std::atomic...nd_explicit(3)

std::atomic...nd_explicit(3) 💅

okay bestie, this one is from the C++ Standard Libary manual (that's the header's spelling, not me 😭), and it's all about std::atomic_fetch_and and std::atomic_fetch_and_explicit. No cap, it's giving atomic bitwise-AND realness.

---

💖 NAME 💖

std::atomic_fetch_and,std::atomic_fetch_and_explicit — yes, literally that. The page name is std::atomic_fetch_and,std::atomic_fetch_and_explicit, and we're gonna get into all of it.

---

✨ Synopsis ✨

Defined in header <atomic>:

Defined in header <atomic>

template< class T >
T atomic_fetch_and( std::atomic<T>* obj,
                    typename std::atomic<T>::value_type arg ) noexcept;
(1) (since C++11)

template< class T >
T atomic_fetch_and( volatile std::atomic<T>* obj,
                    typename std::atomic<T>::value_type arg ) noexcept;
(2) (since C++11)

template< class T >
T atomic_fetch_and_explicit( std::atomic<T>* obj,
                             typename std::atomic<T>::value_type arg,
                             std::memory_order order ) noexcept;
(3) (since C++11)

template< class T >
T atomic_fetch_and_explicit( volatile std::atomic<T>* obj,
                             typename std::atomic<T>::value_type arg,
                             std::memory_order order ) noexcept;
(4) (since C++11)

So what does it actually do? It atomically replaces the value pointed by obj with the result of bitwise AND between the old value of obj and arg. Then it returns the value obj held previously. Obsessed with a clean little fetch-and-return, honestly.

The operation is performed as if the following is executed:

1,2) obj->fetch_and(arg)
3,4) obj->fetch_and(arg, order)

🚨 Important caveat, bestie: if std::atomic<T> has no fetch_and member (this member is only provided for integral types except bool), the program is ill-formed. Not cute, not slay, straight-up ill-formed.

---

🌸 Parameters 🌸

• obj — pointer to the atomic object to modify
• arg — the value to bitwise AND to the value stored in the atomic object
• order — the memory synchronization ordering

---

💖 Return value 💖

The value immediately preceding the effects of this function in the modification order of *obj.

It's literally handing you the old value like "here bestie, this was yours before I changed it."

---

🎀 Example 🎀

Run this code:

// Run this code

#include <atomic>
#include <chrono>
#include <functional>
#include <iostream>
#include <thread>

// Binary semaphore for demonstrative purposes only.
// This is a simple yet meaningful example: atomic operations
// are unnecessary without threads.
class Semaphore
{
    std::atomic_char m_signaled;
public:
    Semaphore(bool initial = false)
    {
        m_signaled = initial;
    }
    // Block until semaphore is signaled
    void take()
    {
        while (!std::atomic_fetch_and(&m_signaled, false))
        {
            std::this_thread::sleep_for(std::chrono::milliseconds(10));
        }
    }

    void put()
    {
        std::atomic_fetch_or(&m_signaled, true);
    }
};

class ThreadedCounter
{
    static const int N = 100;
    static const int REPORT_INTERVAL = 10;
    int m_count;
    bool m_done;
    Semaphore m_count_sem;
    Semaphore m_print_sem;

    void count_up()
    {
        for (m_count = 1; m_count <= N; ++m_count)
            if (m_count % REPORT_INTERVAL == 0)
            {
                if (m_count == N)
                    m_done = true;
                m_print_sem.put(); // signal printing to occur
                m_count_sem.take(); // wait until printing is complete proceeding
            }
        std::cout << "count_up() done\n";
        m_done = true;
        m_print_sem.put();
    }

    void print_count()
    {
        do
        {
            m_print_sem.take();
            std::cout << m_count << '\n';
            m_count_sem.put();
        }
        while (!m_done);
        std::cout << "print_count() done\n";
    }

public:
    ThreadedCounter() : m_done(false) {}
    void run()
    {
        auto print_thread = std::thread(&ThreadedCounter::print_count, this);
        auto count_thread = std::thread(&ThreadedCounter::count_up, this);
        print_thread.join();
        count_thread.join();
    }
};

int main()
{
    ThreadedCounter m_counter;
    m_counter.run();
}

Output 🍓

10
20
30
40
50
60
70
80
90
100
print_count() done
count_up() done

---

🔥 Defect reports 🔥

The following behavior-changing defect reports were applied retroactively to previously published C++ standards.

• P0558R1 — applied to C++11. Behavior as published: exact type match was required because T was deduced from multiple arguments. Correct behavior: T is only deduced from obj.

Retroactive glow-up for the standard, we love to see it.

---

🌈 See also 🌈

• fetch_and — atomically performs bitwise AND between the argument and the value of the atomic object and obtains the value held previously (public member function of std::atomic<T>)
• atomic_fetch_or, atomic_fetch_or_explicit (C++11) — replaces the atomic object with the result of bitwise OR with a non-atomic argument and obtains the previous value of the atomic (function template)
• atomic_fetch_xor, atomic_fetch_xor_explicit (C++11) — replaces the atomic object with the result of bitwise XOR with a non-atomic argument and obtains the previous value of the atomic (function template)
• C documentation for atomic_fetch_and, atomic_fetch_and_explicit

---

xoxo, bestie 💖 — source/version: http://cppreference.com, date: 2024.06.10, page: std::atomic...nd_explicit(3). periodt ✨