1 <?xml version="1.0" encoding="UTF-8"?>
2 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">
3 <html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en">
6 <meta name="description" content="Pod documentation for libev" />
7 <meta name="inputfile" content="<standard input>" />
8 <meta name="outputfile" content="<standard output>" />
9 <meta name="created" content="Fri Nov 23 06:14:47 2007" />
10 <meta name="generator" content="Pod::Xhtml 1.57" />
11 <link rel="stylesheet" href="http://res.tst.eu/pod.css"/></head>
15 <h3 id="TOP">Index</h3>
17 <ul><li><a href="#NAME">NAME</a></li>
18 <li><a href="#SYNOPSIS">SYNOPSIS</a></li>
19 <li><a href="#DESCRIPTION">DESCRIPTION</a></li>
20 <li><a href="#FEATURES">FEATURES</a></li>
21 <li><a href="#CONVENTIONS">CONVENTIONS</a></li>
22 <li><a href="#TIME_REPRESENTATION">TIME REPRESENTATION</a></li>
23 <li><a href="#GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</a></li>
24 <li><a href="#FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</a></li>
25 <li><a href="#ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</a>
26 <ul><li><a href="#ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</a></li>
29 <li><a href="#WATCHER_TYPES">WATCHER TYPES</a>
30 <ul><li><a href="#code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable</a></li>
31 <li><a href="#code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally recurring timeouts</a></li>
32 <li><a href="#code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron</a></li>
33 <li><a href="#code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled</a></li>
34 <li><a href="#code_ev_child_code_wait_for_pid_stat"><code>ev_child</code> - wait for pid status changes</a></li>
35 <li><a href="#code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do</a></li>
36 <li><a href="#code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop</a></li>
39 <li><a href="#OTHER_FUNCTIONS">OTHER FUNCTIONS</a></li>
40 <li><a href="#LIBEVENT_EMULATION">LIBEVENT EMULATION</a></li>
41 <li><a href="#C_SUPPORT">C++ SUPPORT</a></li>
42 <li><a href="#AUTHOR">AUTHOR</a>
47 <h1 id="NAME">NAME</h1><p><a href="#TOP" class="toplink">Top</a></p>
48 <div id="NAME_CONTENT">
49 <p>libev - a high performance full-featured event loop written in C</p>
52 <h1 id="SYNOPSIS">SYNOPSIS</h1><p><a href="#TOP" class="toplink">Top</a></p>
53 <div id="SYNOPSIS_CONTENT">
54 <pre> #include <ev.h>
59 <h1 id="DESCRIPTION">DESCRIPTION</h1><p><a href="#TOP" class="toplink">Top</a></p>
60 <div id="DESCRIPTION_CONTENT">
61 <p>Libev is an event loop: you register interest in certain events (such as a
62 file descriptor being readable or a timeout occuring), and it will manage
63 these event sources and provide your program with events.</p>
64 <p>To do this, it must take more or less complete control over your process
65 (or thread) by executing the <i>event loop</i> handler, and will then
66 communicate events via a callback mechanism.</p>
67 <p>You register interest in certain events by registering so-called <i>event
68 watchers</i>, which are relatively small C structures you initialise with the
69 details of the event, and then hand it over to libev by <i>starting</i> the
73 <h1 id="FEATURES">FEATURES</h1><p><a href="#TOP" class="toplink">Top</a></p>
74 <div id="FEATURES_CONTENT">
75 <p>Libev supports select, poll, the linux-specific epoll and the bsd-specific
76 kqueue mechanisms for file descriptor events, relative timers, absolute
77 timers with customised rescheduling, signal events, process status change
78 events (related to SIGCHLD), and event watchers dealing with the event
79 loop mechanism itself (idle, prepare and check watchers). It also is quite
80 fast (see this <a href="http://libev.schmorp.de/bench.html">benchmark</a> comparing
81 it to libevent for example).</p>
84 <h1 id="CONVENTIONS">CONVENTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p>
85 <div id="CONVENTIONS_CONTENT">
86 <p>Libev is very configurable. In this manual the default configuration
87 will be described, which supports multiple event loops. For more info
88 about various configuration options please have a look at the file
89 <cite>README.embed</cite> in the libev distribution. If libev was configured without
90 support for multiple event loops, then all functions taking an initial
91 argument of name <code>loop</code> (which is always of type <code>struct ev_loop *</code>)
92 will not have this argument.</p>
95 <h1 id="TIME_REPRESENTATION">TIME REPRESENTATION</h1><p><a href="#TOP" class="toplink">Top</a></p>
96 <div id="TIME_REPRESENTATION_CONTENT">
97 <p>Libev represents time as a single floating point number, representing the
98 (fractional) number of seconds since the (POSIX) epoch (somewhere near
99 the beginning of 1970, details are complicated, don't ask). This type is
100 called <code>ev_tstamp</code>, which is what you should use too. It usually aliases
101 to the double type in C.</p>
104 <h1 id="GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p>
105 <div id="GLOBAL_FUNCTIONS_CONTENT">
106 <p>These functions can be called anytime, even before initialising the
107 library in any way.</p>
109 <dt>ev_tstamp ev_time ()</dt>
111 <p>Returns the current time as libev would use it. Please note that the
112 <code>ev_now</code> function is usually faster and also often returns the timestamp
113 you actually want to know.</p>
115 <dt>int ev_version_major ()</dt>
116 <dt>int ev_version_minor ()</dt>
118 <p>You can find out the major and minor version numbers of the library
119 you linked against by calling the functions <code>ev_version_major</code> and
120 <code>ev_version_minor</code>. If you want, you can compare against the global
121 symbols <code>EV_VERSION_MAJOR</code> and <code>EV_VERSION_MINOR</code>, which specify the
