Berkeley sockets

From Wikipedia, the free encyclopedia

The Berkeley sockets application programming interface (API) comprises a library for developing applications in the C programming language that perform inter-process communication, most commonly across a computer network.

Berkeley sockets (also known as the BSD socket API) originated with the 4.2BSD Unix operating system (released in 1983) as an API. Only in 1989, however, could UC Berkeley release versions of its operating system and networking library free from the licensing constraints of AT&T's copyright-protected Unix.

The Berkeley socket API forms the de facto standard abstraction for network sockets. Most other programing languages use an interface similar to the C API.

The STREAMS-based Transport Layer Interface (TLI) API offers an alternative to the socket API. However, the Berkeley socket API predominates convincingly in popularity and in the number of implementations.

Contents

[edit] Berkeley socket interface

The Berkeley socket interface, an API, allows communications between hosts or between processes on one computer, using the concept of an internet socket. It can work with many different I/O devices and drivers, although support for these depends on the operating-system implementation. This interface implementation is implicit for TCP/IP, and it is therefore one of the fundamental technologies underlying the Internet. It was first developed at the University of California, Berkeley for use on Unix systems. All modern operating systems now have some implementation of the Berkeley socket interface, as it became the standard interface for connecting to the Internet.

Programmers can make the socket interfaces accessible at three different levels, most powerfully and fundamentally at the raw socket level. Very few applications need the degree of control over outgoing communications that this provides, so raw sockets support was intended to be available only on computers used for developing Internet-related technologies. In recent years, most operating systems have implemented support for it anyway, including Windows XP.

[edit] The header files

The Berkeley socket development library has many associated header files. They include:

<sys/socket.h>
Core BSD socket functions and data structures.
<netinet/in.h>
AF_INET and AF_INET6 address families. Widely used on the Internet, these include IP addresses and TCP and UDP port numbers.
<sys/un.h>
AF_UNIX address family. Used for local communication between programs running on the same computer. Not used on networks.
<arpa/inet.h>
Functions for manipulating numeric IP addresses.
<netdb.h>
Functions for translating protocol names and host names into numeric addresses. Searches local data as well as DNS.

[edit] TCP

TCP provides the concept of a connection. A process creates a TCP socket by calling the socket() function with the parameters PF_INET or PF_INET6 and SOCK_STREAM (Stream Sockets).

[edit] Server

Setting up a simple TCP server involves the following steps:

  • Creating a TCP socket, with a call to socket().
  • Binding the socket to the listen port, with a call to bind(). Before calling bind(), a programmer must declare a sockaddr_in structure, clear it (with bzero() or memset()), and the sin_family (AF_INET or AF_INET6), and fill its sin_port (the listening port, in network byte order) fields. Converting a short int to network byte order can be done by calling the function htons() (host to network short).
  • Preparing the socket to listen for connections (making it a listening socket), with a call to listen().
  • Accepting incoming connections, via a call to accept(). This blocks until an incoming connection is received, and then returns a socket descriptor for the accepted connection. The initial descriptor remains a listening descriptor, and accept() can be called again at any time with this socket, until it is closed.
  • Communicating with the remote host, which can be done through send() and recv() or write() and read().
  • Eventually closing each socket that was opened, once it is no longer needed, using close(). Note that if there were any calls to fork(), each process must close the sockets it knew about (the kernel keeps track of how many processes have a descriptor open), and two processes should not use the same socket at once.
// Server code in C
 
  #include <sys/types.h>
  #include <sys/socket.h>
  #include <netinet/in.h>
  #include <arpa/inet.h>
  #include <stdio.h>
  #include <stdlib.h>
  #include <strings.h>
  #include <unistd.h>
 
  int main()
  {
    int32_t i32SocketFD = socket(PF_INET, SOCK_STREAM, 0);
 
    if(-1 == i32SocketFD)
    {
      printf("can not create socket");
      exit(-1);
    }
 
    struct sockaddr_in stSockAddr;
    bzero(&stSockAddr, sizeof(stSockAddr));
 
