BSD Sockets API Macintosh Development [Home] [About Us] [People] [Information Systems] [Kerberos for Macintosh] [Applications] [Miscellaneous Documentation] BSD Sockets API int socket (int family, int type, int protocol); Creates a socket. Currently family can only be AF_INET (OT does not support IPv6 so there is no AF_INET6 support). protocol can be PF_INET or PF_UNSPEC (both have the same effect of creating an internet socket. type can be SOCK_STREAM for TCP sockets or SOCK_DGRAM for UDP sockets. socket() returns a socket file descriptor (sockFD) which is a small non-negative integer. This file descriptor number should be used for all other socket operations on that socket. If socket() encounters an error, it will return -1, in which case the application should call GetMITLibError() to get the error code. NOTE: errno is not supported by the Sockets Library. Use ErrorLib. int socket_bind (int sockFD, const struct sockaddr *myAddr, int addrLength); Binds a socket to a local port and/or IP address. Either the port or the IP address in the struct sockaddr structure may be wildcarded in which case the library will select appropriate values. To wildcard the port set the sin_port field of the address to 0. To wildcard the IP, set the sin_addr.s_addr field of the address to INADDR_ANY TCP has a port reuse time limit. If an application attempts to bind to the same local port twice in rapid succession, the second attempt will fail with EADDRINUSE. Just wait and try again. socket_bind() returns 0 on success and -1 on failure. If -1 is returned, call GetMITLibError() to get the error code. Note that the second argument to socket_bind() is actually a struct sockaddr_in* which is cast to the more general struct sockaddr* because internet addresses are used by the Socket Library. Also, it is not necessary to bind a socket prior to connecting it. If a socket is not bound the library will choose the local port and IP. int socket_connect (int sockFD, struct sockaddr *servAddr, int addrLength); Connects a socket to a remote host on a given port. If this is a SOCK_STREAM (TCP) socket, socket_connect() will actually perform TCP negotation to open a connection. If this is a SOCK_DGRAM (UDP) socket, socket_connect() will just store the address for use later with other socket operations (this is important to note because if sockFD refers to a UDP socket, errors will not be reported prior to an attempt to read or write data on the socket). socket_connect() returns 0 on success and -1 on failure. If -1 is returned, call GetMITLibError() to get the error code. Note that the second argument to connect is actually a struct sockaddr_in* which is cast to the more general struct sockaddr* because internet addresses are used by the Socket Library. int socket_select (int maxFDsExamined, fd_set *readFDs, fd_set *writeFDs, fd_set *exceptFDs, struct timeval *timeOut); socket_select() allows for conditions to be checked on one or more sockets. maxFDsExamined should be one greater than value of the largest valued socket descriptor in any of the lists. If you don't wish to calculate this, use FD_SETSIZE instead. For small numbers of sockets, socket_select() will be less efficient if FD_SETSIZE is passed. If the bit corresponding to a socket in an fd_set is set, that socket will be checked for the condition the fd_set corresponds to. If the condition they are checked for is true, they will still be set to true when socket_select() returns (otherwise they will be set to false). Note that the sets are modified by socket_select(): thus they must be reset between each call to the function. fd_sets can be manipulated using the following macros: FD_SET(fd, fdset) Sets socket fd in fdset to true. FD_CLR(fd, fdset) Sets socket fd in fdset to false. FD_ISSET(fd, fdset) Returns true iff socket fd is set to true in fdset. FD_ZERO(fdset) Sets all the sockets in fdset to false. Currently only the readfds condition (whether there is data to read on a socket) is supported. However, in order to stay compatible with most clients, writefds (whether there is room in the kernel buffers to write to a socket) behaves as though writing data will succeed (this is usually fine) and exceptfds behaves as though there are no exception conditions on the socket (exceptfds will always be returned with all sockets set to false). If timeout is NULL, socket_select() blocks until one of the conditions becomes true for one of the sockets. If timeout is non-NULL, socket_select() checks for the amount