C程序循环读取Socket后终端长时间停滞问题求助

问题说明 #

程序通过循环从Socket读取HTTP响应数据并打印到标准输出，目前遇到以下异常：

• 最后一次recv()返回0字节前，所有接收内容都能正常打印
• 倒数第二次recv()执行后，终端会停滞很长时间才完成程序执行，停滞时长与读取的数据量正相关（数据越多，停滞越久）
• 用clock()记录代码片段运行时间，结果无差异，但终端中最后一次printf()的输出需要很久才显示
• 程序运行方式：./exec www.google.com/randtext

代码片段 #

#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <string.h>
#include <sys/types.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <netdb.h>
#include <time.h>
#include <arpa/inet.h>

void error(const char* msg){
    perror(msg);
    exit(0);
}

int main(int argc, char* argv[]){
    int socketfd;
    struct sockaddr_in serv_addr;
    struct hostent* he;
    char sendline[4096];
    char reciveline[4096];

    char* domain = argv[1];
    char* path = strchr(domain, '/');
    *path++ = '\0';
    //printf("host: %s;\npath: %s\n", domain, path);


    he = gethostbyname(domain);
    if(he == NULL){
        error("host error");
    }
    serv_addr.sin_family = AF_INET;     
    serv_addr.sin_port = htons(80);
    serv_addr.sin_addr = *((struct in_addr *)he->h_addr);
    socketfd = socket(AF_INET, SOCK_STREAM, 0);

    bzero(&(serv_addr.sin_zero),8); 
    if (connect(socketfd, (struct sockaddr *)&serv_addr,sizeof(struct sockaddr)) == -1){
       error("Connect");
    }

    memset(sendline, 0, sizeof(sendline));
    //snprintf(sendline, sizeof(sendline), "GET /%s HTTP/1.1\r\nHost: %s\r\n\r\n", path, domain);
    snprintf(sendline, sizeof(sendline), "GET / HTTP/1.1\r\nHost: %s\r\n\r\n", domain);
    printf("%s\n", sendline);
    
    send(socketfd, sendline, strlen(sendline), 0); 
    printf("Data sent.\n\n");


    memset(reciveline,0,sizeof(reciveline));
    int total = sizeof(reciveline)-1;
    int bytes, i = 0;
    do {
        i++;
        bytes = recv(socketfd,reciveline,total,0);

        printf("Amount received : %d\n\n",bytes);
        if (bytes < 0)
            error("ERROR reading response from socket");
        if (bytes == 0){
            printf("\n\nbrokefree!\n\n");
            break;
        }
        reciveline[bytes] = '\0';
        printf("%s", reciveline);
        memset(reciveline,0,sizeof(reciveline));
    } while (1);
    close(socketfd);
    printf("Total loops : %d\n",i);
    return 0;
}

原因分析 #

问题出在HTTP 1.1的默认连接机制上：HTTP 1.1默认启用长连接（Keep-Alive），服务器在发送完响应内容后不会主动关闭TCP连接，而是保持连接等待后续请求。你的程序在读取完所有响应数据后，依然调用recv()阻塞等待服务器发送更多数据，直到服务器的连接超时时间到期才会关闭连接，此时recv()返回0，程序才会继续执行。

而clock()统计的是CPU实际运行时间，recv()阻塞时程序不占用CPU，所以记录的时间没有差异，但终端会表现为长时间停滞。

解决方案 #

方法1：请求服务器关闭连接 #

在HTTP请求头中添加Connection: close，明确告诉服务器发送完响应后立即关闭TCP连接：

snprintf(sendline, sizeof(sendline), "GET / HTTP/1.1\r\nHost: %s\r\nConnection: close\r\n\r\n", domain);

这样服务器发送完响应后会主动关闭连接，recv()会立即返回0，程序不会阻塞。

方法2：解析响应的Content-Length字段 #

读取HTTP响应的头信息，解析出Content-Length的值，计算需要接收的总数据量，当读取的数据达到该长度时，主动跳出循环，不再调用recv()。这种方式更符合HTTP协议规范，适合长连接场景。

方法3：设置recv()超时 #

通过setsockopt()给Socket设置接收超时时间，当recv()阻塞超过指定时间后会返回错误，避免无限等待：

struct timeval timeout;
timeout.tv_sec = 5;  // 5秒超时
timeout.tv_usec = 0;
setsockopt(socketfd, SOL_SOCKET, SO_RCVTIMEO, (char *)&timeout, sizeof(timeout));

修改后的示例代码（方法1） #

#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <string.h>
#include <sys/types.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <netdb.h>
#include <time.h>
#include <arpa/inet.h>

void error(const char* msg){
    perror(msg);
    exit(0);
}

int main(int argc, char* argv[]){
    int socketfd;
    struct sockaddr_in serv_addr;
    struct hostent* he;
    char sendline[4096];
    char reciveline[4096];

    char* domain = argv[1];
    char* path = strchr(domain, '/');
    *path++ = '\0';
    //printf("host: %s;\npath: %s\n", domain, path);


    he = gethostbyname(domain);
    if(he == NULL){
        error("host error");
    }
    serv_addr.sin_family = AF_INET;     
    serv_addr.sin_port = htons(80);
    serv_addr.sin_addr = *((struct in_addr *)he->h_addr);
    socketfd = socket(AF_INET, SOCK_STREAM, 0);

    bzero(&(serv_addr.sin_zero),8); 
    if (connect(socketfd, (struct sockaddr *)&serv_addr,sizeof(struct sockaddr)) == -1){
       error("Connect");
    }

    memset(sendline, 0, sizeof(sendline));
    // 添加Connection: close请求头
    snprintf(sendline, sizeof(sendline), "GET / HTTP/1.1\r\nHost: %s\r\nConnection: close\r\n\r\n", domain);
    printf("%s\n", sendline);
    
    send(socketfd, sendline, strlen(sendline), 0); 
    printf("Data sent.\n\n");


    memset(reciveline,0,sizeof(reciveline));
    int total = sizeof(reciveline)-1;
    int bytes, i = 0;
    do {
        i++;
        bytes = recv(socketfd,reciveline,total,0);

        printf("Amount received : %d\n\n",bytes);
        if (bytes < 0)
            error("ERROR reading response from socket");
        if (bytes == 0){
            printf("\n\nbrokefree!\n\n");
            break;
        }
        reciveline[bytes] = '\0';
        printf("%s", reciveline);
        memset(reciveline,0,sizeof(reciveline));
    } while (1);
    close(socketfd);
    printf("Total loops : %d\n",i);
    return 0;
}

内容的提问来源于stack exchange，提问作者MaxUnluky