    stSockAddr.sin_family = AF_INET;
    stSockAddr.sin_port = htons(1100);
    stSockAddr.sin_addr.s_addr = htonl(INADDR_ANY);
 
    if(-1 == bind(i32SocketFD,(struct sockaddr*) &stSockAddr, sizeof(stSockAddr)))
    {
      printf("error bind failed");
      exit(-1);
    }
 
    if(-1 == listen(i32SocketFD, 10))
    {
      printf("error listen failed");
      exit(-1);
    }
 
    for(; ;)
    {
      int32_t i32ConnectFD = accept(i32SocketFD, NULL, NULL);
 
      if(0 > i32ConnectFD)
      {
        printf("error accept failed");
        exit(-1);
      }
 
     // perform read write operations ...
 
      shutdown(i32ConnectFD, 2);
 
      close(i32ConnectFD);
    }
    return 0;
  }

[edit] Client

Setting up a TCP client involves the following steps:

  • Creating a TCP socket, with a call to socket().
  • Connecting to the server with the use of connect(), passing a sockaddr_in structure with the sin_family set to AF_INET or AF_INET6, sin_port set to the port the endpoint is listening (in network byte order), and sin_addr set to the IPv4 or IPv6 address of the listening server (also in network byte order.)
  • Communicating with the server by using send() and recv() or write() and read().
  • Terminating the connection and cleaning up with a call to close(). Again, if there were any calls to fork(), each process must close() the socket.
// Client code in C
 
  #include <sys/types.h>
  #include <sys/socket.h>
  #include <netinet/in.h>
  #include <arpa/inet.h>
  #include <stdio.h>
  #include <stdlib.h>
  #include <strings.h>
  #include <unistd.h>
 
  int main()
  {
    int32_t i32SocketFD = socket(PF_INET, SOCK_STREAM, 0);
 
    if(-1 == i32SocketFD)
    {
      printf("cannot create socket");
      exit(-1);
    }
 
    struct sockaddr_in stSockAddr;
    bzero(&stSockAddr, sizeof(stSockAddr));
 
    stSockAddr.sin_family = AF_INET;
    stSockAddr.sin_port = htons(1101);
    int32_t i32Res = inet_pton(AF_INET, "192.168.1.3",
        (void *) &stSockAddr.sin_addr);
 
    if(0 > i32Res)
    {
      printf("error: first parameter is not a valid address family");
      exit(-1);
    }
    else if(0 == i32Res)
    {
      printf("char string (second parameter does not contain valid ipaddress");
      exit(-1);
    }
 
    if(-1 == connect(i32SocketFD,(struct sockaddr*) &stSockAddr, sizeof(stSockAddr)))
    {
      printf("connect failed");
      exit(-1);
    }
 
    // perform read write operations ...
 
    shutdown(i32SocketFD, 2);
 
    close(i32SocketFD);
 
    return 0;
  }

[edit] UDP

UDP consists of a connectionless protocol with no guarantee of delivery. UDP packets may arrive out of order, become duplicated and arrive more than once, or even not arrive at all. Due to the minimal guarantees involved, UDP has considerably less overhead than TCP. Being connectionless means that there is no concept of a stream or connection between two hosts, instead, data arrives in datagrams (Datagram Sockets).

UDP address space, the space of UDP port numbers (in ISO terminology, the TSAPs), is completely disjoint from that of TCP ports.

[edit] Server

Code may set up a UDP server on port 7654 as follows:

sock = socket(PF_INET,SOCK_DGRAM,IPPROTO_UDP); 
 
 sa.sin_addr.s_addr = INADDR_ANY;
 sa.sin_port = htons(7654);
 
 bound = bind(sock,(struct sockaddr *)&sa, sizeof(struct sockaddr));
 if (bound < 0)
     fprintf(stderr, "bind(): %s\n",strerror(errno));

bind() binds the socket to an address/port pair.

while (1) {
     printf ("recv test....\n");
     recsize = recvfrom(sock, (void *)Hz, 100, 0, (struct sockaddr *)&sa, fromlen);
     printf ("recsize: %d\n ",recsize);
     if (recsize < 0)
         fprintf(stderr, "%s\n", strerror(errno));
     sleep(1);
     printf("datagram: %s\n",Hz);
 }

This infinite loop receives any UDP datagrams to port 7654 using recvfrom(). It uses the parameters:

  • socket
  • pointer to buffer for data
  • size of buffer
  • flags (same as in read or other receive socket function)
  • address struct of sending peer
  • length of address struct of sending peer.