122 version of the library your program was compiled against.</p>
123 <p>Usually, it's a good idea to terminate if the major versions mismatch,
124 as this indicates an incompatible change. Minor versions are usually
125 compatible to older versions, so a larger minor version alone is usually
128 <dt>unsigned int ev_supported_backends ()</dt>
130 <p>Return the set of all backends (i.e. their corresponding <code>EV_BACKEND_*</code>
131 value) compiled into this binary of libev (independent of their
132 availability on the system you are running on). See <code>ev_default_loop</code> for
133 a description of the set values.</p>
135 <dt>unsigned int ev_recommended_backends ()</dt>
137 <p>Return the set of all backends compiled into this binary of libev and also
138 recommended for this platform. This set is often smaller than the one
139 returned by <code>ev_supported_backends</code>, as for example kqueue is broken on
140 most BSDs and will not be autodetected unless you explicitly request it
141 (assuming you know what you are doing). This is the set of backends that
142 <code>EVFLAG_AUTO</code> will probe for.</p>
144 <dt>ev_set_allocator (void *(*cb)(void *ptr, long size))</dt>
146 <p>Sets the allocation function to use (the prototype is similar to the
147 realloc C function, the semantics are identical). It is used to allocate
148 and free memory (no surprises here). If it returns zero when memory
149 needs to be allocated, the library might abort or take some potentially
150 destructive action. The default is your system realloc function.</p>
151 <p>You could override this function in high-availability programs to, say,
152 free some memory if it cannot allocate memory, to use a special allocator,
153 or even to sleep a while and retry until some memory is available.</p>
155 <dt>ev_set_syserr_cb (void (*cb)(const char *msg));</dt>
157 <p>Set the callback function to call on a retryable syscall error (such
158 as failed select, poll, epoll_wait). The message is a printable string
159 indicating the system call or subsystem causing the problem. If this
160 callback is set, then libev will expect it to remedy the sitution, no
161 matter what, when it returns. That is, libev will generally retry the
162 requested operation, or, if the condition doesn't go away, do bad stuff
168 <h1 id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</h1><p><a href="#TOP" class="toplink">Top</a></p>
169 <div id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP-2">
170 <p>An event loop is described by a <code>struct ev_loop *</code>. The library knows two
171 types of such loops, the <i>default</i> loop, which supports signals and child
172 events, and dynamically created loops which do not.</p>
173 <p>If you use threads, a common model is to run the default event loop
174 in your main thread (or in a separate thread) and for each thread you
175 create, you also create another event loop. Libev itself does no locking
176 whatsoever, so if you mix calls to the same event loop in different
177 threads, make sure you lock (this is usually a bad idea, though, even if
178 done correctly, because it's hideous and inefficient).</p>
180 <dt>struct ev_loop *ev_default_loop (unsigned int flags)</dt>
182 <p>This will initialise the default event loop if it hasn't been initialised
183 yet and return it. If the default loop could not be initialised, returns
184 false. If it already was initialised it simply returns it (and ignores the
185 flags. If that is troubling you, check <code>ev_backend ()</code> afterwards).</p>
186 <p>If you don't know what event loop to use, use the one returned from this
188 <p>The flags argument can be used to specify special behaviour or specific
189 backends to use, and is usually specified as <code>0</code> (or EVFLAG_AUTO).</p>
190 <p>It supports the following flags:</p>
193 <dt><code>EVFLAG_AUTO</code></dt>
195 <p>The default flags value. Use this if you have no clue (it's the right
196 thing, believe me).</p>
198 <dt><code>EVFLAG_NOENV</code></dt>
200 <p>If this flag bit is ored into the flag value (or the program runs setuid
201 or setgid) then libev will <i>not</i> look at the environment variable
202 <code>LIBEV_FLAGS</code>. Otherwise (the default), this environment variable will
203 override the flags completely if it is found in the environment. This is
204 useful to try out specific backends to test their performance, or to work
207 <dt><code>EVBACKEND_SELECT</code> (value 1, portable select backend)</dt>
209 <p>This is your standard select(2) backend. Not <i>completely</i> standard, as
210 libev tries to roll its own fd_set with no limits on the number of fds,
211 but if that fails, expect a fairly low limit on the number of fds when
212 using this backend. It doesn't scale too well (O(highest_fd)), but its usually
213 the fastest backend for a low number of fds.</p>
215 <dt><code>EVBACKEND_POLL</code> (value 2, poll backend, available everywhere except on windows)</dt>
217 <p>And this is your standard poll(2) backend. It's more complicated than
218 select, but handles sparse fds better and has no artificial limit on the
219 number of fds you can use (except it will slow down considerably with a
220 lot of inactive fds). It scales similarly to select, i.e. O(total_fds).</p>
222 <dt><code>EVBACKEND_EPOLL</code> (value 4, Linux)</dt>
224 <p>For few fds, this backend is a bit little slower than poll and select,
225 but it scales phenomenally better. While poll and select usually scale like
226 O(total_fds) where n is the total number of fds (or the highest fd), epoll scales
227 either O(1) or O(active_fds).</p>
228 <p>While stopping and starting an I/O watcher in the same iteration will
