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="Thu Nov 29 21:05:58 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="#EXAMPLE_PROGRAM">EXAMPLE PROGRAM</a></li>
20 <li><a href="#DESCRIPTION">DESCRIPTION</a></li>
21 <li><a href="#FEATURES">FEATURES</a></li>
22 <li><a href="#CONVENTIONS">CONVENTIONS</a></li>
23 <li><a href="#TIME_REPRESENTATION">TIME REPRESENTATION</a></li>
24 <li><a href="#GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</a></li>
25 <li><a href="#FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</a></li>
26 <li><a href="#ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</a>
27 <ul><li><a href="#GENERIC_WATCHER_FUNCTIONS">GENERIC WATCHER FUNCTIONS</a></li>
28 <li><a href="#ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</a></li>
31 <li><a href="#WATCHER_TYPES">WATCHER TYPES</a>
32 <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>
33 <li><a href="#code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally repeating timeouts</a></li>
34 <li><a href="#code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron?</a></li>
35 <li><a href="#code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled!</a></li>
36 <li><a href="#code_ev_child_code_watch_out_for_pro"><code>ev_child</code> - watch out for process status changes</a></li>
37 <li><a href="#code_ev_stat_code_did_the_file_attri"><code>ev_stat</code> - did the file attributes just change?</a></li>
38 <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>
39 <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>
40 <li><a href="#code_ev_embed_code_when_one_backend_"><code>ev_embed</code> - when one backend isn't enough...</a></li>
41 <li><a href="#code_ev_fork_code_the_audacity_to_re"><code>ev_fork</code> - the audacity to resume the event loop after a fork</a></li>
44 <li><a href="#OTHER_FUNCTIONS">OTHER FUNCTIONS</a></li>
45 <li><a href="#LIBEVENT_EMULATION">LIBEVENT EMULATION</a></li>
46 <li><a href="#C_SUPPORT">C++ SUPPORT</a></li>
47 <li><a href="#MACRO_MAGIC">MACRO MAGIC</a></li>
48 <li><a href="#EMBEDDING">EMBEDDING</a>
49 <ul><li><a href="#FILESETS">FILESETS</a>
50 <ul><li><a href="#CORE_EVENT_LOOP">CORE EVENT LOOP</a></li>
51 <li><a href="#LIBEVENT_COMPATIBILITY_API">LIBEVENT COMPATIBILITY API</a></li>
52 <li><a href="#AUTOCONF_SUPPORT">AUTOCONF SUPPORT</a></li>
55 <li><a href="#PREPROCESSOR_SYMBOLS_MACROS">PREPROCESSOR SYMBOLS/MACROS</a></li>
56 <li><a href="#EXAMPLES">EXAMPLES</a></li>
59 <li><a href="#COMPLEXITIES">COMPLEXITIES</a></li>
60 <li><a href="#AUTHOR">AUTHOR</a>
65 <h1 id="NAME">NAME</h1>
66 <div id="NAME_CONTENT">
67 <p>libev - a high performance full-featured event loop written in C</p>
70 <h1 id="SYNOPSIS">SYNOPSIS</h1>
71 <div id="SYNOPSIS_CONTENT">
72 <pre> #include <ev.h>
77 <h1 id="EXAMPLE_PROGRAM">EXAMPLE PROGRAM</h1>
78 <div id="EXAMPLE_PROGRAM_CONTENT">
79 <pre> #include <ev.h>
82 ev_timer timeout_watcher;
84 /* called when data readable on stdin */
86 stdin_cb (EV_P_ struct ev_io *w, int revents)
88 /* puts ("stdin ready"); */
89 ev_io_stop (EV_A_ w); /* just a syntax example */
90 ev_unloop (EV_A_ EVUNLOOP_ALL); /* leave all loop calls */
94 timeout_cb (EV_P_ struct ev_timer *w, int revents)
96 /* puts ("timeout"); */
97 ev_unloop (EV_A_ EVUNLOOP_ONE); /* leave one loop call */
103 struct ev_loop *loop = ev_default_loop (0);
105 /* initialise an io watcher, then start it */
106 ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
107 ev_io_start (loop, &stdin_watcher);
109 /* simple non-repeating 5.5 second timeout */
110 ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
111 ev_timer_start (loop, &timeout_watcher);
113 /* loop till timeout or data ready */
122 <h1 id="DESCRIPTION">DESCRIPTION</h1>
123 <div id="DESCRIPTION_CONTENT">
124 <p>Libev is an event loop: you register interest in certain events (such as a
125 file descriptor being readable or a timeout occuring), and it will manage
126 these event sources and provide your program with events.</p>
127 <p>To do this, it must take more or less complete control over your process
128 (or thread) by executing the <i>event loop</i> handler, and will then
129 communicate events via a callback mechanism.</p>
130 <p>You register interest in certain events by registering so-called <i>event
131 watchers</i>, which are relatively small C structures you initialise with the
132 details of the event, and then hand it over to libev by <i>starting</i> the
136 <h1 id="FEATURES">FEATURES</h1>
137 <div id="FEATURES_CONTENT">
138 <p>Libev supports <code>select</code>, <code>poll</code>, the Linux-specific <code>epoll</code>, the
139 BSD-specific <code>kqueue</code> and the Solaris-specific event port mechanisms
140 for file descriptor events (<code>ev_io</code>), the Linux <code>inotify</code> interface
141 (for <code>ev_stat</code>), relative timers (<code>ev_timer</code>), absolute timers
142 with customised rescheduling (<code>ev_periodic</code>), synchronous signals
143 (<code>ev_signal</code>), process status change events (<code>ev_child</code>), and event
144 watchers dealing with the event loop mechanism itself (<code>ev_idle</code>,
145 <code>ev_embed</code>, <code>ev_prepare</code> and <code>ev_check</code> watchers) as well as
146 file watchers (<code>ev_stat</code>) and even limited support for fork events
147 (<code>ev_fork</code>).</p>
148 <p>It also is quite fast (see this
149 <a href="http://libev.schmorp.de/bench.html">benchmark</a> comparing it to libevent
153 <h1 id="CONVENTIONS">CONVENTIONS</h1>
154 <div id="CONVENTIONS_CONTENT">
155 <p>Libev is very configurable. In this manual the default configuration will
156 be described, which supports multiple event loops. For more info about
157 various configuration options please have a look at <strong>EMBED</strong> section in
158 this manual. If libev was configured without support for multiple event
159 loops, then all functions taking an initial argument of name <code>loop</code>
160 (which is always of type <code>struct ev_loop *</code>) will not have this argument.</p>
163 <h1 id="TIME_REPRESENTATION">TIME REPRESENTATION</h1>
164 <div id="TIME_REPRESENTATION_CONTENT">
165 <p>Libev represents time as a single floating point number, representing the
166 (fractional) number of seconds since the (POSIX) epoch (somewhere near
167 the beginning of 1970, details are complicated, don't ask). This type is
168 called <code>ev_tstamp</code>, which is what you should use too. It usually aliases
169 to the <code>double</code> type in C, and when you need to do any calculations on
170 it, you should treat it as such.</p>
173 <h1 id="GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</h1>
174 <div id="GLOBAL_FUNCTIONS_CONTENT">
175 <p>These functions can be called anytime, even before initialising the
176 library in any way.</p>
178 <dt>ev_tstamp ev_time ()</dt>
180 <p>Returns the current time as libev would use it. Please note that the
181 <code>ev_now</code> function is usually faster and also often returns the timestamp
182 you actually want to know.</p>
184 <dt>int ev_version_major ()</dt>
185 <dt>int ev_version_minor ()</dt>
187 <p>You can find out the major and minor version numbers of the library
188 you linked against by calling the functions <code>ev_version_major</code> and
189 <code>ev_version_minor</code>. If you want, you can compare against the global
190 symbols <code>EV_VERSION_MAJOR</code> and <code>EV_VERSION_MINOR</code>, which specify the
191 version of the library your program was compiled against.</p>
192 <p>Usually, it's a good idea to terminate if the major versions mismatch,
193 as this indicates an incompatible change. Minor versions are usually
194 compatible to older versions, so a larger minor version alone is usually
196 <p>Example: Make sure we haven't accidentally been linked against the wrong
198 <pre> assert (("libev version mismatch",
199 ev_version_major () == EV_VERSION_MAJOR
200 && ev_version_minor () >= EV_VERSION_MINOR));
204 <dt>unsigned int ev_supported_backends ()</dt>
206 <p>Return the set of all backends (i.e. their corresponding <code>EV_BACKEND_*</code>
207 value) compiled into this binary of libev (independent of their
208 availability on the system you are running on). See <code>ev_default_loop</code> for
209 a description of the set values.</p>
210 <p>Example: make sure we have the epoll method, because yeah this is cool and
211 a must have and can we have a torrent of it please!!!11</p>
212 <pre> assert (("sorry, no epoll, no sex",
213 ev_supported_backends () & EVBACKEND_EPOLL));
217 <dt>unsigned int ev_recommended_backends ()</dt>
219 <p>Return the set of all backends compiled into this binary of libev and also
220 recommended for this platform. This set is often smaller than the one
221 returned by <code>ev_supported_backends</code>, as for example kqueue is broken on
222 most BSDs and will not be autodetected unless you explicitly request it
223 (assuming you know what you are doing). This is the set of backends that
224 libev will probe for if you specify no backends explicitly.</p>
226 <dt>unsigned int ev_embeddable_backends ()</dt>
228 <p>Returns the set of backends that are embeddable in other event loops. This
229 is the theoretical, all-platform, value. To find which backends
230 might be supported on the current system, you would need to look at
231 <code>ev_embeddable_backends () & ev_supported_backends ()</code>, likewise for
232 recommended ones.</p>
233 <p>See the description of <code>ev_embed</code> watchers for more info.</p>
235 <dt>ev_set_allocator (void *(*cb)(void *ptr, long size))</dt>
237 <p>Sets the allocation function to use (the prototype is similar - the
238 semantics is identical - to the realloc C function). It is used to
239 allocate and free memory (no surprises here). If it returns zero when
240 memory needs to be allocated, the library might abort or take some
241 potentially destructive action. The default is your system realloc
243 <p>You could override this function in high-availability programs to, say,
244 free some memory if it cannot allocate memory, to use a special allocator,
245 or even to sleep a while and retry until some memory is available.</p>
246 <p>Example: Replace the libev allocator with one that waits a bit and then
249 persistent_realloc (void *ptr, size_t size)
253 void *newptr = realloc (ptr, size);
263 ev_set_allocator (persistent_realloc);
267 <dt>ev_set_syserr_cb (void (*cb)(const char *msg));</dt>
269 <p>Set the callback function to call on a retryable syscall error (such
270 as failed select, poll, epoll_wait). The message is a printable string
271 indicating the system call or subsystem causing the problem. If this
272 callback is set, then libev will expect it to remedy the sitution, no
273 matter what, when it returns. That is, libev will generally retry the
274 requested operation, or, if the condition doesn't go away, do bad stuff
276 <p>Example: This is basically the same thing that libev does internally, too.</p>
278 fatal_error (const char *msg)
285 ev_set_syserr_cb (fatal_error);
292 <h1 id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</h1>
293 <div id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP-2">
294 <p>An event loop is described by a <code>struct ev_loop *</code>. The library knows two
295 types of such loops, the <i>default</i> loop, which supports signals and child
296 events, and dynamically created loops which do not.</p>
297 <p>If you use threads, a common model is to run the default event loop
298 in your main thread (or in a separate thread) and for each thread you
299 create, you also create another event loop. Libev itself does no locking
300 whatsoever, so if you mix calls to the same event loop in different
301 threads, make sure you lock (this is usually a bad idea, though, even if
302 done correctly, because it's hideous and inefficient).</p>
304 <dt>struct ev_loop *ev_default_loop (unsigned int flags)</dt>
306 <p>This will initialise the default event loop if it hasn't been initialised
