O_DIRECT方式读取文件示例

#include <fcntl.h>
#include <stdio.h>
#include <stdlib.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <sys/time.h>
#include <unistd.h>
#include <glog/logging.h>
#include <gflags/gflags.h>

// 读文件类
class CFileReader
{
public:
	CFileReader()
		: _buffer(NULL)
	{
	}

	~CFileReader()
	{
		free(_buffer);
	}

	bool open(const char* filepath)
	{
		// 以O_DIRECT方式打开文件
		int fd = ::open(filepath, O_RDONLY | O_DIRECT);
		if (-1 == fd)
		{
			LOG(ERROR) << "open " << filepath << " error: " << strerror(errno);
			return false;
		}

		// 取得文件大小，以便一次性将文件读取出来
		struct stat st;
		if (-1 == fstat(fd, &st))
		{
			LOG(ERROR) << "stat " << filepath << " error: " << strerror(errno);
			close(fd);
			return false;
		}

		// 分配足以容纳整个文件的Buffer
		// 由于以O_DIRECT方式读取，所以需要按页对齐
		size_t size = st.st_size + (getpagesize() - st.st_size%getpagesize());
                posix_memalign((void**)&_buffer, getpagesize(), size);
		if (NULL == _buffer)
		{
			LOG(ERROR) << "malloc failed";
			close(fd);
			return false;
		}

		// 将整个文件读取_buffer中
		int bytes_read = read(fd, _buffer, size);
		if (-1 == bytes_read)
		{
                        LOG(ERROR) << "read " << filepath << " error: " << strerror(errno);
			close(fd);
			free(_buffer);
			_buffer = NULL;
			return false;
		}
		else if (bytes_read != size)
		{
			// 两组测试输出数据：
			// FileSize(212000000) => AlignSize(212000768) -> RealSize(212000000)
			// FileSize(2120000000) => AlignSize(2120003584) -> RealSize(2120000000)
			printf("FileSize(%d) => AlignSize(%d) -> RealSize(%d)\n", st.st_size, size, bytes_read);
		}

		return true;
	}

	// 从文件中读取一个节点数据
	// offset：偏移量
	// return：返回指向记录的指针
	template <class P>
	const P* get_record(uint64_t offset) const
	{
		return reinterpret_cast<P*>(_buffer + offset);
	}

	// 取得文件所有的记录
	template <class P>
	const P** get_all_record() const
	{
		return reinterpret_cast<P**>(_buffer);
	}

private:
	char* _buffer;
};

// 用于计时
class TimeWatcher
{
public:
	TimeWatcher(const std::string& tip)
		: _tip(tip)
	{
		struct timeval now;
		gettimeofday(&now, NULL);

		_now_msec = (now.tv_sec * 1000) + (now.tv_usec / 1000);
	}

	~TimeWatcher()
	{
		struct timeval now;
		gettimeofday(&now, NULL);

		time_t cur_msec = (now.tv_sec * 1000) + (now.tv_usec / 1000);
		LOG(INFO) << _tip << " spend " << cur_msec - _now_msec << "ms";
	}

private:
	std::string _tip;
	time_t _now_msec;
};

struct User
{
	int32_t age;
	int32_t hight;
	int32_t weight;
	char bitmap[50*4];
};

// 用于生成测试文件
bool make_test_file(const char* filepath, int num_records)
{
	int fd = open(filepath, O_WRONLY | O_CREAT, S_IRUSR | S_IWUSR);
	if (-1 == fd)
	{
		LOG(ERROR) << "open " << filepath << " error: " << strerror(errno);
		return false;
	}

	User user;
	TimeWatcher time_watcher("write");

	for (int i=0; i<num_records; ++i)
	{
		user.age = i;
		user.hight = i + i;
		user.weight = i * i;

		if (-1 == write(fd, &user, sizeof(user)))
		{
			LOG(ERROR) << "write " << filepath << " error: " << strerror(errno);
			close(fd);
			return false;
		}
	}

	close(fd);
	return true;
}

// 模拟随机读取
void random_read(const CFileReader& file_reader, int num_random)
{
	int *index = new int[num_random];

	for (int i=0; i<num_random; ++i)
	{
		srandom(time(NULL));
		index[i] = random() % num_random;
	}

	TimeWatcher time_watcher("randmon read");
	for (int i=0; i<num_random; ++i)
	{
		file_reader.get_record<struct User>(index[i]);
	}
}

// 执行测试
void test()
{
	int num_records1 = 1000000;
	int num_records2 = 10000000;
	std::string file1 = "./file_1000000";
	std::string file2 = "./file_10000000";

	if (make_test_file(file1.c_str(), 1000000)
         && make_test_file(file2.c_str(), 10000000))
	{
		printf("to read, press ENTER to continue ...\n");
		getchar();

		CFileReader file_reader1;

		{
			TimeWatcher time_watcher("open");
			if (!file_reader1.open(file1.c_str()))
			{
				return;
			}
		}

		random_read(file_reader1, 1000000);
		random_read(file_reader1, 3000000);

		CFileReader file_reader2;

		{
			TimeWatcher time_watcher("open");
			if (!file_reader2.open(file2.c_str()))
			{
				return;
			}
		}

		random_read(file_reader2, 1000000);
		random_read(file_reader2, 3000000);
	}
}

int main(int argc, char* argv[])
{
	test();
	return 0;
}