ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/libev/ev.html
(Generate patch)

Comparing libev/ev.html (file contents):
Revision 1.29 by root, Thu Nov 22 12:28:27 2007 UTC vs.
Revision 1.35 by root, Fri Nov 23 16:17:12 2007 UTC

… …
4<head> 4<head>
5 <title>libev</title> 5 <title>libev</title>
6 <meta name="description" content="Pod documentation for libev" /> 6 <meta name="description" content="Pod documentation for libev" />
7 <meta name="inputfile" content="&lt;standard input&gt;" /> 7 <meta name="inputfile" content="&lt;standard input&gt;" />
8 <meta name="outputfile" content="&lt;standard output&gt;" /> 8 <meta name="outputfile" content="&lt;standard output&gt;" />
9 <meta name="created" content="Thu Nov 22 13:26:17 2007" /> 9 <meta name="created" content="Fri Nov 23 17:17:04 2007" />
10 <meta name="generator" content="Pod::Xhtml 1.57" /> 10 <meta name="generator" content="Pod::Xhtml 1.57" />
11<link rel="stylesheet" href="http://res.tst.eu/pod.css"/></head> 11<link rel="stylesheet" href="http://res.tst.eu/pod.css"/></head>
12<body> 12<body>
13<div class="pod"> 13<div class="pod">
14<!-- INDEX START --> 14<!-- INDEX START -->
… …
96<div id="TIME_REPRESENTATION_CONTENT"> 96<div id="TIME_REPRESENTATION_CONTENT">
97<p>Libev represents time as a single floating point number, representing the 97<p>Libev represents time as a single floating point number, representing the
98(fractional) number of seconds since the (POSIX) epoch (somewhere near 98(fractional) number of seconds since the (POSIX) epoch (somewhere near
99the beginning of 1970, details are complicated, don't ask). This type is 99the beginning of 1970, details are complicated, don't ask). This type is
100called <code>ev_tstamp</code>, which is what you should use too. It usually aliases 100called <code>ev_tstamp</code>, which is what you should use too. It usually aliases
101to the double type in C.</p> 101to the <code>double</code> type in C, and when you need to do any calculations on
102it, you should treat it as such.</p>
103
104
105
106
102 107
103</div> 108</div>
104<h1 id="GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p> 109<h1 id="GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p>
105<div id="GLOBAL_FUNCTIONS_CONTENT"> 110<div id="GLOBAL_FUNCTIONS_CONTENT">
106<p>These functions can be called anytime, even before initialising the 111<p>These functions can be called anytime, even before initialising the
… …
122version of the library your program was compiled against.</p> 127version of the library your program was compiled against.</p>
123 <p>Usually, it's a good idea to terminate if the major versions mismatch, 128 <p>Usually, it's a good idea to terminate if the major versions mismatch,
124as this indicates an incompatible change. Minor versions are usually 129as this indicates an incompatible change. Minor versions are usually
125compatible to older versions, so a larger minor version alone is usually 130compatible to older versions, so a larger minor version alone is usually
126not a problem.</p> 131not a problem.</p>
132 <p>Example: make sure we haven't accidentally been linked against the wrong
133version:</p>
134<pre> assert ((&quot;libev version mismatch&quot;,
135 ev_version_major () == EV_VERSION_MAJOR
136 &amp;&amp; ev_version_minor () &gt;= EV_VERSION_MINOR));
137
138</pre>
139 </dd>
140 <dt>unsigned int ev_supported_backends ()</dt>
141 <dd>
142 <p>Return the set of all backends (i.e. their corresponding <code>EV_BACKEND_*</code>
143value) compiled into this binary of libev (independent of their
144availability on the system you are running on). See <code>ev_default_loop</code> for
145a description of the set values.</p>
146 <p>Example: make sure we have the epoll method, because yeah this is cool and
147a must have and can we have a torrent of it please!!!11</p>
148<pre> assert ((&quot;sorry, no epoll, no sex&quot;,
149 ev_supported_backends () &amp; EVBACKEND_EPOLL));
150
151</pre>
152 </dd>
153 <dt>unsigned int ev_recommended_backends ()</dt>
154 <dd>
155 <p>Return the set of all backends compiled into this binary of libev and also
156recommended for this platform. This set is often smaller than the one
157returned by <code>ev_supported_backends</code>, as for example kqueue is broken on
158most BSDs and will not be autodetected unless you explicitly request it
159(assuming you know what you are doing). This is the set of backends that
160libev will probe for if you specify no backends explicitly.</p>
127 </dd> 161 </dd>
128 <dt>ev_set_allocator (void *(*cb)(void *ptr, long size))</dt> 162 <dt>ev_set_allocator (void *(*cb)(void *ptr, long size))</dt>
129 <dd> 163 <dd>
130 <p>Sets the allocation function to use (the prototype is similar to the 164 <p>Sets the allocation function to use (the prototype is similar to the
131realloc C function, the semantics are identical). It is used to allocate 165realloc C function, the semantics are identical). It is used to allocate
… …
133needs to be allocated, the library might abort or take some potentially 167needs to be allocated, the library might abort or take some potentially
134destructive action. The default is your system realloc function.</p> 168destructive action. The default is your system realloc function.</p>
135 <p>You could override this function in high-availability programs to, say, 169 <p>You could override this function in high-availability programs to, say,
136free some memory if it cannot allocate memory, to use a special allocator, 170free some memory if it cannot allocate memory, to use a special allocator,
137or even to sleep a while and retry until some memory is available.</p> 171or even to sleep a while and retry until some memory is available.</p>
172 <p>Example: replace the libev allocator with one that waits a bit and then
173retries: better than mine).</p>
174<pre> static void *
175 persistent_realloc (void *ptr, long size)
176 {
177 for (;;)
178 {
179 void *newptr = realloc (ptr, size);
180
181 if (newptr)
182 return newptr;
183
184 sleep (60);
185 }
186 }
187
188 ...