307 yet and return it. If the default loop could not be initialised, returns
308 false. If it already was initialised it simply returns it (and ignores the
309 flags. If that is troubling you, check <code>ev_backend ()</code> afterwards).</p>
310 <p>If you don't know what event loop to use, use the one returned from this
312 <p>The flags argument can be used to specify special behaviour or specific
313 backends to use, and is usually specified as <code>0</code> (or <code>EVFLAG_AUTO</code>).</p>
314 <p>The following flags are supported:</p>
317 <dt><code>EVFLAG_AUTO</code></dt>
319 <p>The default flags value. Use this if you have no clue (it's the right
320 thing, believe me).</p>
322 <dt><code>EVFLAG_NOENV</code></dt>
324 <p>If this flag bit is ored into the flag value (or the program runs setuid
325 or setgid) then libev will <i>not</i> look at the environment variable
326 <code>LIBEV_FLAGS</code>. Otherwise (the default), this environment variable will
327 override the flags completely if it is found in the environment. This is
328 useful to try out specific backends to test their performance, or to work
331 <dt><code>EVFLAG_FORKCHECK</code></dt>
333 <p>Instead of calling <code>ev_default_fork</code> or <code>ev_loop_fork</code> manually after
334 a fork, you can also make libev check for a fork in each iteration by
335 enabling this flag.</p>
336 <p>This works by calling <code>getpid ()</code> on every iteration of the loop,
337 and thus this might slow down your event loop if you do a lot of loop
338 iterations and little real work, but is usually not noticable (on my
339 Linux system for example, <code>getpid</code> is actually a simple 5-insn sequence
340 without a syscall and thus <i>very</i> fast, but my Linux system also has
341 <code>pthread_atfork</code> which is even faster).</p>
342 <p>The big advantage of this flag is that you can forget about fork (and
343 forget about forgetting to tell libev about forking) when you use this
345 <p>This flag setting cannot be overriden or specified in the <code>LIBEV_FLAGS</code>
346 environment variable.</p>
348 <dt><code>EVBACKEND_SELECT</code> (value 1, portable select backend)</dt>
350 <p>This is your standard select(2) backend. Not <i>completely</i> standard, as
351 libev tries to roll its own fd_set with no limits on the number of fds,
352 but if that fails, expect a fairly low limit on the number of fds when
353 using this backend. It doesn't scale too well (O(highest_fd)), but its usually
354 the fastest backend for a low number of fds.</p>
356 <dt><code>EVBACKEND_POLL</code> (value 2, poll backend, available everywhere except on windows)</dt>
358 <p>And this is your standard poll(2) backend. It's more complicated than
359 select, but handles sparse fds better and has no artificial limit on the
360 number of fds you can use (except it will slow down considerably with a
361 lot of inactive fds). It scales similarly to select, i.e. O(total_fds).</p>
363 <dt><code>EVBACKEND_EPOLL</code> (value 4, Linux)</dt>
365 <p>For few fds, this backend is a bit little slower than poll and select,
366 but it scales phenomenally better. While poll and select usually scale like
367 O(total_fds) where n is the total number of fds (or the highest fd), epoll scales
368 either O(1) or O(active_fds).</p>
369 <p>While stopping and starting an I/O watcher in the same iteration will
370 result in some caching, there is still a syscall per such incident
371 (because the fd could point to a different file description now), so its
372 best to avoid that. Also, dup()ed file descriptors might not work very
373 well if you register events for both fds.</p>
374 <p>Please note that epoll sometimes generates spurious notifications, so you
375 need to use non-blocking I/O or other means to avoid blocking when no data
376 (or space) is available.</p>
378 <dt><code>EVBACKEND_KQUEUE</code> (value 8, most BSD clones)</dt>
380 <p>Kqueue deserves special mention, as at the time of this writing, it
381 was broken on all BSDs except NetBSD (usually it doesn't work with
382 anything but sockets and pipes, except on Darwin, where of course its
383 completely useless). For this reason its not being "autodetected"
384 unless you explicitly specify it explicitly in the flags (i.e. using
385 <code>EVBACKEND_KQUEUE</code>).</p>
386 <p>It scales in the same way as the epoll backend, but the interface to the
387 kernel is more efficient (which says nothing about its actual speed, of
388 course). While starting and stopping an I/O watcher does not cause an
389 extra syscall as with epoll, it still adds up to four event changes per
390 incident, so its best to avoid that.</p>
392 <dt><code>EVBACKEND_DEVPOLL</code> (value 16, Solaris 8)</dt>
394 <p>This is not implemented yet (and might never be).</p>
396 <dt><code>EVBACKEND_PORT</code> (value 32, Solaris 10)</dt>
398 <p>This uses the Solaris 10 port mechanism. As with everything on Solaris,
399 it's really slow, but it still scales very well (O(active_fds)).</p>
400 <p>Please note that solaris ports can result in a lot of spurious
401 notifications, so you need to use non-blocking I/O or other means to avoid
402 blocking when no data (or space) is available.</p>
404 <dt><code>EVBACKEND_ALL</code></dt>
406 <p>Try all backends (even potentially broken ones that wouldn't be tried
407 with <code>EVFLAG_AUTO</code>). Since this is a mask, you can do stuff such as
408 <code>EVBACKEND_ALL & ~EVBACKEND_KQUEUE</code>.</p>
412 <p>If one or more of these are ored into the flags value, then only these
413 backends will be tried (in the reverse order as given here). If none are
414 specified, most compiled-in backend will be tried, usually in reverse
415 order of their flag values :)</p>
416 <p>The most typical usage is like this:</p>
417 <pre> if (!ev_default_loop (0))
418 fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
421 <p>Restrict libev to the select and poll backends, and do not allow
422 environment settings to be taken into account:</p>
423 <pre> ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
426 <p>Use whatever libev has to offer, but make sure that kqueue is used if
427 available (warning, breaks stuff, best use only with your own private
428 event loop and only if you know the OS supports your types of fds):</p>
429 <pre> ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
433 <dt>struct ev_loop *ev_loop_new (unsigned int flags)</dt>
435 <p>Similar to <code>ev_default_loop</code>, but always creates a new event loop that is
436 always distinct from the default loop. Unlike the default loop, it cannot
437 handle signal and child watchers, and attempts to do so will be greeted by
438 undefined behaviour (or a failed assertion if assertions are enabled).</p>
439 <p>Example: Try to create a event loop that uses epoll and nothing else.</p>
440 <pre> struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
442 fatal ("no epoll found here, maybe it hides under your chair");
446 <dt>ev_default_destroy ()</dt>
448 <p>Destroys the default loop again (frees all memory and kernel state
449 etc.). None of the active event watchers will be stopped in the normal
450 sense, so e.g. <code>ev_is_active</code> might still return true. It is your
451 responsibility to either stop all watchers cleanly yoursef <i>before</i>
452 calling this function, or cope with the fact afterwards (which is usually
453 the easiest thing, youc na just ignore the watchers and/or <code>free ()</code> them
456 <dt>ev_loop_destroy (loop)</dt>
458 <p>Like <code>ev_default_destroy</code>, but destroys an event loop created by an
459 earlier call to <code>ev_loop_new</code>.</p>
461 <dt>ev_default_fork ()</dt>
463 <p>This function reinitialises the kernel state for backends that have
464 one. Despite the name, you can call it anytime, but it makes most sense
465 after forking, in either the parent or child process (or both, but that
466 again makes little sense).</p>
467 <p>You <i>must</i> call this function in the child process after forking if and
468 only if you want to use the event library in both processes. If you just
469 fork+exec, you don't have to call it.</p>
470 <p>The function itself is quite fast and it's usually not a problem to call
471 it just in case after a fork. To make this easy, the function will fit in
472 quite nicely into a call to <code>pthread_atfork</code>:</p>
473 <pre> pthread_atfork (0, 0, ev_default_fork);
476 <p>At the moment, <code>EVBACKEND_SELECT</code> and <code>EVBACKEND_POLL</code> are safe to use
477 without calling this function, so if you force one of those backends you
478 do not need to care.</p>
480 <dt>ev_loop_fork (loop)</dt>
482 <p>Like <code>ev_default_fork</code>, but acts on an event loop created by
483 <code>ev_loop_new</code>. Yes, you have to call this on every allocated event loop
484 after fork, and how you do this is entirely your own problem.</p>
486 <dt>unsigned int ev_backend (loop)</dt>
488 <p>Returns one of the <code>EVBACKEND_*</code> flags indicating the event backend in
491 <dt>ev_tstamp ev_now (loop)</dt>
493 <p>Returns the current "event loop time", which is the time the event loop
494 received events and started processing them. This timestamp does not
495 change as long as callbacks are being processed, and this is also the base
496 time used for relative timers. You can treat it as the timestamp of the
497 event occuring (or more correctly, libev finding out about it).</p>
499 <dt>ev_loop (loop, int flags)</dt>
501 <p>Finally, this is it, the event handler. This function usually is called
502 after you initialised all your watchers and you want to start handling
504 <p>If the flags argument is specified as <code>0</code>, it will not return until
505 either no event watchers are active anymore or <code>ev_unloop</code> was called.</p>
506 <p>Please note that an explicit <code>ev_unloop</code> is usually better than
507 relying on all watchers to be stopped when deciding when a program has
508 finished (especially in interactive programs), but having a program that
509 automatically loops as long as it has to and no longer by virtue of
510 relying on its watchers stopping correctly is a thing of beauty.</p>
511 <p>A flags value of <code>EVLOOP_NONBLOCK</code> will look for new events, will handle
512 those events and any outstanding ones, but will not block your process in
513 case there are no events and will return after one iteration of the loop.</p>
514 <p>A flags value of <code>EVLOOP_ONESHOT</code> will look for new events (waiting if
515 neccessary) and will handle those and any outstanding ones. It will block
516 your process until at least one new event arrives, and will return after
517 one iteration of the loop. This is useful if you are waiting for some
518 external event in conjunction with something not expressible using other
519 libev watchers. However, a pair of <code>ev_prepare</code>/<code>ev_check</code> watchers is
520 usually a better approach for this kind of thing.</p>
521 <p>Here are the gory details of what <code>ev_loop</code> does:</p>
522 <pre> * If there are no active watchers (reference count is zero), return.
523 - Queue prepare watchers and then call all outstanding watchers.
524 - If we have been forked, recreate the kernel state.
525 - Update the kernel state with all outstanding changes.
526 - Update the "event loop time".
527 - Calculate for how long to block.
528 - Block the process, waiting for any events.
529 - Queue all outstanding I/O (fd) events.
530 - Update the "event loop time" and do time jump handling.
531 - Queue all outstanding timers.
532 - Queue all outstanding periodics.
533 - If no events are pending now, queue all idle watchers.
534 - Queue all check watchers.
535 - Call all queued watchers in reverse order (i.e. check watchers first).
536 Signals and child watchers are implemented as I/O watchers, and will
537 be handled here by queueing them when their watcher gets executed.
538 - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
539 were used, return, otherwise continue with step *.
