Transcription

1 Basics of Socket Programming Please check the referenced links for the further description and examples. 1

2

3 Procedures for Socket Implementation 1. Create the server application (e.g. a simple shellscript) 2. Test the server application at the command line 3. Decide on a port number and service name 4. Declare the port and name in /etc/services 5. Create a file for this service in /etc/xinetd.d or modify the file /etc/inetd.conf 6. Restart xinetd or inetd 7. Telnet into the service, and see the server app's output 8. Write your client program 3

4 An example of a shell script on server #!/bin/bash /bin/echo -n "Hello World:" /usr/bin/tee /tmp/log.log /bin/date /usr/bin/tee -a /tmp/log.log 4

5 Results from using telnet as a client to link to the server [myuid@mydesk myuid]$ telnet Trying Connected to Escape character is '^]'. Hello World:Tue Jan 23 13:33:55 EST 2001 Connection closed by foreign host. [myuid@mydesk myuid]$ 5

6 TCP socket (Linux) TCP Server Socket ( ) TCP Client Socket ( ) Connect ( ) send ( ) recv ( ) Close ( ) 建立連線 資料 ( 請求 ) 資料 ( 回應 ) < 檔案結尾 > 的通知 Bind ( ) Listen ( ) Accept ( ) 等待客戶端的連線 recv ( ) 處理請求 send ( ) Read ( ) Close ( )

7 Create a Socket l 進行網路傳輸前, 程式必須先呼叫 socket 函式 l Socket 函式 #include <sys/types.h> #include <sys/socket.h> int socket (int family, int type, int protocol) ;

8 Parameters AF_INET AF_INET6 family IPv4 protocols IPv6 protocols Description SOCK_STREAM SOCK_DGRAM type Description stream socket (TCP) datagram socket (UDP) protocol 0 Normally set to 0 Description

9 Connect function l TCP 客戶端使用 connect 函式來建立與 TCP 伺服端的連線 #include <sys/socket.h> #include <sys/types.h> int connect (int sockfd, const struct sockaddr_in *servaddr, int addrlen) ; sockfd 是 socket 函式傳回的 socket descriptor 而第二 三個參數則是 socket 位址結構的指標及其長度

10 socket structures l struct sockaddr: Holds socket address information for many types of sockets l struct sockaddr_in: A parallel structure that makes it easy to reference elements of the socket address

11 Bind function l bind 函式會將本機協定的 IP 和 Port 指定給 socket; 在 Internet 協定中的協定位址是 32 位元的 IPv4 位址 或 128 位元的 IPv6 位址, 加上 16 位元的 TCP 或 UDP 埠號 #include <sys/socket.h> int bind (int sockfd, const struct sockaddr_in *myaddr, socklen_t addlen) ; 第二個參數是特定協定位址的指標 第三個參數則是該位址結構的長度 returns 0 if successful or -1 on error.

12 Example struct sockaddr_in serv; serv.sin_family = AF_INET; serv.sin_addr.s_addr = htonl( ); serv.sin_port = htons(3000); bind(serv_sd, &serv, sizeof(serv));

13 Listen l 只有 TCP 伺服端會呼叫 listen 函式 函式的第 2 個參數會指定 kernel 應該為此 socket 暫存到佇列中的最大連線數目 #include <sys/socket.h> int listen (int sockfd, int backlog) ;

14 Accept l TCP 伺服端會呼叫 accept, 從已完成連線佇列最前端傳回下個已完成的連線 如果已完成連線佇列是空的, 行程就會進入睡眠 #include <sys/socket.h> int accept (int sockfd, struct sockaddr_in *cliaddr, socklen_t *addrlen) ; cliaddr 與 addrlen 參數是用來傳回連線客戶端的協定位址 ;addrlen 是值一結果參數 : 在呼叫前, 將 *addrlen 指向的整數值設為 cliaddr 指向之 socket 位址結構的長度 ; 在傳回時, 整數值會包含 kernel 在 socket 位址結構中實際存放的位元組數目

15 accept styles There are basically three styles of using accept: 1. Iterating server: Only one socket is opened at a time. When the processing on that connection is completed, the socket is closed, and next connection can be accepted. 2. Forking server: After an accept, a child process is forked off to handle the connection. Variation: the child processes are preforked and are passed the socketid. Concurrent single server: use select to simultaneously wait on all open socketids, and waking up the process only when new data arrives. 3. Concurrent single server: use select to simultaneously wait on all open socketids, and waking up the process only when new data arrives. 15

16 Pro and Con of Accept styles Iterating server is basically a low performance technique since only one connection is open at a time. Forking servers enable using multiple processors. But they make sharing state difficult, unless performed with threads. Threads, however present a very fragile programming environment. Concurrent single server: reduces context switches relative to forking processes and complexity relative to threads. But does not benefit from multiprocessors. 16

17 Send l TCP 伺服端或客戶端呼叫 send 傳送資料 #include <unistd.h> int count=send( int sockfd, char *buffer, int buflen, int flags ) count: 傳送資料的大小 buffer: 資料傳送的指標 buflen: 資料長度 flag: 通常設為 0 Or write int write(int sockfd, char * buffer, int buflen);

18 recv l TCP 伺服端或客戶端呼叫 recv 接收資料 #include <unistd.h> Int count=recv( int sockfd, char *buffer, int buflen, int flags ) Count: 接收到的資料大小 buffer: 資料接收後儲存位址的指標 buflen: 資料長度 flag: 通常設為 0 Or read int read(int sockfd, char * buffer, int buflen);

19 Close l 用來關閉 socket, 並且終止 TCP 連線 #include <unistd.h> int close (int sockfd) ;

20 UDP socket (Linux) UDP 伺服端 Socket ( ) UDP 客戶端 socket ( ) 慣用埠 Bind ( ) recvfrom ( ) sendto ( ) recvfrom ( ) 資料 ( 請求 ) 資料 ( 回應 ) 暫停直到收到客戶端的資料包 處理請求 close ( ) sendto ( )

21 Socket 在 TCP 與 UDP 上的差異 l 由於 TCP 與 UDP 這兩個傳輸層本身的差異 UDP 是非連線式 不可靠的資料包協定 TCP 為連線導向式 可靠的位元組串流 l 因此使用 TCP 與使用 UDP 所撰寫的應用程式之間, 有些基本的差異 l UDP 並不會建立與伺服端的連線, 而只是使用 sendto 函式將資料包傳送給伺服端 : 這函式只需目的位址為參數 l 而伺服端也不接受來自客戶端的連線, 而是呼叫 recvfrom 函式, 等待資料由客戶端送達, 並傳回客戶端的協定位址和資料包, 使伺服端傳送回應給正確客戶端

22 revfrom 與 sendto 函式 l 這兩個函式類似標準的 read 與 write 函式 #include <sys/socket.h> int recvfrom (int sockfd, const void *buff, size_t nbytes, int flags, struct sockaddr *from, socklen_t *addrlen) ; int sendto (int sockfd, const void *buff, size_t nbytes, int flags, const struct sockaddr *to, socklen_t addlen) ; 前三個參數 sockfd buff nbytes 與 read 及 write 的前三個參數相同, 分別是描述子 指向讀取或寫入緩衝區的指標以及讀取或寫入的位元組收數目 sendto 的 to 參數是 socket 位址結構, 指定資料所要送往的協定位址, addrlen 則是指定這個 socket 位址結構的長度 ;

23 1st Homework u Go through the article for building a socket program as in the following web site: u Refer to Socket Programming Tutorial & Introduction, such as u Prepare for the demonstration on March