利用命名管道创建进程池——技术实践指南

第1步：创建管道

首先，我们需要准备一些命名管道来构建进程池。为此，我编写了一个函数CreateChannels，用于创建一批命名管道。这是一个典型的案例：

#pragma once#include 
   
    #include 
    
     #include 
     
      #include 
      
       #include 
       
        #include 
        
         #include 
         
          #include 
          
           #include 
           
            #include 
            
             #include 
             
              const int MY_CHANNEL_NUMBER = 10;const std::string base_name = "../my_fifo";void CreateChannels(std::vector
              
               & channels) { for (int i = 0; i < MY_CHANNEL_NUMBER; ++i) { std::stringstream ss; ss << base_name << "_" << i; std::string name = ss.str(); if (mkfifo(name.c_str(), 0666) == -1) { if (errno != EEXIST) { perror("create channels fail"); exit(EXIT_FAILURE); } else { std::cout << "channel has existed" << std::endl; channels.push_back(Channel(name)); } } else { channels.push_back(Channel(name)); } }}

进一步优化：检查错误状态

为了确保管道不存在重复，我们可以检查errno状态：

#pragma once#include 
   
    #include 
    
     #include 
     
      #include 
      
       #include 
       
        #include 
        
         #include 
         
          #include 
          
           #include 
           
            #include 
            
             #include 
             
              const int MY_CHANNEL_NUMBER = 10;const std::string base_name = "../my_fifo";void CreateChannels(std::vector
              
               & channels) { for (int i = 0; i < MY_CHANNEL_NUMBER; ++i) { std::stringstream ss; ss << base_name << "_" << i; std::string name = ss.str(); int fd = mkfifo(name.c_str(), 0666); if (fd == -1) { if (errno != EEXIST) { perror("create channels fail"); exit(EXIT_FAILURE); } else { std::cout << "channel has existed" << std::endl; channels.push_back(Channel(name)); } } else { channels.push_back(Channel(name)); } }}

第2步：批量删除管道

类似地，我们可以编写一个函数DeletePipes，以删除已经创建好的管道：

void DeletePipes(std::vector
   
    & channels) {    for (auto& e : channels) {        if (unlink(e.fifo_path.c_str()) == -1) {            if (errno == ENOENT) {                std::cout << e.fifo_path << " not exists" << std::endl;            } else {                perror("unlink fail");                exit(EXIT_FAILURE);            }        }    }}

封装管道管理

为了更好地管理和编写管道代码，我建议将管道封装到一个类中。这样可以提高代码的可维护性和可扩展性：

struct Channel {    Channel(const std::string& fifo_name) : fifo_path(fifo_name) {}    std::string fifo_path;};void CreateChannels(std::vector
   
    & channels) {    int retry = 0;    while (retry < 10 && channels.size() < MY_CHANNEL_NUMBER) {        std::vector
    
      names;        for (int i = 0; i < MY_CHANNEL_NUMBER; ++i) {            std::stringstream ss;            ss << base_name << "_" << i;            names.push_back(ss.str());        }        for (auto& name : names) {            int fd = mkfifo(name.c_str(), 0666);            if (fd == -1) {                if (errno != EEXIST) {                    perror("create channels fail");                    exit(EXIT_FAILURE);                } else {                    channels.push_back(Channel(name));                }            } else {                channels.push_back(Channel(name));            }        }        retry++;        std::cout << "Channels being created... " << (100 - retry * 10) << "%完成" << std::endl;        sleep(1);    }}

第4步：模拟进程通信

为了模拟进程间通信，我们需要用一个主进程来管理子进程的工作。这通常涉及到fork和exec等系统调用：

void DoingWork(std::vector
   
    & channels, const std::string& work_name) {    auto channel = ChooseChannel(channels);    pid_t pid = fork();    if (pid == 0) {        if (access(channel.fifo_path.c_str(), F_OK) == -1) {            std::cout << "工作流程被终止：" << channel.fifo_path << std::endl;            exit(EXIT_SUCCESS);        }        int rfd = open(channel.fifo_path.c_str(), O_RDONLY);        if (rfd < 0) {            perror("无法打开读管道");            exit(EXIT_FAILURE);        }        char buffer[1024];        ssize_t nread = read(rfd, buffer, sizeof(buffer));        if (nread <= 0) {            std::cout << "读取失败：" << channel.fifo_path << std::endl;            close(rfd);            exit(EXIT_FAILURE);        }        Work(buffer);        close(rfd);    } else if (pid > 0) {        int wfd = open(channel.fifo_path.c_str(), O_WRONLY);        if (wfd < 0) {            perror("无法打开写管道");            exit(EXIT_FAILURE);        }        ssize_t nwritten = write(wfd, work_name.c_str(), work_name.size());        if (nwritten < 0) {            std::cout << "写入失败：" << channel.fifo_path << std::endl;            close(wfd);            exit(EXIT_FAILURE);        }        close(wfd);        waitpid(pid, nullptr, W_UNKennigandroid blocker);    }}