542 <p>Example: Queue some jobs and then loop until no events are outsanding
544 <pre> ... queue jobs here, make sure they register event watchers as long
545 ... as they still have work to do (even an idle watcher will do..)
546 ev_loop (my_loop, 0);
551 <dt>ev_unloop (loop, how)</dt>
553 <p>Can be used to make a call to <code>ev_loop</code> return early (but only after it
554 has processed all outstanding events). The <code>how</code> argument must be either
555 <code>EVUNLOOP_ONE</code>, which will make the innermost <code>ev_loop</code> call return, or
556 <code>EVUNLOOP_ALL</code>, which will make all nested <code>ev_loop</code> calls return.</p>
558 <dt>ev_ref (loop)</dt>
559 <dt>ev_unref (loop)</dt>
561 <p>Ref/unref can be used to add or remove a reference count on the event
562 loop: Every watcher keeps one reference, and as long as the reference
563 count is nonzero, <code>ev_loop</code> will not return on its own. If you have
564 a watcher you never unregister that should not keep <code>ev_loop</code> from
565 returning, ev_unref() after starting, and ev_ref() before stopping it. For
566 example, libev itself uses this for its internal signal pipe: It is not
567 visible to the libev user and should not keep <code>ev_loop</code> from exiting if
568 no event watchers registered by it are active. It is also an excellent
569 way to do this for generic recurring timers or from within third-party
570 libraries. Just remember to <i>unref after start</i> and <i>ref before stop</i>.</p>
571 <p>Example: Create a signal watcher, but keep it from keeping <code>ev_loop</code>
572 running when nothing else is active.</p>
573 <pre> struct ev_signal exitsig;
574 ev_signal_init (&exitsig, sig_cb, SIGINT);
575 ev_signal_start (loop, &exitsig);
579 <p>Example: For some weird reason, unregister the above signal handler again.</p>
581 ev_signal_stop (loop, &exitsig);
592 <h1 id="ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</h1>
593 <div id="ANATOMY_OF_A_WATCHER_CONTENT">
594 <p>A watcher is a structure that you create and register to record your
595 interest in some event. For instance, if you want to wait for STDIN to
596 become readable, you would create an <code>ev_io</code> watcher for that:</p>
597 <pre> static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)
600 ev_unloop (loop, EVUNLOOP_ALL);
603 struct ev_loop *loop = ev_default_loop (0);
604 struct ev_io stdin_watcher;
605 ev_init (&stdin_watcher, my_cb);
606 ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
607 ev_io_start (loop, &stdin_watcher);
611 <p>As you can see, you are responsible for allocating the memory for your
612 watcher structures (and it is usually a bad idea to do this on the stack,
613 although this can sometimes be quite valid).</p>
614 <p>Each watcher structure must be initialised by a call to <code>ev_init
615 (watcher *, callback)</code>, which expects a callback to be provided. This
616 callback gets invoked each time the event occurs (or, in the case of io
617 watchers, each time the event loop detects that the file descriptor given
618 is readable and/or writable).</p>
619 <p>Each watcher type has its own <code>ev_<type>_set (watcher *, ...)</code> macro
620 with arguments specific to this watcher type. There is also a macro
621 to combine initialisation and setting in one call: <code>ev_<type>_init
622 (watcher *, callback, ...)</code>.</p>
623 <p>To make the watcher actually watch out for events, you have to start it
624 with a watcher-specific start function (<code>ev_<type>_start (loop, watcher
625 *)</code>), and you can stop watching for events at any time by calling the
626 corresponding stop function (<code>ev_<type>_stop (loop, watcher *)</code>.</p>
627 <p>As long as your watcher is active (has been started but not stopped) you
628 must not touch the values stored in it. Most specifically you must never
629 reinitialise it or call its <code>set</code> macro.</p>
630 <p>Each and every callback receives the event loop pointer as first, the
631 registered watcher structure as second, and a bitset of received events as
633 <p>The received events usually include a single bit per event type received
634 (you can receive multiple events at the same time). The possible bit masks
637 <dt><code>EV_READ</code></dt>
638 <dt><code>EV_WRITE</code></dt>
640 <p>The file descriptor in the <code>ev_io</code> watcher has become readable and/or
643 <dt><code>EV_TIMEOUT</code></dt>
645 <p>The <code>ev_timer</code> watcher has timed out.</p>
647 <dt><code>EV_PERIODIC</code></dt>
649 <p>The <code>ev_periodic</code> watcher has timed out.</p>
651 <dt><code>EV_SIGNAL</code></dt>
653 <p>The signal specified in the <code>ev_signal</code> watcher has been received by a thread.</p>
655 <dt><code>EV_CHILD</code></dt>
657 <p>The pid specified in the <code>ev_child</code> watcher has received a status change.</p>
659 <dt><code>EV_STAT</code></dt>
661 <p>The path specified in the <code>ev_stat</code> watcher changed its attributes somehow.</p>
663 <dt><code>EV_IDLE</code></dt>
665 <p>The <code>ev_idle</code> watcher has determined that you have nothing better to do.</p>
667 <dt><code>EV_PREPARE</code></dt>
668 <dt><code>EV_CHECK</code></dt>
670 <p>All <code>ev_prepare</code> watchers are invoked just <i>before</i> <code>ev_loop</code> starts
671 to gather new events, and all <code>ev_check</code> watchers are invoked just after
672 <code>ev_loop</code> has gathered them, but before it invokes any callbacks for any
673 received events. Callbacks of both watcher types can start and stop as
674 many watchers as they want, and all of them will be taken into account
675 (for example, a <code>ev_prepare</code> watcher might start an idle watcher to keep
676 <code>ev_loop</code> from blocking).</p>
678 <dt><code>EV_EMBED</code></dt>
680 <p>The embedded event loop specified in the <code>ev_embed</code> watcher needs attention.</p>
682 <dt><code>EV_FORK</code></dt>
684 <p>The event loop has been resumed in the child process after fork (see
685 <code>ev_fork</code>).</p>
687 <dt><code>EV_ERROR</code></dt>
689 <p>An unspecified error has occured, the watcher has been stopped. This might
690 happen because the watcher could not be properly started because libev
691 ran out of memory, a file descriptor was found to be closed or any other
692 problem. You best act on it by reporting the problem and somehow coping
693 with the watcher being stopped.</p>
694 <p>Libev will usually signal a few "dummy" events together with an error,
695 for example it might indicate that a fd is readable or writable, and if
696 your callbacks is well-written it can just attempt the operation and cope
697 with the error from read() or write(). This will not work in multithreaded
698 programs, though, so beware.</p>
703 <h2 id="GENERIC_WATCHER_FUNCTIONS">GENERIC WATCHER FUNCTIONS</h2>
704 <div id="GENERIC_WATCHER_FUNCTIONS_CONTENT">
705 <p>In the following description, <code>TYPE</code> stands for the watcher type,
706 e.g. <code>timer</code> for <code>ev_timer</code> watchers and <code>io</code> for <code>ev_io</code> watchers.</p>
708 <dt><code>ev_init</code> (ev_TYPE *watcher, callback)</dt>
710 <p>This macro initialises the generic portion of a watcher. The contents
711 of the watcher object can be arbitrary (so <code>malloc</code> will do). Only
712 the generic parts of the watcher are initialised, you <i>need</i> to call
713 the type-specific <code>ev_TYPE_set</code> macro afterwards to initialise the
714 type-specific parts. For each type there is also a <code>ev_TYPE_init</code> macro
715 which rolls both calls into one.</p>
716 <p>You can reinitialise a watcher at any time as long as it has been stopped
717 (or never started) and there are no pending events outstanding.</p>
718 <p>The callback is always of type <code>void (*)(ev_loop *loop, ev_TYPE *watcher,
719 int revents)</code>.</p>
721 <dt><code>ev_TYPE_set</code> (ev_TYPE *, [args])</dt>
723 <p>This macro initialises the type-specific parts of a watcher. You need to
724 call <code>ev_init</code> at least once before you call this macro, but you can
725 call <code>ev_TYPE_set</code> any number of times. You must not, however, call this
726 macro on a watcher that is active (it can be pending, however, which is a
727 difference to the <code>ev_init</code> macro).</p>
728 <p>Although some watcher types do not have type-specific arguments
729 (e.g. <code>ev_prepare</code>) you still need to call its <code>set</code> macro.</p>
731 <dt><code>ev_TYPE_init</code> (ev_TYPE *watcher, callback, [args])</dt>
733 <p>This convinience macro rolls both <code>ev_init</code> and <code>ev_TYPE_set</code> macro
734 calls into a single call. This is the most convinient method to initialise
735 a watcher. The same limitations apply, of course.</p>
737 <dt><code>ev_TYPE_start</code> (loop *, ev_TYPE *watcher)</dt>
739 <p>Starts (activates) the given watcher. Only active watchers will receive
740 events. If the watcher is already active nothing will happen.</p>
742 <dt><code>ev_TYPE_stop</code> (loop *, ev_TYPE *watcher)</dt>
744 <p>Stops the given watcher again (if active) and clears the pending
745 status. It is possible that stopped watchers are pending (for example,
746 non-repeating timers are being stopped when they become pending), but
747 <code>ev_TYPE_stop</code> ensures that the watcher is neither active nor pending. If
748 you want to free or reuse the memory used by the watcher it is therefore a
749 good idea to always call its <code>ev_TYPE_stop</code> function.</p>
751 <dt>bool ev_is_active (ev_TYPE *watcher)</dt>
753 <p>Returns a true value iff the watcher is active (i.e. it has been started
754 and not yet been stopped). As long as a watcher is active you must not modify
757 <dt>bool ev_is_pending (ev_TYPE *watcher)</dt>
759 <p>Returns a true value iff the watcher is pending, (i.e. it has outstanding
760 events but its callback has not yet been invoked). As long as a watcher
761 is pending (but not active) you must not call an init function on it (but
762 <code>ev_TYPE_set</code> is safe) and you must make sure the watcher is available to
763 libev (e.g. you cnanot <code>free ()</code> it).</p>
765 <dt>callback ev_cb (ev_TYPE *watcher)</dt>
767 <p>Returns the callback currently set on the watcher.</p>
769 <dt>ev_cb_set (ev_TYPE *watcher, callback)</dt>
771 <p>Change the callback. You can change the callback at virtually any time
772 (modulo threads).</p>
781 <h2 id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</h2>
782 <div id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH-2">
783 <p>Each watcher has, by default, a member <code>void *data</code> that you can change
784 and read at any time, libev will completely ignore it. This can be used
785 to associate arbitrary data with your watcher. If you need more data and
786 don't want to allocate memory and store a pointer to it in that data
787 member, you can also "subclass" the watcher type and provide your own
794 struct whatever *mostinteresting;
798 <p>And since your callback will be called with a pointer to the watcher, you
799 can cast it back to your own type:</p>
800 <pre> static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)
802 struct my_io *w = (struct my_io *)w_;
807 <p>More interesting and less C-conformant ways of casting your callback type
808 instead have been omitted.</p>
809 <p>Another common scenario is having some data structure with multiple
811 <pre> struct my_biggy
819 <p>In this case getting the pointer to <code>my_biggy</code> is a bit more complicated,
820 you need to use <code>offsetof</code>:</p>
821 <pre> #include <stddef.h>
824 t1_cb (EV_P_ struct ev_timer *w, int revents)
826 struct my_biggy big = (struct my_biggy *
827 (((char *)w) - offsetof (struct my_biggy, t1));
831 t2_cb (EV_P_ struct ev_timer *w, int revents)
833 struct my_biggy big = (struct my_biggy *
834 (((char *)w) - offsetof (struct my_biggy, t2));
843 <h1 id="WATCHER_TYPES">WATCHER TYPES</h1>
844 <div id="WATCHER_TYPES_CONTENT">
845 <p>This section describes each watcher in detail, but will not repeat
846 information given in the last section. Any initialisation/set macros,
847 functions and members specific to the watcher type are explained.</p>
