1 |
=head1 NAME |
2 |
|
3 |
libev - a high performance full-featured event loop written in C |
4 |
|
5 |
=head1 SYNOPSIS |
6 |
|
7 |
#include <ev.h> |
8 |
|
9 |
=head1 EXAMPLE PROGRAM |
10 |
|
11 |
#include <ev.h> |
12 |
|
13 |
ev_io stdin_watcher; |
14 |
ev_timer timeout_watcher; |
15 |
|
16 |
/* called when data readable on stdin */ |
17 |
static void |
18 |
stdin_cb (EV_P_ struct ev_io *w, int revents) |
19 |
{ |
20 |
/* puts ("stdin ready"); */ |
21 |
ev_io_stop (EV_A_ w); /* just a syntax example */ |
22 |
ev_unloop (EV_A_ EVUNLOOP_ALL); /* leave all loop calls */ |
23 |
} |
24 |
|
25 |
static void |
26 |
timeout_cb (EV_P_ struct ev_timer *w, int revents) |
27 |
{ |
28 |
/* puts ("timeout"); */ |
29 |
ev_unloop (EV_A_ EVUNLOOP_ONE); /* leave one loop call */ |
30 |
} |
31 |
|
32 |
int |
33 |
main (void) |
34 |
{ |
35 |
struct ev_loop *loop = ev_default_loop (0); |
36 |
|
37 |
/* initialise an io watcher, then start it */ |
38 |
ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ); |
39 |
ev_io_start (loop, &stdin_watcher); |
40 |
|
41 |
/* simple non-repeating 5.5 second timeout */ |
42 |
ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.); |
43 |
ev_timer_start (loop, &timeout_watcher); |
44 |
|
45 |
/* loop till timeout or data ready */ |
46 |
ev_loop (loop, 0); |
47 |
|
48 |
return 0; |
49 |
} |
50 |
|
51 |
=head1 DESCRIPTION |
52 |
|
53 |
The newest version of this document is also available as a html-formatted |
54 |
web page you might find easier to navigate when reading it for the first |
55 |
time: L<http://cvs.schmorp.de/libev/ev.html>. |
56 |
|
57 |
Libev is an event loop: you register interest in certain events (such as a |
58 |
file descriptor being readable or a timeout occurring), and it will manage |
59 |
these event sources and provide your program with events. |
60 |
|
61 |
To do this, it must take more or less complete control over your process |
62 |
(or thread) by executing the I<event loop> handler, and will then |
63 |
communicate events via a callback mechanism. |
64 |
|
65 |
You register interest in certain events by registering so-called I<event |
66 |
watchers>, which are relatively small C structures you initialise with the |
67 |
details of the event, and then hand it over to libev by I<starting> the |
68 |
watcher. |
69 |
|
70 |
=head1 FEATURES |
71 |
|
72 |
Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the |
73 |
BSD-specific C<kqueue> and the Solaris-specific event port mechanisms |
74 |
for file descriptor events (C<ev_io>), the Linux C<inotify> interface |
75 |
(for C<ev_stat>), relative timers (C<ev_timer>), absolute timers |
76 |
with customised rescheduling (C<ev_periodic>), synchronous signals |
77 |
(C<ev_signal>), process status change events (C<ev_child>), and event |
78 |
watchers dealing with the event loop mechanism itself (C<ev_idle>, |
79 |
C<ev_embed>, C<ev_prepare> and C<ev_check> watchers) as well as |
80 |
file watchers (C<ev_stat>) and even limited support for fork events |
81 |
(C<ev_fork>). |
82 |
|
83 |
It also is quite fast (see this |
84 |
L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent |
85 |
for example). |
86 |
|
87 |
=head1 CONVENTIONS |
88 |
|
89 |
Libev is very configurable. In this manual the default configuration will |
90 |
be described, which supports multiple event loops. For more info about |
91 |
various configuration options please have a look at B<EMBED> section in |
92 |
this manual. If libev was configured without support for multiple event |
93 |
loops, then all functions taking an initial argument of name C<loop> |
94 |
(which is always of type C<struct ev_loop *>) will not have this argument. |
95 |
|
96 |
=head1 TIME REPRESENTATION |
97 |
|
98 |
Libev represents time as a single floating point number, representing the |
99 |
(fractional) number of seconds since the (POSIX) epoch (somewhere near |
100 |
the beginning of 1970, details are complicated, don't ask). This type is |
101 |
called C<ev_tstamp>, which is what you should use too. It usually aliases |
102 |
to the C<double> type in C, and when you need to do any calculations on |
103 |
it, you should treat it as some floatingpoint value. Unlike the name |
104 |
component C<stamp> might indicate, it is also used for time differences |
105 |
throughout libev. |
106 |
|
107 |
=head1 GLOBAL FUNCTIONS |
108 |
|
109 |
These functions can be called anytime, even before initialising the |
110 |
library in any way. |
111 |
|
112 |
=over 4 |
113 |
|
114 |
=item ev_tstamp ev_time () |
115 |
|
116 |
Returns the current time as libev would use it. Please note that the |
117 |
C<ev_now> function is usually faster and also often returns the timestamp |
118 |
you actually want to know. |
119 |
|
120 |
=item ev_sleep (ev_tstamp interval) |
121 |
|
122 |
Sleep for the given interval: The current thread will be blocked until |
123 |
either it is interrupted or the given time interval has passed. Basically |
124 |
this is a subsecond-resolution C<sleep ()>. |
125 |
|
126 |
=item int ev_version_major () |
127 |
|
128 |
=item int ev_version_minor () |
129 |
|
130 |
You can find out the major and minor ABI version numbers of the library |
131 |
you linked against by calling the functions C<ev_version_major> and |
132 |
C<ev_version_minor>. If you want, you can compare against the global |
133 |
symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the |
134 |
version of the library your program was compiled against. |
135 |
|
136 |
These version numbers refer to the ABI version of the library, not the |
137 |
release version. |
138 |
|
139 |
Usually, it's a good idea to terminate if the major versions mismatch, |
140 |
as this indicates an incompatible change. Minor versions are usually |
141 |
compatible to older versions, so a larger minor version alone is usually |
142 |
not a problem. |
143 |
|
144 |
Example: Make sure we haven't accidentally been linked against the wrong |
145 |
version. |
146 |
|
147 |
assert (("libev version mismatch", |
148 |
ev_version_major () == EV_VERSION_MAJOR |
149 |
&& ev_version_minor () >= EV_VERSION_MINOR)); |
150 |
|
151 |
=item unsigned int ev_supported_backends () |
152 |
|
153 |
Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*> |
154 |
value) compiled into this binary of libev (independent of their |
155 |
availability on the system you are running on). See C<ev_default_loop> for |
156 |
a description of the set values. |
157 |
|
158 |
Example: make sure we have the epoll method, because yeah this is cool and |
159 |
a must have and can we have a torrent of it please!!!11 |
160 |
|
161 |
assert (("sorry, no epoll, no sex", |
162 |
ev_supported_backends () & EVBACKEND_EPOLL)); |
163 |
|
164 |
=item unsigned int ev_recommended_backends () |
165 |
|
166 |
Return the set of all backends compiled into this binary of libev and also |
167 |
recommended for this platform. This set is often smaller than the one |
168 |
returned by C<ev_supported_backends>, as for example kqueue is broken on |
169 |
most BSDs and will not be autodetected unless you explicitly request it |
170 |
(assuming you know what you are doing). This is the set of backends that |
171 |
libev will probe for if you specify no backends explicitly. |
172 |
|
173 |
=item unsigned int ev_embeddable_backends () |
174 |
|
175 |
Returns the set of backends that are embeddable in other event loops. This |
176 |
is the theoretical, all-platform, value. To find which backends |
177 |
might be supported on the current system, you would need to look at |
178 |
C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for |
179 |
recommended ones. |
180 |
|
181 |
See the description of C<ev_embed> watchers for more info. |
182 |
|
183 |
=item ev_set_allocator (void *(*cb)(void *ptr, long size)) |
184 |
|
185 |
Sets the allocation function to use (the prototype is similar - the |
186 |
semantics is identical - to the realloc C function). It is used to |
187 |
allocate and free memory (no surprises here). If it returns zero when |
188 |
memory needs to be allocated, the library might abort or take some |
189 |
potentially destructive action. The default is your system realloc |
190 |
function. |
191 |
|
192 |
You could override this function in high-availability programs to, say, |
193 |
free some memory if it cannot allocate memory, to use a special allocator, |
194 |
or even to sleep a while and retry until some memory is available. |
195 |
|
196 |
Example: Replace the libev allocator with one that waits a bit and then |
197 |
retries). |
198 |
|
199 |
static void * |
200 |
persistent_realloc (void *ptr, size_t size) |
201 |
{ |
202 |
for (;;) |
203 |
{ |
204 |
void *newptr = realloc (ptr, size); |
205 |
|
206 |
if (newptr) |
207 |
return newptr; |
208 |
|
209 |
sleep (60); |
210 |
} |
211 |
} |
212 |
|
213 |
... |
214 |
ev_set_allocator (persistent_realloc); |
215 |
|
216 |
=item ev_set_syserr_cb (void (*cb)(const char *msg)); |
217 |
|
218 |
Set the callback function to call on a retryable syscall error (such |
219 |
as failed select, poll, epoll_wait). The message is a printable string |
220 |
indicating the system call or subsystem causing the problem. If this |
221 |
callback is set, then libev will expect it to remedy the sitution, no |
222 |
matter what, when it returns. That is, libev will generally retry the |
223 |
requested operation, or, if the condition doesn't go away, do bad stuff |
224 |
(such as abort). |
225 |
|
226 |
Example: This is basically the same thing that libev does internally, too. |
227 |
|
228 |
static void |
229 |
fatal_error (const char *msg) |
230 |
{ |
231 |
perror (msg); |
232 |
abort (); |
233 |
} |
234 |
|
235 |
... |
236 |
ev_set_syserr_cb (fatal_error); |
237 |
|
238 |
=back |
239 |
|
240 |
=head1 FUNCTIONS CONTROLLING THE EVENT LOOP |
241 |
|
242 |
An event loop is described by a C<struct ev_loop *>. The library knows two |
243 |
types of such loops, the I<default> loop, which supports signals and child |
244 |
events, and dynamically created loops which do not. |
245 |
|
246 |
If you use threads, a common model is to run the default event loop |
247 |
in your main thread (or in a separate thread) and for each thread you |
248 |
create, you also create another event loop. Libev itself does no locking |
249 |
whatsoever, so if you mix calls to the same event loop in different |
250 |
threads, make sure you lock (this is usually a bad idea, though, even if |
251 |
done correctly, because it's hideous and inefficient). |
252 |
|
253 |
=over 4 |
254 |
|
255 |
=item struct ev_loop *ev_default_loop (unsigned int flags) |
256 |
|
257 |
This will initialise the default event loop if it hasn't been initialised |
258 |
yet and return it. If the default loop could not be initialised, returns |
259 |
false. If it already was initialised it simply returns it (and ignores the |
260 |
flags. If that is troubling you, check C<ev_backend ()> afterwards). |
261 |
|
262 |
If you don't know what event loop to use, use the one returned from this |
263 |
function. |
264 |
|
265 |
The flags argument can be used to specify special behaviour or specific |
266 |
backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>). |
267 |
|
268 |
The following flags are supported: |
269 |
|
270 |
=over 4 |
271 |
|
272 |
=item C<EVFLAG_AUTO> |
273 |
|
274 |
The default flags value. Use this if you have no clue (it's the right |
275 |
thing, believe me). |
276 |
|
277 |
=item C<EVFLAG_NOENV> |
278 |
|
279 |
If this flag bit is ored into the flag value (or the program runs setuid |
280 |
or setgid) then libev will I<not> look at the environment variable |
281 |
C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will |
282 |
override the flags completely if it is found in the environment. This is |
283 |
useful to try out specific backends to test their performance, or to work |
284 |
around bugs. |
285 |
|
286 |
=item C<EVFLAG_FORKCHECK> |
287 |
|
288 |
Instead of calling C<ev_default_fork> or C<ev_loop_fork> manually after |
289 |
a fork, you can also make libev check for a fork in each iteration by |
290 |
enabling this flag. |
291 |
|
292 |
This works by calling C<getpid ()> on every iteration of the loop, |
293 |
and thus this might slow down your event loop if you do a lot of loop |
294 |
iterations and little real work, but is usually not noticeable (on my |
295 |
Linux system for example, C<getpid> is actually a simple 5-insn sequence |
296 |
without a syscall and thus I<very> fast, but my Linux system also has |
297 |
C<pthread_atfork> which is even faster). |
298 |
|
299 |
The big advantage of this flag is that you can forget about fork (and |
300 |
forget about forgetting to tell libev about forking) when you use this |
301 |
flag. |
302 |
|
303 |
This flag setting cannot be overriden or specified in the C<LIBEV_FLAGS> |
304 |
environment variable. |
305 |
|
306 |
=item C<EVBACKEND_SELECT> (value 1, portable select backend) |
307 |
|
308 |
This is your standard select(2) backend. Not I<completely> standard, as |
309 |
libev tries to roll its own fd_set with no limits on the number of fds, |
310 |
but if that fails, expect a fairly low limit on the number of fds when |
311 |
using this backend. It doesn't scale too well (O(highest_fd)), but its usually |
312 |
the fastest backend for a low number of fds. |
313 |
|
314 |
=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows) |
315 |
|
316 |
And this is your standard poll(2) backend. It's more complicated than |
317 |
select, but handles sparse fds better and has no artificial limit on the |
318 |
number of fds you can use (except it will slow down considerably with a |
319 |
lot of inactive fds). It scales similarly to select, i.e. O(total_fds). |
320 |
|
321 |
=item C<EVBACKEND_EPOLL> (value 4, Linux) |
322 |
|
323 |
For few fds, this backend is a bit little slower than poll and select, |
324 |
but it scales phenomenally better. While poll and select usually scale |
325 |
like O(total_fds) where n is the total number of fds (or the highest fd), |
326 |
epoll scales either O(1) or O(active_fds). The epoll design has a number |
327 |
of shortcomings, such as silently dropping events in some hard-to-detect |
328 |
cases and rewiring a syscall per fd change, no fork support and bad |
329 |
support for dup: |
330 |
|
331 |
While stopping, setting and starting an I/O watcher in the same iteration |
332 |
will result in some caching, there is still a syscall per such incident |
333 |
(because the fd could point to a different file description now), so its |
334 |
best to avoid that. Also, C<dup ()>'ed file descriptors might not work |
335 |
very well if you register events for both fds. |
336 |
|
337 |
Please note that epoll sometimes generates spurious notifications, so you |
338 |
