Copyright (c) 1983 Regents of the University of California. All rights reserved. The Berkeley software License Agreement specifies the terms and conditions for redistribution. Copyright 1989 AT&T Copyright (c) 2000, Sun Microsystems, Inc. All Rights Reserved.
cc [ flag ... ] file ... -lsocket -lnsl [ library ... ] #include <netdb.h> struct protoent *getprotobyname(const char *name);
struct protoent *getprotobyname_r(const char *name, struct protoent *result, char *buffer, int buflen);
struct protoent *getprotobynumber(int proto);
struct protoent *getprotobynumber_r(int proto, struct protoent *result, char *buffer, int buflen);
struct protoent *getprotoent(void);
struct protoent *getprotoent_r(struct protoent *result, char *buffer, int buflen);
int setprotoent(int stayopen);
int endprotoent(void);
These functions return a protocol entry. Two types of interfaces are supported: reentrant (getprotobyname_r(), getprotobynumber_r(), and getprotoent_r()) and non-reentrant (getprotobyname(), getprotobynumber(), and getprotoent()). The reentrant functions can be used in single-threaded applications and are safe for multithreaded applications, making them the preferred interfaces.
The reentrant routines require additional parameters which are used to return results data. result is a pointer to a struct protoent structure and will be where the returned results will be stored. buffer is used as storage space for elements of the returned results. buflen is the size of buffer and should be large enough to contain all returned data. buflen must be at least 1024 bytes.
getprotobyname_r(), getprotobynumber_r(), and getprotoent_r() each return a protocol entry.
The entry may come from one of the following sources: the protocols file (see protocols(5)), the NIS maps ``protocols.byname'' and ``protocols.bynumber''. The sources and their lookup order are specified in the /etc/nsswitch.conf file (see nsswitch.conf(5) for details). Some name services such as NIS will return only one name for a host, whereas others such as DNS will return all aliases.
The getprotobyname_r() and getprotobynumber_r() functions sequentially search from the beginning of the file until a matching protocol name or protocol number is found, or until an EOF is encountered.
getprotobyname() and getprotobynumber() have the same functionality as getprotobyname_r() and getprotobynumber_r() except that a static buffer is used to store returned results. These functions are Unsafe in a multithreaded application.
getprotoent_r() enumerates protocol entries: successive calls to getprotoent_r() will return either successive protocol entries or NULL. Enumeration might not be supported by some sources. If multiple threads call getprotoent_r(), each will retrieve a subset of the protocol database.
getprotent() has the same functionality as getprotent_r() except that a static buffer is used to store returned results. This routine is unsafe in a multithreaded application.
setprotoent() "rewinds" to the beginning of the enumeration of protocol entries. If the stayopen flag is non-zero, resources such as open file descriptors are not deallocated after each call to getprotobynumber_r() and getprotobyname_r(). Calls to getprotobyname_r() , The getprotobyname(), getprotobynumber_r(), and getprotobynumber() functions might leave the enumeration in an indeterminate state, so setprotoent() should be called before the first call to getprotoent_r() or getprotoent(). The setprotoent() function has process-wide scope, and ``rewinds'' the protocol entries for all threads calling getprotoent_r() as well as main-thread calls to getprotoent().
The endprotoent() function can be called to indicate that protocol processing is complete; the system may then close any open protocols file, deallocate storage, and so forth. It is legitimate, but possibly less efficient, to call more protocol functions after endprotoent().
The internal representation of a protocol entry is a protoent structure defined in <netdb.h> with the following members:
char *p_name; char **p_aliases; int p_proto;
The getprotobyname_r(), getprotobyname(), getprotobynumber_r(), and getprotobynumber() functions return a pointer to a struct protoent if they successfully locate the requested entry; otherwise they return NULL.
The getprotoent_r() and getprotoent() functions return a pointer to a struct protoent if they successfully enumerate an entry; otherwise they return NULL, indicating the end of the enumeration.
The getprotobyname_r(), getprotobynumber_r(), and getprotoent_r() functions will fail if: ERANGE
The length of the buffer supplied by the caller is not large enough to store the result.
See attributes(7) for descriptions of the following attributes:
| ATTRIBUTE TYPE ATTRIBUTE VALUE |
| MT-Level See NOTES below. |
Intro (3), netdb.h (3HEAD), nsswitch.conf (5), protocols (5), attributes (7)
Although getprotobyname_r(), getprotobynumber_r(), and getprotoent_r() are not mentioned by POSIX 1003.1:2001, they were added to complete the functionality provided by similar thread-safe functions.
When compiling multithreaded applications, see Intro(3), Notes On Multithread Applications, for information about the use of the _REENTRANT flag.
The getprotobyname_r(), getprotobynumber_r(), and getprotoent_r() functions are reentrant and multithread safe. The reentrant interfaces can be used in single-threaded as well as multithreaded applications and are therefore the preferred interfaces.
The getprotobyname(), getprotobyaddr(), and getprotoent() functions use static storage, so returned data must be copied if it is to be saved. Because of their use of static storage for returned data, these functions are not safe for multithreaded applications.
The setprotoent() and endprotoent() functions have process-wide scope, and are therefore not safe in multi-threaded applications.
Use of getprotoent_r() and getprotoent() is discouraged; enumeration is well-defined for the protocols file and is supported (albeit inefficiently) for NIS but in general may not be well-defined. The semantics of enumeration are discussed in nsswitch.conf(5).
Only the Internet protocols are currently understood.