229 result in some caching, there is still a syscall per such incident
230 (because the fd could point to a different file description now), so its
231 best to avoid that. Also, dup()ed file descriptors might not work very
232 well if you register events for both fds.</p>
234 <dt><code>EVBACKEND_KQUEUE</code> (value 8, most BSD clones)</dt>
236 <p>Kqueue deserves special mention, as at the time of this writing, it
237 was broken on all BSDs except NetBSD (usually it doesn't work with
238 anything but sockets and pipes, except on Darwin, where of course its
239 completely useless). For this reason its not being "autodetected" unless
240 you explicitly specify the flags (i.e. you don't use EVFLAG_AUTO).</p>
241 <p>It scales in the same way as the epoll backend, but the interface to the
242 kernel is more efficient (which says nothing about its actual speed, of
243 course). While starting and stopping an I/O watcher does not cause an
244 extra syscall as with epoll, it still adds up to four event changes per
245 incident, so its best to avoid that.</p>
247 <dt><code>EVBACKEND_DEVPOLL</code> (value 16, Solaris 8)</dt>
249 <p>This is not implemented yet (and might never be).</p>
251 <dt><code>EVBACKEND_PORT</code> (value 32, Solaris 10)</dt>
253 <p>This uses the Solaris 10 port mechanism. As with everything on Solaris,
254 it's really slow, but it still scales very well (O(active_fds)).</p>
256 <dt><code>EVBACKEND_ALL</code></dt>
258 <p>Try all backends (even potentially broken ones that wouldn't be tried
259 with <code>EVFLAG_AUTO</code>). Since this is a mask, you can do stuff such as
260 <code>EVBACKEND_ALL & ~EVBACKEND_KQUEUE</code>.</p>
264 <p>If one or more of these are ored into the flags value, then only these
265 backends will be tried (in the reverse order as given here). If none are
266 specified, most compiled-in backend will be tried, usually in reverse
267 order of their flag values :)</p>
269 <dt>struct ev_loop *ev_loop_new (unsigned int flags)</dt>
271 <p>Similar to <code>ev_default_loop</code>, but always creates a new event loop that is
272 always distinct from the default loop. Unlike the default loop, it cannot
273 handle signal and child watchers, and attempts to do so will be greeted by
274 undefined behaviour (or a failed assertion if assertions are enabled).</p>
276 <dt>ev_default_destroy ()</dt>
278 <p>Destroys the default loop again (frees all memory and kernel state
279 etc.). This stops all registered event watchers (by not touching them in
280 any way whatsoever, although you cannot rely on this :).</p>
282 <dt>ev_loop_destroy (loop)</dt>
284 <p>Like <code>ev_default_destroy</code>, but destroys an event loop created by an
285 earlier call to <code>ev_loop_new</code>.</p>
287 <dt>ev_default_fork ()</dt>
289 <p>This function reinitialises the kernel state for backends that have
290 one. Despite the name, you can call it anytime, but it makes most sense
291 after forking, in either the parent or child process (or both, but that
292 again makes little sense).</p>
293 <p>You <i>must</i> call this function in the child process after forking if and
294 only if you want to use the event library in both processes. If you just
295 fork+exec, you don't have to call it.</p>
296 <p>The function itself is quite fast and it's usually not a problem to call
297 it just in case after a fork. To make this easy, the function will fit in
298 quite nicely into a call to <code>pthread_atfork</code>:</p>
299 <pre> pthread_atfork (0, 0, ev_default_fork);
302 <p>At the moment, <code>EVBACKEND_SELECT</code> and <code>EVBACKEND_POLL</code> are safe to use
303 without calling this function, so if you force one of those backends you
304 do not need to care.</p>
306 <dt>ev_loop_fork (loop)</dt>
308 <p>Like <code>ev_default_fork</code>, but acts on an event loop created by
309 <code>ev_loop_new</code>. Yes, you have to call this on every allocated event loop
310 after fork, and how you do this is entirely your own problem.</p>
312 <dt>unsigned int ev_backend (loop)</dt>
314 <p>Returns one of the <code>EVBACKEND_*</code> flags indicating the event backend in
317 <dt>ev_tstamp ev_now (loop)</dt>
319 <p>Returns the current "event loop time", which is the time the event loop
320 got events and started processing them. This timestamp does not change
321 as long as callbacks are being processed, and this is also the base time
322 used for relative timers. You can treat it as the timestamp of the event
323 occuring (or more correctly, the mainloop finding out about it).</p>
325 <dt>ev_loop (loop, int flags)</dt>
327 <p>Finally, this is it, the event handler. This function usually is called
328 after you initialised all your watchers and you want to start handling
330 <p>If the flags argument is specified as 0, it will not return until either
331 no event watchers are active anymore or <code>ev_unloop</code> was called.</p>
332 <p>A flags value of <code>EVLOOP_NONBLOCK</code> will look for new events, will handle
333 those events and any outstanding ones, but will not block your process in
334 case there are no events and will return after one iteration of the loop.</p>
335 <p>A flags value of <code>EVLOOP_ONESHOT</code> will look for new events (waiting if
336 neccessary) and will handle those and any outstanding ones. It will block
337 your process until at least one new event arrives, and will return after
338 one iteration of the loop.</p>
339 <p>This flags value could be used to implement alternative looping
340 constructs, but the <code>prepare</code> and <code>check</code> watchers provide a better and
341 more generic mechanism.</p>
342 <p>Here are the gory details of what ev_loop does:</p>
343 <pre> 1. If there are no active watchers (reference count is zero), return.
