add it to your include directories

• using CPM

CPMAddPackage("gh:soehrl/[email protected]")
target_link_libraries(your-target PRIVATE sync::sync)

sync::mutex<T>

In contrast to std::mutex, sync::mutex<T> takes an additional type parameter T that represents the data that is protected by the mutex. When constructing the mutex, you have to supply an initial value for T. Accessing the value of T can be done by calling mutex::lock() or mutex::try_lock() which return a RAII-style lock guard. The lock guard then provides the * and -> operator to access the value protected by the mutex. This makes it very clear which data is protected by which mutex and makes it hard to accidentally access the data without locking the corresponding mutex. Especially, it prevents the issue where you create a std::unique_lock but forget to give it a name, so the lock is immediately released.

std::mutex

sync::mutex<T>

#include <chrono> #include <thread> #include <mutex> int g_num = 0; // protected by g_num_mutex std::mutex g_num_mutex; using namespace std::literals::chrono_literals; void increment(int id) { for (int i = 0; i < 1000; ++i) { std::unique_lock<std::mutex> lock(g_num_mutex); ++g_num; std::this_thread::sleep_for(10ms); } } int main() { std::thread t1{increment, 0}; std::thread t2{increment, 1}; t1.join(); t2.join(); }

#include <chrono> #include <thread> #include "sync/mutex.hpp" sync::mutex<int> g_num(0); using namespace std::literals::chrono_literals; void increment(int id) { for (int i = 0; i < 1000; ++i) { *g_num.lock() += 1; std::this_thread::sleep_for(10ms); } } int main() { std::thread t1{increment, 0}; std::thread t2{increment, 1}; t1.join(); t2.join(); }

sync::read_write_lock

A small wrapper around an std::shared_timed_lock. It works similarly to sync::mutex but it allows to lock the mutex either for reading (std::shared_lock) via read_write_lock::read() or for writing (std::unique_lock) via read_write_lock::write(). The lock guard return for reading gives access to a const T& while the lock guard returned for writing gives access to a non-const T&. At any time there can be either multiple readers or a single writer for one read_write_lock. This assumes, that it is safe to access const methods from multiple threads at a time. According to the citation of the C++ standard from the accepted answer of this stackoverflow post this is true for the C++ standard library, however, you should be careful when using it with third-party libraries or your own types.

Usage:

sync::read_write_lock<int> rwlock(100);

SECTION("Simultaneous reads") {
    auto value = rwlock.read();
    REQUIRE(*value == 100);

    auto value2 = rwlock.try_read();
    REQUIRE(value2.owns_lock());
    REQUIRE(*value2 == 100);
}

SECTION("Try write while read") {
    auto value = rwlock.read();
    REQUIRE(*value == 100);

    auto value2 = rwlock.try_write();
    REQUIRE(!value2.owns_lock());
}

SECTION("Try read while write") {
    auto value = rwlock.write();
    REQUIRE(value.owns_lock());
    REQUIRE(*value == 100);

    auto value2 = rwlock.try_read();
    REQUIRE(!value2.owns_lock());
}