848 <p>Members are additionally marked with either <i>[read-only]</i>, meaning that,
849 while the watcher is active, you can look at the member and expect some
850 sensible content, but you must not modify it (you can modify it while the
851 watcher is stopped to your hearts content), or <i>[read-write]</i>, which
852 means you can expect it to have some sensible content while the watcher
853 is active, but you can also modify it. Modifying it may not do something
854 sensible or take immediate effect (or do anything at all), but libev will
855 not crash or malfunction in any way.</p>
862 <h2 id="code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable?</h2>
863 <div id="code_ev_io_code_is_this_file_descrip-2">
864 <p>I/O watchers check whether a file descriptor is readable or writable
865 in each iteration of the event loop, or, more precisely, when reading
866 would not block the process and writing would at least be able to write
867 some data. This behaviour is called level-triggering because you keep
868 receiving events as long as the condition persists. Remember you can stop
869 the watcher if you don't want to act on the event and neither want to
870 receive future events.</p>
871 <p>In general you can register as many read and/or write event watchers per
872 fd as you want (as long as you don't confuse yourself). Setting all file
873 descriptors to non-blocking mode is also usually a good idea (but not
874 required if you know what you are doing).</p>
875 <p>You have to be careful with dup'ed file descriptors, though. Some backends
876 (the linux epoll backend is a notable example) cannot handle dup'ed file
877 descriptors correctly if you register interest in two or more fds pointing
878 to the same underlying file/socket/etc. description (that is, they share
879 the same underlying "file open").</p>
880 <p>If you must do this, then force the use of a known-to-be-good backend
881 (at the time of this writing, this includes only <code>EVBACKEND_SELECT</code> and
882 <code>EVBACKEND_POLL</code>).</p>
883 <p>Another thing you have to watch out for is that it is quite easy to
884 receive "spurious" readyness notifications, that is your callback might
885 be called with <code>EV_READ</code> but a subsequent <code>read</code>(2) will actually block
886 because there is no data. Not only are some backends known to create a
887 lot of those (for example solaris ports), it is very easy to get into
888 this situation even with a relatively standard program structure. Thus
889 it is best to always use non-blocking I/O: An extra <code>read</code>(2) returning
890 <code>EAGAIN</code> is far preferable to a program hanging until some data arrives.</p>
891 <p>If you cannot run the fd in non-blocking mode (for example you should not
892 play around with an Xlib connection), then you have to seperately re-test
893 wether a file descriptor is really ready with a known-to-be good interface
894 such as poll (fortunately in our Xlib example, Xlib already does this on
895 its own, so its quite safe to use).</p>
897 <dt>ev_io_init (ev_io *, callback, int fd, int events)</dt>
898 <dt>ev_io_set (ev_io *, int fd, int events)</dt>
900 <p>Configures an <code>ev_io</code> watcher. The <code>fd</code> is the file descriptor to
901 rceeive events for and events is either <code>EV_READ</code>, <code>EV_WRITE</code> or
902 <code>EV_READ | EV_WRITE</code> to receive the given events.</p>
904 <dt>int fd [read-only]</dt>
906 <p>The file descriptor being watched.</p>
908 <dt>int events [read-only]</dt>
910 <p>The events being watched.</p>
913 <p>Example: Call <code>stdin_readable_cb</code> when STDIN_FILENO has become, well
914 readable, but only once. Since it is likely line-buffered, you could
915 attempt to read a whole line in the callback.</p>
917 stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)
919 ev_io_stop (loop, w);
920 .. read from stdin here (or from w->fd) and haqndle any I/O errors
924 struct ev_loop *loop = ev_default_init (0);
925 struct ev_io stdin_readable;
926 ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
927 ev_io_start (loop, &stdin_readable);
936 <h2 id="code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally repeating timeouts</h2>
937 <div id="code_ev_timer_code_relative_and_opti-2">
938 <p>Timer watchers are simple relative timers that generate an event after a
939 given time, and optionally repeating in regular intervals after that.</p>
940 <p>The timers are based on real time, that is, if you register an event that
941 times out after an hour and you reset your system clock to last years
942 time, it will still time out after (roughly) and hour. "Roughly" because
943 detecting time jumps is hard, and some inaccuracies are unavoidable (the
944 monotonic clock option helps a lot here).</p>
945 <p>The relative timeouts are calculated relative to the <code>ev_now ()</code>
946 time. This is usually the right thing as this timestamp refers to the time
947 of the event triggering whatever timeout you are modifying/starting. If
948 you suspect event processing to be delayed and you <i>need</i> to base the timeout
949 on the current time, use something like this to adjust for this:</p>
950 <pre> ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
953 <p>The callback is guarenteed to be invoked only when its timeout has passed,
954 but if multiple timers become ready during the same loop iteration then
955 order of execution is undefined.</p>
957 <dt>ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)</dt>
958 <dt>ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)</dt>
960 <p>Configure the timer to trigger after <code>after</code> seconds. If <code>repeat</code> is
961 <code>0.</code>, then it will automatically be stopped. If it is positive, then the
962 timer will automatically be configured to trigger again <code>repeat</code> seconds
963 later, again, and again, until stopped manually.</p>
964 <p>The timer itself will do a best-effort at avoiding drift, that is, if you
965 configure a timer to trigger every 10 seconds, then it will trigger at
966 exactly 10 second intervals. If, however, your program cannot keep up with
967 the timer (because it takes longer than those 10 seconds to do stuff) the
968 timer will not fire more than once per event loop iteration.</p>
970 <dt>ev_timer_again (loop)</dt>
972 <p>This will act as if the timer timed out and restart it again if it is
973 repeating. The exact semantics are:</p>
974 <p>If the timer is pending, its pending status is cleared.</p>
975 <p>If the timer is started but nonrepeating, stop it (as if it timed out).</p>
976 <p>If the timer is repeating, either start it if necessary (with the
977 <code>repeat</code> value), or reset the running timer to the <code>repeat</code> value.</p>
978 <p>This sounds a bit complicated, but here is a useful and typical
979 example: Imagine you have a tcp connection and you want a so-called idle
980 timeout, that is, you want to be called when there have been, say, 60
981 seconds of inactivity on the socket. The easiest way to do this is to
982 configure an <code>ev_timer</code> with a <code>repeat</code> value of <code>60</code> and then call
983 <code>ev_timer_again</code> each time you successfully read or write some data. If
984 you go into an idle state where you do not expect data to travel on the
985 socket, you can <code>ev_timer_stop</code> the timer, and <code>ev_timer_again</code> will
986 automatically restart it if need be.</p>
987 <p>That means you can ignore the <code>after</code> value and <code>ev_timer_start</code>
988 altogether and only ever use the <code>repeat</code> value and <code>ev_timer_again</code>:</p>
989 <pre> ev_timer_init (timer, callback, 0., 5.);
990 ev_timer_again (loop, timer);
992 timer->again = 17.;
993 ev_timer_again (loop, timer);
995 timer->again = 10.;
996 ev_timer_again (loop, timer);
999 <p>This is more slightly efficient then stopping/starting the timer each time
1000 you want to modify its timeout value.</p>
1002 <dt>ev_tstamp repeat [read-write]</dt>
1004 <p>The current <code>repeat</code> value. Will be used each time the watcher times out
1005 or <code>ev_timer_again</code> is called and determines the next timeout (if any),
1006 which is also when any modifications are taken into account.</p>
1009 <p>Example: Create a timer that fires after 60 seconds.</p>
1011 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1013 .. one minute over, w is actually stopped right here
1016 struct ev_timer mytimer;
1017 ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
1018 ev_timer_start (loop, &mytimer);
1021 <p>Example: Create a timeout timer that times out after 10 seconds of
1024 timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1026 .. ten seconds without any activity
1029 struct ev_timer mytimer;
1030 ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
1031 ev_timer_again (&mytimer); /* start timer */
1034 // and in some piece of code that gets executed on any "activity":
1035 // reset the timeout to start ticking again at 10 seconds
1036 ev_timer_again (&mytimer);
1044 <h2 id="code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron?</h2>
1045 <div id="code_ev_periodic_code_to_cron_or_not-2">
1046 <p>Periodic watchers are also timers of a kind, but they are very versatile
1047 (and unfortunately a bit complex).</p>
1048 <p>Unlike <code>ev_timer</code>'s, they are not based on real time (or relative time)
1049 but on wallclock time (absolute time). You can tell a periodic watcher
1050 to trigger "at" some specific point in time. For example, if you tell a
1051 periodic watcher to trigger in 10 seconds (by specifiying e.g. <code>ev_now ()
1052 + 10.</code>) and then reset your system clock to the last year, then it will
1053 take a year to trigger the event (unlike an <code>ev_timer</code>, which would trigger
1054 roughly 10 seconds later and of course not if you reset your system time
1056 <p>They can also be used to implement vastly more complex timers, such as
1057 triggering an event on eahc midnight, local time.</p>
1058 <p>As with timers, the callback is guarenteed to be invoked only when the
1059 time (<code>at</code>) has been passed, but if multiple periodic timers become ready
1060 during the same loop iteration then order of execution is undefined.</p>
1062 <dt>ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)</dt>
1063 <dt>ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)</dt>
1065 <p>Lots of arguments, lets sort it out... There are basically three modes of
1066 operation, and we will explain them from simplest to complex:</p>
1069 <dt>* absolute timer (interval = reschedule_cb = 0)</dt>
1071 <p>In this configuration the watcher triggers an event at the wallclock time
1072 <code>at</code> and doesn't repeat. It will not adjust when a time jump occurs,
1073 that is, if it is to be run at January 1st 2011 then it will run when the
1074 system time reaches or surpasses this time.</p>
1076 <dt>* non-repeating interval timer (interval > 0, reschedule_cb = 0)</dt>
1078 <p>In this mode the watcher will always be scheduled to time out at the next
1079 <code>at + N * interval</code> time (for some integer N) and then repeat, regardless
1080 of any time jumps.</p>
1081 <p>This can be used to create timers that do not drift with respect to system
1083 <pre> ev_periodic_set (&periodic, 0., 3600., 0);
1086 <p>This doesn't mean there will always be 3600 seconds in between triggers,
1087 but only that the the callback will be called when the system time shows a
1088 full hour (UTC), or more correctly, when the system time is evenly divisible
1090 <p>Another way to think about it (for the mathematically inclined) is that
1091 <code>ev_periodic</code> will try to run the callback in this mode at the next possible
1092 time where <code>time = at (mod interval)</code>, regardless of any time jumps.</p>
1094 <dt>* manual reschedule mode (reschedule_cb = callback)</dt>
1096 <p>In this mode the values for <code>interval</code> and <code>at</code> are both being
1097 ignored. Instead, each time the periodic watcher gets scheduled, the
1098 reschedule callback will be called with the watcher as first, and the
1099 current time as second argument.</p>
1100 <p>NOTE: <i>This callback MUST NOT stop or destroy any periodic watcher,
