unix网络编程笔记---7.TCP的C/S实例3wait与waitpid

1. 说明
    第5.10的wait与waitpid
    当client有5个连接同时关闭时,
    server端的wait来不及处理同时发生的5个sigchld,导致僵死进程产生。
1.1 现象

cong@msi:/work/test/tcpip/7tcp/client$ ./client
client.c:main[59]: connect success
client.c:main[59]: connect success
client.c:main[59]: connect success
client.c:main[59]: connect success
client.c:main[59]: connect success
cong@msi:/work/test/tcpip/7tcp/client$

当client连接上时，共执行了5次fork,但是只收到了3次sigchld

cong@msi:/work/test/tcpip/7tcp/serv$ ./tcpserv
tcpserv.c:main[50]: accept:127.0.0.1,port=45388
tcpserv.c:main[50]: accept:127.0.0.1,port=45389
tcpserv.c:main[58]: fork           --> fork1
tcpserv.c:main[50]: accept:127.0.0.1,port=45390
tcpserv.c:main[58]: fork           --> fork2
tcpserv.c:main[50]: accept:127.0.0.1,port=45391
tcpserv.c:main[58]: fork           --> fork3
tcpserv.c:main[50]: accept:127.0.0.1,port=45392
tcpserv.c:main[58]: fork           --> fork4
tcpserv.c:main[58]: fork           --> fork5
tcpserv.c:sig_chld[9]: child 16165 terminated     --> term1
tcpserv.c:sig_chld[9]: child 16166 terminated     --> term2
tcpserv.c:sig_chld[9]: child 16167 terminated     --> term3

ps查看发现，正好有2个处于僵死状态的子进程

cong@msi:/work/test/tcpip/7tcp/client$ ps -u
USER       PID %CPU %MEM    VSZ   RSS TTY      STAT START TIME COMMAND
cong     16197  0.0  0.0   4196   504 pts/1    S+ 18:00   0:00 ./tcpserv
cong     16202  0.0  0.0      0     0 pts/1    Z+ 18:01   0:00 [tcpserv] <defunct>
cong     16203  0.0  0.0      0     0 pts/1    Z+ 18:01   0:00 [tcpserv] <defunct>

---

2. 代码
2.1 修改了的client端

cong@msi:/work/test/tcpip/7tcp/client$ cat client.c
#include "utils.h"

//#define TIME_SERV_ADDR "192.168.4.98"
#define TIME_SERV_ADDR "127.0.0.1"
#if 0
ssize_t readline(int fd, void* vptr, size_t maxlen)
{
    ssize_t n, rc;
    char c,*ptr;
    ptr = vptr;
    for(n=1; n<maxlen; n++)
    {
again:
        if((rc=read(fd,&c,1)) == 1)
        {
            *ptr++ = c;
            if(c == '\n')
                break;
        } else if(rc == 0) {
            *ptr = 0;
            return (n-1);
        }else {
            if( errno == EINTR)
                goto again;
            return -1;
        }
    }
}
#endif
ssize_t readline(int fd, void* vptr, size_t maxlen)
{
    ssize_t n, rc; 
    char c,*ptr;
    ptr = vptr;
    for(n=0; n
again:
        if((rc=read(fd,&c,1)) == 1)
        {
            *ptr++ = c;
            if(c == '\n')
                break;
        } else if(rc == 0) {
            *ptr = 0;
            return n;
        }else {
            if( errno == EINTR)
                goto again;
            return -1; 
        }
    }   
    *ptr = '\0';
    return n;
}
void str_cli(FILE* fp, int sockfd)
{
    char sendline[MAXLINE], recvline[MAXLINE];
    while(fgets(sendline, MAXLINE, fp) != NULL)
    {
        write(sockfd, sendline, strlen(sendline));
        if(readline(sockfd, recvline, MAXLINE) == 0)
            dbmsg("server terminated");
        fputs(recvline, stdout);
    }

}

int main ( int argc, char *argv[] )
{
    int i;
    int sockfd, n;
    struct sockaddr_in servaddr;
    char recvline[1024];
    for(i=0; i<5; i++)     //产生了5个client
    {
        sockfd = socket(AF_INET, SOCK_STREAM, 0);
        if(sockfd < 0)
            return -1;
        bzero(&servaddr, sizeof(servaddr));
        servaddr.sin_family = AF_INET;
        servaddr.sin_port = htons(13999);
        inet_pton(AF_INET, TIME_SERV_ADDR, &servaddr.sin_addr);
        connect(sockfd, (const struct sockaddr*)&servaddr, sizeof(servaddr));
        dbmsg("connect success");
    }
    str_cli(stdin, sockfd);   //str_cli返回后，进程退出，这会发送5个FIN
    return EXIT_SUCCESS;
}

2.2 server.c出的问题分析

cong@msi:/work/test/tcpip/7tcp/serv$ cat tcpserv.c
#include "utils.h"