need to use non-blocking I/O or other means to avoid blocking when no data |
339 |
(or space) is available. |
340 |
|
341 |
=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones) |
342 |
|
343 |
Kqueue deserves special mention, as at the time of this writing, it |
344 |
was broken on all BSDs except NetBSD (usually it doesn't work reliably |
345 |
with anything but sockets and pipes, except on Darwin, where of course |
346 |
it's completely useless). For this reason it's not being "autodetected" |
347 |
unless you explicitly specify it explicitly in the flags (i.e. using |
348 |
C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough) |
349 |
system like NetBSD. |
350 |
|
351 |
You still can embed kqueue into a normal poll or select backend and use it |
352 |
only for sockets (after having made sure that sockets work with kqueue on |
353 |
the target platform). See C<ev_embed> watchers for more info. |
354 |
|
355 |
It scales in the same way as the epoll backend, but the interface to the |
356 |
kernel is more efficient (which says nothing about its actual speed, of |
357 |
course). While stopping, setting and starting an I/O watcher does never |
358 |
cause an extra syscall as with C<EVBACKEND_EPOLL>, it still adds up to |
359 |
two event changes per incident, support for C<fork ()> is very bad and it |
360 |
drops fds silently in similarly hard-to-detect cases. |
361 |
|
362 |
=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8) |
363 |
|
364 |
This is not implemented yet (and might never be). |
365 |
|
366 |
=item C<EVBACKEND_PORT> (value 32, Solaris 10) |
367 |
|
368 |
This uses the Solaris 10 event port mechanism. As with everything on Solaris, |
369 |
it's really slow, but it still scales very well (O(active_fds)). |
370 |
|
371 |
Please note that solaris event ports can deliver a lot of spurious |
372 |
notifications, so you need to use non-blocking I/O or other means to avoid |
373 |
blocking when no data (or space) is available. |
374 |
|
375 |
=item C<EVBACKEND_ALL> |
376 |
|
377 |
Try all backends (even potentially broken ones that wouldn't be tried |
378 |
with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as |
379 |
C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>. |
380 |
|
381 |
=back |
382 |
|
383 |
If one or more of these are ored into the flags value, then only these |
384 |
backends will be tried (in the reverse order as given here). If none are |
385 |
specified, most compiled-in backend will be tried, usually in reverse |
386 |
order of their flag values :) |
387 |
|
388 |
The most typical usage is like this: |
389 |
|
390 |
if (!ev_default_loop (0)) |
391 |
fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?"); |
392 |
|
393 |
Restrict libev to the select and poll backends, and do not allow |
394 |
environment settings to be taken into account: |
395 |
|
396 |
ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV); |
397 |
|
398 |
Use whatever libev has to offer, but make sure that kqueue is used if |
399 |
available (warning, breaks stuff, best use only with your own private |
400 |
event loop and only if you know the OS supports your types of fds): |
401 |
|
402 |
ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE); |
403 |
|
404 |
=item struct ev_loop *ev_loop_new (unsigned int flags) |
405 |
|
406 |
Similar to C<ev_default_loop>, but always creates a new event loop that is |
407 |
always distinct from the default loop. Unlike the default loop, it cannot |
408 |
handle signal and child watchers, and attempts to do so will be greeted by |
409 |
undefined behaviour (or a failed assertion if assertions are enabled). |
410 |
|
411 |
Example: Try to create a event loop that uses epoll and nothing else. |
412 |
|
413 |
struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV); |
414 |
if (!epoller) |
415 |
fatal ("no epoll found here, maybe it hides under your chair"); |
416 |
|
417 |
=item ev_default_destroy () |
418 |
|
419 |
Destroys the default loop again (frees all memory and kernel state |
420 |
etc.). None of the active event watchers will be stopped in the normal |
421 |
sense, so e.g. C<ev_is_active> might still return true. It is your |
422 |
responsibility to either stop all watchers cleanly yoursef I<before> |
423 |
calling this function, or cope with the fact afterwards (which is usually |
424 |
the easiest thing, you can just ignore the watchers and/or C<free ()> them |
425 |
for example). |
426 |
|
427 |
Note that certain global state, such as signal state, will not be freed by |
428 |
this function, and related watchers (such as signal and child watchers) |
429 |
would need to be stopped manually. |
430 |
|
431 |
In general it is not advisable to call this function except in the |
432 |
rare occasion where you really need to free e.g. the signal handling |
433 |
pipe fds. If you need dynamically allocated loops it is better to use |
434 |
C<ev_loop_new> and C<ev_loop_destroy>). |
435 |
|
436 |
=item ev_loop_destroy (loop) |
437 |
|
438 |
Like C<ev_default_destroy>, but destroys an event loop created by an |
439 |
earlier call to C<ev_loop_new>. |
440 |
|
441 |
=item ev_default_fork () |
442 |
|
443 |
This function reinitialises the kernel state for backends that have |
444 |
one. Despite the name, you can call it anytime, but it makes most sense |
445 |
after forking, in either the parent or child process (or both, but that |
446 |
again makes little sense). |
447 |
|
448 |
You I<must> call this function in the child process after forking if and |
449 |
only if you want to use the event library in both processes. If you just |
450 |
fork+exec, you don't have to call it. |
451 |
|
452 |
The function itself is quite fast and it's usually not a problem to call |
453 |
it just in case after a fork. To make this easy, the function will fit in |
454 |
quite nicely into a call to C<pthread_atfork>: |
455 |
|
456 |
pthread_atfork (0, 0, ev_default_fork); |
457 |
|
458 |
At the moment, C<EVBACKEND_SELECT> and C<EVBACKEND_POLL> are safe to use |
459 |
without calling this function, so if you force one of those backends you |
460 |
do not need to care. |
461 |
|
462 |
=item ev_loop_fork (loop) |
463 |
|
464 |
Like C<ev_default_fork>, but acts on an event loop created by |
465 |
C<ev_loop_new>. Yes, you have to call this on every allocated event loop |
466 |
after fork, and how you do this is entirely your own problem. |
467 |
|
468 |
=item unsigned int ev_loop_count (loop) |
469 |
|
470 |
Returns the count of loop iterations for the loop, which is identical to |
471 |
the number of times libev did poll for new events. It starts at C<0> and |
472 |
happily wraps around with enough iterations. |
473 |
|
474 |
This value can sometimes be useful as a generation counter of sorts (it |
475 |
"ticks" the number of loop iterations), as it roughly corresponds with |
476 |
C<ev_prepare> and C<ev_check> calls. |
477 |
|
478 |
=item unsigned int ev_backend (loop) |
479 |
|
480 |
Returns one of the C<EVBACKEND_*> flags indicating the event backend in |
481 |
use. |
482 |
|
483 |
=item ev_tstamp ev_now (loop) |
484 |
|
485 |
Returns the current "event loop time", which is the time the event loop |
486 |
received events and started processing them. This timestamp does not |
487 |
change as long as callbacks are being processed, and this is also the base |
488 |
time used for relative timers. You can treat it as the timestamp of the |
489 |
event occurring (or more correctly, libev finding out about it). |
490 |
|
491 |
=item ev_loop (loop, int flags) |
492 |
|
493 |
Finally, this is it, the event handler. This function usually is called |
494 |
after you initialised all your watchers and you want to start handling |
495 |
events. |
496 |
|
497 |
If the flags argument is specified as C<0>, it will not return until |
498 |
either no event watchers are active anymore or C<ev_unloop> was called. |
499 |
|
500 |
Please note that an explicit C<ev_unloop> is usually better than |
501 |
relying on all watchers to be stopped when deciding when a program has |
502 |
finished (especially in interactive programs), but having a program that |
503 |
automatically loops as long as it has to and no longer by virtue of |
504 |
relying on its watchers stopping correctly is a thing of beauty. |
505 |
|
506 |
A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle |
507 |
those events and any outstanding ones, but will not block your process in |
508 |
case there are no events and will return after one iteration of the loop. |
509 |
|
510 |
A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if |
511 |
neccessary) and will handle those and any outstanding ones. It will block |
512 |
your process until at least one new event arrives, and will return after |
513 |
one iteration of the loop. This is useful if you are waiting for some |
514 |
external event in conjunction with something not expressible using other |
515 |
libev watchers. However, a pair of C<ev_prepare>/C<ev_check> watchers is |
516 |
usually a better approach for this kind of thing. |
517 |
|
518 |
Here are the gory details of what C<ev_loop> does: |
519 |
|
520 |
- Before the first iteration, call any pending watchers. |
521 |
* If there are no active watchers (reference count is zero), return. |
522 |
- Queue all prepare watchers and then call all outstanding watchers. |
523 |
- If we have been forked, recreate the kernel state. |
524 |
- Update the kernel state with all outstanding changes. |
525 |
- Update the "event loop time". |
526 |
- Calculate for how long to block. |
527 |
- Block the process, waiting for any events. |
528 |
- Queue all outstanding I/O (fd) events. |
529 |
- Update the "event loop time" and do time jump handling. |
530 |
- Queue all outstanding timers. |
531 |
- Queue all outstanding periodics. |
532 |
- If no events are pending now, queue all idle watchers. |
533 |
- Queue all check watchers. |
534 |
- Call all queued watchers in reverse order (i.e. check watchers first). |
535 |
Signals and child watchers are implemented as I/O watchers, and will |
536 |
be handled here by queueing them when their watcher gets executed. |
537 |
- If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK |
538 |
were used, return, otherwise continue with step *. |
539 |
|
540 |
Example: Queue some jobs and then loop until no events are outsanding |
541 |
anymore. |
542 |
|
543 |
... queue jobs here, make sure they register event watchers as long |
544 |
... as they still have work to do (even an idle watcher will do..) |
545 |
ev_loop (my_loop, 0); |
546 |
... jobs done. yeah! |
547 |
|
548 |
=item ev_unloop (loop, how) |
549 |
|
550 |
Can be used to make a call to C<ev_loop> return early (but only after it |
551 |
has processed all outstanding events). The C<how> argument must be either |
552 |
C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or |
553 |
C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return. |
554 |
|
555 |
=item ev_ref (loop) |
556 |
|
557 |
=item ev_unref (loop) |
558 |
|
559 |
Ref/unref can be used to add or remove a reference count on the event |
560 |
loop: Every watcher keeps one reference, and as long as the reference |
561 |
count is nonzero, C<ev_loop> will not return on its own. If you have |
562 |
a watcher you never unregister that should not keep C<ev_loop> from |
563 |
returning, ev_unref() after starting, and ev_ref() before stopping it. For |
564 |
example, libev itself uses this for its internal signal pipe: It is not |
565 |
visible to the libev user and should not keep C<ev_loop> from exiting if |
566 |
no event watchers registered by it are active. It is also an excellent |
567 |
way to do this for generic recurring timers or from within third-party |
568 |
libraries. Just remember to I<unref after start> and I<ref before stop>. |
569 |
|
570 |
Example: Create a signal watcher, but keep it from keeping C<ev_loop> |
571 |
running when nothing else is active. |
572 |
|
573 |
struct ev_signal exitsig; |
574 |
ev_signal_init (&exitsig, sig_cb, SIGINT); |
575 |
ev_signal_start (loop, &exitsig); |
576 |
evf_unref (loop); |
577 |
|
578 |
Example: For some weird reason, unregister the above signal handler again. |
579 |
|
580 |
ev_ref (loop); |
581 |
ev_signal_stop (loop, &exitsig); |
582 |
|
583 |
=item ev_set_io_collect_interval (loop, ev_tstamp interval) |
584 |
|
585 |
=item ev_set_timeout_collect_interval (loop, ev_tstamp interval) |
586 |
|
587 |
These advanced functions influence the time that libev will spend waiting |
588 |
for events. Both are by default C<0>, meaning that libev will try to |
589 |
invoke timer/periodic callbacks and I/O callbacks with minimum latency. |
590 |
|
591 |
Setting these to a higher value (the C<interval> I<must> be >= C<0>) |
592 |
allows libev to delay invocation of I/O and timer/periodic callbacks to |
593 |
increase efficiency of loop iterations. |
594 |
|
595 |
The background is that sometimes your program runs just fast enough to |
596 |
handle one (or very few) event(s) per loop iteration. While this makes |
597 |
the program responsive, it also wastes a lot of CPU time to poll for new |
598 |
events, especially with backends like C<select ()> which have a high |
599 |
overhead for the actual polling but can deliver many events at once. |
600 |
|
601 |
By setting a higher I<io collect interval> you allow libev to spend more |
602 |
time collecting I/O events, so you can handle more events per iteration, |
603 |
at the cost of increasing latency. Timeouts (both C<ev_periodic> and |
604 |
C<ev_timer>) will be not affected. Setting this to a non-null value will |
605 |
introduce an additional C<ev_sleep ()> call into most loop iterations. |
606 |
|
607 |
Likewise, by setting a higher I<timeout collect interval> you allow libev |
608 |
to spend more time collecting timeouts, at the expense of increased |
609 |
latency (the watcher callback will be called later). C<ev_io> watchers |
610 |
will not be affected. Setting this to a non-null value will not introduce |
611 |
any overhead in libev. |
612 |
|
613 |
Many (busy) programs can usually benefit by setting the io collect |
614 |
interval to a value near C<0.1> or so, which is often enough for |
615 |
interactive servers (of course not for games), likewise for timeouts. It |
616 |
usually doesn't make much sense to set it to a lower value than C<0.01>, |
617 |
as this approsaches the timing granularity of most systems. |
618 |
|
619 |
=back |
620 |
|
621 |
|
622 |
=head1 ANATOMY OF A WATCHER |
623 |
|
624 |
A watcher is a structure that you create and register to record your |
625 |
interest in some event. For instance, if you want to wait for STDIN to |
626 |
become readable, you would create an C<ev_io> watcher for that: |
627 |
|
628 |