344 2. Queue and immediately call all prepare watchers.
345 3. If we have been forked, recreate the kernel state.
346 4. Update the kernel state with all outstanding changes.
347 5. Update the "event loop time".
348 6. Calculate for how long to block.
349 7. Block the process, waiting for events.
350 8. Update the "event loop time" and do time jump handling.
351 9. Queue all outstanding timers.
352 10. Queue all outstanding periodics.
353 11. If no events are pending now, queue all idle watchers.
354 12. Queue all check watchers.
355 13. Call all queued watchers in reverse order (i.e. check watchers first).
356 14. If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
357 was used, return, otherwise continue with step #1.
361 <dt>ev_unloop (loop, how)</dt>
363 <p>Can be used to make a call to <code>ev_loop</code> return early (but only after it
364 has processed all outstanding events). The <code>how</code> argument must be either
365 <code>EVUNLOOP_ONE</code>, which will make the innermost <code>ev_loop</code> call return, or
366 <code>EVUNLOOP_ALL</code>, which will make all nested <code>ev_loop</code> calls return.</p>
368 <dt>ev_ref (loop)</dt>
369 <dt>ev_unref (loop)</dt>
371 <p>Ref/unref can be used to add or remove a reference count on the event
372 loop: Every watcher keeps one reference, and as long as the reference
373 count is nonzero, <code>ev_loop</code> will not return on its own. If you have
374 a watcher you never unregister that should not keep <code>ev_loop</code> from
375 returning, ev_unref() after starting, and ev_ref() before stopping it. For
376 example, libev itself uses this for its internal signal pipe: It is not
377 visible to the libev user and should not keep <code>ev_loop</code> from exiting if
378 no event watchers registered by it are active. It is also an excellent
379 way to do this for generic recurring timers or from within third-party
380 libraries. Just remember to <i>unref after start</i> and <i>ref before stop</i>.</p>
385 <h1 id="ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</h1><p><a href="#TOP" class="toplink">Top</a></p>
386 <div id="ANATOMY_OF_A_WATCHER_CONTENT">
387 <p>A watcher is a structure that you create and register to record your
388 interest in some event. For instance, if you want to wait for STDIN to
389 become readable, you would create an <code>ev_io</code> watcher for that:</p>
390 <pre> static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)
393 ev_unloop (loop, EVUNLOOP_ALL);
396 struct ev_loop *loop = ev_default_loop (0);
397 struct ev_io stdin_watcher;
398 ev_init (&stdin_watcher, my_cb);
399 ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
400 ev_io_start (loop, &stdin_watcher);
404 <p>As you can see, you are responsible for allocating the memory for your
405 watcher structures (and it is usually a bad idea to do this on the stack,
406 although this can sometimes be quite valid).</p>
407 <p>Each watcher structure must be initialised by a call to <code>ev_init
408 (watcher *, callback)</code>, which expects a callback to be provided. This
409 callback gets invoked each time the event occurs (or, in the case of io
410 watchers, each time the event loop detects that the file descriptor given
411 is readable and/or writable).</p>
412 <p>Each watcher type has its own <code>ev_<type>_set (watcher *, ...)</code> macro
413 with arguments specific to this watcher type. There is also a macro
414 to combine initialisation and setting in one call: <code>ev_<type>_init
415 (watcher *, callback, ...)</code>.</p>
416 <p>To make the watcher actually watch out for events, you have to start it
417 with a watcher-specific start function (<code>ev_<type>_start (loop, watcher
418 *)</code>), and you can stop watching for events at any time by calling the
419 corresponding stop function (<code>ev_<type>_stop (loop, watcher *)</code>.</p>
420 <p>As long as your watcher is active (has been started but not stopped) you
421 must not touch the values stored in it. Most specifically you must never
422 reinitialise it or call its set macro.</p>
423 <p>You can check whether an event is active by calling the <code>ev_is_active
424 (watcher *)</code> macro. To see whether an event is outstanding (but the
425 callback for it has not been called yet) you can use the <code>ev_is_pending
426 (watcher *)</code> macro.</p>
427 <p>Each and every callback receives the event loop pointer as first, the
428 registered watcher structure as second, and a bitset of received events as
430 <p>The received events usually include a single bit per event type received
431 (you can receive multiple events at the same time). The possible bit masks
434 <dt><code>EV_READ</code></dt>
435 <dt><code>EV_WRITE</code></dt>
437 <p>The file descriptor in the <code>ev_io</code> watcher has become readable and/or
440 <dt><code>EV_TIMEOUT</code></dt>
442 <p>The <code>ev_timer</code> watcher has timed out.</p>
444 <dt><code>EV_PERIODIC</code></dt>
446 <p>The <code>ev_periodic</code> watcher has timed out.</p>
448 <dt><code>EV_SIGNAL</code></dt>
450 <p>The signal specified in the <code>ev_signal</code> watcher has been received by a thread.</p>
452 <dt><code>EV_CHILD</code></dt>
454 <p>The pid specified in the <code>ev_child</code> watcher has received a status change.</p>