189 ev_set_allocator (persistent_realloc);
190
191</pre>
138 </dd> 192 </dd>
139 <dt>ev_set_syserr_cb (void (*cb)(const char *msg));</dt> 193 <dt>ev_set_syserr_cb (void (*cb)(const char *msg));</dt>
140 <dd> 194 <dd>
141 <p>Set the callback function to call on a retryable syscall error (such 195 <p>Set the callback function to call on a retryable syscall error (such
142as failed select, poll, epoll_wait). The message is a printable string 196as failed select, poll, epoll_wait). The message is a printable string
143indicating the system call or subsystem causing the problem. If this 197indicating the system call or subsystem causing the problem. If this
144callback is set, then libev will expect it to remedy the sitution, no 198callback is set, then libev will expect it to remedy the sitution, no
145matter what, when it returns. That is, libev will generally retry the 199matter what, when it returns. That is, libev will generally retry the
146requested operation, or, if the condition doesn't go away, do bad stuff 200requested operation, or, if the condition doesn't go away, do bad stuff
147(such as abort).</p> 201(such as abort).</p>
202 <p>Example: do the same thing as libev does internally:</p>
203<pre> static void
204 fatal_error (const char *msg)
205 {
206 perror (msg);
207 abort ();
208 }
209
210 ...
211 ev_set_syserr_cb (fatal_error);
212
213</pre>
148 </dd> 214 </dd>
149</dl> 215</dl>
150 216
151</div> 217</div>
152<h1 id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</h1><p><a href="#TOP" class="toplink">Top</a></p> 218<h1 id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</h1><p><a href="#TOP" class="toplink">Top</a></p>
… …
164 <dt>struct ev_loop *ev_default_loop (unsigned int flags)</dt> 230 <dt>struct ev_loop *ev_default_loop (unsigned int flags)</dt>
165 <dd> 231 <dd>
166 <p>This will initialise the default event loop if it hasn't been initialised 232 <p>This will initialise the default event loop if it hasn't been initialised
167yet and return it. If the default loop could not be initialised, returns 233yet and return it. If the default loop could not be initialised, returns
168false. If it already was initialised it simply returns it (and ignores the 234false. If it already was initialised it simply returns it (and ignores the
169flags).</p> 235flags. If that is troubling you, check <code>ev_backend ()</code> afterwards).</p>
170 <p>If you don't know what event loop to use, use the one returned from this 236 <p>If you don't know what event loop to use, use the one returned from this
171function.</p> 237function.</p>
172 <p>The flags argument can be used to specify special behaviour or specific 238 <p>The flags argument can be used to specify special behaviour or specific
173backends to use, and is usually specified as 0 (or EVFLAG_AUTO).</p> 239backends to use, and is usually specified as <code>0</code> (or <code>EVFLAG_AUTO</code>).</p>
174 <p>It supports the following flags:</p> 240 <p>The following flags are supported:</p>
175 <p> 241 <p>
176 <dl> 242 <dl>
177 <dt><code>EVFLAG_AUTO</code></dt> 243 <dt><code>EVFLAG_AUTO</code></dt>
178 <dd> 244 <dd>
179 <p>The default flags value. Use this if you have no clue (it's the right 245 <p>The default flags value. Use this if you have no clue (it's the right
… …
186<code>LIBEV_FLAGS</code>. Otherwise (the default), this environment variable will 252<code>LIBEV_FLAGS</code>. Otherwise (the default), this environment variable will
187override the flags completely if it is found in the environment. This is 253override the flags completely if it is found in the environment. This is
188useful to try out specific backends to test their performance, or to work 254useful to try out specific backends to test their performance, or to work
189around bugs.</p> 255around bugs.</p>
190 </dd> 256 </dd>
191 <dt><code>EVMETHOD_SELECT</code> (value 1, portable select backend)</dt> 257 <dt><code>EVBACKEND_SELECT</code> (value 1, portable select backend)</dt>
192 <dd> 258 <dd>
193 <p>This is your standard select(2) backend. Not <i>completely</i> standard, as 259 <p>This is your standard select(2) backend. Not <i>completely</i> standard, as
194libev tries to roll its own fd_set with no limits on the number of fds, 260libev tries to roll its own fd_set with no limits on the number of fds,
195but if that fails, expect a fairly low limit on the number of fds when 261but if that fails, expect a fairly low limit on the number of fds when
196using this backend. It doesn't scale too well (O(highest_fd)), but its usually 262using this backend. It doesn't scale too well (O(highest_fd)), but its usually
197the fastest backend for a low number of fds.</p> 263the fastest backend for a low number of fds.</p>
198 </dd> 264 </dd>
199 <dt><code>EVMETHOD_POLL</code> (value 2, poll backend, available everywhere except on windows)</dt> 265 <dt><code>EVBACKEND_POLL</code> (value 2, poll backend, available everywhere except on windows)</dt>
200 <dd> 266 <dd>
201 <p>And this is your standard poll(2) backend. It's more complicated than 267 <p>And this is your standard poll(2) backend. It's more complicated than
202select, but handles sparse fds better and has no artificial limit on the 268select, but handles sparse fds better and has no artificial limit on the
203number of fds you can use (except it will slow down considerably with a 269number of fds you can use (except it will slow down considerably with a