static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents) |
629 |
{ |
630 |
ev_io_stop (w); |
631 |
ev_unloop (loop, EVUNLOOP_ALL); |
632 |
} |
633 |
|
634 |
struct ev_loop *loop = ev_default_loop (0); |
635 |
struct ev_io stdin_watcher; |
636 |
ev_init (&stdin_watcher, my_cb); |
637 |
ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ); |
638 |
ev_io_start (loop, &stdin_watcher); |
639 |
ev_loop (loop, 0); |
640 |
|
641 |
As you can see, you are responsible for allocating the memory for your |
642 |
watcher structures (and it is usually a bad idea to do this on the stack, |
643 |
although this can sometimes be quite valid). |
644 |
|
645 |
Each watcher structure must be initialised by a call to C<ev_init |
646 |
(watcher *, callback)>, which expects a callback to be provided. This |
647 |
callback gets invoked each time the event occurs (or, in the case of io |
648 |
watchers, each time the event loop detects that the file descriptor given |
649 |
is readable and/or writable). |
650 |
|
651 |
Each watcher type has its own C<< ev_<type>_set (watcher *, ...) >> macro |
652 |
with arguments specific to this watcher type. There is also a macro |
653 |
to combine initialisation and setting in one call: C<< ev_<type>_init |
654 |
(watcher *, callback, ...) >>. |
655 |
|
656 |
To make the watcher actually watch out for events, you have to start it |
657 |
with a watcher-specific start function (C<< ev_<type>_start (loop, watcher |
658 |
*) >>), and you can stop watching for events at any time by calling the |
659 |
corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>. |
660 |
|
661 |
As long as your watcher is active (has been started but not stopped) you |
662 |
must not touch the values stored in it. Most specifically you must never |
663 |
reinitialise it or call its C<set> macro. |
664 |
|
665 |
Each and every callback receives the event loop pointer as first, the |
666 |
registered watcher structure as second, and a bitset of received events as |
667 |
third argument. |
668 |
|
669 |
The received events usually include a single bit per event type received |
670 |
(you can receive multiple events at the same time). The possible bit masks |
671 |
are: |
672 |
|
673 |
=over 4 |
674 |
|
675 |
=item C<EV_READ> |
676 |
|
677 |
=item C<EV_WRITE> |
678 |
|
679 |
The file descriptor in the C<ev_io> watcher has become readable and/or |
680 |
writable. |
681 |
|
682 |
=item C<EV_TIMEOUT> |
683 |
|
684 |
The C<ev_timer> watcher has timed out. |
685 |
|
686 |
=item C<EV_PERIODIC> |
687 |
|
688 |
The C<ev_periodic> watcher has timed out. |
689 |
|
690 |
=item C<EV_SIGNAL> |
691 |
|
692 |
The signal specified in the C<ev_signal> watcher has been received by a thread. |
693 |
|
694 |
=item C<EV_CHILD> |
695 |
|
696 |
The pid specified in the C<ev_child> watcher has received a status change. |
697 |
|
698 |
=item C<EV_STAT> |
699 |
|
700 |
The path specified in the C<ev_stat> watcher changed its attributes somehow. |
701 |
|
702 |
=item C<EV_IDLE> |
703 |
|
704 |
The C<ev_idle> watcher has determined that you have nothing better to do. |
705 |
|
706 |
=item C<EV_PREPARE> |
707 |
|
708 |
=item C<EV_CHECK> |
709 |
|
710 |
All C<ev_prepare> watchers are invoked just I<before> C<ev_loop> starts |
711 |
to gather new events, and all C<ev_check> watchers are invoked just after |
712 |
C<ev_loop> has gathered them, but before it invokes any callbacks for any |
713 |
received events. Callbacks of both watcher types can start and stop as |
714 |
many watchers as they want, and all of them will be taken into account |
715 |
(for example, a C<ev_prepare> watcher might start an idle watcher to keep |
716 |
C<ev_loop> from blocking). |
717 |
|
718 |
=item C<EV_EMBED> |
719 |
|
720 |
The embedded event loop specified in the C<ev_embed> watcher needs attention. |
721 |
|
722 |
=item C<EV_FORK> |
723 |
|
724 |
The event loop has been resumed in the child process after fork (see |
725 |
C<ev_fork>). |
726 |
|
727 |
=item C<EV_ERROR> |
728 |
|
729 |
An unspecified error has occured, the watcher has been stopped. This might |
730 |
happen because the watcher could not be properly started because libev |
731 |
ran out of memory, a file descriptor was found to be closed or any other |
732 |
problem. You best act on it by reporting the problem and somehow coping |
733 |
with the watcher being stopped. |
734 |
|
735 |
Libev will usually signal a few "dummy" events together with an error, |
736 |
for example it might indicate that a fd is readable or writable, and if |
737 |
your callbacks is well-written it can just attempt the operation and cope |
738 |
with the error from read() or write(). This will not work in multithreaded |
739 |
programs, though, so beware. |
740 |
|
741 |
=back |
742 |
|
743 |
=head2 GENERIC WATCHER FUNCTIONS |
744 |
|
745 |
In the following description, C<TYPE> stands for the watcher type, |
746 |
e.g. C<timer> for C<ev_timer> watchers and C<io> for C<ev_io> watchers. |
747 |
|
748 |
=over 4 |
749 |
|
750 |
=item C<ev_init> (ev_TYPE *watcher, callback) |
751 |
|
752 |
This macro initialises the generic portion of a watcher. The contents |
753 |
of the watcher object can be arbitrary (so C<malloc> will do). Only |
754 |
the generic parts of the watcher are initialised, you I<need> to call |
755 |
the type-specific C<ev_TYPE_set> macro afterwards to initialise the |
756 |
type-specific parts. For each type there is also a C<ev_TYPE_init> macro |
757 |
which rolls both calls into one. |
758 |
|
759 |
You can reinitialise a watcher at any time as long as it has been stopped |
760 |
(or never started) and there are no pending events outstanding. |
761 |
|
762 |
The callback is always of type C<void (*)(ev_loop *loop, ev_TYPE *watcher, |
763 |
int revents)>. |
764 |
|
765 |
=item C<ev_TYPE_set> (ev_TYPE *, [args]) |
766 |
|
767 |
This macro initialises the type-specific parts of a watcher. You need to |
768 |
call C<ev_init> at least once before you call this macro, but you can |
769 |
call C<ev_TYPE_set> any number of times. You must not, however, call this |
770 |
macro on a watcher that is active (it can be pending, however, which is a |
771 |
difference to the C<ev_init> macro). |
772 |
|
773 |
Although some watcher types do not have type-specific arguments |
774 |
(e.g. C<ev_prepare>) you still need to call its C<set> macro. |
775 |
|
776 |
=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args]) |
777 |
|
778 |
This convinience macro rolls both C<ev_init> and C<ev_TYPE_set> macro |
779 |
calls into a single call. This is the most convinient method to initialise |
780 |
a watcher. The same limitations apply, of course. |
781 |
|
782 |
=item C<ev_TYPE_start> (loop *, ev_TYPE *watcher) |
783 |
|
784 |
Starts (activates) the given watcher. Only active watchers will receive |
785 |
events. If the watcher is already active nothing will happen. |
786 |
|
787 |
=item C<ev_TYPE_stop> (loop *, ev_TYPE *watcher) |
788 |
|
789 |
Stops the given watcher again (if active) and clears the pending |
790 |
status. It is possible that stopped watchers are pending (for example, |
791 |
non-repeating timers are being stopped when they become pending), but |
792 |
C<ev_TYPE_stop> ensures that the watcher is neither active nor pending. If |
793 |
you want to free or reuse the memory used by the watcher it is therefore a |
794 |
good idea to always call its C<ev_TYPE_stop> function. |
795 |
|
796 |
=item bool ev_is_active (ev_TYPE *watcher) |
797 |
|
798 |
Returns a true value iff the watcher is active (i.e. it has been started |
799 |
and not yet been stopped). As long as a watcher is active you must not modify |
800 |
it. |
801 |
|
802 |
=item bool ev_is_pending (ev_TYPE *watcher) |
803 |
|
804 |
Returns a true value iff the watcher is pending, (i.e. it has outstanding |
805 |
events but its callback has not yet been invoked). As long as a watcher |
806 |
is pending (but not active) you must not call an init function on it (but |
807 |
C<ev_TYPE_set> is safe), you must not change its priority, and you must |
808 |
make sure the watcher is available to libev (e.g. you cannot C<free ()> |
809 |
it). |
810 |
|
811 |
=item callback ev_cb (ev_TYPE *watcher) |
812 |
|
813 |
Returns the callback currently set on the watcher. |
814 |
|
815 |
=item ev_cb_set (ev_TYPE *watcher, callback) |
816 |
|
817 |
Change the callback. You can change the callback at virtually any time |
818 |
(modulo threads). |
819 |
|
820 |
=item ev_set_priority (ev_TYPE *watcher, priority) |
821 |
|
822 |
=item int ev_priority (ev_TYPE *watcher) |
823 |
|
824 |
Set and query the priority of the watcher. The priority is a small |
825 |
integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI> |
826 |
(default: C<-2>). Pending watchers with higher priority will be invoked |
827 |
before watchers with lower priority, but priority will not keep watchers |
828 |
from being executed (except for C<ev_idle> watchers). |
829 |
|
830 |
This means that priorities are I<only> used for ordering callback |
831 |
invocation after new events have been received. This is useful, for |
832 |
example, to reduce latency after idling, or more often, to bind two |
833 |
watchers on the same event and make sure one is called first. |
834 |
|
835 |
If you need to suppress invocation when higher priority events are pending |
836 |
you need to look at C<ev_idle> watchers, which provide this functionality. |
837 |
|
838 |
You I<must not> change the priority of a watcher as long as it is active or |
839 |
pending. |
840 |
|
841 |
The default priority used by watchers when no priority has been set is |
842 |
always C<0>, which is supposed to not be too high and not be too low :). |
843 |
|
844 |
Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is |
845 |
fine, as long as you do not mind that the priority value you query might |
846 |
or might not have been adjusted to be within valid range. |
847 |
|
848 |
=item ev_invoke (loop, ev_TYPE *watcher, int revents) |
849 |
|
850 |
Invoke the C<watcher> with the given C<loop> and C<revents>. Neither |
851 |
C<loop> nor C<revents> need to be valid as long as the watcher callback |
852 |
can deal with that fact. |
853 |
|
854 |
=item int ev_clear_pending (loop, ev_TYPE *watcher) |
855 |
|
856 |
If the watcher is pending, this function returns clears its pending status |
857 |
and returns its C<revents> bitset (as if its callback was invoked). If the |
858 |
watcher isn't pending it does nothing and returns C<0>. |
859 |
|
860 |
=back |
861 |
|
862 |
|
863 |
=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER |
864 |
|
865 |
Each watcher has, by default, a member C<void *data> that you can change |
866 |
and read at any time, libev will completely ignore it. This can be used |
867 |
to associate arbitrary data with your watcher. If you need more data and |
868 |
don't want to allocate memory and store a pointer to it in that data |
869 |
member, you can also "subclass" the watcher type and provide your own |
870 |
data: |
871 |
|
872 |
struct my_io |
873 |
{ |
874 |
struct ev_io io; |
875 |
int otherfd; |
876 |
void *somedata; |
877 |
struct whatever *mostinteresting; |
878 |
} |
879 |
|
880 |
And since your callback will be called with a pointer to the watcher, you |
881 |
can cast it back to your own type: |
882 |
|
883 |
static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents) |
884 |
{ |
885 |
struct my_io *w = (struct my_io *)w_; |
886 |
... |
887 |
} |
888 |
|
889 |
More interesting and less C-conformant ways of casting your callback type |
890 |
instead have been omitted. |
891 |
|
892 |
Another common scenario is having some data structure with multiple |
893 |
watchers: |
894 |
|
895 |
struct my_biggy |
896 |
{ |
897 |
int some_data; |
898 |
ev_timer t1; |
899 |
ev_timer t2; |
900 |
} |
901 |
|
902 |
In this case getting the pointer to C<my_biggy> is a bit more complicated, |
903 |
you need to use C<offsetof>: |
904 |
|
905 |
#include <stddef.h> |
906 |
|
907 |
static void |
908 |
t1_cb (EV_P_ struct ev_timer *w, int revents) |
909 |
{ |
910 |
struct my_biggy big = (struct my_biggy * |
911 |
(((char *)w) - offsetof (struct my_biggy, t1)); |
912 |
} |
913 |
|
914 |
static void |
915 |
t2_cb (EV_P_ struct ev_timer *w, int revents) |
916 |
{ |
917 |
struct my_biggy big = (struct my_biggy * |
918 |
(((char *)w) - offsetof (struct my_biggy, t2)); |
919 |
} |
920 |
|
921 |
|
922 |
=head1 WATCHER TYPES |
923 |
|
924 |
This section describes each watcher in detail, but will not repeat |
925 |
information given in the last section. Any initialisation/set macros, |
926 |
functions and members specific to the watcher type are explained. |
927 |
|
928 |
Members are additionally marked with either I<[read-only]>, meaning that, |
929 |
while the watcher is active, you can look at the member and expect some |
930 |
sensible content, but you must not modify it (you can modify it while the |
931 |
watcher is stopped to your hearts content), or I<[read-write]>, which |
932 |
means you can expect it to have some sensible content while the watcher |
933 |
is active, but you can also modify it. Modifying it may not do something |
934 |
sensible or take immediate effect (or do anything at all), but libev will |
935 |
not crash or malfunction in any way. |
936 |
|
937 |
|
938 |
=head2 C<ev_io> - is this file descriptor readable or writable? |
939 |
|
940 |
I/O watchers check whether a file descriptor is readable or writable |
941 |
in each iteration of the event loop, or, more precisely, when reading |
942 |
would not block the process and writing would at least be able to write |
943 |
some data. This behaviour is called level-triggering because you keep |
944 |
receiving events as long as the condition persists. Remember you can stop |
945 |
the watcher if you don't want to act on the event and neither want to |
946 |
receive future events. |
947 |
|
948 |
In general you can register as many read and/or write event watchers per |
949 |
fd as you want (as long as you don't confuse yourself). Setting all file |
950 |
descriptors to non-blocking mode is also usually a good idea (but not |
951 |
required if you know what you are doing). |
952 |
|
953 |
You have to be careful with dup'ed file descriptors, though. Some backends |
954 |
(the linux epoll backend is a notable example) cannot handle dup'ed file |
955 |
descriptors correctly if you register interest in two or more fds pointing |
956 |
to the same underlying file/socket/etc. description (that is, they share |
957 |
the same underlying "file open"). |
958 |
|
959 |
If you must do this, then force the use of a known-to-be-good backend |
960 |
(at the time of this writing, this includes only C<EVBACKEND_SELECT> and |
961 |
C<EVBACKEND_POLL>). |
962 |
|
963 |