456 <dt><code>EV_IDLE</code></dt>
458 <p>The <code>ev_idle</code> watcher has determined that you have nothing better to do.</p>
460 <dt><code>EV_PREPARE</code></dt>
461 <dt><code>EV_CHECK</code></dt>
463 <p>All <code>ev_prepare</code> watchers are invoked just <i>before</i> <code>ev_loop</code> starts
464 to gather new events, and all <code>ev_check</code> watchers are invoked just after
465 <code>ev_loop</code> has gathered them, but before it invokes any callbacks for any
466 received events. Callbacks of both watcher types can start and stop as
467 many watchers as they want, and all of them will be taken into account
468 (for example, a <code>ev_prepare</code> watcher might start an idle watcher to keep
469 <code>ev_loop</code> from blocking).</p>
471 <dt><code>EV_ERROR</code></dt>
473 <p>An unspecified error has occured, the watcher has been stopped. This might
474 happen because the watcher could not be properly started because libev
475 ran out of memory, a file descriptor was found to be closed or any other
476 problem. You best act on it by reporting the problem and somehow coping
477 with the watcher being stopped.</p>
478 <p>Libev will usually signal a few "dummy" events together with an error,
479 for example it might indicate that a fd is readable or writable, and if
480 your callbacks is well-written it can just attempt the operation and cope
481 with the error from read() or write(). This will not work in multithreaded
482 programs, though, so beware.</p>
487 <h2 id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</h2>
488 <div id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH-2">
489 <p>Each watcher has, by default, a member <code>void *data</code> that you can change
490 and read at any time, libev will completely ignore it. This can be used
491 to associate arbitrary data with your watcher. If you need more data and
492 don't want to allocate memory and store a pointer to it in that data
493 member, you can also "subclass" the watcher type and provide your own
500 struct whatever *mostinteresting;
504 <p>And since your callback will be called with a pointer to the watcher, you
505 can cast it back to your own type:</p>
506 <pre> static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)
508 struct my_io *w = (struct my_io *)w_;
513 <p>More interesting and less C-conformant ways of catsing your callback type
514 have been omitted....</p>
521 <h1 id="WATCHER_TYPES">WATCHER TYPES</h1><p><a href="#TOP" class="toplink">Top</a></p>
522 <div id="WATCHER_TYPES_CONTENT">
523 <p>This section describes each watcher in detail, but will not repeat
524 information given in the last section.</p>
527 <h2 id="code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable</h2>
528 <div id="code_ev_io_code_is_this_file_descrip-2">
529 <p>I/O watchers check whether a file descriptor is readable or writable
530 in each iteration of the event loop (This behaviour is called
531 level-triggering because you keep receiving events as long as the
532 condition persists. Remember you can stop the watcher if you don't want to
533 act on the event and neither want to receive future events).</p>
534 <p>In general you can register as many read and/or write event watchers per
535 fd as you want (as long as you don't confuse yourself). Setting all file
536 descriptors to non-blocking mode is also usually a good idea (but not
537 required if you know what you are doing).</p>
538 <p>You have to be careful with dup'ed file descriptors, though. Some backends
539 (the linux epoll backend is a notable example) cannot handle dup'ed file
540 descriptors correctly if you register interest in two or more fds pointing
541 to the same underlying file/socket etc. description (that is, they share
542 the same underlying "file open").</p>
543 <p>If you must do this, then force the use of a known-to-be-good backend
544 (at the time of this writing, this includes only <code>EVBACKEND_SELECT</code> and
545 <code>EVBACKEND_POLL</code>).</p>
547 <dt>ev_io_init (ev_io *, callback, int fd, int events)</dt>
548 <dt>ev_io_set (ev_io *, int fd, int events)</dt>
550 <p>Configures an <code>ev_io</code> watcher. The fd is the file descriptor to rceeive
551 events for and events is either <code>EV_READ</code>, <code>EV_WRITE</code> or <code>EV_READ |
552 EV_WRITE</code> to receive the given events.</p>
557 <h2 id="code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally recurring timeouts</h2>
558 <div id="code_ev_timer_code_relative_and_opti-2">
559 <p>Timer watchers are simple relative timers that generate an event after a
560 given time, and optionally repeating in regular intervals after that.</p>
561 <p>The timers are based on real time, that is, if you register an event that
562 times out after an hour and you reset your system clock to last years
563 time, it will still time out after (roughly) and hour. "Roughly" because
564 detecting time jumps is hard, and some inaccuracies are unavoidable (the
565 monotonic clock option helps a lot here).</p>
566 <p>The relative timeouts are calculated relative to the <code>ev_now ()</code>
567 time. This is usually the right thing as this timestamp refers to the time
568 of the event triggering whatever timeout you are modifying/starting. If
569 you suspect event processing to be delayed and you <i>need</i> to base the timeout