204lot of inactive fds). It scales similarly to select, i.e. O(total_fds).</p> 270lot of inactive fds). It scales similarly to select, i.e. O(total_fds).</p>
205 </dd> 271 </dd>
206 <dt><code>EVMETHOD_EPOLL</code> (value 4, Linux)</dt> 272 <dt><code>EVBACKEND_EPOLL</code> (value 4, Linux)</dt>
207 <dd> 273 <dd>
208 <p>For few fds, this backend is a bit little slower than poll and select, 274 <p>For few fds, this backend is a bit little slower than poll and select,
209but it scales phenomenally better. While poll and select usually scale like 275but it scales phenomenally better. While poll and select usually scale like
210O(total_fds) where n is the total number of fds (or the highest fd), epoll scales 276O(total_fds) where n is the total number of fds (or the highest fd), epoll scales
211either O(1) or O(active_fds).</p> 277either O(1) or O(active_fds).</p>
212 <p>While stopping and starting an I/O watcher in the same iteration will 278 <p>While stopping and starting an I/O watcher in the same iteration will
213result in some caching, there is still a syscall per such incident 279result in some caching, there is still a syscall per such incident
214(because the fd could point to a different file description now), so its 280(because the fd could point to a different file description now), so its
215best to avoid that. Also, dup()ed file descriptors might not work very 281best to avoid that. Also, dup()ed file descriptors might not work very
216well if you register events for both fds.</p> 282well if you register events for both fds.</p>
283 <p>Please note that epoll sometimes generates spurious notifications, so you
284need to use non-blocking I/O or other means to avoid blocking when no data
285(or space) is available.</p>
217 </dd> 286 </dd>
218 <dt><code>EVMETHOD_KQUEUE</code> (value 8, most BSD clones)</dt> 287 <dt><code>EVBACKEND_KQUEUE</code> (value 8, most BSD clones)</dt>
219 <dd> 288 <dd>
220 <p>Kqueue deserves special mention, as at the time of this writing, it 289 <p>Kqueue deserves special mention, as at the time of this writing, it
221was broken on all BSDs except NetBSD (usually it doesn't work with 290was broken on all BSDs except NetBSD (usually it doesn't work with
222anything but sockets and pipes, except on Darwin, where of course its 291anything but sockets and pipes, except on Darwin, where of course its
223completely useless). For this reason its not being &quot;autodetected&quot; unless 292completely useless). For this reason its not being &quot;autodetected&quot;
224you explicitly specify the flags (i.e. you don't use EVFLAG_AUTO).</p> 293unless you explicitly specify it explicitly in the flags (i.e. using
294<code>EVBACKEND_KQUEUE</code>).</p>
225 <p>It scales in the same way as the epoll backend, but the interface to the 295 <p>It scales in the same way as the epoll backend, but the interface to the
226kernel is more efficient (which says nothing about its actual speed, of 296kernel is more efficient (which says nothing about its actual speed, of
227course). While starting and stopping an I/O watcher does not cause an 297course). While starting and stopping an I/O watcher does not cause an
228extra syscall as with epoll, it still adds up to four event changes per 298extra syscall as with epoll, it still adds up to four event changes per
229incident, so its best to avoid that.</p> 299incident, so its best to avoid that.</p>
230 </dd> 300 </dd>
231 <dt><code>EVMETHOD_DEVPOLL</code> (value 16, Solaris 8)</dt> 301 <dt><code>EVBACKEND_DEVPOLL</code> (value 16, Solaris 8)</dt>
232 <dd> 302 <dd>
233 <p>This is not implemented yet (and might never be).</p> 303 <p>This is not implemented yet (and might never be).</p>
234 </dd> 304 </dd>
235 <dt><code>EVMETHOD_PORT</code> (value 32, Solaris 10)</dt> 305 <dt><code>EVBACKEND_PORT</code> (value 32, Solaris 10)</dt>
236 <dd> 306 <dd>
237 <p>This uses the Solaris 10 port mechanism. As with everything on Solaris, 307 <p>This uses the Solaris 10 port mechanism. As with everything on Solaris,
238it's really slow, but it still scales very well (O(active_fds)).</p> 308it's really slow, but it still scales very well (O(active_fds)).</p>
309 <p>Please note that solaris ports can result in a lot of spurious
310notifications, so you need to use non-blocking I/O or other means to avoid
311blocking when no data (or space) is available.</p>
239 </dd> 312 </dd>
240 <dt><code>EVMETHOD_ALL</code></dt> 313 <dt><code>EVBACKEND_ALL</code></dt>
241 <dd> 314 <dd>
242 <p>Try all backends (even potentially broken ones). Since this is a mask, you 315 <p>Try all backends (even potentially broken ones that wouldn't be tried
243can do stuff like <code>EVMETHOD_ALL &amp; ~EVMETHOD_KQUEUE</code>.</p> 316with <code>EVFLAG_AUTO</code>). Since this is a mask, you can do stuff such as
317<code>EVBACKEND_ALL &amp; ~EVBACKEND_KQUEUE</code>.</p>
244 </dd> 318 </dd>
245 </dl> 319 </dl>
246 </p> 320 </p>
247 <p>If one or more of these are ored into the flags value, then only these 321 <p>If one or more of these are ored into the flags value, then only these
248backends will be tried (in the reverse order as given here). If none are 322backends will be tried (in the reverse order as given here). If none are
249specified, most compiled-in backend will be tried, usually in reverse 323specified, most compiled-in backend will be tried, usually in reverse