1101 ever, or make any event loop modifications</i>. If you need to stop it,
1102 return <code>now + 1e30</code> (or so, fudge fudge) and stop it afterwards (e.g. by
1103 starting a prepare watcher).</p>
1104 <p>Its prototype is <code>ev_tstamp (*reschedule_cb)(struct ev_periodic *w,
1105 ev_tstamp now)</code>, e.g.:</p>
1106 <pre> static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)
1112 <p>It must return the next time to trigger, based on the passed time value
1113 (that is, the lowest time value larger than to the second argument). It
1114 will usually be called just before the callback will be triggered, but
1115 might be called at other times, too.</p>
1116 <p>NOTE: <i>This callback must always return a time that is later than the
1117 passed <code>now</code> value</i>. Not even <code>now</code> itself will do, it <i>must</i> be larger.</p>
1118 <p>This can be used to create very complex timers, such as a timer that
1119 triggers on each midnight, local time. To do this, you would calculate the
1120 next midnight after <code>now</code> and return the timestamp value for this. How
1121 you do this is, again, up to you (but it is not trivial, which is the main
1122 reason I omitted it as an example).</p>
1127 <dt>ev_periodic_again (loop, ev_periodic *)</dt>
1129 <p>Simply stops and restarts the periodic watcher again. This is only useful
1130 when you changed some parameters or the reschedule callback would return
1131 a different time than the last time it was called (e.g. in a crond like
1132 program when the crontabs have changed).</p>
1134 <dt>ev_tstamp interval [read-write]</dt>
1136 <p>The current interval value. Can be modified any time, but changes only
1137 take effect when the periodic timer fires or <code>ev_periodic_again</code> is being
1140 <dt>ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]</dt>
1142 <p>The current reschedule callback, or <code>0</code>, if this functionality is
1143 switched off. Can be changed any time, but changes only take effect when
1144 the periodic timer fires or <code>ev_periodic_again</code> is being called.</p>
1147 <p>Example: Call a callback every hour, or, more precisely, whenever the
1148 system clock is divisible by 3600. The callback invocation times have
1149 potentially a lot of jittering, but good long-term stability.</p>
1151 clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)
1153 ... its now a full hour (UTC, or TAI or whatever your clock follows)
1156 struct ev_periodic hourly_tick;
1157 ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
1158 ev_periodic_start (loop, &hourly_tick);
1161 <p>Example: The same as above, but use a reschedule callback to do it:</p>
1162 <pre> #include <math.h>
1165 my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)
1167 return fmod (now, 3600.) + 3600.;
1170 ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
1173 <p>Example: Call a callback every hour, starting now:</p>
1174 <pre> struct ev_periodic hourly_tick;
1175 ev_periodic_init (&hourly_tick, clock_cb,
1176 fmod (ev_now (loop), 3600.), 3600., 0);
1177 ev_periodic_start (loop, &hourly_tick);
1185 <h2 id="code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled!</h2>
1186 <div id="code_ev_signal_code_signal_me_when_a-2">
1187 <p>Signal watchers will trigger an event when the process receives a specific
1188 signal one or more times. Even though signals are very asynchronous, libev
1189 will try it's best to deliver signals synchronously, i.e. as part of the
1190 normal event processing, like any other event.</p>
1191 <p>You can configure as many watchers as you like per signal. Only when the
1192 first watcher gets started will libev actually register a signal watcher
1193 with the kernel (thus it coexists with your own signal handlers as long
1194 as you don't register any with libev). Similarly, when the last signal
1195 watcher for a signal is stopped libev will reset the signal handler to
1196 SIG_DFL (regardless of what it was set to before).</p>
1198 <dt>ev_signal_init (ev_signal *, callback, int signum)</dt>
1199 <dt>ev_signal_set (ev_signal *, int signum)</dt>
1201 <p>Configures the watcher to trigger on the given signal number (usually one
1202 of the <code>SIGxxx</code> constants).</p>
1204 <dt>int signum [read-only]</dt>
1206 <p>The signal the watcher watches out for.</p>
1215 <h2 id="code_ev_child_code_watch_out_for_pro"><code>ev_child</code> - watch out for process status changes</h2>
1216 <div id="code_ev_child_code_watch_out_for_pro-2">
1217 <p>Child watchers trigger when your process receives a SIGCHLD in response to
1218 some child status changes (most typically when a child of yours dies).</p>
1220 <dt>ev_child_init (ev_child *, callback, int pid)</dt>
1221 <dt>ev_child_set (ev_child *, int pid)</dt>
1223 <p>Configures the watcher to wait for status changes of process <code>pid</code> (or
1224 <i>any</i> process if <code>pid</code> is specified as <code>0</code>). The callback can look
1225 at the <code>rstatus</code> member of the <code>ev_child</code> watcher structure to see
1226 the status word (use the macros from <code>sys/wait.h</code> and see your systems
1227 <code>waitpid</code> documentation). The <code>rpid</code> member contains the pid of the
1228 process causing the status change.</p>
1230 <dt>int pid [read-only]</dt>
1232 <p>The process id this watcher watches out for, or <code>0</code>, meaning any process id.</p>
1234 <dt>int rpid [read-write]</dt>
1236 <p>The process id that detected a status change.</p>
1238 <dt>int rstatus [read-write]</dt>
1240 <p>The process exit/trace status caused by <code>rpid</code> (see your systems
1241 <code>waitpid</code> and <code>sys/wait.h</code> documentation for details).</p>
1244 <p>Example: Try to exit cleanly on SIGINT and SIGTERM.</p>
1246 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
1248 ev_unloop (loop, EVUNLOOP_ALL);
1251 struct ev_signal signal_watcher;
1252 ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
1253 ev_signal_start (loop, &sigint_cb);
1261 <h2 id="code_ev_stat_code_did_the_file_attri"><code>ev_stat</code> - did the file attributes just change?</h2>
1262 <div id="code_ev_stat_code_did_the_file_attri-2">
1263 <p>This watches a filesystem path for attribute changes. That is, it calls
1264 <code>stat</code> regularly (or when the OS says it changed) and sees if it changed
1265 compared to the last time, invoking the callback if it did.</p>
1266 <p>The path does not need to exist: changing from "path exists" to "path does
1267 not exist" is a status change like any other. The condition "path does
1268 not exist" is signified by the <code>st_nlink</code> field being zero (which is
1269 otherwise always forced to be at least one) and all the other fields of
1270 the stat buffer having unspecified contents.</p>
1271 <p>The path <i>should</i> be absolute and <i>must not</i> end in a slash. If it is
1272 relative and your working directory changes, the behaviour is undefined.</p>
1273 <p>Since there is no standard to do this, the portable implementation simply
1274 calls <code>stat (2)</code> regularly on the path to see if it changed somehow. You
1275 can specify a recommended polling interval for this case. If you specify
1276 a polling interval of <code>0</code> (highly recommended!) then a <i>suitable,
1277 unspecified default</i> value will be used (which you can expect to be around
1278 five seconds, although this might change dynamically). Libev will also
1279 impose a minimum interval which is currently around <code>0.1</code>, but thats
1280 usually overkill.</p>
1281 <p>This watcher type is not meant for massive numbers of stat watchers,
1282 as even with OS-supported change notifications, this can be
1283 resource-intensive.</p>
1284 <p>At the time of this writing, only the Linux inotify interface is
1285 implemented (implementing kqueue support is left as an exercise for the
1286 reader). Inotify will be used to give hints only and should not change the
1287 semantics of <code>ev_stat</code> watchers, which means that libev sometimes needs
1288 to fall back to regular polling again even with inotify, but changes are
1289 usually detected immediately, and if the file exists there will be no
1292 <dt>ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)</dt>
1293 <dt>ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)</dt>
1295 <p>Configures the watcher to wait for status changes of the given
1296 <code>path</code>. The <code>interval</code> is a hint on how quickly a change is expected to
1297 be detected and should normally be specified as <code>0</code> to let libev choose
1298 a suitable value. The memory pointed to by <code>path</code> must point to the same
1299 path for as long as the watcher is active.</p>
1300 <p>The callback will be receive <code>EV_STAT</code> when a change was detected,
1301 relative to the attributes at the time the watcher was started (or the
1302 last change was detected).</p>
1304 <dt>ev_stat_stat (ev_stat *)</dt>
1306 <p>Updates the stat buffer immediately with new values. If you change the
1307 watched path in your callback, you could call this fucntion to avoid
1308 detecting this change (while introducing a race condition). Can also be
1309 useful simply to find out the new values.</p>
1311 <dt>ev_statdata attr [read-only]</dt>
1313 <p>The most-recently detected attributes of the file. Although the type is of
1314 <code>ev_statdata</code>, this is usually the (or one of the) <code>struct stat</code> types
1315 suitable for your system. If the <code>st_nlink</code> member is <code>0</code>, then there
1316 was some error while <code>stat</code>ing the file.</p>
1318 <dt>ev_statdata prev [read-only]</dt>
1320 <p>The previous attributes of the file. The callback gets invoked whenever
1321 <code>prev</code> != <code>attr</code>.</p>
1323 <dt>ev_tstamp interval [read-only]</dt>
1325 <p>The specified interval.</p>
1327 <dt>const char *path [read-only]</dt>
1329 <p>The filesystem path that is being watched.</p>
1332 <p>Example: Watch <code>/etc/passwd</code> for attribute changes.</p>
1334 passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1336 /* /etc/passwd changed in some way */
1337 if (w->attr.st_nlink)
1339 printf ("passwd current size %ld\n", (long)w->attr.st_size);
1340 printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
1341 printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
1344 /* you shalt not abuse printf for puts */
1345 puts ("wow, /etc/passwd is not there, expect problems. "
1346 "if this is windows, they already arrived\n");
1352 ev_stat_init (&passwd, passwd_cb, "/etc/passwd");
1353 ev_stat_start (loop, &passwd);
1361 <h2 id="code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do...</h2>
1362 <div id="code_ev_idle_code_when_you_ve_got_no-2">
1363 <p>Idle watchers trigger events when there are no other events are pending
1364 (prepare, check and other idle watchers do not count). That is, as long
1365 as your process is busy handling sockets or timeouts (or even signals,
1366 imagine) it will not be triggered. But when your process is idle all idle
1367 watchers are being called again and again, once per event loop iteration -
1368 until stopped, that is, or your process receives more events and becomes
1370 <p>The most noteworthy effect is that as long as any idle watchers are
1371 active, the process will not block when waiting for new events.</p>
1372 <p>Apart from keeping your process non-blocking (which is a useful
1373 effect on its own sometimes), idle watchers are a good place to do
1374 "pseudo-background processing", or delay processing stuff to after the
1375 event loop has handled all outstanding events.</p>
1377 <dt>ev_idle_init (ev_signal *, callback)</dt>
1379 <p>Initialises and configures the idle watcher - it has no parameters of any
1380 kind. There is a <code>ev_idle_set</code> macro, but using it is utterly pointless,
1384 <p>Example: Dynamically allocate an <code>ev_idle</code> watcher, start it, and in the
1385 callback, free it. Also, use no error checking, as usual.</p>
1387 idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
1390 // now do something you wanted to do when the program has
1391 // no longer asnything immediate to do.