570 on the current time, use something like this to adjust for this:</p>
571 <pre> ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
574 <p>The callback is guarenteed to be invoked only when its timeout has passed,
575 but if multiple timers become ready during the same loop iteration then
576 order of execution is undefined.</p>
578 <dt>ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)</dt>
579 <dt>ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)</dt>
581 <p>Configure the timer to trigger after <code>after</code> seconds. If <code>repeat</code> is
582 <code>0.</code>, then it will automatically be stopped. If it is positive, then the
583 timer will automatically be configured to trigger again <code>repeat</code> seconds
584 later, again, and again, until stopped manually.</p>
585 <p>The timer itself will do a best-effort at avoiding drift, that is, if you
586 configure a timer to trigger every 10 seconds, then it will trigger at
587 exactly 10 second intervals. If, however, your program cannot keep up with
588 the timer (because it takes longer than those 10 seconds to do stuff) the
589 timer will not fire more than once per event loop iteration.</p>
591 <dt>ev_timer_again (loop)</dt>
593 <p>This will act as if the timer timed out and restart it again if it is
594 repeating. The exact semantics are:</p>
595 <p>If the timer is started but nonrepeating, stop it.</p>
596 <p>If the timer is repeating, either start it if necessary (with the repeat
597 value), or reset the running timer to the repeat value.</p>
598 <p>This sounds a bit complicated, but here is a useful and typical
599 example: Imagine you have a tcp connection and you want a so-called idle
600 timeout, that is, you want to be called when there have been, say, 60
601 seconds of inactivity on the socket. The easiest way to do this is to
602 configure an <code>ev_timer</code> with after=repeat=60 and calling ev_timer_again each
603 time you successfully read or write some data. If you go into an idle
604 state where you do not expect data to travel on the socket, you can stop
605 the timer, and again will automatically restart it if need be.</p>
610 <h2 id="code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron</h2>
611 <div id="code_ev_periodic_code_to_cron_or_not-2">
612 <p>Periodic watchers are also timers of a kind, but they are very versatile
613 (and unfortunately a bit complex).</p>
614 <p>Unlike <code>ev_timer</code>'s, they are not based on real time (or relative time)
615 but on wallclock time (absolute time). You can tell a periodic watcher
616 to trigger "at" some specific point in time. For example, if you tell a
617 periodic watcher to trigger in 10 seconds (by specifiying e.g. c<ev_now ()
618 + 10.>) and then reset your system clock to the last year, then it will
619 take a year to trigger the event (unlike an <code>ev_timer</code>, which would trigger
620 roughly 10 seconds later and of course not if you reset your system time
622 <p>They can also be used to implement vastly more complex timers, such as
623 triggering an event on eahc midnight, local time.</p>
624 <p>As with timers, the callback is guarenteed to be invoked only when the
625 time (<code>at</code>) has been passed, but if multiple periodic timers become ready
626 during the same loop iteration then order of execution is undefined.</p>
628 <dt>ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)</dt>
629 <dt>ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)</dt>
631 <p>Lots of arguments, lets sort it out... There are basically three modes of
632 operation, and we will explain them from simplest to complex:</p>
635 <dt>* absolute timer (interval = reschedule_cb = 0)</dt>
637 <p>In this configuration the watcher triggers an event at the wallclock time
638 <code>at</code> and doesn't repeat. It will not adjust when a time jump occurs,
639 that is, if it is to be run at January 1st 2011 then it will run when the
640 system time reaches or surpasses this time.</p>
642 <dt>* non-repeating interval timer (interval > 0, reschedule_cb = 0)</dt>
644 <p>In this mode the watcher will always be scheduled to time out at the next
645 <code>at + N * interval</code> time (for some integer N) and then repeat, regardless
646 of any time jumps.</p>
647 <p>This can be used to create timers that do not drift with respect to system
649 <pre> ev_periodic_set (&periodic, 0., 3600., 0);
652 <p>This doesn't mean there will always be 3600 seconds in between triggers,
653 but only that the the callback will be called when the system time shows a
654 full hour (UTC), or more correctly, when the system time is evenly divisible
656 <p>Another way to think about it (for the mathematically inclined) is that
657 <code>ev_periodic</code> will try to run the callback in this mode at the next possible
658 time where <code>time = at (mod interval)</code>, regardless of any time jumps.</p>
660 <dt>* manual reschedule mode (reschedule_cb = callback)</dt>
662 <p>In this mode the values for <code>interval</code> and <code>at</code> are both being
663 ignored. Instead, each time the periodic watcher gets scheduled, the
664 reschedule callback will be called with the watcher as first, and the
665 current time as second argument.</p>