Another thing you have to watch out for is that it is quite easy to |
964 |
receive "spurious" readyness notifications, that is your callback might |
965 |
be called with C<EV_READ> but a subsequent C<read>(2) will actually block |
966 |
because there is no data. Not only are some backends known to create a |
967 |
lot of those (for example solaris ports), it is very easy to get into |
968 |
this situation even with a relatively standard program structure. Thus |
969 |
it is best to always use non-blocking I/O: An extra C<read>(2) returning |
970 |
C<EAGAIN> is far preferable to a program hanging until some data arrives. |
971 |
|
972 |
If you cannot run the fd in non-blocking mode (for example you should not |
973 |
play around with an Xlib connection), then you have to seperately re-test |
974 |
whether a file descriptor is really ready with a known-to-be good interface |
975 |
such as poll (fortunately in our Xlib example, Xlib already does this on |
976 |
its own, so its quite safe to use). |
977 |
|
978 |
=head3 The special problem of disappearing file descriptors |
979 |
|
980 |
Some backends (e.g. kqueue, epoll) need to be told about closing a file |
981 |
descriptor (either by calling C<close> explicitly or by any other means, |
982 |
such as C<dup>). The reason is that you register interest in some file |
983 |
descriptor, but when it goes away, the operating system will silently drop |
984 |
this interest. If another file descriptor with the same number then is |
985 |
registered with libev, there is no efficient way to see that this is, in |
986 |
fact, a different file descriptor. |
987 |
|
988 |
To avoid having to explicitly tell libev about such cases, libev follows |
989 |
the following policy: Each time C<ev_io_set> is being called, libev |
990 |
will assume that this is potentially a new file descriptor, otherwise |
991 |
it is assumed that the file descriptor stays the same. That means that |
992 |
you I<have> to call C<ev_io_set> (or C<ev_io_init>) when you change the |
993 |
descriptor even if the file descriptor number itself did not change. |
994 |
|
995 |
This is how one would do it normally anyway, the important point is that |
996 |
the libev application should not optimise around libev but should leave |
997 |
optimisations to libev. |
998 |
|
999 |
=head3 The special problem of dup'ed file descriptors |
1000 |
|
1001 |
Some backends (e.g. epoll), cannot register events for file descriptors, |
1002 |
but only events for the underlying file descriptions. That menas when you |
1003 |
have C<dup ()>'ed file descriptors and register events for them, only one |
1004 |
file descriptor might actually receive events. |
1005 |
|
1006 |
There is no workaorund possible except not registering events |
1007 |
for potentially C<dup ()>'ed file descriptors or to resort to |
1008 |
C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>. |
1009 |
|
1010 |
=head3 The special problem of fork |
1011 |
|
1012 |
Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit |
1013 |
useless behaviour. Libev fully supports fork, but needs to be told about |
1014 |
it in the child. |
1015 |
|
1016 |
To support fork in your programs, you either have to call |
1017 |
C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child, |
1018 |
enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or |
1019 |
C<EVBACKEND_POLL>. |
1020 |
|
1021 |
|
1022 |
=head3 Watcher-Specific Functions |
1023 |
|
1024 |
=over 4 |
1025 |
|
1026 |
=item ev_io_init (ev_io *, callback, int fd, int events) |
1027 |
|
1028 |
=item ev_io_set (ev_io *, int fd, int events) |
1029 |
|
1030 |
Configures an C<ev_io> watcher. The C<fd> is the file descriptor to |
1031 |
rceeive events for and events is either C<EV_READ>, C<EV_WRITE> or |
1032 |
C<EV_READ | EV_WRITE> to receive the given events. |
1033 |
|
1034 |
=item int fd [read-only] |
1035 |
|
1036 |
The file descriptor being watched. |
1037 |
|
1038 |
=item int events [read-only] |
1039 |
|
1040 |
The events being watched. |
1041 |
|
1042 |
=back |
1043 |
|
1044 |
Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well |
1045 |
readable, but only once. Since it is likely line-buffered, you could |
1046 |
attempt to read a whole line in the callback. |
1047 |
|
1048 |
static void |
1049 |
stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents) |
1050 |
{ |
1051 |
ev_io_stop (loop, w); |
1052 |
.. read from stdin here (or from w->fd) and haqndle any I/O errors |
1053 |
} |
1054 |
|
1055 |
... |
1056 |
struct ev_loop *loop = ev_default_init (0); |
1057 |
struct ev_io stdin_readable; |
1058 |
ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ); |
1059 |
ev_io_start (loop, &stdin_readable); |
1060 |
ev_loop (loop, 0); |
1061 |
|
1062 |
|
1063 |
=head2 C<ev_timer> - relative and optionally repeating timeouts |
1064 |
|
1065 |
Timer watchers are simple relative timers that generate an event after a |
1066 |
given time, and optionally repeating in regular intervals after that. |
1067 |
|
1068 |
The timers are based on real time, that is, if you register an event that |
1069 |
times out after an hour and you reset your system clock to last years |
1070 |
time, it will still time out after (roughly) and hour. "Roughly" because |
1071 |
detecting time jumps is hard, and some inaccuracies are unavoidable (the |
1072 |
monotonic clock option helps a lot here). |
1073 |
|
1074 |
The relative timeouts are calculated relative to the C<ev_now ()> |
1075 |
time. This is usually the right thing as this timestamp refers to the time |
1076 |
of the event triggering whatever timeout you are modifying/starting. If |
1077 |
you suspect event processing to be delayed and you I<need> to base the timeout |
1078 |
on the current time, use something like this to adjust for this: |
1079 |
|
1080 |
ev_timer_set (&timer, after + ev_now () - ev_time (), 0.); |
1081 |
|
1082 |
The callback is guarenteed to be invoked only when its timeout has passed, |
1083 |
but if multiple timers become ready during the same loop iteration then |
1084 |
order of execution is undefined. |
1085 |
|
1086 |
=head3 Watcher-Specific Functions and Data Members |
1087 |
|
1088 |
=over 4 |
1089 |
|
1090 |
=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat) |
1091 |
|
1092 |
=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat) |
1093 |
|
1094 |
Configure the timer to trigger after C<after> seconds. If C<repeat> is |
1095 |
C<0.>, then it will automatically be stopped. If it is positive, then the |
1096 |
timer will automatically be configured to trigger again C<repeat> seconds |
1097 |
later, again, and again, until stopped manually. |
1098 |
|
1099 |
The timer itself will do a best-effort at avoiding drift, that is, if you |
1100 |
configure a timer to trigger every 10 seconds, then it will trigger at |
1101 |
exactly 10 second intervals. If, however, your program cannot keep up with |
1102 |
the timer (because it takes longer than those 10 seconds to do stuff) the |
1103 |
timer will not fire more than once per event loop iteration. |
1104 |
|
1105 |
=item ev_timer_again (loop) |
1106 |
|
1107 |
This will act as if the timer timed out and restart it again if it is |
1108 |
repeating. The exact semantics are: |
1109 |
|
1110 |
If the timer is pending, its pending status is cleared. |
1111 |
|
1112 |
If the timer is started but nonrepeating, stop it (as if it timed out). |
1113 |
|
1114 |
If the timer is repeating, either start it if necessary (with the |
1115 |
C<repeat> value), or reset the running timer to the C<repeat> value. |
1116 |
|
1117 |
This sounds a bit complicated, but here is a useful and typical |
1118 |
example: Imagine you have a tcp connection and you want a so-called idle |
1119 |
timeout, that is, you want to be called when there have been, say, 60 |
1120 |
seconds of inactivity on the socket. The easiest way to do this is to |
1121 |
configure an C<ev_timer> with a C<repeat> value of C<60> and then call |
1122 |
C<ev_timer_again> each time you successfully read or write some data. If |
1123 |
you go into an idle state where you do not expect data to travel on the |
1124 |
socket, you can C<ev_timer_stop> the timer, and C<ev_timer_again> will |
1125 |
automatically restart it if need be. |
1126 |
|
1127 |
That means you can ignore the C<after> value and C<ev_timer_start> |
1128 |
altogether and only ever use the C<repeat> value and C<ev_timer_again>: |
1129 |
|
1130 |
ev_timer_init (timer, callback, 0., 5.); |
1131 |
ev_timer_again (loop, timer); |
1132 |
... |
1133 |
timer->again = 17.; |
1134 |
ev_timer_again (loop, timer); |
1135 |
... |
1136 |
timer->again = 10.; |
1137 |
ev_timer_again (loop, timer); |
1138 |
|
1139 |
This is more slightly efficient then stopping/starting the timer each time |
1140 |
you want to modify its timeout value. |
1141 |
|
1142 |
=item ev_tstamp repeat [read-write] |
1143 |
|
1144 |
The current C<repeat> value. Will be used each time the watcher times out |
1145 |
or C<ev_timer_again> is called and determines the next timeout (if any), |
1146 |
which is also when any modifications are taken into account. |
1147 |
|
1148 |
=back |
1149 |
|
1150 |
Example: Create a timer that fires after 60 seconds. |
1151 |
|
1152 |
static void |
1153 |
one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents) |
1154 |
{ |
1155 |
.. one minute over, w is actually stopped right here |
1156 |
} |
1157 |
|
1158 |
struct ev_timer mytimer; |
1159 |
ev_timer_init (&mytimer, one_minute_cb, 60., 0.); |
1160 |
ev_timer_start (loop, &mytimer); |
1161 |
|
1162 |
Example: Create a timeout timer that times out after 10 seconds of |
1163 |
inactivity. |
1164 |
|
1165 |
static void |
1166 |
timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents) |
1167 |
{ |
1168 |
.. ten seconds without any activity |
1169 |
} |
1170 |
|
1171 |
struct ev_timer mytimer; |
1172 |
ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */ |
1173 |
ev_timer_again (&mytimer); /* start timer */ |
1174 |
ev_loop (loop, 0); |
1175 |
|
1176 |
// and in some piece of code that gets executed on any "activity": |
1177 |
// reset the timeout to start ticking again at 10 seconds |
1178 |
ev_timer_again (&mytimer); |
1179 |
|
1180 |
|
1181 |
=head2 C<ev_periodic> - to cron or not to cron? |
1182 |
|
1183 |
Periodic watchers are also timers of a kind, but they are very versatile |
1184 |
(and unfortunately a bit complex). |
1185 |
|
1186 |
Unlike C<ev_timer>'s, they are not based on real time (or relative time) |
1187 |
but on wallclock time (absolute time). You can tell a periodic watcher |
1188 |
to trigger "at" some specific point in time. For example, if you tell a |
1189 |
periodic watcher to trigger in 10 seconds (by specifiying e.g. C<ev_now () |
1190 |
+ 10.>) and then reset your system clock to the last year, then it will |
1191 |
take a year to trigger the event (unlike an C<ev_timer>, which would trigger |
1192 |
roughly 10 seconds later). |
1193 |
|
1194 |
They can also be used to implement vastly more complex timers, such as |
1195 |
triggering an event on each midnight, local time or other, complicated, |
1196 |
rules. |
1197 |
|
1198 |
As with timers, the callback is guarenteed to be invoked only when the |
1199 |
time (C<at>) has been passed, but if multiple periodic timers become ready |
1200 |
during the same loop iteration then order of execution is undefined. |
1201 |
|
1202 |
=head3 Watcher-Specific Functions and Data Members |
1203 |
|
1204 |
=over 4 |
1205 |
|
1206 |
=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb) |
1207 |
|
1208 |
=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb) |
1209 |
|
1210 |
Lots of arguments, lets sort it out... There are basically three modes of |
1211 |
operation, and we will explain them from simplest to complex: |
1212 |
|
1213 |
=over 4 |
1214 |
|
1215 |
=item * absolute timer (at = time, interval = reschedule_cb = 0) |
1216 |
|
1217 |
In this configuration the watcher triggers an event at the wallclock time |
1218 |
C<at> and doesn't repeat. It will not adjust when a time jump occurs, |
1219 |
that is, if it is to be run at January 1st 2011 then it will run when the |
1220 |
system time reaches or surpasses this time. |
1221 |
|
1222 |
=item * non-repeating interval timer (at = offset, interval > 0, reschedule_cb = 0) |
1223 |
|
1224 |
In this mode the watcher will always be scheduled to time out at the next |
1225 |
C<at + N * interval> time (for some integer N, which can also be negative) |
1226 |
and then repeat, regardless of any time jumps. |
1227 |
|
1228 |
This can be used to create timers that do not drift with respect to system |
1229 |
time: |
1230 |
|
1231 |
ev_periodic_set (&periodic, 0., 3600., 0); |
1232 |
|
1233 |
This doesn't mean there will always be 3600 seconds in between triggers, |
1234 |
but only that the the callback will be called when the system time shows a |
1235 |
full hour (UTC), or more correctly, when the system time is evenly divisible |
1236 |
by 3600. |
1237 |
|
1238 |
Another way to think about it (for the mathematically inclined) is that |
1239 |
C<ev_periodic> will try to run the callback in this mode at the next possible |
1240 |
time where C<time = at (mod interval)>, regardless of any time jumps. |
1241 |
|
1242 |
For numerical stability it is preferable that the C<at> value is near |
1243 |
C<ev_now ()> (the current time), but there is no range requirement for |
1244 |
this value. |
1245 |
|
1246 |
=item * manual reschedule mode (at and interval ignored, reschedule_cb = callback) |
1247 |
|
1248 |
In this mode the values for C<interval> and C<at> are both being |
1249 |
ignored. Instead, each time the periodic watcher gets scheduled, the |
1250 |
reschedule callback will be called with the watcher as first, and the |
1251 |
current time as second argument. |
1252 |
|
1253 |
NOTE: I<This callback MUST NOT stop or destroy any periodic watcher, |
1254 |
ever, or make any event loop modifications>. If you need to stop it, |
1255 |
return C<now + 1e30> (or so, fudge fudge) and stop it afterwards (e.g. by |
1256 |
starting an C<ev_prepare> watcher, which is legal). |
1257 |
|
1258 |
Its prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic *w, |
1259 |
ev_tstamp now)>, e.g.: |
1260 |
|
1261 |
static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now) |
1262 |
{ |
1263 |
return now + 60.; |
1264 |
} |
1265 |
|
1266 |
It must return the next time to trigger, based on the passed time value |
1267 |
(that is, the lowest time value larger than to the second argument). It |
1268 |
will usually be called just before the callback will be triggered, but |
1269 |
might be called at other times, too. |
1270 |
|
1271 |
NOTE: I<< This callback must always return a time that is later than the |
1272 |