of time corresponding to timeout and then returns (regardless of whether it has found any conditions to be true). If one of the conditions becomes true for one of the sockets it finds before the timeout has expired, it always returns immediately. To use socket_select() in non-blocking mode call it with a non-null timeOut whose tv_sec and tv_usec fields are both set to zero. socket_select() returns the number of sockets for which the specified conditions are true. If it encounters an error, it will return -1 (in which case GetMITLibError() can be called to retrieve the error code.) int socket_fcntl (int sockFD, int command, int flags); socket_fcntl() sets various options on a socket. Currently the only flag supported is O_NONBLOCK which sets a socket to non-blocking mode. If socket_fcntl() is called with command equal to F_GETFL it will return the current flags for the socket sockFD. The parameter flags is ignored in this case. If the function is called with command equal to F_SETFL it will replace the socket's flags with those specified by flags. The correct way to change a flag on a socket is as illustrated in the following example: flags = socket_fcntl(sockfd, F_GETFL, 0); flags |= O_NONBLOCK; err = socket_fcntl(sockfd, F_SETFL, flags); Getting the current flags and modifying them avoids the potential error of clearing other flags. socket_fcntl() returns 0 on success and -1 on failure. If -1 is returned, call GetMITLibError() to get the error code. Note: the corresponding BSD call, fcntl takes a variable number of arguments while socket_fcntl does not. int socket_getpeername (int sockFD, struct sockaddr *peerAddr, int *addrLength); socket_getpeername() gets the address of the remote host the socket is connected to, if any. If the socket is not connected, GetMITLibError() will return ENOTCONN. socket_getpeername() returns 0 on success and -1 on failure. If -1 is returned, call GetMITLibError() to get the error code. int socket_getsockname (int sockFD, struct sockaddr *localAddr, int *addrLength); socket_getsockname() gets the socket's local address and port. socket_getsockname() returns 0 on success and -1 on failure. If -1 is returned, call GetMITLibError() to get the error code. int socket_read (int sockFD, void *buffer, UInt32 numBytes); Reads data from the socket into buffer. numBytes should be the size of the buffer. socket_read() may not fill the entire buffer. socket_read() returns the amount of data which was read. If there is an error, -1 is returned and GetMITLibError() can be called to retrieve the error code. If 0 is returned, this means that the socket received an EOF (the remote host closed the connection gracefully.) To perform a full read on a socket, continue to call socket_read() until the desired number of bytes have been accumulated. Note that socket_read() may block if no data is available to be read. This condition can be checked using socket_select(). The socket must be connected. Note: the standard library call read() is not supported for sockets. int socket_write (int sockFD, void *buffer, UInt32 numBytes); Writes data to the socket from buffer. numBytes should be the amount of data in the buffer. socket_write() may not write out the entire buffer. socket_write() returns the amount of data which was written. If there is an error, -1 is returned and GetMITLibError() can be called to get the error code. The socket must be connected. int socket_readv (int sockFD, struct iovec *iov, UInt32 iovCount); The socket_readv() function is equivalent to socket_read(), but places the input data into the iovcnt buffers specified by the members of the iov array: iov0, iov1, ..., iov[iovcnt-1]. The iovcnt argument is valid if greater than 0 and less than or equal to IOV_MAX. The iovec structure contains the following members: caddr_t iov_base; int iov_len; Each iovec entry specifies the base address and length of an area in memory where data should be placed. The socket_readv() function always fills an area completely before proceeding to the next. Upon successful completion, socket_readv() marks for update the st_atime field of the file. socket_readv() returns the total amount of data which was read into all buffers. If there is an error, -1 is returned and GetMITLibError() can be called to get the error code. If 0 is returned, this means that the socket got an EOF (the remote host closed the connection gracefully. The socket must be connected. int socket_writev (int sockFD, struct iovec *iov, UInt32 iovCount); The