[edit] Client

A simple demo to send a UDP packet containing "Hello World!" to address 127.0.0.1, port 7654 might look like this:

#include <stdio.h>
 #include <errno.h>
 #include <string.h>
 #include <sys/socket.h>
 #include <sys/types.h>
 #include <netinet/in.h>
 #include <unistd.h>//for close() for socket
 
 int main(int argc, char *argv[])
 {
   int sock;
   struct sockaddr_in sa;
   int bytes_sent, buffer_length;
   char buffer[200];
 
   sprintf(buffer, "Hello World!");
   buffer_length = strlen(buffer) + 1;
 
   sock = socket(PF_INET, SOCK_DGRAM, IPPROTO_UDP);
   if(-1 == sock)//if socket failed to initialize, exit
   {
      printf("Error Creating Socket");
      return 0;
   }
 
   sa.sin_family = AF_INET;
   sa.sin_addr.s_addr = htonl(0x7F000001);
   sa.sin_port = htons(7654);
 
   bytes_sent = sendto(sock, buffer, buffer_length, 0,(struct sockaddr*) &sa, sizeof(struct sockaddr_in) );
   if(bytes_sent < 0)
     printf("Error sending packet: %s\n", strerror(errno) );
 
   close(sock);//close the socket
   return 0;
 }

In this code, buffer provides a pointer to the data to send, and buffer_length specifies the size of the buffer contents.

[edit] Functions

[edit] socket()

socket() creates an endpoint for communication and returns a descriptor. socket() takes three arguments:

  • domain, which specifies the protocol family of the created socket. For example:
    • PF_INET for network protocol IPv4 or
    • PF_INET6 for IPv6.
    • PF_UNIX for local socket (using a file).
  • type, one of:
    • SOCK_STREAM (reliable stream-oriented service or Stream Sockets)
    • SOCK_DGRAM (datagram service or Datagram Sockets)
    • SOCK_SEQPACKET (reliable sequenced packet service), or
    • SOCK_RAW (raw protocols atop the network layer).
  • protocol, usually set to IPPROTO_IP, which is defined as 0, to represent the default transport protocol for the specified domain and type values (TCP for PF_INET or PF_INET6 and SOCK_STREAM, UDP for those PF_ values and SOCK_DGRAM), but which can also explicitly specify a protocol. These protocols are specified in <netinet/in.h>.

The function returns -1 if an error occurred. Otherwise, it returns an integer representing the newly-assigned descriptor.

[edit] Prototype

#include <sys/types.h>
#include <sys/socket.h>
int socket(int domain, int type, int protocol);

[edit] gethostbyname() and gethostbyaddr()

The gethostbyname() and gethostbyaddr() functions each return a pointer to an object of type struct hostent, which describes an internet host referenced by name or by address, respectively. This structure contains either the information obtained from a name server (ex: named), or broken-out fields from a line in /etc/hosts. If the local name server is not running these routines do a lookup in /etc/hosts. The functions take the following arguments:

  • name, which specifies the name of the host. For example: www.wikipedia.org
  • addr, which specifies a pointer to a struct in_addr containing the address of the host.
  • len, which specifies the length, in bytes, of addr.
  • type, which specifies the domain type of the host address. For example: AF_INET

The functions return a NULL pointer in case of error, in which case the external integer h_errno may be checked so see whether this is a temporary failure or an invalid or unknown host. Otherwise a valid struct hostent * is returned.

[edit] Prototypes

struct hostent *gethostbyname(const char *name);
struct hostent *gethostbyaddr(const void *addr, int len, int type);

[edit] connect()

connect() It returns an integer representing the error code: 0 represents success, while -1 represents an error.