passed C<now> value >>. Not even C<now> itself will do, it I<must> be larger. |
1273 |
|
1274 |
This can be used to create very complex timers, such as a timer that |
1275 |
triggers on each midnight, local time. To do this, you would calculate the |
1276 |
next midnight after C<now> and return the timestamp value for this. How |
1277 |
you do this is, again, up to you (but it is not trivial, which is the main |
1278 |
reason I omitted it as an example). |
1279 |
|
1280 |
=back |
1281 |
|
1282 |
=item ev_periodic_again (loop, ev_periodic *) |
1283 |
|
1284 |
Simply stops and restarts the periodic watcher again. This is only useful |
1285 |
when you changed some parameters or the reschedule callback would return |
1286 |
a different time than the last time it was called (e.g. in a crond like |
1287 |
program when the crontabs have changed). |
1288 |
|
1289 |
=item ev_tstamp offset [read-write] |
1290 |
|
1291 |
When repeating, this contains the offset value, otherwise this is the |
1292 |
absolute point in time (the C<at> value passed to C<ev_periodic_set>). |
1293 |
|
1294 |
Can be modified any time, but changes only take effect when the periodic |
1295 |
timer fires or C<ev_periodic_again> is being called. |
1296 |
|
1297 |
=item ev_tstamp interval [read-write] |
1298 |
|
1299 |
The current interval value. Can be modified any time, but changes only |
1300 |
take effect when the periodic timer fires or C<ev_periodic_again> is being |
1301 |
called. |
1302 |
|
1303 |
=item ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write] |
1304 |
|
1305 |
The current reschedule callback, or C<0>, if this functionality is |
1306 |
switched off. Can be changed any time, but changes only take effect when |
1307 |
the periodic timer fires or C<ev_periodic_again> is being called. |
1308 |
|
1309 |
=item ev_tstamp at [read-only] |
1310 |
|
1311 |
When active, contains the absolute time that the watcher is supposed to |
1312 |
trigger next. |
1313 |
|
1314 |
=back |
1315 |
|
1316 |
Example: Call a callback every hour, or, more precisely, whenever the |
1317 |
system clock is divisible by 3600. The callback invocation times have |
1318 |
potentially a lot of jittering, but good long-term stability. |
1319 |
|
1320 |
static void |
1321 |
clock_cb (struct ev_loop *loop, struct ev_io *w, int revents) |
1322 |
{ |
1323 |
... its now a full hour (UTC, or TAI or whatever your clock follows) |
1324 |
} |
1325 |
|
1326 |
struct ev_periodic hourly_tick; |
1327 |
ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0); |
1328 |
ev_periodic_start (loop, &hourly_tick); |
1329 |
|
1330 |
Example: The same as above, but use a reschedule callback to do it: |
1331 |
|
1332 |
#include <math.h> |
1333 |
|
1334 |
static ev_tstamp |
1335 |
my_scheduler_cb (struct ev_periodic *w, ev_tstamp now) |
1336 |
{ |
1337 |
return fmod (now, 3600.) + 3600.; |
1338 |
} |
1339 |
|
1340 |
ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb); |
1341 |
|
1342 |
Example: Call a callback every hour, starting now: |
1343 |
|
1344 |
struct ev_periodic hourly_tick; |
1345 |
ev_periodic_init (&hourly_tick, clock_cb, |
1346 |
fmod (ev_now (loop), 3600.), 3600., 0); |
1347 |
ev_periodic_start (loop, &hourly_tick); |
1348 |
|
1349 |
|
1350 |
=head2 C<ev_signal> - signal me when a signal gets signalled! |
1351 |
|
1352 |
Signal watchers will trigger an event when the process receives a specific |
1353 |
signal one or more times. Even though signals are very asynchronous, libev |
1354 |
will try it's best to deliver signals synchronously, i.e. as part of the |
1355 |
normal event processing, like any other event. |
1356 |
|
1357 |
You can configure as many watchers as you like per signal. Only when the |
1358 |
first watcher gets started will libev actually register a signal watcher |
1359 |
with the kernel (thus it coexists with your own signal handlers as long |
1360 |
as you don't register any with libev). Similarly, when the last signal |
1361 |
watcher for a signal is stopped libev will reset the signal handler to |
1362 |
SIG_DFL (regardless of what it was set to before). |
1363 |
|
1364 |
=head3 Watcher-Specific Functions and Data Members |
1365 |
|
1366 |
=over 4 |
1367 |
|
1368 |
=item ev_signal_init (ev_signal *, callback, int signum) |
1369 |
|
1370 |
=item ev_signal_set (ev_signal *, int signum) |
1371 |
|
1372 |
Configures the watcher to trigger on the given signal number (usually one |
1373 |
of the C<SIGxxx> constants). |
1374 |
|
1375 |
=item int signum [read-only] |
1376 |
|
1377 |
The signal the watcher watches out for. |
1378 |
|
1379 |
=back |
1380 |
|
1381 |
|
1382 |
=head2 C<ev_child> - watch out for process status changes |
1383 |
|
1384 |
Child watchers trigger when your process receives a SIGCHLD in response to |
1385 |
some child status changes (most typically when a child of yours dies). |
1386 |
|
1387 |
=head3 Watcher-Specific Functions and Data Members |
1388 |
|
1389 |
=over 4 |
1390 |
|
1391 |
=item ev_child_init (ev_child *, callback, int pid) |
1392 |
|
1393 |
=item ev_child_set (ev_child *, int pid) |
1394 |
|
1395 |
Configures the watcher to wait for status changes of process C<pid> (or |
1396 |
I<any> process if C<pid> is specified as C<0>). The callback can look |
1397 |
at the C<rstatus> member of the C<ev_child> watcher structure to see |
1398 |
the status word (use the macros from C<sys/wait.h> and see your systems |
1399 |
C<waitpid> documentation). The C<rpid> member contains the pid of the |
1400 |
process causing the status change. |
1401 |
|
1402 |
=item int pid [read-only] |
1403 |
|
1404 |
The process id this watcher watches out for, or C<0>, meaning any process id. |
1405 |
|
1406 |
=item int rpid [read-write] |
1407 |
|
1408 |
The process id that detected a status change. |
1409 |
|
1410 |
=item int rstatus [read-write] |
1411 |
|
1412 |
The process exit/trace status caused by C<rpid> (see your systems |
1413 |
C<waitpid> and C<sys/wait.h> documentation for details). |
1414 |
|
1415 |
=back |
1416 |
|
1417 |
Example: Try to exit cleanly on SIGINT and SIGTERM. |
1418 |
|
1419 |
static void |
1420 |
sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents) |
1421 |
{ |
1422 |
ev_unloop (loop, EVUNLOOP_ALL); |
1423 |
} |
1424 |
|
1425 |
struct ev_signal signal_watcher; |
1426 |
ev_signal_init (&signal_watcher, sigint_cb, SIGINT); |
1427 |
ev_signal_start (loop, &sigint_cb); |
1428 |
|
1429 |
|
1430 |
=head2 C<ev_stat> - did the file attributes just change? |
1431 |
|
1432 |
This watches a filesystem path for attribute changes. That is, it calls |
1433 |
C<stat> regularly (or when the OS says it changed) and sees if it changed |
1434 |
compared to the last time, invoking the callback if it did. |
1435 |
|
1436 |
The path does not need to exist: changing from "path exists" to "path does |
1437 |
not exist" is a status change like any other. The condition "path does |
1438 |
not exist" is signified by the C<st_nlink> field being zero (which is |
1439 |
otherwise always forced to be at least one) and all the other fields of |
1440 |
the stat buffer having unspecified contents. |
1441 |
|
1442 |
The path I<should> be absolute and I<must not> end in a slash. If it is |
1443 |
relative and your working directory changes, the behaviour is undefined. |
1444 |
|
1445 |
Since there is no standard to do this, the portable implementation simply |
1446 |
calls C<stat (2)> regularly on the path to see if it changed somehow. You |
1447 |
can specify a recommended polling interval for this case. If you specify |
1448 |
a polling interval of C<0> (highly recommended!) then a I<suitable, |
1449 |
unspecified default> value will be used (which you can expect to be around |
1450 |
five seconds, although this might change dynamically). Libev will also |
1451 |
impose a minimum interval which is currently around C<0.1>, but thats |
1452 |
usually overkill. |
1453 |
|
1454 |
This watcher type is not meant for massive numbers of stat watchers, |
1455 |
as even with OS-supported change notifications, this can be |
1456 |
resource-intensive. |
1457 |
|
1458 |
At the time of this writing, only the Linux inotify interface is |
1459 |
implemented (implementing kqueue support is left as an exercise for the |
1460 |
reader). Inotify will be used to give hints only and should not change the |
1461 |
semantics of C<ev_stat> watchers, which means that libev sometimes needs |
1462 |
to fall back to regular polling again even with inotify, but changes are |
1463 |
usually detected immediately, and if the file exists there will be no |
1464 |
polling. |
1465 |
|
1466 |
=head3 Watcher-Specific Functions and Data Members |
1467 |
|
1468 |
=over 4 |
1469 |
|
1470 |
=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval) |
1471 |
|
1472 |
=item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval) |
1473 |
|
1474 |
Configures the watcher to wait for status changes of the given |
1475 |
C<path>. The C<interval> is a hint on how quickly a change is expected to |
1476 |
be detected and should normally be specified as C<0> to let libev choose |
1477 |
a suitable value. The memory pointed to by C<path> must point to the same |
1478 |
path for as long as the watcher is active. |
1479 |
|
1480 |
The callback will be receive C<EV_STAT> when a change was detected, |
1481 |
relative to the attributes at the time the watcher was started (or the |
1482 |
last change was detected). |
1483 |
|
1484 |
=item ev_stat_stat (ev_stat *) |
1485 |
|
1486 |
Updates the stat buffer immediately with new values. If you change the |
1487 |
watched path in your callback, you could call this fucntion to avoid |
1488 |
detecting this change (while introducing a race condition). Can also be |
1489 |
useful simply to find out the new values. |
1490 |
|
1491 |
=item ev_statdata attr [read-only] |
1492 |
|
1493 |
The most-recently detected attributes of the file. Although the type is of |
1494 |
C<ev_statdata>, this is usually the (or one of the) C<struct stat> types |
1495 |
suitable for your system. If the C<st_nlink> member is C<0>, then there |
1496 |
was some error while C<stat>ing the file. |
1497 |
|
1498 |
=item ev_statdata prev [read-only] |
1499 |
|
1500 |
The previous attributes of the file. The callback gets invoked whenever |
1501 |
C<prev> != C<attr>. |
1502 |
|
1503 |
=item ev_tstamp interval [read-only] |
1504 |
|
1505 |
The specified interval. |
1506 |
|
1507 |
=item const char *path [read-only] |
1508 |
|
1509 |
The filesystem path that is being watched. |
1510 |
|
1511 |
=back |
1512 |
|
1513 |
Example: Watch C</etc/passwd> for attribute changes. |
1514 |
|
1515 |
static void |
1516 |
passwd_cb (struct ev_loop *loop, ev_stat *w, int revents) |
1517 |
{ |
1518 |
/* /etc/passwd changed in some way */ |
1519 |
if (w->attr.st_nlink) |
1520 |
{ |
1521 |
printf ("passwd current size %ld\n", (long)w->attr.st_size); |
1522 |
printf ("passwd current atime %ld\n", (long)w->attr.st_mtime); |
1523 |
printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime); |
1524 |
} |
1525 |
else |
1526 |
/* you shalt not abuse printf for puts */ |
1527 |
puts ("wow, /etc/passwd is not there, expect problems. " |
1528 |
"if this is windows, they already arrived\n"); |
1529 |
} |
1530 |
|
1531 |
... |
1532 |
ev_stat passwd; |
1533 |
|
1534 |
ev_stat_init (&passwd, passwd_cb, "/etc/passwd"); |
1535 |
ev_stat_start (loop, &passwd); |
1536 |
|
1537 |
|
1538 |
=head2 C<ev_idle> - when you've got nothing better to do... |
1539 |
|
1540 |
Idle watchers trigger events when no other events of the same or higher |
1541 |
priority are pending (prepare, check and other idle watchers do not |
1542 |
count). |
1543 |
|
1544 |
That is, as long as your process is busy handling sockets or timeouts |
1545 |
(or even signals, imagine) of the same or higher priority it will not be |
1546 |
triggered. But when your process is idle (or only lower-priority watchers |
1547 |
are pending), the idle watchers are being called once per event loop |
1548 |
iteration - until stopped, that is, or your process receives more events |
1549 |
and becomes busy again with higher priority stuff. |
1550 |
|
1551 |
The most noteworthy effect is that as long as any idle watchers are |
1552 |
active, the process will not block when waiting for new events. |
1553 |
|
1554 |
Apart from keeping your process non-blocking (which is a useful |
1555 |
effect on its own sometimes), idle watchers are a good place to do |
1556 |
"pseudo-background processing", or delay processing stuff to after the |
1557 |
event loop has handled all outstanding events. |
1558 |
|
1559 |
=head3 Watcher-Specific Functions and Data Members |
1560 |
|
1561 |
=over 4 |
1562 |
|
1563 |
=item ev_idle_init (ev_signal *, callback) |
1564 |
|
1565 |
Initialises and configures the idle watcher - it has no parameters of any |
1566 |
kind. There is a C<ev_idle_set> macro, but using it is utterly pointless, |
1567 |
believe me. |
1568 |
|
1569 |
=back |
1570 |
|
1571 |
Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the |
1572 |
callback, free it. Also, use no error checking, as usual. |
1573 |
|
1574 |
static void |
1575 |
idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents) |
1576 |
{ |
1577 |
free (w); |
1578 |
// now do something you wanted to do when the program has |
1579 |
// no longer asnything immediate to do. |
1580 |
} |
1581 |
|
1582 |
struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle)); |
1583 |
ev_idle_init (idle_watcher, idle_cb); |
1584 |
ev_idle_start (loop, idle_cb); |
1585 |
|
1586 |
|
1587 |
=head2 C<ev_prepare> and C<ev_check> - customise your event loop! |
1588 |
|
1589 |
Prepare and check watchers are usually (but not always) used in tandem: |
1590 |
prepare watchers get invoked before the process blocks and check watchers |
1591 |
afterwards. |
1592 |
|
1593 |
You I<must not> call C<ev_loop> or similar functions that enter |
1594 |
the current event loop from either C<ev_prepare> or C<ev_check> |
1595 |
watchers. Other loops than the current one are fine, however. The |
1596 |
rationale behind this is that you do not need to check for recursion in |
1597 |
those watchers, i.e. the sequence will always be C<ev_prepare>, blocking, |
1598 |
C<ev_check> so if you have one watcher of each kind they will always be |
1599 |
called in pairs bracketing the blocking call. |
1600 |
|
1601 |
Their main purpose is to integrate other event mechanisms into libev and |
1602 |
their use is somewhat advanced. This could be used, for example, to track |
1603 |
variable changes, implement your own watchers, integrate net-snmp or a |
1604 |
coroutine library and lots more. They are also occasionally useful if |
1605 |
you cache some data and want to flush it before blocking (for example, |
1606 |