1394 struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
1395 ev_idle_init (idle_watcher, idle_cb);
1396 ev_idle_start (loop, idle_cb);
1404 <h2 id="code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop!</h2>
1405 <div id="code_ev_prepare_code_and_code_ev_che-2">
1406 <p>Prepare and check watchers are usually (but not always) used in tandem:
1407 prepare watchers get invoked before the process blocks and check watchers
1409 <p>You <i>must not</i> call <code>ev_loop</code> or similar functions that enter
1410 the current event loop from either <code>ev_prepare</code> or <code>ev_check</code>
1411 watchers. Other loops than the current one are fine, however. The
1412 rationale behind this is that you do not need to check for recursion in
1413 those watchers, i.e. the sequence will always be <code>ev_prepare</code>, blocking,
1414 <code>ev_check</code> so if you have one watcher of each kind they will always be
1415 called in pairs bracketing the blocking call.</p>
1416 <p>Their main purpose is to integrate other event mechanisms into libev and
1417 their use is somewhat advanced. This could be used, for example, to track
1418 variable changes, implement your own watchers, integrate net-snmp or a
1419 coroutine library and lots more. They are also occasionally useful if
1420 you cache some data and want to flush it before blocking (for example,
1421 in X programs you might want to do an <code>XFlush ()</code> in an <code>ev_prepare</code>
1423 <p>This is done by examining in each prepare call which file descriptors need
1424 to be watched by the other library, registering <code>ev_io</code> watchers for
1425 them and starting an <code>ev_timer</code> watcher for any timeouts (many libraries
1426 provide just this functionality). Then, in the check watcher you check for
1427 any events that occured (by checking the pending status of all watchers
1428 and stopping them) and call back into the library. The I/O and timer
1429 callbacks will never actually be called (but must be valid nevertheless,
1430 because you never know, you know?).</p>
1431 <p>As another example, the Perl Coro module uses these hooks to integrate
1432 coroutines into libev programs, by yielding to other active coroutines
1433 during each prepare and only letting the process block if no coroutines
1434 are ready to run (it's actually more complicated: it only runs coroutines
1435 with priority higher than or equal to the event loop and one coroutine
1436 of lower priority, but only once, using idle watchers to keep the event
1437 loop from blocking if lower-priority coroutines are active, thus mapping
1438 low-priority coroutines to idle/background tasks).</p>
1440 <dt>ev_prepare_init (ev_prepare *, callback)</dt>
1441 <dt>ev_check_init (ev_check *, callback)</dt>
1443 <p>Initialises and configures the prepare or check watcher - they have no
1444 parameters of any kind. There are <code>ev_prepare_set</code> and <code>ev_check_set</code>
1445 macros, but using them is utterly, utterly and completely pointless.</p>
1448 <p>Example: To include a library such as adns, you would add IO watchers
1449 and a timeout watcher in a prepare handler, as required by libadns, and
1450 in a check watcher, destroy them and call into libadns. What follows is
1451 pseudo-code only of course:</p>
1452 <pre> static ev_io iow [nfd];
1456 io_cb (ev_loop *loop, ev_io *w, int revents)
1458 // set the relevant poll flags
1459 // could also call adns_processreadable etc. here
1460 struct pollfd *fd = (struct pollfd *)w->data;
1461 if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
1462 if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
1465 // create io watchers for each fd and a timer before blocking
1467 adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)
1469 int timeout = 3600000;truct pollfd fds [nfd];
1470 // actual code will need to loop here and realloc etc.
1471 adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
1473 /* the callback is illegal, but won't be called as we stop during check */
1474 ev_timer_init (&tw, 0, timeout * 1e-3);
1475 ev_timer_start (loop, &tw);
1477 // create on ev_io per pollfd
1478 for (int i = 0; i < nfd; ++i)
1480 ev_io_init (iow + i, io_cb, fds [i].fd,
1481 ((fds [i].events & POLLIN ? EV_READ : 0)
1482 | (fds [i].events & POLLOUT ? EV_WRITE : 0)));
1484 fds [i].revents = 0;
1485 iow [i].data = fds + i;
1486 ev_io_start (loop, iow + i);
1490 // stop all watchers after blocking
1492 adns_check_cb (ev_loop *loop, ev_check *w, int revents)
1494 ev_timer_stop (loop, &tw);
1496 for (int i = 0; i < nfd; ++i)
1497 ev_io_stop (loop, iow + i);
1499 adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
1508 <h2 id="code_ev_embed_code_when_one_backend_"><code>ev_embed</code> - when one backend isn't enough...</h2>
1509 <div id="code_ev_embed_code_when_one_backend_-2">
1510 <p>This is a rather advanced watcher type that lets you embed one event loop
1511 into another (currently only <code>ev_io</code> events are supported in the embedded
1512 loop, other types of watchers might be handled in a delayed or incorrect
1513 fashion and must not be used).</p>
1514 <p>There are primarily two reasons you would want that: work around bugs and
1516 <p>As an example for a bug workaround, the kqueue backend might only support
1517 sockets on some platform, so it is unusable as generic backend, but you
1518 still want to make use of it because you have many sockets and it scales
1519 so nicely. In this case, you would create a kqueue-based loop and embed it
1520 into your default loop (which might use e.g. poll). Overall operation will
1521 be a bit slower because first libev has to poll and then call kevent, but
1522 at least you can use both at what they are best.</p>
1523 <p>As for prioritising I/O: rarely you have the case where some fds have
1524 to be watched and handled very quickly (with low latency), and even
1525 priorities and idle watchers might have too much overhead. In this case
1526 you would put all the high priority stuff in one loop and all the rest in
1527 a second one, and embed the second one in the first.</p>
1528 <p>As long as the watcher is active, the callback will be invoked every time
1529 there might be events pending in the embedded loop. The callback must then
1530 call <code>ev_embed_sweep (mainloop, watcher)</code> to make a single sweep and invoke
1531 their callbacks (you could also start an idle watcher to give the embedded
1532 loop strictly lower priority for example). You can also set the callback
1533 to <code>0</code>, in which case the embed watcher will automatically execute the
1534 embedded loop sweep.</p>
1535 <p>As long as the watcher is started it will automatically handle events. The
1536 callback will be invoked whenever some events have been handled. You can
1537 set the callback to <code>0</code> to avoid having to specify one if you are not
1538 interested in that.</p>
1539 <p>Also, there have not currently been made special provisions for forking:
1540 when you fork, you not only have to call <code>ev_loop_fork</code> on both loops,
1541 but you will also have to stop and restart any <code>ev_embed</code> watchers
1543 <p>Unfortunately, not all backends are embeddable, only the ones returned by
1544 <code>ev_embeddable_backends</code> are, which, unfortunately, does not include any
1546 <p>So when you want to use this feature you will always have to be prepared
1547 that you cannot get an embeddable loop. The recommended way to get around
1548 this is to have a separate variables for your embeddable loop, try to
1549 create it, and if that fails, use the normal loop for everything:</p>
1550 <pre> struct ev_loop *loop_hi = ev_default_init (0);
1551 struct ev_loop *loop_lo = 0;
1552 struct ev_embed embed;
1554 // see if there is a chance of getting one that works
1555 // (remember that a flags value of 0 means autodetection)
1556 loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
1557 ? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
1560 // if we got one, then embed it, otherwise default to loop_hi
1563 ev_embed_init (&embed, 0, loop_lo);
1564 ev_embed_start (loop_hi, &embed);
1571 <dt>ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
1572 <dt>ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
1574 <p>Configures the watcher to embed the given loop, which must be
1575 embeddable. If the callback is <code>0</code>, then <code>ev_embed_sweep</code> will be
1576 invoked automatically, otherwise it is the responsibility of the callback
1577 to invoke it (it will continue to be called until the sweep has been done,
1578 if you do not want thta, you need to temporarily stop the embed watcher).</p>
1580 <dt>ev_embed_sweep (loop, ev_embed *)</dt>
1582 <p>Make a single, non-blocking sweep over the embedded loop. This works
1583 similarly to <code>ev_loop (embedded_loop, EVLOOP_NONBLOCK)</code>, but in the most
1584 apropriate way for embedded loops.</p>
1586 <dt>struct ev_loop *loop [read-only]</dt>
1588 <p>The embedded event loop.</p>
1597 <h2 id="code_ev_fork_code_the_audacity_to_re"><code>ev_fork</code> - the audacity to resume the event loop after a fork</h2>
1598 <div id="code_ev_fork_code_the_audacity_to_re-2">
1599 <p>Fork watchers are called when a <code>fork ()</code> was detected (usually because
1600 whoever is a good citizen cared to tell libev about it by calling
1601 <code>ev_default_fork</code> or <code>ev_loop_fork</code>). The invocation is done before the
1602 event loop blocks next and before <code>ev_check</code> watchers are being called,
1603 and only in the child after the fork. If whoever good citizen calling
1604 <code>ev_default_fork</code> cheats and calls it in the wrong process, the fork
1605 handlers will be invoked, too, of course.</p>
1607 <dt>ev_fork_init (ev_signal *, callback)</dt>
1609 <p>Initialises and configures the fork watcher - it has no parameters of any
1610 kind. There is a <code>ev_fork_set</code> macro, but using it is utterly pointless,
1620 <h1 id="OTHER_FUNCTIONS">OTHER FUNCTIONS</h1>
1621 <div id="OTHER_FUNCTIONS_CONTENT">
1622 <p>There are some other functions of possible interest. Described. Here. Now.</p>
1624 <dt>ev_once (loop, int fd, int events, ev_tstamp timeout, callback)</dt>
1626 <p>This function combines a simple timer and an I/O watcher, calls your
1627 callback on whichever event happens first and automatically stop both
1628 watchers. This is useful if you want to wait for a single event on an fd
1629 or timeout without having to allocate/configure/start/stop/free one or
1630 more watchers yourself.</p>
1631 <p>If <code>fd</code> is less than 0, then no I/O watcher will be started and events
1632 is being ignored. Otherwise, an <code>ev_io</code> watcher for the given <code>fd</code> and
1633 <code>events</code> set will be craeted and started.</p>
1634 <p>If <code>timeout</code> is less than 0, then no timeout watcher will be
1635 started. Otherwise an <code>ev_timer</code> watcher with after = <code>timeout</code> (and
1636 repeat = 0) will be started. While <code>0</code> is a valid timeout, it is of
1638 <p>The callback has the type <code>void (*cb)(int revents, void *arg)</code> and gets
1639 passed an <code>revents</code> set like normal event callbacks (a combination of
1640 <code>EV_ERROR</code>, <code>EV_READ</code>, <code>EV_WRITE</code> or <code>EV_TIMEOUT</code>) and the <code>arg</code>
1641 value passed to <code>ev_once</code>:</p>
1642 <pre> static void stdin_ready (int revents, void *arg)
1644 if (revents & EV_TIMEOUT)
1645 /* doh, nothing entered */;
1646 else if (revents & EV_READ)
1647 /* stdin might have data for us, joy! */;
1650 ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
1654 <dt>ev_feed_event (ev_loop *, watcher *, int revents)</dt>
1656 <p>Feeds the given event set into the event loop, as if the specified event
1657 had happened for the specified watcher (which must be a pointer to an
1658 initialised but not necessarily started event watcher).</p>
1660 <dt>ev_feed_fd_event (ev_loop *, int fd, int revents)</dt>
1662 <p>Feed an event on the given fd, as if a file descriptor backend detected
1663 the given events it.</p>
1665 <dt>ev_feed_signal_event (ev_loop *loop, int signum)</dt>
1667 <p>Feed an event as if the given signal occured (<code>loop</code> must be the default
1677 <h1 id="LIBEVENT_EMULATION">LIBEVENT EMULATION</h1>
1678 <div id="LIBEVENT_EMULATION_CONTENT">
1679 <p>Libev offers a compatibility emulation layer for libevent. It cannot
1680 emulate the internals of libevent, so here are some usage hints:</p>
1682 <dt>* Use it by including <event.h>, as usual.</dt>
1683 <dt>* The following members are fully supported: ev_base, ev_callback,
1684 ev_arg, ev_fd, ev_res, ev_events.</dt>
1685 <dt>* Avoid using ev_flags and the EVLIST_*-macros, while it is
1686 maintained by libev, it does not work exactly the same way as in libevent (consider
1687 it a private API).</dt>
1688 <dt>* Priorities are not currently supported. Initialising priorities
1689 will fail and all watchers will have the same priority, even though there
1690 is an ev_pri field.</dt>
1691 <dt>* Other members are not supported.</dt>
1692 <dt>* The libev emulation is <i>not</i> ABI compatible to libevent, you need
1693 to use the libev header file and library.</dt>
1697 <h1 id="C_SUPPORT">C++ SUPPORT</h1>
1698 <div id="C_SUPPORT_CONTENT">
1699 <p>Libev comes with some simplistic wrapper classes for C++ that mainly allow
1700 you to use some convinience methods to start/stop watchers and also change
1701 the callback model to a model using method callbacks on objects.</p>
1703 <pre> #include <ev++.h>
1706 <p>(it is not installed by default). This automatically includes <cite>ev.h</cite>
1707 and puts all of its definitions (many of them macros) into the global
1708 namespace. All C++ specific things are put into the <code>ev</code> namespace.</p>
1709 <p>It should support all the same embedding options as <cite>ev.h</cite>, most notably
1710 <code>EV_MULTIPLICITY</code>.</p>
1711 <p>Here is a list of things available in the <code>ev</code> namespace:</p>
1713 <dt><code>ev::READ</code>, <code>ev::WRITE</code> etc.</dt>
1715 <p>These are just enum values with the same values as the <code>EV_READ</code> etc.