Certain types of sockets are connectionless, most commonly user datagram protocol sockets. For these sockets, connect takes on a special meaning: the default target for sending and receiving data gets set to the given address, allowing the use of functions such as send() and recv() on connectionless sockets.

[edit] Prototype

#include <sys/types.h>
#include <sys/socket.h>
int connect(int sockfd, const struct sockaddr *serv_addr, socklen_t addrlen);

[edit] bind()

bind() assigns a socket an address. When a socket is created using socket(), it is given an address family, but not assigned an address. Before a socket may accept incoming connections, it must be bound. bind() takes three arguments:

  • sockfd, a descriptor representing the socket to perform the bind on
  • serv_addr, a pointer to a sockaddr structure representing the address to bind to.
  • addrlen, a socklen_t field representing the length of the sockaddr structure.

It returns 0 on success and -1 if an error occurs.

[edit] Prototype

#include <sys/types.h>
#include <sys/socket.h>
int bind(int sockfd, struct sockaddr *my_addr, socklen_t addrlen);

[edit] listen()

listen() prepares a bound socket to accept incoming connections. This function is only applicable to the SOCK_STREAM and SOCK_SEQPACKET socket types. It takes two arguments:

  • sockfd, a valid socket descriptor.
  • backlog, an integer representing the number of pending connections that can be queued up at any one time. The operating system usually places a cap on this value.

Once a connection is accepted, it is dequeued. On success, 0 is returned. If an error occurs, -1 is returned.

[edit] Prototype

#include <sys/socket.h>
int listen(int sockfd, int backlog);

[edit] accept()

accept() is used to accept a connection request from a remote host. It takes the following arguments:

  • sockfd, the descriptor of the listening socket to accept the connection from.
  • cliaddr, a pointer to the sockaddr structure that accept() should put the client's address information into.
  • addrlen, a pointer to the socklen_t integer that will indicate to accept() how large the sockaddr structure pointed to by cliaddr is. When accept() returns, the socklen_t integer then indicates how many bytes of the cliaddr structure were actually used.

The function returns a socket corresponding to the accepted connection, or -1 if an error occurs.

[edit] Prototype

#include <sys/types.h>
#include <sys/socket.h>
int accept(int sockfd, struct sockaddr *cliaddr, socklen_t *addrlen);

[edit] Blocking vs. nonblocking

Berkeley sockets can operate in one of two modes: blocking or non-blocking. A blocking socket will not return control until it has sent (or received) all the data specified for the operation. This is true only in Linux systems. In other systems, such as FreeBSD, it is normal for a blocking socket not to send all the data. Always check the return value to find out how many bytes have been sent/received. It’s up to you to send the rest of the string[1]. It also may cause problems if a socket continues to listen: a program may hang as the socket waits for data that may never arrive.

A socket is typically set to blocking or nonblocking mode using the fcntl() or ioctl() functions.

[edit] Cleaning up

The system will not release the resources allocated by the socket() call until a close() call occurs. This is especially important if the connect() call fails and may be retried. Each call to socket() must have a matching call to close() in all possible execution paths. Include <unistd.h> for the close function.

When an issue to the close() system call is made only the interface to the socket gets closed, not the socket itself. It is up to the kernel to close the socket. Sometimes, for really technical reasons, the socket is kept alive for a few minutes after you close it. It is normal, for example for the socket to go into a TIME_WAIT state, on the server side, for a few minutes. The official standard says that it should be 4 minutes.[2]

[edit] See also

[edit] References

The "de jure" standard definition of the Sockets interface is contained in the POSIX standard, known as:

  • IEEE Std. 1003.1-2001 Standard for Information Technology -- Portable Operating System Interface (POSIX).
  • Open Group Technical Standard: Base Specifications, Issue 6, December 2001.
  • ISO/IEC 9945:2002

Information about this standard and ongoing work on it is available from the Austin website.

The IPv6 extensions to the base socket API are documented in RFC 3493 and RFC 3542.

[edit] External links

This article was originally based on material from the Free On-line Dictionary of Computing, which is licensed under the GFDL.