in X programs you might want to do an C<XFlush ()> in an C<ev_prepare> |
1607 |
watcher). |
1608 |
|
1609 |
This is done by examining in each prepare call which file descriptors need |
1610 |
to be watched by the other library, registering C<ev_io> watchers for |
1611 |
them and starting an C<ev_timer> watcher for any timeouts (many libraries |
1612 |
provide just this functionality). Then, in the check watcher you check for |
1613 |
any events that occured (by checking the pending status of all watchers |
1614 |
and stopping them) and call back into the library. The I/O and timer |
1615 |
callbacks will never actually be called (but must be valid nevertheless, |
1616 |
because you never know, you know?). |
1617 |
|
1618 |
As another example, the Perl Coro module uses these hooks to integrate |
1619 |
coroutines into libev programs, by yielding to other active coroutines |
1620 |
during each prepare and only letting the process block if no coroutines |
1621 |
are ready to run (it's actually more complicated: it only runs coroutines |
1622 |
with priority higher than or equal to the event loop and one coroutine |
1623 |
of lower priority, but only once, using idle watchers to keep the event |
1624 |
loop from blocking if lower-priority coroutines are active, thus mapping |
1625 |
low-priority coroutines to idle/background tasks). |
1626 |
|
1627 |
It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>) |
1628 |
priority, to ensure that they are being run before any other watchers |
1629 |
after the poll. Also, C<ev_check> watchers (and C<ev_prepare> watchers, |
1630 |
too) should not activate ("feed") events into libev. While libev fully |
1631 |
supports this, they will be called before other C<ev_check> watchers |
1632 |
did their job. As C<ev_check> watchers are often used to embed other |
1633 |
(non-libev) event loops those other event loops might be in an unusable |
1634 |
state until their C<ev_check> watcher ran (always remind yourself to |
1635 |
coexist peacefully with others). |
1636 |
|
1637 |
=head3 Watcher-Specific Functions and Data Members |
1638 |
|
1639 |
=over 4 |
1640 |
|
1641 |
=item ev_prepare_init (ev_prepare *, callback) |
1642 |
|
1643 |
=item ev_check_init (ev_check *, callback) |
1644 |
|
1645 |
Initialises and configures the prepare or check watcher - they have no |
1646 |
parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set> |
1647 |
macros, but using them is utterly, utterly and completely pointless. |
1648 |
|
1649 |
=back |
1650 |
|
1651 |
There are a number of principal ways to embed other event loops or modules |
1652 |
into libev. Here are some ideas on how to include libadns into libev |
1653 |
(there is a Perl module named C<EV::ADNS> that does this, which you could |
1654 |
use for an actually working example. Another Perl module named C<EV::Glib> |
1655 |
embeds a Glib main context into libev, and finally, C<Glib::EV> embeds EV |
1656 |
into the Glib event loop). |
1657 |
|
1658 |
Method 1: Add IO watchers and a timeout watcher in a prepare handler, |
1659 |
and in a check watcher, destroy them and call into libadns. What follows |
1660 |
is pseudo-code only of course. This requires you to either use a low |
1661 |
priority for the check watcher or use C<ev_clear_pending> explicitly, as |
1662 |
the callbacks for the IO/timeout watchers might not have been called yet. |
1663 |
|
1664 |
static ev_io iow [nfd]; |
1665 |
static ev_timer tw; |
1666 |
|
1667 |
static void |
1668 |
io_cb (ev_loop *loop, ev_io *w, int revents) |
1669 |
{ |
1670 |
} |
1671 |
|
1672 |
// create io watchers for each fd and a timer before blocking |
1673 |
static void |
1674 |
adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents) |
1675 |
{ |
1676 |
int timeout = 3600000; |
1677 |
struct pollfd fds [nfd]; |
1678 |
// actual code will need to loop here and realloc etc. |
1679 |
adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ())); |
1680 |
|
1681 |
/* the callback is illegal, but won't be called as we stop during check */ |
1682 |
ev_timer_init (&tw, 0, timeout * 1e-3); |
1683 |
ev_timer_start (loop, &tw); |
1684 |
|
1685 |
// create one ev_io per pollfd |
1686 |
for (int i = 0; i < nfd; ++i) |
1687 |
{ |
1688 |
ev_io_init (iow + i, io_cb, fds [i].fd, |
1689 |
((fds [i].events & POLLIN ? EV_READ : 0) |
1690 |
| (fds [i].events & POLLOUT ? EV_WRITE : 0))); |
1691 |
|
1692 |
fds [i].revents = 0; |
1693 |
ev_io_start (loop, iow + i); |
1694 |
} |
1695 |
} |
1696 |
|
1697 |
// stop all watchers after blocking |
1698 |
static void |
1699 |
adns_check_cb (ev_loop *loop, ev_check *w, int revents) |
1700 |
{ |
1701 |
ev_timer_stop (loop, &tw); |
1702 |
|
1703 |
for (int i = 0; i < nfd; ++i) |
1704 |
{ |
1705 |
// set the relevant poll flags |
1706 |
// could also call adns_processreadable etc. here |
1707 |
struct pollfd *fd = fds + i; |
1708 |
int revents = ev_clear_pending (iow + i); |
1709 |
if (revents & EV_READ ) fd->revents |= fd->events & POLLIN; |
1710 |
if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT; |
1711 |
|
1712 |
// now stop the watcher |
1713 |
ev_io_stop (loop, iow + i); |
1714 |
} |
1715 |
|
1716 |
adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop)); |
1717 |
} |
1718 |
|
1719 |
Method 2: This would be just like method 1, but you run C<adns_afterpoll> |
1720 |
in the prepare watcher and would dispose of the check watcher. |
1721 |
|
1722 |
Method 3: If the module to be embedded supports explicit event |
1723 |
notification (adns does), you can also make use of the actual watcher |
1724 |
callbacks, and only destroy/create the watchers in the prepare watcher. |
1725 |
|
1726 |
static void |
1727 |
timer_cb (EV_P_ ev_timer *w, int revents) |
1728 |
{ |
1729 |
adns_state ads = (adns_state)w->data; |
1730 |
update_now (EV_A); |
1731 |
|
1732 |
adns_processtimeouts (ads, &tv_now); |
1733 |
} |
1734 |
|
1735 |
static void |
1736 |
io_cb (EV_P_ ev_io *w, int revents) |
1737 |
{ |
1738 |
adns_state ads = (adns_state)w->data; |
1739 |
update_now (EV_A); |
1740 |
|
1741 |
if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now); |
1742 |
if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now); |
1743 |
} |
1744 |
|
1745 |
// do not ever call adns_afterpoll |
1746 |
|
1747 |
Method 4: Do not use a prepare or check watcher because the module you |
1748 |
want to embed is too inflexible to support it. Instead, youc na override |
1749 |
their poll function. The drawback with this solution is that the main |
1750 |
loop is now no longer controllable by EV. The C<Glib::EV> module does |
1751 |
this. |
1752 |
|
1753 |
static gint |
1754 |
event_poll_func (GPollFD *fds, guint nfds, gint timeout) |
1755 |
{ |
1756 |
int got_events = 0; |
1757 |
|
1758 |
for (n = 0; n < nfds; ++n) |
1759 |
// create/start io watcher that sets the relevant bits in fds[n] and increment got_events |
1760 |
|
1761 |
if (timeout >= 0) |
1762 |
// create/start timer |
1763 |
|
1764 |
// poll |
1765 |
ev_loop (EV_A_ 0); |
1766 |
|
1767 |
// stop timer again |
1768 |
if (timeout >= 0) |
1769 |
ev_timer_stop (EV_A_ &to); |
1770 |
|
1771 |
// stop io watchers again - their callbacks should have set |
1772 |
for (n = 0; n < nfds; ++n) |
1773 |
ev_io_stop (EV_A_ iow [n]); |
1774 |
|
1775 |
return got_events; |
1776 |
} |
1777 |
|
1778 |
|
1779 |
=head2 C<ev_embed> - when one backend isn't enough... |
1780 |
|
1781 |
This is a rather advanced watcher type that lets you embed one event loop |
1782 |
into another (currently only C<ev_io> events are supported in the embedded |
1783 |
loop, other types of watchers might be handled in a delayed or incorrect |
1784 |
fashion and must not be used). |
1785 |
|
1786 |
There are primarily two reasons you would want that: work around bugs and |
1787 |
prioritise I/O. |
1788 |
|
1789 |
As an example for a bug workaround, the kqueue backend might only support |
1790 |
sockets on some platform, so it is unusable as generic backend, but you |
1791 |
still want to make use of it because you have many sockets and it scales |
1792 |
so nicely. In this case, you would create a kqueue-based loop and embed it |
1793 |
into your default loop (which might use e.g. poll). Overall operation will |
1794 |
be a bit slower because first libev has to poll and then call kevent, but |
1795 |
at least you can use both at what they are best. |
1796 |
|
1797 |
As for prioritising I/O: rarely you have the case where some fds have |
1798 |
to be watched and handled very quickly (with low latency), and even |
1799 |
priorities and idle watchers might have too much overhead. In this case |
1800 |
you would put all the high priority stuff in one loop and all the rest in |
1801 |
a second one, and embed the second one in the first. |
1802 |
|
1803 |
As long as the watcher is active, the callback will be invoked every time |
1804 |
there might be events pending in the embedded loop. The callback must then |
1805 |
call C<ev_embed_sweep (mainloop, watcher)> to make a single sweep and invoke |
1806 |
their callbacks (you could also start an idle watcher to give the embedded |
1807 |
loop strictly lower priority for example). You can also set the callback |
1808 |
to C<0>, in which case the embed watcher will automatically execute the |
1809 |
embedded loop sweep. |
1810 |
|
1811 |
As long as the watcher is started it will automatically handle events. The |
1812 |
callback will be invoked whenever some events have been handled. You can |
1813 |
set the callback to C<0> to avoid having to specify one if you are not |
1814 |
interested in that. |
1815 |
|
1816 |
Also, there have not currently been made special provisions for forking: |
1817 |
when you fork, you not only have to call C<ev_loop_fork> on both loops, |
1818 |
but you will also have to stop and restart any C<ev_embed> watchers |
1819 |
yourself. |
1820 |
|
1821 |
Unfortunately, not all backends are embeddable, only the ones returned by |
1822 |
C<ev_embeddable_backends> are, which, unfortunately, does not include any |
1823 |
portable one. |
1824 |
|
1825 |
So when you want to use this feature you will always have to be prepared |
1826 |
that you cannot get an embeddable loop. The recommended way to get around |
1827 |
this is to have a separate variables for your embeddable loop, try to |
1828 |
create it, and if that fails, use the normal loop for everything: |
1829 |
|
1830 |
struct ev_loop *loop_hi = ev_default_init (0); |
1831 |
struct ev_loop *loop_lo = 0; |
1832 |
struct ev_embed embed; |
1833 |
|
1834 |
// see if there is a chance of getting one that works |
1835 |
// (remember that a flags value of 0 means autodetection) |
1836 |
loop_lo = ev_embeddable_backends () & ev_recommended_backends () |
1837 |
? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ()) |
1838 |
: 0; |
1839 |
|
1840 |
// if we got one, then embed it, otherwise default to loop_hi |
1841 |
if (loop_lo) |
1842 |
{ |
1843 |
ev_embed_init (&embed, 0, loop_lo); |
1844 |
ev_embed_start (loop_hi, &embed); |
1845 |
} |
1846 |
else |
1847 |
loop_lo = loop_hi; |
1848 |
|
1849 |
=head3 Watcher-Specific Functions and Data Members |
1850 |
|
1851 |
=over 4 |
1852 |
|
1853 |
=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop) |
1854 |
|
1855 |
=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop) |
1856 |
|
1857 |
Configures the watcher to embed the given loop, which must be |
1858 |
embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be |
1859 |
invoked automatically, otherwise it is the responsibility of the callback |
1860 |
to invoke it (it will continue to be called until the sweep has been done, |
1861 |
if you do not want thta, you need to temporarily stop the embed watcher). |
1862 |
|
1863 |
=item ev_embed_sweep (loop, ev_embed *) |
1864 |
|
1865 |
Make a single, non-blocking sweep over the embedded loop. This works |
1866 |
similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most |
1867 |
apropriate way for embedded loops. |
1868 |
|
1869 |
=item struct ev_loop *other [read-only] |
1870 |
|
1871 |
The embedded event loop. |
1872 |
|
1873 |
=back |
1874 |
|
1875 |
|
1876 |
=head2 C<ev_fork> - the audacity to resume the event loop after a fork |
1877 |
|
1878 |
Fork watchers are called when a C<fork ()> was detected (usually because |
1879 |
whoever is a good citizen cared to tell libev about it by calling |
1880 |
C<ev_default_fork> or C<ev_loop_fork>). The invocation is done before the |
1881 |
event loop blocks next and before C<ev_check> watchers are being called, |
1882 |
and only in the child after the fork. If whoever good citizen calling |
1883 |
C<ev_default_fork> cheats and calls it in the wrong process, the fork |
1884 |
handlers will be invoked, too, of course. |
1885 |
|
1886 |
=head3 Watcher-Specific Functions and Data Members |
1887 |
|
1888 |
=over 4 |
1889 |
|
1890 |
=item ev_fork_init (ev_signal *, callback) |
1891 |
|
1892 |
Initialises and configures the fork watcher - it has no parameters of any |
1893 |
kind. There is a C<ev_fork_set> macro, but using it is utterly pointless, |
1894 |
believe me. |
1895 |
|
1896 |
=back |
1897 |
|
1898 |
|
1899 |
=head1 OTHER FUNCTIONS |
1900 |
|
1901 |
There are some other functions of possible interest. Described. Here. Now. |
1902 |
|
1903 |
=over 4 |
1904 |
|
1905 |
=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback) |
1906 |
|
1907 |
This function combines a simple timer and an I/O watcher, calls your |
1908 |
callback on whichever event happens first and automatically stop both |
1909 |
watchers. This is useful if you want to wait for a single event on an fd |
1910 |
or timeout without having to allocate/configure/start/stop/free one or |
1911 |
more watchers yourself. |
1912 |
|
1913 |
If C<fd> is less than 0, then no I/O watcher will be started and events |
1914 |
is being ignored. Otherwise, an C<ev_io> watcher for the given C<fd> and |
1915 |
C<events> set will be craeted and started. |
1916 |
|
1917 |
If C<timeout> is less than 0, then no timeout watcher will be |
1918 |
started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and |
1919 |
repeat = 0) will be started. While C<0> is a valid timeout, it is of |
1920 |
dubious value. |
1921 |
|
1922 |
The callback has the type C<void (*cb)(int revents, void *arg)> and gets |
1923 |
passed an C<revents> set like normal event callbacks (a combination of |
1924 |
C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg> |
1925 |
value passed to C<ev_once>: |
1926 |
|
1927 |
static void stdin_ready (int revents, void *arg) |
1928 |
{ |
1929 |
if (revents & EV_TIMEOUT) |
1930 |
/* doh, nothing entered */; |
1931 |
else if (revents & EV_READ) |
1932 |
/* stdin might have data for us, joy! */; |
1933 |
} |
1934 |
|
1935 |
ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0); |
1936 |
|
