UNIX下fork()与dup2()实现进程循环时提前终止问题排查

Let me break down the most common causes for your shell's main loop exiting after one command, and walk you through actionable fixes for each scenario.

Top Culprits to Investigate #

1. Forgetting to Exit the Child Process After Exec Failure #

This is the #1 mistake I see in beginner shell implementations. If execvp() fails (e.g., invalid command, malformed arguments), the child process won't just vanish—it will continue executing the parent's code, including the main loop. This creates two concurrent shell processes, and it often looks like the original loop terminated when it's actually the child process taking over input.

Fix: Always call exit() immediately after execvp() (since execvp() only returns if it fails):

pid_t pid = fork();
if (pid == 0) {
    // Parse your command into the args array here
    execvp(args[0], args);
    // If we reach this line, execvp has failed
    perror("Failed to execute command");
    exit(EXIT_FAILURE); // Critical: stops the child from entering the main loop
}

2. Unflushed Standard Output (Hidden Prompt) #

Shell prompts like > use line-buffered output by default. If you don't manually flush stdout, the prompt won't appear until a newline is printed, making it seem like the loop terminated when it's actually waiting for input.

Fix: Add fflush(stdout) right after printing your prompt:

while (1) {
    printf("> ");
    fflush(stdout); // Forces the prompt to show up immediately
    char cmd[256];
    if (fgets(cmd, sizeof(cmd), stdin) == NULL) break;
    
    // Handle exit command
    if (strcmp(cmd, "exit\n") == 0) break;
    
    // Parse and execute your command...
}

3. Unhandled Errors in wait() #

If wait() returns an error (e.g., unexpected signal interrupt), your code might be exiting the loop prematurely instead of handling the error and continuing.

Fix: Check the return value of wait() and log errors without breaking the loop:

else if (pid > 0) {
    int status;
    pid_t wait_result = wait(&status);
    if (wait_result == -1) {
        perror("wait() failed");
        // Don't break the loop—keep the shell running
    }
}

4. Accidental Loop Termination Logic #

Double-check your main loop structure to ensure you aren't hitting an unintended break or return statement after executing a command. For example, make sure command parsing or execution code isn't accidentally triggering an exit condition meant only for the exit command.

Quick Debugging Trick #

Add debug prints at the start and end of your main loop to confirm whether the loop is actually terminating or just not showing the prompt:

while (1) {
    printf("[DEBUG] Entering main loop\n");
    printf("> ");
    fflush(stdout);
    
    // Command handling code...
    
    printf("[DEBUG] Finished command, looping back\n");
}

If you see the "looping back" message but no prompt, the flush issue is your problem. If you don't see either, your loop is breaking unexpectedly.

内容的提问来源于stack exchange，提问作者user9431482