The following simple code C++ example can be used for investigation of how GCC thread sanitizer works:

#include <mutex>
#include <atomic>
#include <iostream>
#include <thread>

std::mutex mutex;
int a = 3;
const size_t size = 1000 * 1000;
std::atomic<int> b(1);

void testA()
{
	for (size_t counter = 0; counter < size; counter++)
	{
		++b;
		std::unique_lock<std::mutex> lock(mutex);
		++a;
	}
}

void testB()
{
	for (size_t counter = 0; counter < size; counter++)
	{
		--b;
		std::unique_lock<std::mutex> lock(mutex);
		--a;
	}
}

int main()
{
	std::thread t1(testA);
	std::thread t2(testB);
	t1.join();
	t2.join();
}

(more…)

The code below demonstrates why it is not guaranteed that 4-byte value being written by another thread is read either as original or final, but it can be read “partially written”:

static constexpr int offset=2;
alignas(64) char vars[64+4-offset];
static volatile unsigned * const p = reinterpret_cast<unsigned *>(&vars[64-offset]);

unsigned getVar()
{
    return *p;
}

void loop()
{
    while(true)
    {
        *p = -1;
        *p = 0;
    }
}

#include <thread>
#include <iostream>
#include <iomanip>
#include <cstdlib>
#include <map>

int main()
{
    std::thread thread(loop);
    std::map<unsigned,int> xs;
    for(int i=0;i<10000000;++i)
    {
        const auto x=getVar();
        ++xs[x];
    }
    for(const auto& x : xs)
        std::cout << std::setfill('0') << std::setw(8) << std::hex << x.first << ": " << std::dec << x.second << " times\n";
    std::exit(0); // exit, killing the thread without abnormal termination via std::terminate
}

(more…)

In the following C++ code the values of ‘z’ and ‘n’ are undefined, because they are the result of an operation with signed integer arithmetic overflow (‘x’ and ‘y’ are first implicitly converted to signed int). The value of ‘w’ is implementation defined, because it is the result of a conversion:

#include <iostream>
#include <bitset>

int main(int argc, char *argv[])
{
    unsigned short x = 65535, y = x;
    unsigned short z = x * y;
    unsigned int n = x * y;
    std::cerr << "z = " << std::bitset<16>(z) << ", n = " << std::bitset<32>(n) << ", sizeof(int) = " << sizeof(int) << std::endl;

    short w = 0x80000000;
    
    return 0;
}

see Numeric conversions section of Implicit conversions article.

The following C++ code compares the performance of std::atomic and std::mutex:

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

const size_t size = 100000000;
std::mutex mutex;
bool var = false;

typedef std::chrono::high_resolution_clock Clock;

void testA()
{
    std::atomic<bool> sync(true);
    const auto start_time = Clock::now();
    for (size_t counter = 0; counter < size; counter++)
    {
        var = sync.load();
        //sync.store(true);
        //sync.exchange(true);
    }
    const auto end_time = Clock::now();
    std::cout << 1e-6*std::chrono::duration_cast<std::chrono::microseconds>(end_time - start_time).count() << " s\n";
}

(more…)