1937 |
=item ev_feed_event (ev_loop *, watcher *, int revents) |
1938 |
|
1939 |
Feeds the given event set into the event loop, as if the specified event |
1940 |
had happened for the specified watcher (which must be a pointer to an |
1941 |
initialised but not necessarily started event watcher). |
1942 |
|
1943 |
=item ev_feed_fd_event (ev_loop *, int fd, int revents) |
1944 |
|
1945 |
Feed an event on the given fd, as if a file descriptor backend detected |
1946 |
the given events it. |
1947 |
|
1948 |
=item ev_feed_signal_event (ev_loop *loop, int signum) |
1949 |
|
1950 |
Feed an event as if the given signal occured (C<loop> must be the default |
1951 |
loop!). |
1952 |
|
1953 |
=back |
1954 |
|
1955 |
|
1956 |
=head1 LIBEVENT EMULATION |
1957 |
|
1958 |
Libev offers a compatibility emulation layer for libevent. It cannot |
1959 |
emulate the internals of libevent, so here are some usage hints: |
1960 |
|
1961 |
=over 4 |
1962 |
|
1963 |
=item * Use it by including <event.h>, as usual. |
1964 |
|
1965 |
=item * The following members are fully supported: ev_base, ev_callback, |
1966 |
ev_arg, ev_fd, ev_res, ev_events. |
1967 |
|
1968 |
=item * Avoid using ev_flags and the EVLIST_*-macros, while it is |
1969 |
maintained by libev, it does not work exactly the same way as in libevent (consider |
1970 |
it a private API). |
1971 |
|
1972 |
=item * Priorities are not currently supported. Initialising priorities |
1973 |
will fail and all watchers will have the same priority, even though there |
1974 |
is an ev_pri field. |
1975 |
|
1976 |
=item * Other members are not supported. |
1977 |
|
1978 |
=item * The libev emulation is I<not> ABI compatible to libevent, you need |
1979 |
to use the libev header file and library. |
1980 |
|
1981 |
=back |
1982 |
|
1983 |
=head1 C++ SUPPORT |
1984 |
|
1985 |
Libev comes with some simplistic wrapper classes for C++ that mainly allow |
1986 |
you to use some convinience methods to start/stop watchers and also change |
1987 |
the callback model to a model using method callbacks on objects. |
1988 |
|
1989 |
To use it, |
1990 |
|
1991 |
#include <ev++.h> |
1992 |
|
1993 |
This automatically includes F<ev.h> and puts all of its definitions (many |
1994 |
of them macros) into the global namespace. All C++ specific things are |
1995 |
put into the C<ev> namespace. It should support all the same embedding |
1996 |
options as F<ev.h>, most notably C<EV_MULTIPLICITY>. |
1997 |
|
1998 |
Care has been taken to keep the overhead low. The only data member the C++ |
1999 |
classes add (compared to plain C-style watchers) is the event loop pointer |
2000 |
that the watcher is associated with (or no additional members at all if |
2001 |
you disable C<EV_MULTIPLICITY> when embedding libev). |
2002 |
|
2003 |
Currently, functions, and static and non-static member functions can be |
2004 |
used as callbacks. Other types should be easy to add as long as they only |
2005 |
need one additional pointer for context. If you need support for other |
2006 |
types of functors please contact the author (preferably after implementing |
2007 |
it). |
2008 |
|
2009 |
Here is a list of things available in the C<ev> namespace: |
2010 |
|
2011 |
=over 4 |
2012 |
|
2013 |
=item C<ev::READ>, C<ev::WRITE> etc. |
2014 |
|
2015 |
These are just enum values with the same values as the C<EV_READ> etc. |
2016 |
macros from F<ev.h>. |
2017 |
|
2018 |
=item C<ev::tstamp>, C<ev::now> |
2019 |
|
2020 |
Aliases to the same types/functions as with the C<ev_> prefix. |
2021 |
|
2022 |
=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc. |
2023 |
|
2024 |
For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of |
2025 |
the same name in the C<ev> namespace, with the exception of C<ev_signal> |
2026 |
which is called C<ev::sig> to avoid clashes with the C<signal> macro |
2027 |
defines by many implementations. |
2028 |
|
2029 |
All of those classes have these methods: |
2030 |
|
2031 |
=over 4 |
2032 |
|
2033 |
=item ev::TYPE::TYPE () |
2034 |
|
2035 |
=item ev::TYPE::TYPE (struct ev_loop *) |
2036 |
|
2037 |
=item ev::TYPE::~TYPE |
2038 |
|
2039 |
The constructor (optionally) takes an event loop to associate the watcher |
2040 |
with. If it is omitted, it will use C<EV_DEFAULT>. |
2041 |
|
2042 |
The constructor calls C<ev_init> for you, which means you have to call the |
2043 |
C<set> method before starting it. |
2044 |
|
2045 |
It will not set a callback, however: You have to call the templated C<set> |
2046 |
method to set a callback before you can start the watcher. |
2047 |
|
2048 |
(The reason why you have to use a method is a limitation in C++ which does |
2049 |
not allow explicit template arguments for constructors). |
2050 |
|
2051 |
The destructor automatically stops the watcher if it is active. |
2052 |
|
2053 |
=item w->set<class, &class::method> (object *) |
2054 |
|
2055 |
This method sets the callback method to call. The method has to have a |
2056 |
signature of C<void (*)(ev_TYPE &, int)>, it receives the watcher as |
2057 |
first argument and the C<revents> as second. The object must be given as |
2058 |
parameter and is stored in the C<data> member of the watcher. |
2059 |
|
2060 |
This method synthesizes efficient thunking code to call your method from |
2061 |
the C callback that libev requires. If your compiler can inline your |
2062 |
callback (i.e. it is visible to it at the place of the C<set> call and |
2063 |
your compiler is good :), then the method will be fully inlined into the |
2064 |
thunking function, making it as fast as a direct C callback. |
2065 |
|
2066 |
Example: simple class declaration and watcher initialisation |
2067 |
|
2068 |
struct myclass |
2069 |
{ |
2070 |
void io_cb (ev::io &w, int revents) { } |
2071 |
} |
2072 |
|
2073 |
myclass obj; |
2074 |
ev::io iow; |
2075 |
iow.set <myclass, &myclass::io_cb> (&obj); |
2076 |
|
2077 |
=item w->set<function> (void *data = 0) |
2078 |
|
2079 |
Also sets a callback, but uses a static method or plain function as |
2080 |
callback. The optional C<data> argument will be stored in the watcher's |
2081 |
C<data> member and is free for you to use. |
2082 |
|
2083 |
The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>. |
2084 |
|
2085 |
See the method-C<set> above for more details. |
2086 |
|
2087 |
Example: |
2088 |
|
2089 |
static void io_cb (ev::io &w, int revents) { } |
2090 |
iow.set <io_cb> (); |
2091 |
|
2092 |
=item w->set (struct ev_loop *) |
2093 |
|
2094 |
Associates a different C<struct ev_loop> with this watcher. You can only |
2095 |
do this when the watcher is inactive (and not pending either). |
2096 |
|
2097 |
=item w->set ([args]) |
2098 |
|
2099 |
Basically the same as C<ev_TYPE_set>, with the same args. Must be |
2100 |
called at least once. Unlike the C counterpart, an active watcher gets |
2101 |
automatically stopped and restarted when reconfiguring it with this |
2102 |
method. |
2103 |
|
2104 |
=item w->start () |
2105 |
|
2106 |
Starts the watcher. Note that there is no C<loop> argument, as the |
2107 |
constructor already stores the event loop. |
2108 |
|
2109 |
=item w->stop () |
2110 |
|
2111 |
Stops the watcher if it is active. Again, no C<loop> argument. |
2112 |
|
2113 |
=item w->again () (C<ev::timer>, C<ev::periodic> only) |
2114 |
|
2115 |
For C<ev::timer> and C<ev::periodic>, this invokes the corresponding |
2116 |
C<ev_TYPE_again> function. |
2117 |
|
2118 |
=item w->sweep () (C<ev::embed> only) |
2119 |
|
2120 |
Invokes C<ev_embed_sweep>. |
2121 |
|
2122 |
=item w->update () (C<ev::stat> only) |
2123 |
|
2124 |
Invokes C<ev_stat_stat>. |
2125 |
|
2126 |
=back |
2127 |
|
2128 |
=back |
2129 |
|
2130 |
Example: Define a class with an IO and idle watcher, start one of them in |
2131 |
the constructor. |
2132 |
|
2133 |
class myclass |
2134 |
{ |
2135 |
ev_io io; void io_cb (ev::io &w, int revents); |
2136 |
ev_idle idle void idle_cb (ev::idle &w, int revents); |
2137 |
|
2138 |
myclass (); |
2139 |
} |
2140 |
|
2141 |
myclass::myclass (int fd) |
2142 |
{ |
2143 |
io .set <myclass, &myclass::io_cb > (this); |
2144 |
idle.set <myclass, &myclass::idle_cb> (this); |
2145 |
|
2146 |
io.start (fd, ev::READ); |
2147 |
} |
2148 |
|
2149 |
|
2150 |
=head1 MACRO MAGIC |
2151 |
|
2152 |
Libev can be compiled with a variety of options, the most fundamantal |
2153 |
of which is C<EV_MULTIPLICITY>. This option determines whether (most) |
2154 |
functions and callbacks have an initial C<struct ev_loop *> argument. |
2155 |
|
2156 |
To make it easier to write programs that cope with either variant, the |
2157 |
following macros are defined: |
2158 |
|
2159 |
=over 4 |
2160 |
|
2161 |
=item C<EV_A>, C<EV_A_> |
2162 |
|
2163 |
This provides the loop I<argument> for functions, if one is required ("ev |
2164 |
loop argument"). The C<EV_A> form is used when this is the sole argument, |
2165 |
C<EV_A_> is used when other arguments are following. Example: |
2166 |
|
2167 |
ev_unref (EV_A); |
2168 |
ev_timer_add (EV_A_ watcher); |
2169 |
ev_loop (EV_A_ 0); |
2170 |
|
2171 |
It assumes the variable C<loop> of type C<struct ev_loop *> is in scope, |
2172 |
which is often provided by the following macro. |
2173 |
|
2174 |
=item C<EV_P>, C<EV_P_> |
2175 |
|
2176 |
This provides the loop I<parameter> for functions, if one is required ("ev |
2177 |
loop parameter"). The C<EV_P> form is used when this is the sole parameter, |
2178 |
C<EV_P_> is used when other parameters are following. Example: |
2179 |
|
2180 |
// this is how ev_unref is being declared |
2181 |
static void ev_unref (EV_P); |
2182 |
|
2183 |
// this is how you can declare your typical callback |
2184 |
static void cb (EV_P_ ev_timer *w, int revents) |
2185 |
|
2186 |
It declares a parameter C<loop> of type C<struct ev_loop *>, quite |
2187 |
suitable for use with C<EV_A>. |
2188 |
|
2189 |
=item C<EV_DEFAULT>, C<EV_DEFAULT_> |
2190 |
|
2191 |
Similar to the other two macros, this gives you the value of the default |
2192 |
loop, if multiple loops are supported ("ev loop default"). |
2193 |
|
2194 |
=back |
2195 |
|
2196 |
Example: Declare and initialise a check watcher, utilising the above |
2197 |
macros so it will work regardless of whether multiple loops are supported |
2198 |
or not. |
2199 |
|
2200 |
static void |
2201 |
check_cb (EV_P_ ev_timer *w, int revents) |
2202 |
{ |
2203 |
ev_check_stop (EV_A_ w); |
2204 |
} |
2205 |
|
2206 |
ev_check check; |
2207 |
ev_check_init (&check, check_cb); |
2208 |
ev_check_start (EV_DEFAULT_ &check); |
2209 |
ev_loop (EV_DEFAULT_ 0); |
2210 |
|
2211 |
=head1 EMBEDDING |
2212 |
|
2213 |
Libev can (and often is) directly embedded into host |
2214 |
applications. Examples of applications that embed it include the Deliantra |
2215 |
Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe) |
2216 |
and rxvt-unicode. |
2217 |
|
2218 |
The goal is to enable you to just copy the necessary files into your |
2219 |
source directory without having to change even a single line in them, so |
2220 |
you can easily upgrade by simply copying (or having a checked-out copy of |
2221 |
libev somewhere in your source tree). |
2222 |
|
2223 |
=head2 FILESETS |
2224 |
|
2225 |
Depending on what features you need you need to include one or more sets of files |
2226 |
in your app. |
2227 |
|
2228 |
=head3 CORE EVENT LOOP |
2229 |
|
2230 |
To include only the libev core (all the C<ev_*> functions), with manual |
2231 |
configuration (no autoconf): |
2232 |
|
2233 |
#define EV_STANDALONE 1 |
2234 |
#include "ev.c" |
2235 |
|
2236 |
This will automatically include F<ev.h>, too, and should be done in a |
2237 |
single C source file only to provide the function implementations. To use |
2238 |
it, do the same for F<ev.h> in all files wishing to use this API (best |
2239 |
done by writing a wrapper around F<ev.h> that you can include instead and |
2240 |
where you can put other configuration options): |
2241 |
|
2242 |
#define EV_STANDALONE 1 |
2243 |
#include "ev.h" |
2244 |
|
2245 |
Both header files and implementation files can be compiled with a C++ |
2246 |
compiler (at least, thats a stated goal, and breakage will be treated |
2247 |
as a bug). |
2248 |
|
2249 |
You need the following files in your source tree, or in a directory |
2250 |
in your include path (e.g. in libev/ when using -Ilibev): |
2251 |
|
2252 |
ev.h |
2253 |
ev.c |
2254 |
ev_vars.h |
2255 |
ev_wrap.h |
2256 |
|
2257 |
ev_win32.c required on win32 platforms only |
2258 |
|
2259 |
ev_select.c only when select backend is enabled (which is enabled by default) |
2260 |
ev_poll.c only when poll backend is enabled (disabled by default) |
2261 |
ev_epoll.c only when the epoll backend is enabled (disabled by default) |
2262 |
ev_kqueue.c only when the kqueue backend is enabled (disabled by default) |
2263 |
ev_port.c only when the solaris port backend is enabled (disabled by default) |
2264 |
|
2265 |
F<ev.c> includes the backend files directly when enabled, so you only need |
2266 |
to compile this single file. |
2267 |
|
2268 |
=head3 LIBEVENT COMPATIBILITY API |
2269 |
|
2270 |
To include the libevent compatibility API, also include: |
2271 |
|
2272 |
#include "event.c" |
2273 |
|
2274 |
in the file including F<ev.c>, and: |
2275 |
|
2276 |
#include "event.h" |
2277 |
|
2278 |
in the files that want to use the libevent API. This also includes F<ev.h>. |
2279 |
|
2280 |
You need the following additional files for this: |
2281 |
|
2282 |
event.h |
2283 |
event.c |
2284 |
|
2285 |
=head3 AUTOCONF SUPPORT |
2286 |
|
2287 |
Instead of using C<EV_STANDALONE=1> and providing your config in |
2288 |
whatever way you want, you can also C<m4_include([libev.m4])> in your |
2289 |
F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then |
2290 |
include F<config.h> and configure itself accordingly. |
2291 |
|
2292 |
For this of course you need the m4 file: |
2293 |
|
2294 |
libev.m4 |
2295 |
|
2296 |
=head2 PREPROCESSOR SYMBOLS/MACROS |
2297 |
|
2298 |
Libev can be configured via a variety of preprocessor symbols you have to define |
2299 |
before including any of its files. The default is not to build for multiplicity |
2300 |
and only include the select backend. |
2301 |
|
2302 |
=over 4 |
2303 |
|
2304 |
=item EV_STANDALONE |
2305 |
|
2306 |
Must always be C<1> if you do not use autoconf configuration, which |
2307 |
keeps libev from including F<config.h>, and it also defines dummy |
2308 |
implementations for some libevent functions (such as logging, which is not |
2309 |
supported). It will also not define any of the structs usually found in |
2310 |
F<event.h> that are not directly supported by the libev core alone. |
2311 |
|
2312 |
=item EV_USE_MONOTONIC |
2313 |
|
2314 |
If defined to be C<1>, libev will try to detect the availability of the |