250order of their flag values :)</p> 324order of their flag values :)</p>
325 <p>The most typical usage is like this:</p>
326<pre> if (!ev_default_loop (0))
327 fatal (&quot;could not initialise libev, bad $LIBEV_FLAGS in environment?&quot;);
328
329</pre>
330 <p>Restrict libev to the select and poll backends, and do not allow
331environment settings to be taken into account:</p>
332<pre> ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
333
334</pre>
335 <p>Use whatever libev has to offer, but make sure that kqueue is used if
336available (warning, breaks stuff, best use only with your own private
337event loop and only if you know the OS supports your types of fds):</p>
338<pre> ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
339
340</pre>
251 </dd> 341 </dd>
252 <dt>struct ev_loop *ev_loop_new (unsigned int flags)</dt> 342 <dt>struct ev_loop *ev_loop_new (unsigned int flags)</dt>
253 <dd> 343 <dd>
254 <p>Similar to <code>ev_default_loop</code>, but always creates a new event loop that is 344 <p>Similar to <code>ev_default_loop</code>, but always creates a new event loop that is
255always distinct from the default loop. Unlike the default loop, it cannot 345always distinct from the default loop. Unlike the default loop, it cannot
256handle signal and child watchers, and attempts to do so will be greeted by 346handle signal and child watchers, and attempts to do so will be greeted by
257undefined behaviour (or a failed assertion if assertions are enabled).</p> 347undefined behaviour (or a failed assertion if assertions are enabled).</p>
348 <p>Example: try to create a event loop that uses epoll and nothing else.</p>
349<pre> struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
350 if (!epoller)
351 fatal (&quot;no epoll found here, maybe it hides under your chair&quot;);
352
353</pre>
258 </dd> 354 </dd>
259 <dt>ev_default_destroy ()</dt> 355 <dt>ev_default_destroy ()</dt>
260 <dd> 356 <dd>
261 <p>Destroys the default loop again (frees all memory and kernel state 357 <p>Destroys the default loop again (frees all memory and kernel state
262etc.). This stops all registered event watchers (by not touching them in 358etc.). This stops all registered event watchers (by not touching them in
… …
271 <dd> 367 <dd>
272 <p>This function reinitialises the kernel state for backends that have 368 <p>This function reinitialises the kernel state for backends that have
273one. Despite the name, you can call it anytime, but it makes most sense 369one. Despite the name, you can call it anytime, but it makes most sense
274after forking, in either the parent or child process (or both, but that 370after forking, in either the parent or child process (or both, but that
275again makes little sense).</p> 371again makes little sense).</p>
276 <p>You <i>must</i> call this function after forking if and only if you want to 372 <p>You <i>must</i> call this function in the child process after forking if and
277use the event library in both processes. If you just fork+exec, you don't 373only if you want to use the event library in both processes. If you just
278have to call it.</p> 374fork+exec, you don't have to call it.</p>
279 <p>The function itself is quite fast and it's usually not a problem to call 375 <p>The function itself is quite fast and it's usually not a problem to call
280it just in case after a fork. To make this easy, the function will fit in 376it just in case after a fork. To make this easy, the function will fit in
281quite nicely into a call to <code>pthread_atfork</code>:</p> 377quite nicely into a call to <code>pthread_atfork</code>:</p>
282<pre> pthread_atfork (0, 0, ev_default_fork); 378<pre> pthread_atfork (0, 0, ev_default_fork);
283 379
284</pre> 380</pre>
381 <p>At the moment, <code>EVBACKEND_SELECT</code> and <code>EVBACKEND_POLL</code> are safe to use
382without calling this function, so if you force one of those backends you
383do not need to care.</p>
285 </dd> 384 </dd>
286 <dt>ev_loop_fork (loop)</dt> 385 <dt>ev_loop_fork (loop)</dt>
287 <dd> 386 <dd>
288 <p>Like <code>ev_default_fork</code>, but acts on an event loop created by 387 <p>Like <code>ev_default_fork</code>, but acts on an event loop created by
289<code>ev_loop_new</code>. Yes, you have to call this on every allocated event loop 388<code>ev_loop_new</code>. Yes, you have to call this on every allocated event loop
290after fork, and how you do this is entirely your own problem.</p> 389after fork, and how you do this is entirely your own problem.</p>
291 </dd> 390 </dd>
292 <dt>unsigned int ev_method (loop)</dt> 391 <dt>unsigned int ev_backend (loop)</dt>
293 <dd> 392 <dd>
294 <p>Returns one of the <code>EVMETHOD_*</code> flags indicating the event backend in 393 <p>Returns one of the <code>EVBACKEND_*</code> flags indicating the event backend in
295use.</p> 394use.</p>
296 </dd> 395 </dd>
297 <dt>ev_tstamp ev_now (loop)</dt> 396 <dt>ev_tstamp ev_now (loop)</dt>
298 <dd> 397 <dd>
299 <p>Returns the current &quot;event loop time&quot;, which is the time the event loop 398 <p>Returns the current &quot;event loop time&quot;, which is the time the event loop
300got events and started processing them. This timestamp does not change 399received events and started processing them. This timestamp does not