666 <p>NOTE: <i>This callback MUST NOT stop or destroy any periodic watcher,
667 ever, or make any event loop modifications</i>. If you need to stop it,
668 return <code>now + 1e30</code> (or so, fudge fudge) and stop it afterwards (e.g. by
669 starting a prepare watcher).</p>
670 <p>Its prototype is <code>ev_tstamp (*reschedule_cb)(struct ev_periodic *w,
671 ev_tstamp now)</code>, e.g.:</p>
672 <pre> static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)
678 <p>It must return the next time to trigger, based on the passed time value
679 (that is, the lowest time value larger than to the second argument). It
680 will usually be called just before the callback will be triggered, but
681 might be called at other times, too.</p>
682 <p>NOTE: <i>This callback must always return a time that is later than the
683 passed <code>now</code> value</i>. Not even <code>now</code> itself will do, it <i>must</i> be larger.</p>
684 <p>This can be used to create very complex timers, such as a timer that
685 triggers on each midnight, local time. To do this, you would calculate the
686 next midnight after <code>now</code> and return the timestamp value for this. How
687 you do this is, again, up to you (but it is not trivial, which is the main
688 reason I omitted it as an example).</p>
693 <dt>ev_periodic_again (loop, ev_periodic *)</dt>
695 <p>Simply stops and restarts the periodic watcher again. This is only useful
696 when you changed some parameters or the reschedule callback would return
697 a different time than the last time it was called (e.g. in a crond like
698 program when the crontabs have changed).</p>
703 <h2 id="code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled</h2>
704 <div id="code_ev_signal_code_signal_me_when_a-2">
705 <p>Signal watchers will trigger an event when the process receives a specific
706 signal one or more times. Even though signals are very asynchronous, libev
707 will try it's best to deliver signals synchronously, i.e. as part of the
708 normal event processing, like any other event.</p>
709 <p>You can configure as many watchers as you like per signal. Only when the
710 first watcher gets started will libev actually register a signal watcher
711 with the kernel (thus it coexists with your own signal handlers as long
712 as you don't register any with libev). Similarly, when the last signal
713 watcher for a signal is stopped libev will reset the signal handler to
714 SIG_DFL (regardless of what it was set to before).</p>
716 <dt>ev_signal_init (ev_signal *, callback, int signum)</dt>
717 <dt>ev_signal_set (ev_signal *, int signum)</dt>
719 <p>Configures the watcher to trigger on the given signal number (usually one
720 of the <code>SIGxxx</code> constants).</p>
725 <h2 id="code_ev_child_code_wait_for_pid_stat"><code>ev_child</code> - wait for pid status changes</h2>
726 <div id="code_ev_child_code_wait_for_pid_stat-2">
727 <p>Child watchers trigger when your process receives a SIGCHLD in response to
728 some child status changes (most typically when a child of yours dies).</p>
730 <dt>ev_child_init (ev_child *, callback, int pid)</dt>
731 <dt>ev_child_set (ev_child *, int pid)</dt>
733 <p>Configures the watcher to wait for status changes of process <code>pid</code> (or
734 <i>any</i> process if <code>pid</code> is specified as <code>0</code>). The callback can look
735 at the <code>rstatus</code> member of the <code>ev_child</code> watcher structure to see
736 the status word (use the macros from <code>sys/wait.h</code> and see your systems
737 <code>waitpid</code> documentation). The <code>rpid</code> member contains the pid of the
738 process causing the status change.</p>
743 <h2 id="code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do</h2>
744 <div id="code_ev_idle_code_when_you_ve_got_no-2">
745 <p>Idle watchers trigger events when there are no other events are pending
746 (prepare, check and other idle watchers do not count). That is, as long
747 as your process is busy handling sockets or timeouts (or even signals,
748 imagine) it will not be triggered. But when your process is idle all idle
749 watchers are being called again and again, once per event loop iteration -
750 until stopped, that is, or your process receives more events and becomes
752 <p>The most noteworthy effect is that as long as any idle watchers are
753 active, the process will not block when waiting for new events.</p>
754 <p>Apart from keeping your process non-blocking (which is a useful
755 effect on its own sometimes), idle watchers are a good place to do
756 "pseudo-background processing", or delay processing stuff to after the
757 event loop has handled all outstanding events.</p>
759 <dt>ev_idle_init (ev_signal *, callback)</dt>
761 <p>Initialises and configures the idle watcher - it has no parameters of any
762 kind. There is a <code>ev_idle_set</code> macro, but using it is utterly pointless,
768 <h2 id="code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop</h2>
769 <div id="code_ev_prepare_code_and_code_ev_che-2">
770 <p>Prepare and check watchers are usually (but not always) used in tandem:
771 prepare watchers get invoked before the process blocks and check watchers
773 <p>Their main purpose is to integrate other event mechanisms into libev. This
774 could be used, for example, to track variable changes, implement your own
