boost::circular_buffer 如何处理覆盖移位

Question

我有 2 个进程：一个生产者和一个“消费者”，它们仍然将值留在缓冲区内，它们将被覆盖。

但是让消费者跟踪是个问题。当缓冲区已满并且值被覆盖时，指向索引 0 的值是刚刚覆盖的值（即下一个最旧的值）之前的值，而刚刚插入的值是最后一个索引，移动所有值介于两者之间。

cb.push_back(0)
cb.push_back(1)
cb.push_back(2)

consumer reads to cb[1], cb[2] should == 2 when next read

cb.push_back(3)

cb[2] now == 1 effectively reading the old value

有趣的是，即使在缓冲区开始被覆盖时，循环缓冲区上的迭代器也会保持相同的值，这可以正常工作，除非在读取时确实到达 end() 迭代器，即使在插入之后它也将始终等于 end() 迭代器更多的值，所以你必须在完成消费后std::prev(iter, 1)，然后当你在插入更多值后再次读取时，请执行std::next(iter, 1)，这样你就不会读取你已经读取的值。

Answer 1

我相信 circular_buffer 的存在正是为了从您那里抽象出迭代器的定位。

缓冲区 循环的事实对您来说并不重要：它只是一个队列接口。

如何使用circular_buffer想要在这个例子中可以很清楚的看到：http://www.boost.org/doc/libs/1_60_0/libs/circular_buffer/example/circular_buffer_sum_example.cpp

如果你想要那种程度的控制，你要么

• 想要使用更简单的容器原语并构建自己的逻辑

• 您可以将有界缓冲区写在循环缓冲区之上。一个完整的例子在这里：http://www.boost.org/doc/libs/1_60_0/libs/circular_buffer/test/bounded_buffer_comparison.cpp

  explanation 提到：

  有界缓冲区通常用于生产者-消费者模式 [...]

  [...]

  有界缓冲区::pop_back() 方法不会删除项目，但项目会留在循环缓冲区中，然后在循环缓冲区被替换时将其替换为新的（由生产者插入）满的。这种技术比通过调用 circular_buffer 的 circular_buffer::pop_back() 方法显式删除项目更有效。

听起来应该对你有很大帮助。

更新

这是一个适用于共享内存的演示：

#define BOOST_CB_DISABLE_DEBUG

#include <boost/circular_buffer.hpp>
#include <boost/thread/thread.hpp>
#include <boost/call_traits.hpp>
#include <boost/bind.hpp>
#include <boost/interprocess/allocators/allocator.hpp>
#include <boost/interprocess/managed_shared_memory.hpp>
#include <boost/interprocess/sync/interprocess_condition.hpp>
#include <boost/interprocess/sync/interprocess_mutex.hpp>
#include <iostream>

const unsigned long QUEUE_SIZE     = 1000L;
const unsigned long TOTAL_ELEMENTS = QUEUE_SIZE * 1000L;

namespace bip = boost::interprocess;

template <class T, class Alloc, typename CV = boost::condition_variable, typename Mutex = boost::mutex>
class bounded_buffer {
public:
    typedef boost::circular_buffer<T, Alloc> container_type;
    typedef typename container_type::size_type                  size_type;
    typedef typename container_type::value_type                 value_type;
    typedef typename container_type::allocator_type             allocator_type;
    typedef typename boost::call_traits<value_type>::param_type param_type;

    bounded_buffer(size_type capacity, Alloc alloc = Alloc()) : m_unread(0), m_container(capacity, alloc) {}

    void push_front(param_type item) {
        boost::unique_lock<Mutex> lock(m_mutex);

        m_not_full.wait(lock, boost::bind(&bounded_buffer::is_not_full, this));
        m_container.push_front(item);
        ++m_unread;
        lock.unlock();

        m_not_empty.notify_one();
    }

    void pop_back(value_type* pItem) {
        boost::unique_lock<Mutex> lock(m_mutex);

        m_not_empty.wait(lock, boost::bind(&bounded_buffer::is_not_empty, this));
        *pItem = m_container[--m_unread];
        lock.unlock();

        m_not_full.notify_one();
    }

private:
    bounded_buffer(const bounded_buffer&);              // Disabled copy constructor
    bounded_buffer& operator = (const bounded_buffer&); // Disabled assign operator

    bool is_not_empty() const { return m_unread > 0; }
    bool is_not_full() const { return m_unread < m_container.capacity(); }

    size_type m_unread;
    container_type m_container;
    Mutex m_mutex;
    CV m_not_empty;
    CV m_not_full;
};

namespace Shared {
    using segment = bip::managed_shared_memory;
    using smgr    = segment::segment_manager;
    template <typename T> using alloc = bip::allocator<T, smgr>;
    template <typename T> using bounded_buffer = ::bounded_buffer<T, alloc<T>, bip::interprocess_condition, bip::interprocess_mutex >;
}

template<class Buffer>
class Consumer {

    typedef typename Buffer::value_type value_type;
    Buffer* m_container;
    value_type m_item;

public:
    Consumer(Buffer* buffer) : m_container(buffer) {}

    void operator()() {
        for (unsigned long i = 0L; i < TOTAL_ELEMENTS; ++i) {
            m_container->pop_back(&m_item);
        }
    }
};

template<class Buffer>
class Producer {

    typedef typename Buffer::value_type value_type;
    Buffer* m_container;

public:
    Producer(Buffer* buffer) : m_container(buffer) {}

    void operator()() {
        for (unsigned long i = 0L; i < TOTAL_ELEMENTS; ++i) {
            m_container->push_front(value_type());
        }
    }
};

int main(int argc, char**) {
    using Buffer = Shared::bounded_buffer<int>;

    if (argc>1) {
        std::cout << "Creating shared buffer\n";
        Shared::segment mem(bip::create_only, "test_bounded_buffer", 10<<20); // 10 MiB
        Buffer* buffer = mem.find_or_construct<Buffer>("shared_buffer")(QUEUE_SIZE, mem.get_segment_manager());

        assert(buffer);

        // Initialize the buffer with some values before launching producer and consumer threads.
        for (unsigned long i = QUEUE_SIZE / 2L; i > 0; --i) {
            buffer->push_front(BOOST_DEDUCED_TYPENAME Buffer::value_type());
        }

        std::cout << "running producer\n";
        Producer<Buffer> producer(buffer);
        boost::thread(producer).join();
    } else {
        std::cout << "Opening shared buffer\n";

        Shared::segment mem(bip::open_only, "test_bounded_buffer");
        Buffer* buffer = mem.find_or_construct<Buffer>("shared_buffer")(QUEUE_SIZE, mem.get_segment_manager());

        assert(buffer);

        std::cout << "running consumer\n";
        Consumer<Buffer> consumer(buffer);
        boost::thread(consumer).join();
    }
}

当你运行两个进程时：

time (./test producer & sleep .1; ./test; wait)
Creating shared buffer
running producer
Opening shared buffer
running consumer

real    0m0.594s
user    0m0.372s
sys 0m0.600s