301as long as callbacks are being processed, and this is also the base time 400change as long as callbacks are being processed, and this is also the base
302used for relative timers. You can treat it as the timestamp of the event 401time used for relative timers. You can treat it as the timestamp of the
303occuring (or more correctly, the mainloop finding out about it).</p> 402event occuring (or more correctly, libev finding out about it).</p>
304 </dd> 403 </dd>
305 <dt>ev_loop (loop, int flags)</dt> 404 <dt>ev_loop (loop, int flags)</dt>
306 <dd> 405 <dd>
307 <p>Finally, this is it, the event handler. This function usually is called 406 <p>Finally, this is it, the event handler. This function usually is called
308after you initialised all your watchers and you want to start handling 407after you initialised all your watchers and you want to start handling
309events.</p> 408events.</p>
310 <p>If the flags argument is specified as 0, it will not return until either 409 <p>If the flags argument is specified as <code>0</code>, it will not return until
311no event watchers are active anymore or <code>ev_unloop</code> was called.</p> 410either no event watchers are active anymore or <code>ev_unloop</code> was called.</p>
411 <p>Please note that an explicit <code>ev_unloop</code> is usually better than
412relying on all watchers to be stopped when deciding when a program has
413finished (especially in interactive programs), but having a program that
414automatically loops as long as it has to and no longer by virtue of
415relying on its watchers stopping correctly is a thing of beauty.</p>
312 <p>A flags value of <code>EVLOOP_NONBLOCK</code> will look for new events, will handle 416 <p>A flags value of <code>EVLOOP_NONBLOCK</code> will look for new events, will handle
313those events and any outstanding ones, but will not block your process in 417those events and any outstanding ones, but will not block your process in
314case there are no events and will return after one iteration of the loop.</p> 418case there are no events and will return after one iteration of the loop.</p>
315 <p>A flags value of <code>EVLOOP_ONESHOT</code> will look for new events (waiting if 419 <p>A flags value of <code>EVLOOP_ONESHOT</code> will look for new events (waiting if
316neccessary) and will handle those and any outstanding ones. It will block 420neccessary) and will handle those and any outstanding ones. It will block
317your process until at least one new event arrives, and will return after 421your process until at least one new event arrives, and will return after
318one iteration of the loop.</p> 422one iteration of the loop. This is useful if you are waiting for some
319 <p>This flags value could be used to implement alternative looping 423external event in conjunction with something not expressible using other
320constructs, but the <code>prepare</code> and <code>check</code> watchers provide a better and 424libev watchers. However, a pair of <code>ev_prepare</code>/<code>ev_check</code> watchers is
321more generic mechanism.</p> 425usually a better approach for this kind of thing.</p>
322 <p>Here are the gory details of what ev_loop does:</p> 426 <p>Here are the gory details of what <code>ev_loop</code> does:</p>
323<pre> 1. If there are no active watchers (reference count is zero), return. 427<pre> * If there are no active watchers (reference count is zero), return.
324 2. Queue and immediately call all prepare watchers. 428 - Queue prepare watchers and then call all outstanding watchers.
325 3. If we have been forked, recreate the kernel state. 429 - If we have been forked, recreate the kernel state.
326 4. Update the kernel state with all outstanding changes. 430 - Update the kernel state with all outstanding changes.
327 5. Update the &quot;event loop time&quot;. 431 - Update the &quot;event loop time&quot;.
328 6. Calculate for how long to block. 432 - Calculate for how long to block.
329 7. Block the process, waiting for events. 433 - Block the process, waiting for any events.
434 - Queue all outstanding I/O (fd) events.
330 8. Update the &quot;event loop time&quot; and do time jump handling. 435 - Update the &quot;event loop time&quot; and do time jump handling.
331 9. Queue all outstanding timers. 436 - Queue all outstanding timers.
332 10. Queue all outstanding periodics. 437 - Queue all outstanding periodics.
333 11. If no events are pending now, queue all idle watchers. 438 - If no events are pending now, queue all idle watchers.
334 12. Queue all check watchers. 439 - Queue all check watchers.
335 13. Call all queued watchers in reverse order (i.e. check watchers first). 440 - Call all queued watchers in reverse order (i.e. check watchers first).
441 Signals and child watchers are implemented as I/O watchers, and will
442 be handled here by queueing them when their watcher gets executed.
336 14. If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK 443 - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
337 was used, return, otherwise continue with step #1. 444 were used, return, otherwise continue with step *.
445
446</pre>
447 <p>Example: queue some jobs and then loop until no events are outsanding
448anymore.</p>
449<pre> ... queue jobs here, make sure they register event watchers as long
450 ... as they still have work to do (even an idle watcher will do..)
451 ev_loop (my_loop, 0);
452 ... jobs done. yeah!