功能扩展：类封装

为了更高效地管理进程池，我建议将这些功能封装到一个主程序类中，便于管理和扩展：

class MainProcess {public:    MainProcess() = default;    void createChannels() {        for (int i = 0; i < 10; ++i) {            std::stringstream ss;            ss << "../my_fifo_" << i;            if (mkfifo(ss.str().c_str(), 0666) == 0) {                channels.push_back(Channel(ss.str()));            } else {                if (errno != EEXIST) {                    bs:: cout << "创建管道失败：" << ss.str() << std::endl;                    exit(EXIT_FAILURE);                } else {                    channels.push_back(Channel(ss.str()));                }            }        }    }private:    std::vector
   
     channels;};

模拟进程任务调度

为了验证进程池的功能，我编写了一个简单的任务调度逻辑：

std::string SeleteWork() {    std::random Device rd;    std::mt19937 gen(rd());    std::uniform_int_distribution
   
     dis(0, 4);    int task_number = dis(gen);    static const std::vector
    
      task_list = {        "下载视频",        "解压视频",        "转换视频格式",        "观看视频",        "学习编程"    };    return task_list[task_number];}

完整代码示例

你可能会遇到一些实际编码问题，让我帮你解决！

_SDUTIL源文件：#pragma once#include 
   
    #include 
    
     #include 
     
      #include 
      
       #include 
       
        #include 
        
         #include 
         
          #include 
          
           #include 
           
            #include 
            
             #include 
             
              #include 
              
               #include 
               
                #include 
                
                 #include 
                 
                  struct Channel { Channel(std::string fifo_name) : fifo_path(fifo_name) {} std::string fifo_path;};class MainProcess {public: MainProcess() = default; void createChannels() { std::vector
                  
                    channels; const int ChannelNumber = 5; for(int i = 0; i < ChannelNumber; ++i) { std::string name = "../myfifo_" + std::to_string(i); if (mkfifo(name.c_str(), 0666) == 0) { channels.push_back(Channel(name)); } else { if (errno != EEXIST) { std::cerr << "failed to create fifo: " << name << std::endl; exit(EXIT_FAILURE); } else { std::cout << "fifo exists: " << name << std::endl; channels.push_back(Channel(name)); } } } channels.size() = ChannelNumber; std::cout << "已创建" << ChannelNumber << "个管道" << std::endl; } void deleteChannels() { for (auto& c : channels) { if (unlink(c.fifo_path.c_str()) == -1) { if (errno == ENOENT) { std::cout << "删除成功：" << c.fifo_path << std::endl; } else { std::cerr << "删除失败：" << c.fifo_path << std::endl; } } } std::cout << "所有管道已被删除" << std::endl; } const Channel& chooseChannel() { static std::random_device rd; static std::mt19937 gen(rd()); std::uniform_int_distribution
                   
                     dis(0, channels.size() - 1); return channels[dis(gen)]; } const std::string& selectTask() { static std::random_device rd; static std::mt19937 gen(rd()); std::uniform_int_distribution
                    
                      dis(0, 4); int choose = dis(gen); static const std::vector
                     
                       taskList = { "下载视频", "解压视频", "转换视频格式", "观看视频", "学习编程" }; return taskList[choose]; } void doingWork() { std::string task = selectTask(); std::cout << "启动执行任务：" << task << std::endl; pid_t pid = fork(); if (pid == 0) { // 子进程读取任务并执行 std::string fifo_path = chooseChannel().fifo_path; if (access(fifo_path.c_str(), F_OK) == -1) { std::cerr << " fifo path not exists: " << fifo_path << std::endl; return EXIT_SUCCESS; } int rfd = ::open(fifo_path.c_str(), O_RDONLY); if (rfd == -1) { std::cerr << "无法打开读取端" << std::endl; return EXIT_SUCCESS; } char buffer[1024]; ssize_t bytes_read = ::read(rfd, buffer, sizeof(buffer)); if (bytes_read <= 0) { std::cerr << "读取失败：" << fifo_path << std::endl; return EXIT_SUCCESS; } buffer[bytes_read] = '\0'; Work(buffer); close(rfd); } else { // 父进程写入任务 int wfd = ::open(chooseChannel().fifo_path.c_str(), O_WRONLY); if (wfd == -1) { std::cerr << "无法打开写进入道" << std::endl; return EXIT_SUCCESS; } bytes_written = ::write(wfd, task.c_str(), task.size()); if (bytes_written < 0) { std::cerr << "写入失败：" << chooseChannel().fifo_path << std::endl; return EXIT_SUCCESS; } close(wfd); waitpid(pid, nullptr, W_UNBLOCKED); } }};

总结

通过以上步骤，我们成功创建并管理了一个基于命名管道的进程池系统。这种方法通过封装管理管道，实现了进程间的安全通信和资源共享。在实际应用中，可以根据具体需求进行调整，例如处理更复杂的任务和错误情况。此外，也可以将这些功能扩展到更多的操作系统命令和功能中。希望本文能为大家提供一个清晰的指导，帮助您成功实现进程池管理！

上一篇：Linux——动态库

下一篇：Linux——共享内存

发表评论

关于作者

喝酒易醉，品茶养心，人生如梦，品茶悟道，何以解忧？唯有杜康！

-- 愿君每日到此一游！