#define LISTENQ 10
void sig_chld(int signo)
{
    pid_t pid;
    int stat;
    pid = wait(&stat);              //3.同时产生5个SIGCHLD，会同时调用wait
    dbmsg("child %d terminated", pid); //但是wait是阻塞的，当wait在处理sigchld信号时再有sigchld信号过来
    return ;                         //wait是没有办法处理的，所以导致了僵死进程产生
}
void str_echo(int sockfd)
{
    ssize_t n;
    char buf[MAXLINE];
again:
    while( (n=read(sockfd, buf, MAXLINE)) > 0)
        write(sockfd, buf, n);

    if( (n<0) && (errno==EINTR) )
        goto again;
    else if(n < 0)
        dbmsg("read_error");
}

int main ( int argc, char *argv[] )
{
    int listenfd, connfd;
    struct sockaddr_in servaddr, cliaddr;
    signal(SIGCHLD, sig_chld);
    pid_t childpid;
    char buf[1024];
    int len;
    listenfd = socket(AF_INET, SOCK_STREAM, 0);
    if(listenfd< 0)
        return -1;
    bzero(&servaddr, sizeof(servaddr));
    servaddr.sin_family = AF_INET;
    servaddr.sin_port = htons(13999);
    servaddr.sin_addr.s_addr = htonl(INADDR_ANY);
    //inet_pton(AF_INET, TIME_SERV_ADDR, &servaddr.sin_addr);

    bind(listenfd, (const struct sockaddr*)&servaddr, sizeof(servaddr));
    listen(listenfd, LISTENQ);
    while(1)
    {
        len = sizeof(cliaddr);
        //connfd = accept(listenfd, (struct sockaddr*)NULL, NULL);
        connfd = accept(listenfd, (struct sockaddr*)&cliaddr,(socklen_t*)&len);
        dbmsg("accept:%s,port=%d", inet_ntop(AF_INET,&cliaddr.sin_addr,buf, sizeof(buf)), ntohs(cliaddr.sin_port));
        if(errno == EINTR)
        {
            dbmsg("EINTR");
            continue;
        }
        if( (childpid=fork()) == 0)
        {
            dbmsg("fork");
            close(listenfd);
            str_echo(connfd);    //1.client:str_cli返回后，进程退出，这会发送5个FIN
            exit(0);             //2.这会导致serve的5个进程同时终止，并同时产生5个SIGCHLD
        }
        close(connfd);
    }

    return EXIT_SUCCESS;
}

2.3 源码
client.rar (下载后改名为client.tar.gz)
serv_old.rar (下载后改名为serv_old.tar.gz)

---

3. 修改后的server.c

cong@msi:/work/test/tcpip/7tcp/serv$ cat tcpserv.c
#include "utils.h"

#define LISTENQ 10
void sig_chld(int signo)
{
    pid_t pid;
    int stat;
    //pid = wait(&stat);                           //WNOHANG: return immediately if no child has exited
    while( (pid = waitpid(-1,&stat,WNOHANG)) > 0)  //3.同时产生5个SIGCHLD，会同时调用waitpid,此时waitpid是非阻塞的
        dbmsg("child %d terminated", pid);         //接收到sigchld信号时会立即返回并等侍下一个sigchld信号的到来,不会丢信号
    return ;
}
void str_echo(int sockfd)
{
    ssize_t n;
    char buf[MAXLINE];
again:
    while( (n=read(sockfd, buf, MAXLINE)) > 0)
        write(sockfd, buf, n);

    if( (n<0) && (errno==EINTR) )
        goto again;
    else if(n < 0)
        dbmsg("read_error");
}

int main ( int argc, char *argv[] )
{
    int listenfd, connfd;
    struct sockaddr_in servaddr, cliaddr;
    signal(SIGCHLD, sig_chld);
    pid_t childpid;
    char buf[1024];
    int len;
    listenfd = socket(AF_INET, SOCK_STREAM, 0);
    if(listenfd< 0)
        return -1;
    bzero(&servaddr, sizeof(servaddr));
    servaddr.sin_family = AF_INET;
    servaddr.sin_port = htons(13999);
    servaddr.sin_addr.s_addr = htonl(INADDR_ANY);
    //inet_pton(AF_INET, TIME_SERV_ADDR, &servaddr.sin_addr);

    bind(listenfd, (const struct sockaddr*)&servaddr, sizeof(servaddr));
    listen(listenfd, LISTENQ);
    while(1)
    {
        len = sizeof(cliaddr);
        //connfd = accept(listenfd, (struct sockaddr*)NULL, NULL);
        connfd = accept(listenfd, (struct sockaddr*)&cliaddr,(socklen_t*)&len);
        dbmsg("accept:%s,port=%d", inet_ntop(AF_INET,&cliaddr.sin_addr,buf, sizeof(buf)), ntohs(cliaddr.sin_port));
        if(errno == EINTR)
        {
            dbmsg("EINTR");
            continue;
        }
        if( (childpid=fork()) == 0)
        {
            dbmsg("fork");
            close(listenfd);   //1.client:str_cli返回后，进程退出，这会发送5个FIN
            str_echo(connfd);  //2.这会导致serve的5个进程同时终止，并同时产生5个SIGCHLD
            exit(0);
        }
        close(connfd);
    }

    return EXIT_SUCCESS;
}

client.rar (下载后改名为client.tar.gz)

serv_new.rar (下载后改名为serv_new.tar.gz)