338 453
339</pre> 454</pre>
340 </dd> 455 </dd>
341 <dt>ev_unloop (loop, how)</dt> 456 <dt>ev_unloop (loop, how)</dt>
342 <dd> 457 <dd>
… …
356example, libev itself uses this for its internal signal pipe: It is not 471example, libev itself uses this for its internal signal pipe: It is not
357visible to the libev user and should not keep <code>ev_loop</code> from exiting if 472visible to the libev user and should not keep <code>ev_loop</code> from exiting if
358no event watchers registered by it are active. It is also an excellent 473no event watchers registered by it are active. It is also an excellent
359way to do this for generic recurring timers or from within third-party 474way to do this for generic recurring timers or from within third-party
360libraries. Just remember to <i>unref after start</i> and <i>ref before stop</i>.</p> 475libraries. Just remember to <i>unref after start</i> and <i>ref before stop</i>.</p>
476 <p>Example: create a signal watcher, but keep it from keeping <code>ev_loop</code>
477running when nothing else is active.</p>
478<pre> struct dv_signal exitsig;
479 ev_signal_init (&amp;exitsig, sig_cb, SIGINT);
480 ev_signal_start (myloop, &amp;exitsig);
481 evf_unref (myloop);
482
483</pre>
484 <p>Example: for some weird reason, unregister the above signal handler again.</p>
485<pre> ev_ref (myloop);
486 ev_signal_stop (myloop, &amp;exitsig);
487
488</pre>
361 </dd> 489 </dd>
362</dl> 490</dl>
363 491
364</div> 492</div>
365<h1 id="ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</h1><p><a href="#TOP" class="toplink">Top</a></p> 493<h1 id="ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</h1><p><a href="#TOP" class="toplink">Top</a></p>
… …
397with a watcher-specific start function (<code>ev_&lt;type&gt;_start (loop, watcher 525with a watcher-specific start function (<code>ev_&lt;type&gt;_start (loop, watcher
398*)</code>), and you can stop watching for events at any time by calling the 526*)</code>), and you can stop watching for events at any time by calling the
399corresponding stop function (<code>ev_&lt;type&gt;_stop (loop, watcher *)</code>.</p> 527corresponding stop function (<code>ev_&lt;type&gt;_stop (loop, watcher *)</code>.</p>
400<p>As long as your watcher is active (has been started but not stopped) you 528<p>As long as your watcher is active (has been started but not stopped) you
401must not touch the values stored in it. Most specifically you must never 529must not touch the values stored in it. Most specifically you must never
402reinitialise it or call its set method.</p> 530reinitialise it or call its set macro.</p>
403<p>You can check whether an event is active by calling the <code>ev_is_active 531<p>You can check whether an event is active by calling the <code>ev_is_active
404(watcher *)</code> macro. To see whether an event is outstanding (but the 532(watcher *)</code> macro. To see whether an event is outstanding (but the
405callback for it has not been called yet) you can use the <code>ev_is_pending 533callback for it has not been called yet) you can use the <code>ev_is_pending
406(watcher *)</code> macro.</p> 534(watcher *)</code> macro.</p>
407<p>Each and every callback receives the event loop pointer as first, the 535<p>Each and every callback receives the event loop pointer as first, the
… …
500</div> 628</div>
501<h1 id="WATCHER_TYPES">WATCHER TYPES</h1><p><a href="#TOP" class="toplink">Top</a></p> 629<h1 id="WATCHER_TYPES">WATCHER TYPES</h1><p><a href="#TOP" class="toplink">Top</a></p>
502<div id="WATCHER_TYPES_CONTENT"> 630<div id="WATCHER_TYPES_CONTENT">
