Analyzing the Patron Saint of Linux

The biggest differences are as follows:
• (a) The daemon process has completely separated from the terminal console, but the background program has not completely separated from the terminal (results will still be output to the terminal before the terminal is closed);
• (b) The daemon process will not be affected when closing the terminal console, but the background program will stop when the user exits. It needs to be run in the nohup command & format to avoid the impact;
• (c) The session group, current directory, and file descriptors of the daemon process are independent. Running in the background is just a fork of the terminal to allow the program to execute in the background. None of this has changed;

Daemon is a special process that runs in the background. It is separated from the terminal. This prevents the process from being interrupted by signals generated by any terminal. The information generated during the execution of the process is not stored in any terminal. displayed on. A daemon process periodically performs certain tasks or waits for processing of certain events that occur. Most Linux servers are implemented using daemon processes.

1. 1. Shielding some signals related to the control terminal operation is to prevent the control terminal from being disturbed and exiting or hanging before the daemon process is started normally. code show as below:

      /* 处理可能的终端信号 */
      signal(SIGTTOU, SIG_IGN);
      signal(SIGTTIN, SIG_IGN);
      signal(SIGTSTP, SIG_IGN);
      signal(SIGHUP , SIG_IGN);
      

      Copy after login
   2. Run in the background.

      /* 是父进程，结束父进程，子进程继续 */
      if(fork())
          exit(0);
      

      Copy after login
   3. Leave the control terminal and process group:
      • (1) A process belongs to a process group, and the process group number (PGID) is the process number (PID) of the process group leader
      • (2) Processes in the same process group share a control terminal. This control terminal defaults to the terminal where the process was created
      • (3) The control terminal and process group associated with a process are usually inherited from the parent process. Therefore, the child process is still affected by the terminal of the parent process, because the signals generated by the terminal will be sent to all the foreground process groups. process.

      Based on the above reasons, it is necessary to completely get rid of the influence of the terminal for the child process. It is necessary to call setsid() to make the child process the new session leader. The code is as follows:

      setsid();

      Copy after login

      After the setsid() call is successful, the process calling this function becomes the new session group leader and new process group leader, and is separated from the original process group. Due to the exclusivity of the session process to the controlling terminal, the process is detached from the controlling terminal at the same time.

   4. To prohibit the process from reopening the control terminal, the method used is to create a child process again and let the parent process exit. The child process is no longer the session leader, thus achieving the goal. code show as below:

      /* 结束第一子进程，第二子进程继续 */
      if(fork())
          exit(0);
      

      Copy after login
   5. Close the open file descriptor. Because a process inherits open file descriptors from the parent process that created it, they are generally no longer needed. If it is not closed, system resources will be wasted. code show as below:

      #define NOFILE  256
      
      for(i=0; i

      Copy after login

   6. Change the current working directory. While a process is active, the file system containing its working directory cannot be unmounted. Therefore, you need to change the daemon's working directory to an appropriate directory. code show as below:

      chdir("/tmp");

      Copy after login
   7. Reset file creation mask. A process inherits the file creation mask from the parent process that created it. It may modify the access permissions of files created by the daemon. code show as below:

      umask(0);
      

      Copy after login
   8. Handle the SIGCHLD signal (child process exit signal). If you do not wait for the child process to end, the child process will become a zombie process and occupy system kernel resources.

      /* 将子进程退出信号设为SIG_IGN，让系统帮助回收进程资源 */
      signal(SIGCHLD, SIG_IGN);

      Copy after login

      The overall code is as follows:

      #define NOFILE      256
      
      void DaemonMode()
      {
          int num = 0;
          int fd0, fd1, fd2;
      
          /* 屏蔽可能的信号 */
          signal(SIGTTOU, SIG_IGN);
          signal(SIGTTIN, SIG_IGN);
          signal(SIGTSTP, SIG_IGN);
          signal(SIGHUP , SIG_IGN);
      
          if(fork())
              exit(0);
      
          setsid();
      
          if(fork())
              exit(0);
      
          chdir("/tmp/httpd");
      
          umask(0);
      
          for(; num

      Copy after login

If the calling process is already the leader of a process group, this function returns an error. In order to prevent this situation, fork() is usually called to create a child process, and then its parent process is terminated, while the child process continues, and this function is called in the child process.

If the process calling this function is not a process group leader, this function will create a new session, and the process calling the setsid() function will become the leader process of the new session, and will be associated with the session group and process group of its parent process. Detach. Because the session is exclusive to the controlling terminal, the process is simultaneously detached from the controlling terminal.

The above is the detailed content of Analyzing the Patron Saint of Linux. For more information, please follow other related articles on the PHP Chinese website!