Question

我一直在我们的项目上运行Valgrind，并且valgrind报告说，尽管在函数的底部调用了freeaddrinfo，但是从getaddrinfo的调用中已经丢失了内存。知道可能导致这种情况的原因吗？

int tcp_connect(char *address, char *port)
{
//printf("%s\n ", address);
    int status, sockfd;
    struct addrinfo hints, *servinfo, *p;

    memset(&hints, 0, sizeof hints);
    hints.ai_family = AF_UNSPEC;
    hints.ai_socktype= SOCK_STREAM;
    status = getaddrinfo(address,port,&hints,&servinfo);
    if(status != 0){
        printf("tcpconnect: getaddrinfo failed\n");
    }
    for(p = servinfo; p != NULL; p = p->ai_next) {
        sockfd = socket(p->ai_family, p->ai_socktype, p->ai_protocol);
        if(sockfd == -1){
            printf("tcpconnect: socket failed\n");
            continue;
        }

        status = connect(sockfd, p->ai_addr, p->ai_addrlen);
        if(status == -1){
            close(sockfd);
            printf("tcpconnect: connect failed\n");
            continue;
        }
        break;
    }

    if (p == NULL){
        printf("tcpconnect: could not get a valid socket file descriptor\n");
        return -1;
    }

    freeaddrinfo(servinfo);
    //return socket
    return sockfd;
}

valgrind输出在这里：

==7803== 384 bytes in 8 blocks are definitely lost in loss record 30 of 37
==7803==    at 0x4023CC8: malloc (vg_replace_malloc.c:236)
==7803==    by 0x413FE46: gaih_inet (in /lib/libc-2.7.so)
==7803==    by 0x4141CA1: getaddrinfo (in /lib/libc-2.7.so)
==7803==    by 0x804D20D: tcp_connect (tcpconnect.c:18)
==7803==    by 0x804D505: update_radio_data_t (radiodata.c:69)
==7803==    by 0x4073368: start_thread (in /lib/libpthread-2.7.so)
==7803==    by 0x4159CFD: clone (in /lib/libc-2.7.so)

我对此内存泄漏感到困惑，因为我确保在底部调用freeaddrinfo。这可能是Valgrind或freeaddrinfo中的错误吗？

Answer 1

您的失败路径（if (p == NULL)）未拨打freeaddrinfo，我会在检查前将呼叫移至此处。

Answer 2

if (p == NULL){
    printf("tcpconnect: could not get a valid socket file descriptor\n");
    return -1;
}

如果此条件为真，则不会调用freeaddrinfo(servinfo)。

Valgrind报告说getaddrinfo正在泄漏内存？

2 个答案: