OSF

NAME

exit, atexit, _exit − Terminates a process

LIBRARY

Standard C Library (libc.a): atexit(), _exit()

SYNOPSIS

#include <stdlib.h> void exit (
        int status ); void _exit (
        int status ); int atexit (
        void  (∗function) ( void ) );

PARAMETERS

statusIndicates the status of the process. 

functionPoints to a function that is called at normal process termination for cleanup processing.  A push-down stack of functions is kept, such that the last function registered is the first function called.  Any function which is registered more than once will be repeated.  Up to 32 functions can be specified with atexit(). 

DESCRIPTION

The atexit() function registers functions to be called at normal process termination for cleanup processing. 

The exit() function terminates the calling process after calling the Standard I/O Library _cleanup() function to flush any buffered output.  Then it calls any functions registered previously for the process by the atexit() function, in the reverse order to that in which they were registered.  In addition, the exit() function flushes all open output streams, closes all open streams, and removes all files created by the tmpfile() function.  Finally, it calls the _exit() function, which completes process termination and does not return. 

The _exit() function terminates the calling process and causes the following to occur:

       •All of the file descriptors, directory streams, and message catalog descriptors open in the calling process are closed.  Since the exit() function terminates the process, any errors encountered during these close operations go unreported. 

       •Terminating a process by exiting does not terminate its child processes.  Instead, the parent process ID of all of the calling process child processes and zombie child processes is set to the process ID of init.  The init process thus inherits each of these processes, catches the SIGCHLD signals they generate, and calls the wait() function for each of them. 

       •If the parent process of the calling process is running a wait() or waitpid() function, it is notified of the termination of the calling process and the low-order 8 bits (that is, bits 0377 or 0xFF) of the status parameter are made available to it. 

       •If the parent process is not running a wait() or waitpid() function when the child process terminates, it may do so later on, and the child’s status will be returned to it at that time.  Meanwhile, the child process is transformed into a zombie process, and its parent process is sent a SIGCHLD signal to notify it of the termination of a child process.  A zombie process is a process that occupies a slot in the process table, but has no other space allocated to it either in user or kernel space. The process table slot that it occupies is partially overlaid with time accounting information to be used by the times() function. (See the sys/proc.h header file.)  A process remains a zombie until its parent issues one of the wait functions. At this time, the zombie is laid to rest, and its process table entry is released. 

       •The parent process is sent a SIGCHLD signal when a child terminates; however, since the default action for this signal is to ignore it, the signal usually is not seen.  If an exiting child’s parent is ignoring the SIGCHLD signal, the child’s parent process ID is changed to that of the initialization process, init, which will catch the SIGCHLD signal and call the wait() function. 

       •If the process is a controlling process, a SIGHUP signal is sent to each process in the foreground process group of the controlling terminal belonging to the calling process.  The controlling terminal is disassociated from the session, allowing it to be acquired by a new controlling process. 

       •If the exit of a process causes a process group to become orphaned, and if any member of the newly orphaned process group is stopped, then a SIGHUP signal is sent to each newly orphaned process. 

       •Each attached shared memory segment is detached and the value of shm_nattach in the data structure associated with its shared memory identifier is decremented by 1. 

       •For each semaphore for which the calling process has set a semadj value, that semadj value is added to the semval of the specified semaphore.  (The semop() function provides information about semaphore operations.) 

       •If the process has a process lock, text lock, or data lock, an unlock is performed. (See the plock() function.) 

       •An accounting record is written on the accounting file if the system accounting routine is enabled. (The acct() function provides information about enabling accounting routines.) 

       •Locks set by the fcntl(), flock(),and lockf() functions are removed. 

If a thread calls the _exit() function, the entire process exits and all threads within the process are terminated. 

NOTES

The system init process is used to assist cleanup of terminating processes. If the code for the init process is replaced, the program must be prepared to accept SIGCHLD signals and issue a wait() function for each. 

AES Support Level:
Full use

RETURN VALUES

The exit() function and _exit() function do not return.  The atexit() function returns 0 (zero) if successful, and a nonzero value if there has been an attempt to register more exit() functions than can be held in the atexit() array. 

RELATED INFORMATION

Functions: acct(2), sigaction(2), times(3), wait(2), sigvec(2)