Copyright (c)1996-2006 by Hartmut Brandt
All rights reserved.

Author: harti@freebsd.org <Hartmut Brandt>

Redistribution of this software and documentation and use in source and
binary forms, with or without modification, are permitted provided that
the following conditions are met:

1. Redistributions of source code or documentation must retain the above
copyright notice, this list of conditions and the following disclaimer.
2. Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.

THIS SOFTWARE AND DOCUMENTATION IS PROVIDED BY THE AUTHOR
AND ITS CONTRIBUTORS ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES,
INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL
THE AUTHOR OR ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA,
OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

$Begemot: libbegemot/rpoll.man,v 1.4 2004/09/21 15:59:00 brandt Exp $

rpoll 3 "8 Dec 2006" "BEGEMOT" "BEGEMOT Library"
NAME
rpoll - callback functions for file descriptors and timers
SYNOPSIS

"# include <rpoll.h>"

"typedef void (*poll_f)(int " "fd" ", int " "mask" ", void *" "arg" ");"

"typedef void (*timer_f)(int " "tid" ", void *" "arg" ");"

"int poll_register(int " "fd" ", poll_f " "func" ", void *" "arg" ", int " "mask" ");"

"void poll_unregister(int " "handle" ");"

"int poll_start_timer(u_int " "msecs" ", int " "repeat" ", timer_f " "func" "," "void *" "arg" ");"

"void poll_stop_timer(int " "handle" ");"

"int poll_start_utimer(unsigned long long " "usecs" ", int " "repeat" ", "timer_f " "func" ", void *" "arg" ");"

"void poll_dispatch(int " "wait" ");"

DESCRIPTION
Many programs need to read from several file descriptors at the same time. Typically in these programs one of select (3c) or poll (2) is used. These calls are however clumsy to use and the usage of one of these calls is probably not portable to other systems - not all systems support both calls.

The rpoll (l) family of functions is designed to overcome these restrictions. They support the well known and understood technique of event driven programing and, in addition to select (3c) and poll (2) also support timers.

Each event on a file descriptor or each timer event is translated into a call to a user defined callback function. These functions need to be registered. A file descriptor is registered with poll_register . fd is the file descriptor to watch, mask is an event mask. It may be any combination of POLL_IN to get informed when input on the file descriptor is possible, POLL_OUT to get informed when output is possible or POLL_EXCEPT to get informed when an exceptional condition occures. An example of an exceptional condition is the arrival of urgent data. (Note, that an end of file condition is signaled via POLL_IN). func is the user function to be called and arg is a user supplied argument for this function. The callback functions is called with the file descriptor, a mask describing the actual events (from the set supplied in the registration) and the user argument. poll_register returns a handle, which may be used later to de-register the file descriptor. A file descriptor may be registered more than once, if the function, the user arguments or both differ in the call to poll_register . If func and arg are the same, then no new registration is done, instead the event mask of the registration is changed to reflect the new mask.

A registered file descriptor may be de-registered by calling poll_unregister with the handle returned by poll_register .

A timer is created with poll_start_timer or poll_start_utimer . msecs is the number of milliseconds in poll_start_timer while usecs is the number of microseconds in poll_start_utimer , after which the timer event will be generated. If the functions use the poll (2) system call, then usecs is rounded to milliseconds and poll_start_timer is called. repeat selects one-short behavior (if 0) or a repeatable timer (if not 0). A one-short timer will automatically unregistered after expiry. func is the user function which will be called with a timer id and the user supplied arg . poll_start_timer and poll_start_utimer return a timer id, which may be used to cancel the timer with poll_stop_timer . A one-short timer should be canceled only if it has not yet fired.

poll_dispatch must be called to actually dispatch events. wait is a flag, which should be 0, if only a poll should be done. In this case, the function returns, after polling the registered file descriptors and timers. If wait is not 0, poll_dispatch waits until an event occures. All events are dispatch (i.e. callback functions called) and poll_dispatch returns.

Typical use is:

while(1)
 poll_dispatch(1);
"SEE ALSO"
poll (2), select (3C)
"RETURN VALUES"
poll_register , poll_start_timer and poll_start_utimer return a handle which may be used to unregister the file descriptor or cancel the timer.

Both functions and poll_dispatch call xrealloc (l) and can end in panic (l).

"ERRORS"
System call or memory allocation errors are fatal and are handle by calling panic (l). The one exception is a return of EINTR from select (3c) or poll (2) in poll_dispatch . In this case poll_dispatch simply returns.
"BUGS"
Obscure sequences of poll_start_timer and poll_stop_timer in callback functions may probably break the code.

The semantics of POLL_EXCEPT are not clear.

AUTHORS
Hartmut Brandt, harti@freebsd.org