503<p>This section describes each watcher in detail, but will not repeat 631<p>This section describes each watcher in detail, but will not repeat
504information given in the last section.</p> 632information given in the last section.</p>
633
634
635
636
505 637
506</div> 638</div>
507<h2 id="code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable</h2> 639<h2 id="code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable</h2>
508<div id="code_ev_io_code_is_this_file_descrip-2"> 640<div id="code_ev_io_code_is_this_file_descrip-2">
509<p>I/O watchers check whether a file descriptor is readable or writable 641<p>I/O watchers check whether a file descriptor is readable or writable
… …
519(the linux epoll backend is a notable example) cannot handle dup'ed file 651(the linux epoll backend is a notable example) cannot handle dup'ed file
520descriptors correctly if you register interest in two or more fds pointing 652descriptors correctly if you register interest in two or more fds pointing
521to the same underlying file/socket etc. description (that is, they share 653to the same underlying file/socket etc. description (that is, they share
522the same underlying &quot;file open&quot;).</p> 654the same underlying &quot;file open&quot;).</p>
523<p>If you must do this, then force the use of a known-to-be-good backend 655<p>If you must do this, then force the use of a known-to-be-good backend
524(at the time of this writing, this includes only EVMETHOD_SELECT and 656(at the time of this writing, this includes only <code>EVBACKEND_SELECT</code> and
525EVMETHOD_POLL).</p> 657<code>EVBACKEND_POLL</code>).</p>
526<dl> 658<dl>
527 <dt>ev_io_init (ev_io *, callback, int fd, int events)</dt> 659 <dt>ev_io_init (ev_io *, callback, int fd, int events)</dt>
528 <dt>ev_io_set (ev_io *, int fd, int events)</dt> 660 <dt>ev_io_set (ev_io *, int fd, int events)</dt>
529 <dd> 661 <dd>
530 <p>Configures an <code>ev_io</code> watcher. The fd is the file descriptor to rceeive 662 <p>Configures an <code>ev_io</code> watcher. The fd is the file descriptor to rceeive
531events for and events is either <code>EV_READ</code>, <code>EV_WRITE</code> or <code>EV_READ | 663events for and events is either <code>EV_READ</code>, <code>EV_WRITE</code> or <code>EV_READ |
532EV_WRITE</code> to receive the given events.</p> 664EV_WRITE</code> to receive the given events.</p>
665 <p>Please note that most of the more scalable backend mechanisms (for example
666epoll and solaris ports) can result in spurious readyness notifications
667for file descriptors, so you practically need to use non-blocking I/O (and
668treat callback invocation as hint only), or retest separately with a safe
669interface before doing I/O (XLib can do this), or force the use of either
670<code>EVBACKEND_SELECT</code> or <code>EVBACKEND_POLL</code>, which don't suffer from this
671problem. Also note that it is quite easy to have your callback invoked
672when the readyness condition is no longer valid even when employing
673typical ways of handling events, so its a good idea to use non-blocking
674I/O unconditionally.</p>
533 </dd> 675 </dd>
534</dl> 676</dl>
677<p>Example: call <code>stdin_readable_cb</code> when STDIN_FILENO has become, well
678readable, but only once. Since it is likely line-buffered, you could
679attempt to read a whole line in the callback:</p>
680<pre> static void
681 stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)
682 {
683 ev_io_stop (loop, w);
684 .. read from stdin here (or from w-&gt;fd) and haqndle any I/O errors
685 }
686
687 ...
