c++ multithreading学习

2016-09-01

C++

g++ compile

g++ -std=c++11 launch-thread.cpp -o launch-thread -lpthread

basic thread management

lanch a thread

we can lanch a thread by passing a parameter which can be function, lamda expression and function call operator

#include<iostream>
#include<thread>

void helloWorld(){
	std::cout<<"hello world from function"<<std::endl;
}

class hello{
	public:
		void operator()(){
			std::cout<<"hello world from function call operator"<<std::endl;
		}
};

int main(){
	std::thread testFunction(helloWorld);

	std::thread testLamda([]{
		std::cout<<"hello world from lamda"<<std::endl;
	});

	hello x;
	std::thread testCall(x);

	testFunction.join();
	testLamda.join();
	testCall.join();

	return 0;
}

wait all let it run

using join(), local thread will wait until thread to finished.

using detach(), thread will continue run even if local thread is destroyed

Waiting in exceptional circumstances

thread_guard will be destroyed before thread, so join() will be called in destructor

#include<iostream>
#include<thread>

class thread_guard{
	std::thread &t;
	public:
		thread_guard(std::thread &t_):t(t_){
		}
		~thread_guard(){
			if(t.joinable()){
				t.join();
			}
		}
};

int main(){
	std::thread t([](){
		std::cout<<"hello"<<std::endl;
	});
	thread_guard g(t);

	return 0;
}

passing argument

argument is passed by pass a copy into internal storage. If pass a reference, we should use std::ref.

#include<iostream>
#include<thread>
void func(int &x){
	x++;
	x++;
}
int main(){
	int x=10;
	std::thread t(func,std::ref(x));
	t.join();
	std::cout<<x<<std::endl;
	return 0;
}

thread management

hardware_concurrency

This function returns an indication of the number of threads that can truly run concurrently for a given execution of a program. On a multicore system it might be the number of CPU cores, for example. demo show as follow:

#include <iostream>
#include <thread>
int main(){
	std::cout<<std::thread::hardware_concurrency()<<std::endl;
	return 0;
}

thread id

the identifier for a thread can be obtained from its associated std::thread object by calling the get_id() member function;the identifier for the current thread can be obtained by calling std::this_thread::get_id(). demo show as follow:

#include <iostream>
#include <thread>
int main(){
	std::cout<<std::this_thread::get_id()<<std::endl;
	std::thread a([]{
		std::cout<<"hello world"<<std::endl;
	});
	std::cout<<a.get_id()<<std::endl;
	a.join();
	return 0;
}

race condition

几个线程的操作同时修改某个对象产生的问题。

解决这个问题有三种方法：当某个线程在修改时，不允许其他线程介入；lock-free programming（复杂）;transaction(c++不支持)