1716 macros from <cite>ev.h</cite>.</p>
1718 <dt><code>ev::tstamp</code>, <code>ev::now</code></dt>
1720 <p>Aliases to the same types/functions as with the <code>ev_</code> prefix.</p>
1722 <dt><code>ev::io</code>, <code>ev::timer</code>, <code>ev::periodic</code>, <code>ev::idle</code>, <code>ev::sig</code> etc.</dt>
1724 <p>For each <code>ev_TYPE</code> watcher in <cite>ev.h</cite> there is a corresponding class of
1725 the same name in the <code>ev</code> namespace, with the exception of <code>ev_signal</code>
1726 which is called <code>ev::sig</code> to avoid clashes with the <code>signal</code> macro
1727 defines by many implementations.</p>
1728 <p>All of those classes have these methods:</p>
1731 <dt>ev::TYPE::TYPE (object *, object::method *)</dt>
1732 <dt>ev::TYPE::TYPE (object *, object::method *, struct ev_loop *)</dt>
1733 <dt>ev::TYPE::~TYPE</dt>
1735 <p>The constructor takes a pointer to an object and a method pointer to
1736 the event handler callback to call in this class. The constructor calls
1737 <code>ev_init</code> for you, which means you have to call the <code>set</code> method
1738 before starting it. If you do not specify a loop then the constructor
1739 automatically associates the default loop with this watcher.</p>
1740 <p>The destructor automatically stops the watcher if it is active.</p>
1742 <dt>w->set (struct ev_loop *)</dt>
1744 <p>Associates a different <code>struct ev_loop</code> with this watcher. You can only
1745 do this when the watcher is inactive (and not pending either).</p>
1747 <dt>w->set ([args])</dt>
1749 <p>Basically the same as <code>ev_TYPE_set</code>, with the same args. Must be
1750 called at least once. Unlike the C counterpart, an active watcher gets
1751 automatically stopped and restarted.</p>
1753 <dt>w->start ()</dt>
1755 <p>Starts the watcher. Note that there is no <code>loop</code> argument as the
1756 constructor already takes the loop.</p>
1758 <dt>w->stop ()</dt>
1760 <p>Stops the watcher if it is active. Again, no <code>loop</code> argument.</p>
1762 <dt>w->again () <code>ev::timer</code>, <code>ev::periodic</code> only</dt>
1764 <p>For <code>ev::timer</code> and <code>ev::periodic</code>, this invokes the corresponding
1765 <code>ev_TYPE_again</code> function.</p>
1767 <dt>w->sweep () <code>ev::embed</code> only</dt>
1769 <p>Invokes <code>ev_embed_sweep</code>.</p>
1771 <dt>w->update () <code>ev::stat</code> only</dt>
1773 <p>Invokes <code>ev_stat_stat</code>.</p>
1779 <p>Example: Define a class with an IO and idle watcher, start one of them in
1780 the constructor.</p>
1783 ev_io io; void io_cb (ev::io &w, int revents);
1784 ev_idle idle void idle_cb (ev::idle &w, int revents);
1789 myclass::myclass (int fd)
1790 : io (this, &myclass::io_cb),
1791 idle (this, &myclass::idle_cb)
1793 io.start (fd, ev::READ);
1802 <h1 id="MACRO_MAGIC">MACRO MAGIC</h1>
1803 <div id="MACRO_MAGIC_CONTENT">
1804 <p>Libev can be compiled with a variety of options, the most fundemantal is
1805 <code>EV_MULTIPLICITY</code>. This option determines wether (most) functions and
1806 callbacks have an initial <code>struct ev_loop *</code> argument.</p>
1807 <p>To make it easier to write programs that cope with either variant, the
1808 following macros are defined:</p>
1810 <dt><code>EV_A</code>, <code>EV_A_</code></dt>
1812 <p>This provides the loop <i>argument</i> for functions, if one is required ("ev
1813 loop argument"). The <code>EV_A</code> form is used when this is the sole argument,
1814 <code>EV_A_</code> is used when other arguments are following. Example:</p>
1815 <pre> ev_unref (EV_A);
1816 ev_timer_add (EV_A_ watcher);
1820 <p>It assumes the variable <code>loop</code> of type <code>struct ev_loop *</code> is in scope,
1821 which is often provided by the following macro.</p>
1823 <dt><code>EV_P</code>, <code>EV_P_</code></dt>
1825 <p>This provides the loop <i>parameter</i> for functions, if one is required ("ev
1826 loop parameter"). The <code>EV_P</code> form is used when this is the sole parameter,
1827 <code>EV_P_</code> is used when other parameters are following. Example:</p>
1828 <pre> // this is how ev_unref is being declared
1829 static void ev_unref (EV_P);
1831 // this is how you can declare your typical callback
1832 static void cb (EV_P_ ev_timer *w, int revents)
1835 <p>It declares a parameter <code>loop</code> of type <code>struct ev_loop *</code>, quite
1836 suitable for use with <code>EV_A</code>.</p>
1838 <dt><code>EV_DEFAULT</code>, <code>EV_DEFAULT_</code></dt>
1840 <p>Similar to the other two macros, this gives you the value of the default
1841 loop, if multiple loops are supported ("ev loop default").</p>
1844 <p>Example: Declare and initialise a check watcher, utilising the above
1845 macros so it will work regardless of wether multiple loops are supported
1848 check_cb (EV_P_ ev_timer *w, int revents)
1850 ev_check_stop (EV_A_ w);
1854 ev_check_init (&check, check_cb);
1855 ev_check_start (EV_DEFAULT_ &check);
1856 ev_loop (EV_DEFAULT_ 0);
1861 <h1 id="EMBEDDING">EMBEDDING</h1>
1862 <div id="EMBEDDING_CONTENT">
1863 <p>Libev can (and often is) directly embedded into host
1864 applications. Examples of applications that embed it include the Deliantra
1865 Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
1866 and rxvt-unicode.</p>
1867 <p>The goal is to enable you to just copy the neecssary files into your
1868 source directory without having to change even a single line in them, so
1869 you can easily upgrade by simply copying (or having a checked-out copy of
1870 libev somewhere in your source tree).</p>
1873 <h2 id="FILESETS">FILESETS</h2>
1874 <div id="FILESETS_CONTENT">
1875 <p>Depending on what features you need you need to include one or more sets of files
1879 <h3 id="CORE_EVENT_LOOP">CORE EVENT LOOP</h3>
1880 <div id="CORE_EVENT_LOOP_CONTENT">
1881 <p>To include only the libev core (all the <code>ev_*</code> functions), with manual
1882 configuration (no autoconf):</p>
1883 <pre> #define EV_STANDALONE 1
1884 #include "ev.c"
1887 <p>This will automatically include <cite>ev.h</cite>, too, and should be done in a
1888 single C source file only to provide the function implementations. To use
1889 it, do the same for <cite>ev.h</cite> in all files wishing to use this API (best
1890 done by writing a wrapper around <cite>ev.h</cite> that you can include instead and
1891 where you can put other configuration options):</p>
1892 <pre> #define EV_STANDALONE 1
1893 #include "ev.h"
1896 <p>Both header files and implementation files can be compiled with a C++
1897 compiler (at least, thats a stated goal, and breakage will be treated
1899 <p>You need the following files in your source tree, or in a directory
1900 in your include path (e.g. in libev/ when using -Ilibev):</p>
1906 ev_win32.c required on win32 platforms only
1908 ev_select.c only when select backend is enabled (which is enabled by default)
1909 ev_poll.c only when poll backend is enabled (disabled by default)
1910 ev_epoll.c only when the epoll backend is enabled (disabled by default)
1911 ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
1912 ev_port.c only when the solaris port backend is enabled (disabled by default)
1915 <p><cite>ev.c</cite> includes the backend files directly when enabled, so you only need
1916 to compile this single file.</p>
1919 <h3 id="LIBEVENT_COMPATIBILITY_API">LIBEVENT COMPATIBILITY API</h3>
1920 <div id="LIBEVENT_COMPATIBILITY_API_CONTENT">
1921 <p>To include the libevent compatibility API, also include:</p>
1922 <pre> #include "event.c"
1925 <p>in the file including <cite>ev.c</cite>, and:</p>
1926 <pre> #include "event.h"
1929 <p>in the files that want to use the libevent API. This also includes <cite>ev.h</cite>.</p>
1930 <p>You need the following additional files for this:</p>
1937 <h3 id="AUTOCONF_SUPPORT">AUTOCONF SUPPORT</h3>
1938 <div id="AUTOCONF_SUPPORT_CONTENT">
1939 <p>Instead of using <code>EV_STANDALONE=1</code> and providing your config in
1940 whatever way you want, you can also <code>m4_include([libev.m4])</code> in your
1941 <cite>configure.ac</cite> and leave <code>EV_STANDALONE</code> undefined. <cite>ev.c</cite> will then
1942 include <cite>config.h</cite> and configure itself accordingly.</p>
1943 <p>For this of course you need the m4 file:</p>
1949 <h2 id="PREPROCESSOR_SYMBOLS_MACROS">PREPROCESSOR SYMBOLS/MACROS</h2>
1950 <div id="PREPROCESSOR_SYMBOLS_MACROS_CONTENT">
1951 <p>Libev can be configured via a variety of preprocessor symbols you have to define
1952 before including any of its files. The default is not to build for multiplicity
1953 and only include the select backend.</p>
1955 <dt>EV_STANDALONE</dt>
1957 <p>Must always be <code>1</code> if you do not use autoconf configuration, which
1958 keeps libev from including <cite>config.h</cite>, and it also defines dummy
1959 implementations for some libevent functions (such as logging, which is not
1960 supported). It will also not define any of the structs usually found in
1961 <cite>event.h</cite> that are not directly supported by the libev core alone.</p>
1963 <dt>EV_USE_MONOTONIC</dt>
1965 <p>If defined to be <code>1</code>, libev will try to detect the availability of the
1966 monotonic clock option at both compiletime and runtime. Otherwise no use
1967 of the monotonic clock option will be attempted. If you enable this, you
1968 usually have to link against librt or something similar. Enabling it when
1969 the functionality isn't available is safe, though, althoguh you have
1970 to make sure you link against any libraries where the <code>clock_gettime</code>
1971 function is hiding in (often <cite>-lrt</cite>).</p>
1973 <dt>EV_USE_REALTIME</dt>
1975 <p>If defined to be <code>1</code>, libev will try to detect the availability of the
1976 realtime clock option at compiletime (and assume its availability at
1977 runtime if successful). Otherwise no use of the realtime clock option will
1978 be attempted. This effectively replaces <code>gettimeofday</code> by <code>clock_get