775 watchers, integrate net-snmp or a coroutine library and lots more.</p>
776 <p>This is done by examining in each prepare call which file descriptors need
777 to be watched by the other library, registering <code>ev_io</code> watchers for
778 them and starting an <code>ev_timer</code> watcher for any timeouts (many libraries
779 provide just this functionality). Then, in the check watcher you check for
780 any events that occured (by checking the pending status of all watchers
781 and stopping them) and call back into the library. The I/O and timer
782 callbacks will never actually be called (but must be valid nevertheless,
783 because you never know, you know?).</p>
784 <p>As another example, the Perl Coro module uses these hooks to integrate
785 coroutines into libev programs, by yielding to other active coroutines
786 during each prepare and only letting the process block if no coroutines
787 are ready to run (it's actually more complicated: it only runs coroutines
788 with priority higher than or equal to the event loop and one coroutine
789 of lower priority, but only once, using idle watchers to keep the event
790 loop from blocking if lower-priority coroutines are active, thus mapping
791 low-priority coroutines to idle/background tasks).</p>
793 <dt>ev_prepare_init (ev_prepare *, callback)</dt>
794 <dt>ev_check_init (ev_check *, callback)</dt>
796 <p>Initialises and configures the prepare or check watcher - they have no
797 parameters of any kind. There are <code>ev_prepare_set</code> and <code>ev_check_set</code>
798 macros, but using them is utterly, utterly and completely pointless.</p>
803 <h1 id="OTHER_FUNCTIONS">OTHER FUNCTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p>
804 <div id="OTHER_FUNCTIONS_CONTENT">
805 <p>There are some other functions of possible interest. Described. Here. Now.</p>
807 <dt>ev_once (loop, int fd, int events, ev_tstamp timeout, callback)</dt>
809 <p>This function combines a simple timer and an I/O watcher, calls your
810 callback on whichever event happens first and automatically stop both
811 watchers. This is useful if you want to wait for a single event on an fd
812 or timeout without having to allocate/configure/start/stop/free one or
813 more watchers yourself.</p>
814 <p>If <code>fd</code> is less than 0, then no I/O watcher will be started and events
815 is being ignored. Otherwise, an <code>ev_io</code> watcher for the given <code>fd</code> and
816 <code>events</code> set will be craeted and started.</p>
817 <p>If <code>timeout</code> is less than 0, then no timeout watcher will be
818 started. Otherwise an <code>ev_timer</code> watcher with after = <code>timeout</code> (and
819 repeat = 0) will be started. While <code>0</code> is a valid timeout, it is of
821 <p>The callback has the type <code>void (*cb)(int revents, void *arg)</code> and gets
822 passed an <code>revents</code> set like normal event callbacks (a combination of
823 <code>EV_ERROR</code>, <code>EV_READ</code>, <code>EV_WRITE</code> or <code>EV_TIMEOUT</code>) and the <code>arg</code>
824 value passed to <code>ev_once</code>:</p>
825 <pre> static void stdin_ready (int revents, void *arg)
827 if (revents & EV_TIMEOUT)
828 /* doh, nothing entered */;
829 else if (revents & EV_READ)
830 /* stdin might have data for us, joy! */;
833 ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
837 <dt>ev_feed_event (loop, watcher, int events)</dt>
839 <p>Feeds the given event set into the event loop, as if the specified event
840 had happened for the specified watcher (which must be a pointer to an
841 initialised but not necessarily started event watcher).</p>
843 <dt>ev_feed_fd_event (loop, int fd, int revents)</dt>
845 <p>Feed an event on the given fd, as if a file descriptor backend detected
846 the given events it.</p>
848 <dt>ev_feed_signal_event (loop, int signum)</dt>
850 <p>Feed an event as if the given signal occured (loop must be the default loop!).</p>
855 <h1 id="LIBEVENT_EMULATION">LIBEVENT EMULATION</h1><p><a href="#TOP" class="toplink">Top</a></p>
856 <div id="LIBEVENT_EMULATION_CONTENT">
857 <p>Libev offers a compatibility emulation layer for libevent. It cannot
858 emulate the internals of libevent, so here are some usage hints:</p>
860 <dt>* Use it by including <event.h>, as usual.</dt>
861 <dt>* The following members are fully supported: ev_base, ev_callback,
862 ev_arg, ev_fd, ev_res, ev_events.</dt>
863 <dt>* Avoid using ev_flags and the EVLIST_*-macros, while it is
864 maintained by libev, it does not work exactly the same way as in libevent (consider
865 it a private API).</dt>
866 <dt>* Priorities are not currently supported. Initialising priorities
867 will fail and all watchers will have the same priority, even though there
868 is an ev_pri field.</dt>
869 <dt>* Other members are not supported.</dt>
870 <dt>* The libev emulation is <i>not</i> ABI compatible to libevent, you need
871 to use the libev header file and library.</dt>
875 <h1 id="C_SUPPORT">C++ SUPPORT</h1><p><a href="#TOP" class="toplink">Top</a></p>
876 <div id="C_SUPPORT_CONTENT">
880 <h1 id="AUTHOR">AUTHOR</h1><p><a href="#TOP" class="toplink">Top</a></p>
881 <div id="AUTHOR_CONTENT">
882 <p>Marc Lehmann <libev@schmorp.de>.</p>
projects / software / libev.git / blob