2315 |
monotonic clock option at both compiletime and runtime. Otherwise no use |
2316 |
of the monotonic clock option will be attempted. If you enable this, you |
2317 |
usually have to link against librt or something similar. Enabling it when |
2318 |
the functionality isn't available is safe, though, although you have |
2319 |
to make sure you link against any libraries where the C<clock_gettime> |
2320 |
function is hiding in (often F<-lrt>). |
2321 |
|
2322 |
=item EV_USE_REALTIME |
2323 |
|
2324 |
If defined to be C<1>, libev will try to detect the availability of the |
2325 |
realtime clock option at compiletime (and assume its availability at |
2326 |
runtime if successful). Otherwise no use of the realtime clock option will |
2327 |
be attempted. This effectively replaces C<gettimeofday> by C<clock_get |
2328 |
(CLOCK_REALTIME, ...)> and will not normally affect correctness. See the |
2329 |
note about libraries in the description of C<EV_USE_MONOTONIC>, though. |
2330 |
|
2331 |
=item EV_USE_NANOSLEEP |
2332 |
|
2333 |
If defined to be C<1>, libev will assume that C<nanosleep ()> is available |
2334 |
and will use it for delays. Otherwise it will use C<select ()>. |
2335 |
|
2336 |
=item EV_USE_SELECT |
2337 |
|
2338 |
If undefined or defined to be C<1>, libev will compile in support for the |
2339 |
C<select>(2) backend. No attempt at autodetection will be done: if no |
2340 |
other method takes over, select will be it. Otherwise the select backend |
2341 |
will not be compiled in. |
2342 |
|
2343 |
=item EV_SELECT_USE_FD_SET |
2344 |
|
2345 |
If defined to C<1>, then the select backend will use the system C<fd_set> |
2346 |
structure. This is useful if libev doesn't compile due to a missing |
2347 |
C<NFDBITS> or C<fd_mask> definition or it misguesses the bitset layout on |
2348 |
exotic systems. This usually limits the range of file descriptors to some |
2349 |
low limit such as 1024 or might have other limitations (winsocket only |
2350 |
allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation, might |
2351 |
influence the size of the C<fd_set> used. |
2352 |
|
2353 |
=item EV_SELECT_IS_WINSOCKET |
2354 |
|
2355 |
When defined to C<1>, the select backend will assume that |
2356 |
select/socket/connect etc. don't understand file descriptors but |
2357 |
wants osf handles on win32 (this is the case when the select to |
2358 |
be used is the winsock select). This means that it will call |
2359 |
C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise, |
2360 |
it is assumed that all these functions actually work on fds, even |
2361 |
on win32. Should not be defined on non-win32 platforms. |
2362 |
|
2363 |
=item EV_USE_POLL |
2364 |
|
2365 |
If defined to be C<1>, libev will compile in support for the C<poll>(2) |
2366 |
backend. Otherwise it will be enabled on non-win32 platforms. It |
2367 |
takes precedence over select. |
2368 |
|
2369 |
=item EV_USE_EPOLL |
2370 |
|
2371 |
If defined to be C<1>, libev will compile in support for the Linux |
2372 |
C<epoll>(7) backend. Its availability will be detected at runtime, |
2373 |
otherwise another method will be used as fallback. This is the |
2374 |
preferred backend for GNU/Linux systems. |
2375 |
|
2376 |
=item EV_USE_KQUEUE |
2377 |
|
2378 |
If defined to be C<1>, libev will compile in support for the BSD style |
2379 |
C<kqueue>(2) backend. Its actual availability will be detected at runtime, |
2380 |
otherwise another method will be used as fallback. This is the preferred |
2381 |
backend for BSD and BSD-like systems, although on most BSDs kqueue only |
2382 |
supports some types of fds correctly (the only platform we found that |
2383 |
supports ptys for example was NetBSD), so kqueue might be compiled in, but |
2384 |
not be used unless explicitly requested. The best way to use it is to find |
2385 |
out whether kqueue supports your type of fd properly and use an embedded |
2386 |
kqueue loop. |
2387 |
|
2388 |
=item EV_USE_PORT |
2389 |
|
2390 |
If defined to be C<1>, libev will compile in support for the Solaris |
2391 |
10 port style backend. Its availability will be detected at runtime, |
2392 |
otherwise another method will be used as fallback. This is the preferred |
2393 |
backend for Solaris 10 systems. |
2394 |
|
2395 |
=item EV_USE_DEVPOLL |
2396 |
|
2397 |
reserved for future expansion, works like the USE symbols above. |
2398 |
|
2399 |
=item EV_USE_INOTIFY |
2400 |
|
2401 |
If defined to be C<1>, libev will compile in support for the Linux inotify |
2402 |
interface to speed up C<ev_stat> watchers. Its actual availability will |
2403 |
be detected at runtime. |
2404 |
|
2405 |
=item EV_H |
2406 |
|
2407 |
The name of the F<ev.h> header file used to include it. The default if |
2408 |
undefined is C<< <ev.h> >> in F<event.h> and C<"ev.h"> in F<ev.c>. This |
2409 |
can be used to virtually rename the F<ev.h> header file in case of conflicts. |
2410 |
|
2411 |
=item EV_CONFIG_H |
2412 |
|
2413 |
If C<EV_STANDALONE> isn't C<1>, this variable can be used to override |
2414 |
F<ev.c>'s idea of where to find the F<config.h> file, similarly to |
2415 |
C<EV_H>, above. |
2416 |
|
2417 |
=item EV_EVENT_H |
2418 |
|
2419 |
Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea |
2420 |
of how the F<event.h> header can be found. |
2421 |
|
2422 |
=item EV_PROTOTYPES |
2423 |
|
2424 |
If defined to be C<0>, then F<ev.h> will not define any function |
2425 |
prototypes, but still define all the structs and other symbols. This is |
2426 |
occasionally useful if you want to provide your own wrapper functions |
2427 |
around libev functions. |
2428 |
|
2429 |
=item EV_MULTIPLICITY |
2430 |
|
2431 |
If undefined or defined to C<1>, then all event-loop-specific functions |
2432 |
will have the C<struct ev_loop *> as first argument, and you can create |
2433 |
additional independent event loops. Otherwise there will be no support |
2434 |
for multiple event loops and there is no first event loop pointer |
2435 |
argument. Instead, all functions act on the single default loop. |
2436 |
|
2437 |
=item EV_MINPRI |
2438 |
|
2439 |
=item EV_MAXPRI |
2440 |
|
2441 |
The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to |
2442 |
C<EV_MAXPRI>, but otherwise there are no non-obvious limitations. You can |
2443 |
provide for more priorities by overriding those symbols (usually defined |
2444 |
to be C<-2> and C<2>, respectively). |
2445 |
|
2446 |
When doing priority-based operations, libev usually has to linearly search |
2447 |
all the priorities, so having many of them (hundreds) uses a lot of space |
2448 |
and time, so using the defaults of five priorities (-2 .. +2) is usually |
2449 |
fine. |
2450 |
|
2451 |
If your embedding app does not need any priorities, defining these both to |
2452 |
C<0> will save some memory and cpu. |
2453 |
|
2454 |
=item EV_PERIODIC_ENABLE |
2455 |
|
2456 |
If undefined or defined to be C<1>, then periodic timers are supported. If |
2457 |
defined to be C<0>, then they are not. Disabling them saves a few kB of |
2458 |
code. |
2459 |
|
2460 |
=item EV_IDLE_ENABLE |
2461 |
|
2462 |
If undefined or defined to be C<1>, then idle watchers are supported. If |
2463 |
defined to be C<0>, then they are not. Disabling them saves a few kB of |
2464 |
code. |
2465 |
|
2466 |
=item EV_EMBED_ENABLE |
2467 |
|
2468 |
If undefined or defined to be C<1>, then embed watchers are supported. If |
2469 |
defined to be C<0>, then they are not. |
2470 |
|
2471 |
=item EV_STAT_ENABLE |
2472 |
|
2473 |
If undefined or defined to be C<1>, then stat watchers are supported. If |
2474 |
defined to be C<0>, then they are not. |
2475 |
|
2476 |
=item EV_FORK_ENABLE |
2477 |
|
2478 |
If undefined or defined to be C<1>, then fork watchers are supported. If |
2479 |
defined to be C<0>, then they are not. |
2480 |
|
2481 |
=item EV_MINIMAL |
2482 |
|
2483 |
If you need to shave off some kilobytes of code at the expense of some |
2484 |
speed, define this symbol to C<1>. Currently only used for gcc to override |
2485 |
some inlining decisions, saves roughly 30% codesize of amd64. |
2486 |
|
2487 |
=item EV_PID_HASHSIZE |
2488 |
|
2489 |
C<ev_child> watchers use a small hash table to distribute workload by |
2490 |
pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more |
2491 |
than enough. If you need to manage thousands of children you might want to |
2492 |
increase this value (I<must> be a power of two). |
2493 |
|
2494 |
=item EV_INOTIFY_HASHSIZE |
2495 |
|
2496 |
C<ev_staz> watchers use a small hash table to distribute workload by |
2497 |
inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>), |
2498 |
usually more than enough. If you need to manage thousands of C<ev_stat> |
2499 |
watchers you might want to increase this value (I<must> be a power of |
2500 |
two). |
2501 |
|
2502 |
=item EV_COMMON |
2503 |
|
2504 |
By default, all watchers have a C<void *data> member. By redefining |
2505 |
this macro to a something else you can include more and other types of |
2506 |
members. You have to define it each time you include one of the files, |
2507 |
though, and it must be identical each time. |
2508 |
|
2509 |
For example, the perl EV module uses something like this: |
2510 |
|
2511 |
#define EV_COMMON \ |
2512 |
SV *self; /* contains this struct */ \ |
2513 |
SV *cb_sv, *fh /* note no trailing ";" */ |
2514 |
|
2515 |
=item EV_CB_DECLARE (type) |
2516 |
|
2517 |
=item EV_CB_INVOKE (watcher, revents) |
2518 |
|
2519 |
=item ev_set_cb (ev, cb) |
2520 |
|
2521 |
Can be used to change the callback member declaration in each watcher, |
2522 |
and the way callbacks are invoked and set. Must expand to a struct member |
2523 |
definition and a statement, respectively. See the F<ev.h> header file for |
2524 |
their default definitions. One possible use for overriding these is to |
2525 |
avoid the C<struct ev_loop *> as first argument in all cases, or to use |
2526 |
method calls instead of plain function calls in C++. |
2527 |
|
2528 |
=head2 EXPORTED API SYMBOLS |
2529 |
|
2530 |
If you need to re-export the API (e.g. via a dll) and you need a list of |
2531 |
exported symbols, you can use the provided F<Symbol.*> files which list |
2532 |
all public symbols, one per line: |
2533 |
|
2534 |
Symbols.ev for libev proper |
2535 |
Symbols.event for the libevent emulation |
2536 |
|
2537 |
This can also be used to rename all public symbols to avoid clashes with |
2538 |
multiple versions of libev linked together (which is obviously bad in |
2539 |
itself, but sometimes it is inconvinient to avoid this). |
2540 |
|
2541 |
A sed command like this will create wrapper C<#define>'s that you need to |
2542 |
include before including F<ev.h>: |
2543 |
|
2544 |
<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h |
2545 |
|
2546 |
This would create a file F<wrap.h> which essentially looks like this: |
2547 |
|
2548 |
#define ev_backend myprefix_ev_backend |
2549 |
#define ev_check_start myprefix_ev_check_start |
2550 |
#define ev_check_stop myprefix_ev_check_stop |
2551 |
... |
2552 |
|
2553 |
=head2 EXAMPLES |
2554 |
|
2555 |
For a real-world example of a program the includes libev |
2556 |
verbatim, you can have a look at the EV perl module |
2557 |
(L<http://software.schmorp.de/pkg/EV.html>). It has the libev files in |
2558 |
the F<libev/> subdirectory and includes them in the F<EV/EVAPI.h> (public |
2559 |
interface) and F<EV.xs> (implementation) files. Only the F<EV.xs> file |
2560 |
will be compiled. It is pretty complex because it provides its own header |
2561 |
file. |
2562 |
|
2563 |
The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file |
2564 |
that everybody includes and which overrides some configure choices: |
2565 |
|
2566 |
#define EV_MINIMAL 1 |
2567 |
#define EV_USE_POLL 0 |
2568 |
#define EV_MULTIPLICITY 0 |
2569 |
#define EV_PERIODIC_ENABLE 0 |
2570 |
#define EV_STAT_ENABLE 0 |
2571 |
#define EV_FORK_ENABLE 0 |
2572 |
#define EV_CONFIG_H <config.h> |
2573 |
#define EV_MINPRI 0 |
2574 |
#define EV_MAXPRI 0 |
2575 |
|
2576 |
#include "ev++.h" |
2577 |
|
2578 |
And a F<ev_cpp.C> implementation file that contains libev proper and is compiled: |
2579 |
|
2580 |
#include "ev_cpp.h" |
2581 |
#include "ev.c" |
2582 |
|
2583 |
|
2584 |
=head1 COMPLEXITIES |
2585 |
|
2586 |
In this section the complexities of (many of) the algorithms used inside |
2587 |
libev will be explained. For complexity discussions about backends see the |
2588 |
documentation for C<ev_default_init>. |
2589 |
|
2590 |
All of the following are about amortised time: If an array needs to be |
2591 |
extended, libev needs to realloc and move the whole array, but this |
2592 |
happens asymptotically never with higher number of elements, so O(1) might |
2593 |
mean it might do a lengthy realloc operation in rare cases, but on average |
2594 |
it is much faster and asymptotically approaches constant time. |
2595 |
|
2596 |
=over 4 |
2597 |
|
2598 |
=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers) |
2599 |
|
2600 |
This means that, when you have a watcher that triggers in one hour and |
2601 |
there are 100 watchers that would trigger before that then inserting will |
2602 |
have to skip those 100 watchers. |
2603 |
|
2604 |
=item Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers) |
2605 |
|
2606 |
That means that for changing a timer costs less than removing/adding them |
2607 |
as only the relative motion in the event queue has to be paid for. |
2608 |
|
2609 |
=item Starting io/check/prepare/idle/signal/child watchers: O(1) |
2610 |
|
2611 |
These just add the watcher into an array or at the head of a list. |
2612 |
=item Stopping check/prepare/idle watchers: O(1) |
2613 |
|
2614 |
=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE)) |
2615 |
|
2616 |
These watchers are stored in lists then need to be walked to find the |
2617 |
correct watcher to remove. The lists are usually short (you don't usually |
2618 |
have many watchers waiting for the same fd or signal). |
2619 |
|
2620 |
=item Finding the next timer per loop iteration: O(1) |
2621 |
|
2622 |
=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd) |
2623 |
|
2624 |
A change means an I/O watcher gets started or stopped, which requires |
2625 |
libev to recalculate its status (and possibly tell the kernel). |
2626 |
|
2627 |
=item Activating one watcher: O(1) |
2628 |
|
2629 |
=item Priority handling: O(number_of_priorities) |
2630 |
|
2631 |
Priorities are implemented by allocating some space for each |
2632 |
priority. When doing priority-based operations, libev usually has to |
2633 |
linearly search all the priorities. |
2634 |
|
2635 |
=back |
2636 |
|
2637 |
|
2638 |
=head1 AUTHOR |
2639 |
|
2640 |
Marc Lehmann <libev@schmorp.de>. |
2641 |
|