1979 (CLOCK_REALTIME, ...)</code> and will not normally affect correctness. See tzhe note about libraries
1980 in the description of <code>EV_USE_MONOTONIC</code>, though.</p>
1982 <dt>EV_USE_SELECT</dt>
1984 <p>If undefined or defined to be <code>1</code>, libev will compile in support for the
1985 <code>select</code>(2) backend. No attempt at autodetection will be done: if no
1986 other method takes over, select will be it. Otherwise the select backend
1987 will not be compiled in.</p>
1989 <dt>EV_SELECT_USE_FD_SET</dt>
1991 <p>If defined to <code>1</code>, then the select backend will use the system <code>fd_set</code>
1992 structure. This is useful if libev doesn't compile due to a missing
1993 <code>NFDBITS</code> or <code>fd_mask</code> definition or it misguesses the bitset layout on
1994 exotic systems. This usually limits the range of file descriptors to some
1995 low limit such as 1024 or might have other limitations (winsocket only
1996 allows 64 sockets). The <code>FD_SETSIZE</code> macro, set before compilation, might
1997 influence the size of the <code>fd_set</code> used.</p>
1999 <dt>EV_SELECT_IS_WINSOCKET</dt>
2001 <p>When defined to <code>1</code>, the select backend will assume that
2002 select/socket/connect etc. don't understand file descriptors but
2003 wants osf handles on win32 (this is the case when the select to
2004 be used is the winsock select). This means that it will call
2005 <code>_get_osfhandle</code> on the fd to convert it to an OS handle. Otherwise,
2006 it is assumed that all these functions actually work on fds, even
2007 on win32. Should not be defined on non-win32 platforms.</p>
2009 <dt>EV_USE_POLL</dt>
2011 <p>If defined to be <code>1</code>, libev will compile in support for the <code>poll</code>(2)
2012 backend. Otherwise it will be enabled on non-win32 platforms. It
2013 takes precedence over select.</p>
2015 <dt>EV_USE_EPOLL</dt>
2017 <p>If defined to be <code>1</code>, libev will compile in support for the Linux
2018 <code>epoll</code>(7) backend. Its availability will be detected at runtime,
2019 otherwise another method will be used as fallback. This is the
2020 preferred backend for GNU/Linux systems.</p>
2022 <dt>EV_USE_KQUEUE</dt>
2024 <p>If defined to be <code>1</code>, libev will compile in support for the BSD style
2025 <code>kqueue</code>(2) backend. Its actual availability will be detected at runtime,
2026 otherwise another method will be used as fallback. This is the preferred
2027 backend for BSD and BSD-like systems, although on most BSDs kqueue only
2028 supports some types of fds correctly (the only platform we found that
2029 supports ptys for example was NetBSD), so kqueue might be compiled in, but
2030 not be used unless explicitly requested. The best way to use it is to find
2031 out whether kqueue supports your type of fd properly and use an embedded
2034 <dt>EV_USE_PORT</dt>
2036 <p>If defined to be <code>1</code>, libev will compile in support for the Solaris
2037 10 port style backend. Its availability will be detected at runtime,
2038 otherwise another method will be used as fallback. This is the preferred
2039 backend for Solaris 10 systems.</p>
2041 <dt>EV_USE_DEVPOLL</dt>
2043 <p>reserved for future expansion, works like the USE symbols above.</p>
2045 <dt>EV_USE_INOTIFY</dt>
2047 <p>If defined to be <code>1</code>, libev will compile in support for the Linux inotify
2048 interface to speed up <code>ev_stat</code> watchers. Its actual availability will
2049 be detected at runtime.</p>
2053 <p>The name of the <cite>ev.h</cite> header file used to include it. The default if
2054 undefined is <code><ev.h></code> in <cite>event.h</cite> and <code>"ev.h"</code> in <cite>ev.c</cite>. This
2055 can be used to virtually rename the <cite>ev.h</cite> header file in case of conflicts.</p>
2057 <dt>EV_CONFIG_H</dt>
2059 <p>If <code>EV_STANDALONE</code> isn't <code>1</code>, this variable can be used to override
2060 <cite>ev.c</cite>'s idea of where to find the <cite>config.h</cite> file, similarly to
2061 <code>EV_H</code>, above.</p>
2065 <p>Similarly to <code>EV_H</code>, this macro can be used to override <cite>event.c</cite>'s idea
2066 of how the <cite>event.h</cite> header can be found.</p>
2068 <dt>EV_PROTOTYPES</dt>
2070 <p>If defined to be <code>0</code>, then <cite>ev.h</cite> will not define any function
2071 prototypes, but still define all the structs and other symbols. This is
2072 occasionally useful if you want to provide your own wrapper functions
2073 around libev functions.</p>
2075 <dt>EV_MULTIPLICITY</dt>
2077 <p>If undefined or defined to <code>1</code>, then all event-loop-specific functions
2078 will have the <code>struct ev_loop *</code> as first argument, and you can create
2079 additional independent event loops. Otherwise there will be no support
2080 for multiple event loops and there is no first event loop pointer
2081 argument. Instead, all functions act on the single default loop.</p>
2083 <dt>EV_PERIODIC_ENABLE</dt>
2085 <p>If undefined or defined to be <code>1</code>, then periodic timers are supported. If
2086 defined to be <code>0</code>, then they are not. Disabling them saves a few kB of
2089 <dt>EV_EMBED_ENABLE</dt>
2091 <p>If undefined or defined to be <code>1</code>, then embed watchers are supported. If
2092 defined to be <code>0</code>, then they are not.</p>
2094 <dt>EV_STAT_ENABLE</dt>
2096 <p>If undefined or defined to be <code>1</code>, then stat watchers are supported. If
2097 defined to be <code>0</code>, then they are not.</p>
2099 <dt>EV_FORK_ENABLE</dt>
2101 <p>If undefined or defined to be <code>1</code>, then fork watchers are supported. If
2102 defined to be <code>0</code>, then they are not.</p>
2106 <p>If you need to shave off some kilobytes of code at the expense of some
2107 speed, define this symbol to <code>1</code>. Currently only used for gcc to override
2108 some inlining decisions, saves roughly 30% codesize of amd64.</p>
2110 <dt>EV_PID_HASHSIZE</dt>
2112 <p><code>ev_child</code> watchers use a small hash table to distribute workload by
2113 pid. The default size is <code>16</code> (or <code>1</code> with <code>EV_MINIMAL</code>), usually more
2114 than enough. If you need to manage thousands of children you might want to
2115 increase this value (<i>must</i> be a power of two).</p>
2117 <dt>EV_INOTIFY_HASHSIZE</dt>
2119 <p><code>ev_staz</code> watchers use a small hash table to distribute workload by
2120 inotify watch id. The default size is <code>16</code> (or <code>1</code> with <code>EV_MINIMAL</code>),
2121 usually more than enough. If you need to manage thousands of <code>ev_stat</code>
2122 watchers you might want to increase this value (<i>must</i> be a power of
2127 <p>By default, all watchers have a <code>void *data</code> member. By redefining
2128 this macro to a something else you can include more and other types of
2129 members. You have to define it each time you include one of the files,
2130 though, and it must be identical each time.</p>
2131 <p>For example, the perl EV module uses something like this:</p>
2132 <pre> #define EV_COMMON \
2133 SV *self; /* contains this struct */ \
2134 SV *cb_sv, *fh /* note no trailing ";" */
2138 <dt>EV_CB_DECLARE (type)</dt>
2139 <dt>EV_CB_INVOKE (watcher, revents)</dt>
2140 <dt>ev_set_cb (ev, cb)</dt>
2142 <p>Can be used to change the callback member declaration in each watcher,
2143 and the way callbacks are invoked and set. Must expand to a struct member
2144 definition and a statement, respectively. See the <cite>ev.v</cite> header file for
2145 their default definitions. One possible use for overriding these is to
2146 avoid the <code>struct ev_loop *</code> as first argument in all cases, or to use
2147 method calls instead of plain function calls in C++.</p>
2150 <h2 id="EXAMPLES">EXAMPLES</h2>
2151 <div id="EXAMPLES_CONTENT">
2152 <p>For a real-world example of a program the includes libev
2153 verbatim, you can have a look at the EV perl module
2154 (<a href="http://software.schmorp.de/pkg/EV.html">http://software.schmorp.de/pkg/EV.html</a>). It has the libev files in
2155 the <cite>libev/</cite> subdirectory and includes them in the <cite>EV/EVAPI.h</cite> (public
2156 interface) and <cite>EV.xs</cite> (implementation) files. Only the <cite>EV.xs</cite> file
2157 will be compiled. It is pretty complex because it provides its own header
2159 <p>The usage in rxvt-unicode is simpler. It has a <cite>ev_cpp.h</cite> header file
2160 that everybody includes and which overrides some configure choices:</p>
2161 <pre> #define EV_MINIMAL 1
2162 #define EV_USE_POLL 0
2163 #define EV_MULTIPLICITY 0
2164 #define EV_PERIODIC_ENABLE 0
2165 #define EV_STAT_ENABLE 0
2166 #define EV_FORK_ENABLE 0
2167 #define EV_CONFIG_H <config.h>
2171 #include "ev++.h"
2174 <p>And a <cite>ev_cpp.C</cite> implementation file that contains libev proper and is compiled:</p>
2175 <pre> #include "ev_cpp.h"
2176 #include "ev.c"
2184 <h1 id="COMPLEXITIES">COMPLEXITIES</h1>
2185 <div id="COMPLEXITIES_CONTENT">
2186 <p>In this section the complexities of (many of) the algorithms used inside
2187 libev will be explained. For complexity discussions about backends see the
2188 documentation for <code>ev_default_init</code>.</p>
2191 <dt>Starting and stopping timer/periodic watchers: O(log skipped_other_timers)</dt>
2192 <dt>Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)</dt>
2193 <dt>Starting io/check/prepare/idle/signal/child watchers: O(1)</dt>
2194 <dt>Stopping check/prepare/idle watchers: O(1)</dt>
2195 <dt>Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))</dt>
2196 <dt>Finding the next timer per loop iteration: O(1)</dt>
2197 <dt>Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)</dt>
2198 <dt>Activating one watcher: O(1)</dt>
2207 <h1 id="AUTHOR">AUTHOR</h1>
2208 <div id="AUTHOR_CONTENT">
2209 <p>Marc Lehmann <libev@schmorp.de>.</p>
projects / software / libev.git / blob