TCP编程基础

http://learn.akae.cn/media/ch36s01.html（协议栈）

关于网络序列

因为网络连接可能建立在big-endian和little-endian的机器之间。

详见这篇文章http://splayx.iteye.com/blog/1739223

 

所以端口（port）、地址（ip）有必要从主机的字符顺序转为网络的字符顺序，

网络的字符顺序作为不同机器之间通讯的中间载体。linux（<arpa/inet.h>）提供了4个转化的操作。

uint32_t htonl(uint32_t hostlong);

uint16_t htons(uint16_t hostshort);

uint32_t ntohl(uint32_t netlong);

uint16_t ntohs(uint16_t netshort);

 

n: net bytes order

h: host bytes order

s: short, uint16_t

l: long, uint32_t

 

以上每个函数的作用是显而易见的。

 

下面是一个简单的TCP用到的一些函数。

#include <sys/socket.h>

int listen(int socket, int backlog);

backlog: the size of listening queue

 

#include <sys/socket.h>

int socket(int domain, int type, int protocol);

domain:

AF_UNIX

File system pathnames.

AF_INET

Internet address.

---------------------------

type:

SOCK_STREAM

TCP

SOCK_DGRAM

UDP

SOCK_SEQPACKET

一条长记录可以通过几次接收，并有MSG_EOR作为结束标记

---------------------------

protocol:

如果不用其他协议,可以无视这项，为0就好了。

 

#include <sys/socket.h>

---------------------------

int accept(int socket, struct sockaddr *address,socklen_t *address_len);

---------------------------

int bind(int socket, const struct sockaddr *address,socklen_t address_len);

---------------------------

int connect(int socket, const struct sockaddr *address,socklen_t address_len);

因为不用协议地址的长度可能是不一样的，例如ipv4跟ipv6,

使得结构体sockaddr的大小有差异，所以需要address_len来指定长度。

 

最后使用这些接口建立一个简单的server-client的通讯。

server：

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

const int N = 1 << 20;

char buffer[N];

int main() {
    int sockfd;

    // Create A socket.
    if ((sockfd = socket(AF_INET, SOCK_STREAM, 0)) == -1) {
        fprintf(stderr, "Socket Error:%s\n\a", strerror(errno));
        return 1;
    }

    // Info of server's addr.
    struct sockaddr_in srv_info;
    srv_info.sin_family = AF_INET;
    srv_info.sin_addr.s_addr = inet_addr("127.0.0.1");
    srv_info.sin_port = htons(60530);
    bzero(srv_info.sin_zero, 8);

    // bind the socket to the server's addr
    if (bind(sockfd, (struct sockaddr*)&srv_info, sizeof(struct sockaddr)) == -1) {
        fprintf(stderr, "Bind Error:%s\n\a", strerror(errno));
        return 1;
    }

    // listen to sockfd, with a listening queue of size 10.
    if (listen(sockfd, 10) == -1) {
        fprintf(stderr, "LIsten Error:%s\n\a", strerror(errno));
        return 1;
    }

    socklen_t size = sizeof(struct sockaddr_in);
    struct sockaddr_in rmt_info;

    // Blocked wait for a connect to the server.
    int nsockfd = accept(sockfd, (struct sockaddr *)&rmt_info, &size);
    if (nsockfd == -1) {
        fprintf(stderr, "Accept Error:%s\n\a", strerror(errno));
        return 1;
    }
    while (true) {
        printf("Msg to send:\n");
        scanf("%s", buffer);
        write(nsockfd, buffer, strlen(buffer));
    }
    close(nsockfd);
    close(sockfd);
    return 0;
}

 

client：

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

const int N = 1 << 10;
char buffer[N];

int main() {
    int sockfd;

    // Creat a socket for TCP.
    if ((sockfd = socket(AF_INET, SOCK_STREAM, 0)) == -1) {
        fprintf(stderr, "Socket Error:%s\n\a", strerror(errno));
        return 1;
    }

    // Info of server's addr.
    struct sockaddr_in srv_info;
    srv_info.sin_family = AF_INET;
    srv_info.sin_addr.s_addr = inet_addr("127.0.0.1");
    srv_info.sin_port = htons(60530);
    bzero(srv_info.sin_zero, 8);

    // Use the sockfd to connect the server.
    // Failed, when the server doesn't set up.
    if (connect(sockfd, (struct sockaddr *)&srv_info, sizeof(struct sockaddr)) == -1) {
        fprintf(stderr, "Connect Error:%s\n\a", strerror(errno));
        return 1;
    }
    int nbytes;
    // A blocked read from server
    // And nbytes' data actually read.
    while (true) {
        if ((nbytes = read(sockfd, buffer, N - 1)) == -1) {
            fprintf(stderr, "Read Error:%s\n", strerror(errno));
            return 1;
        }
        buffer[nbytes] = '\0';
        printf("Msg received:\n%s\n", buffer);
    }
    close(sockfd);
    return 0;
}

 

server的输入

Msg to send:

hellosefise

Msg to send:

this is server speaking

Msg to send:

Msg to send:

Msg to send:

Msg to send:

 

client的输出

Msg received:

hellosefise

Msg received:

thisisserverspeaking

 

这里我们看到当server端输入够快时，client端一次性全收到了。

关于tcp内核的行为和用户具体应对策略，甚至更底层的tcp对ip协议的封装，

有很多有意思的地方值得我们去探究。

 

参考

http://pubs.opengroup.org/onlinepubs/7908799/xns/netinetin.h.html

http://pubs.opengroup.org/onlinepubs/7908799/xns/syssocket.h.html