socket_writev() function performs the same action as socket_write(), but gathers the output data from the iovcnt buffers specified by the members of the iov array: iov[0], iov[1], ..., iov[iovcnt-1]. The iovcnt buffer is valid if greater than 0 and less than or equal to IOV_MAX. The iovec structure contains the following members: caddr_t iov_base; int iov_len; Each iovec entry specifies the base address and length of an area in memory from which data should be written. socket_writev() always writes all data from an area before proceeding to the next. socket_writev() returns the amount of data which was written. If there is an error, -1 is returned and GetMITLibError() can be called to get the error code. The socket must be connected. int socket_recv (int sockFD, void *buffer, UInt32 numBytes, int flags); This function is similiar to socket_read() with the addition of a final parameter. socket_recv() reads data from the socket into buffer. numBytes should be the size of the buffer. socket_recv() may not fill the entire buffer. If flags is set to MSG_DONTWAIT, then socket_recv will not block if not data is available. socket_recv() returns the amount of data which was read. If there is an error, -1 is returned and GetMITLibError() can be called to get the error code. If 0 is returned, this means that the socket received an EOF (the remote host closed the connection gracefully. The socket must be connected. int socket_send (int sockFD, void *buffer, UInt32 numBytes, int flags); Similiar to socket_write(), socket_send() writes data to the socket from buffer. numBytes should be the amount of data in the buffer. socket_send() may not write out the entire buffer. If flags is set to MSG_DONTWAIT, then socket_send() will not block waiting for buffers to become free. socket_send() returns the amount of data which was written. If there is an error, -1 is returned and GetMITLibError() can be called to receive the error code. The socket must be connected. int socket_recvfrom (int sockFD, void *buffer, UInt32 numBytes, int flags, struct sockaddr *fromAddr, socklen_t *addrLength); Reads data from the remote host specified by fromAddr into buffer. The socket must be a SOCK_DGRAM (UDP) socket. numBytes should be the size of the buffer. socket_recvfrom() may not fill the entire buffer. If flags is set to MSG_DONTWAIT, then socket_recvfrom() will not block waiting for data. Note that fromAddr will actually be a struct sockaddr_in* socket_recvfrom() returns the amount of data which was read. If there is an error, -1 is returned and GetMITLibError() may be called to get the error code. If 0 is returned, this means that the socket got an EOF (the remote host closed the connection gracefully. int socket_sendto (int sockFD, void *buffer, UInt32 numBytes, int flags, struct sockaddr *toAddr, socklen_t addrLength); Writes data the remote host specified by fromAddr into buffer. The socket must be a SOCK_DGRAM (UDP) socket. numBytes should be the amount of data in the buffer. socket_sendto() may not write out the entire buffer. If flags is set to MSG_DONTWAIT, then socket_sendto() will not block waiting for buffers to become free. Note that toAddr will actually be a struct sockaddr_in* socket_sendto() returns the amount of data which was written. If there is an error, -1 is returned and GetMITLibError() can be called to get the error code. int socket_shutdown (int sockFD, int howTo); socket_shutdown() closes one or both directions of a connected socket. howTo can be SHUT_RD, SHUT_WR or SHUT_RDWR. SHUT_RD tells it to close the reading side of the connection (reads from the socket will no longer be possible). SHUT_WR tells it to close the writing half of the socket (this will cause it to send an orderly disconnect to the remote host, telling that host it no longer has anything to write). SHUT_RDWR tells it to close both halves of the connection (It is still necessary to free the socket with socket_close()). socket_shutdown() returns 0 on success and -1 on failure. If -1 is returned, call GetMITLibError() to get the error code. int socket_close (int sockFD); socket_close() frees a socket's resources, disconnecting it from the remote host, if necessary. Both TCP and UDP sockets should be closed with this function. socket_close() returns 0 on success and -1 on failure. If -1 is returned, call GetMITLibError() to get the error code. Questions or comments? Send mail to macdev@mit.edu Last updated on $Date: 2003/11/19 20:49:26 $ Last modified by $Author: smcguire $