688 struct ev_loop *loop = ev_default_init (0);
689 struct ev_io stdin_readable;
690 ev_io_init (&amp;stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
691 ev_io_start (loop, &amp;stdin_readable);
692 ev_loop (loop, 0);
693
694
695
696
697</pre>
535 698
536</div> 699</div>
537<h2 id="code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally recurring timeouts</h2> 700<h2 id="code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally recurring timeouts</h2>
538<div id="code_ev_timer_code_relative_and_opti-2"> 701<div id="code_ev_timer_code_relative_and_opti-2">
539<p>Timer watchers are simple relative timers that generate an event after a 702<p>Timer watchers are simple relative timers that generate an event after a
… …
583time you successfully read or write some data. If you go into an idle 746time you successfully read or write some data. If you go into an idle
584state where you do not expect data to travel on the socket, you can stop 747state where you do not expect data to travel on the socket, you can stop
585the timer, and again will automatically restart it if need be.</p> 748the timer, and again will automatically restart it if need be.</p>
586 </dd> 749 </dd>
587</dl> 750</dl>
751<p>Example: create a timer that fires after 60 seconds.</p>
752<pre> static void
753 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
754 {
755 .. one minute over, w is actually stopped right here
756 }
757
758 struct ev_timer mytimer;
759 ev_timer_init (&amp;mytimer, one_minute_cb, 60., 0.);
760 ev_timer_start (loop, &amp;mytimer);
761
762</pre>
763<p>Example: create a timeout timer that times out after 10 seconds of
764inactivity.</p>
765<pre> static void
766 timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
767 {
768 .. ten seconds without any activity
769 }
770
771 struct ev_timer mytimer;
772 ev_timer_init (&amp;mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
773 ev_timer_again (&amp;mytimer); /* start timer */
774 ev_loop (loop, 0);
775
776 // and in some piece of code that gets executed on any &quot;activity&quot;:
777 // reset the timeout to start ticking again at 10 seconds
778 ev_timer_again (&amp;mytimer);
779
780
781
782
783</pre>
588 784
589</div> 785</div>
590<h2 id="code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron</h2> 786<h2 id="code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron</h2>
591<div id="code_ev_periodic_code_to_cron_or_not-2"> 787<div id="code_ev_periodic_code_to_cron_or_not-2">
592<p>Periodic watchers are also timers of a kind, but they are very versatile 788<p>Periodic watchers are also timers of a kind, but they are very versatile
… …
676when you changed some parameters or the reschedule callback would return 872when you changed some parameters or the reschedule callback would return
677a different time than the last time it was called (e.g. in a crond like 873a different time than the last time it was called (e.g. in a crond like
678program when the crontabs have changed).</p> 874program when the crontabs have changed).</p>
679 </dd> 875 </dd>
680</dl> 876</dl>
877<p>Example: call a callback every hour, or, more precisely, whenever the
878system clock is divisible by 3600. The callback invocation times have
879potentially a lot of jittering, but good long-term stability.</p>
880<pre> static void
881 clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)
882 {
883 ... its now a full hour (UTC, or TAI or whatever your clock follows)
884 }
885
886 struct ev_periodic hourly_tick;
887 ev_periodic_init (&amp;hourly_tick, clock_cb, 0., 3600., 0);
888 ev_periodic_start (loop, &amp;hourly_tick);
889
890</pre>
891<p>Example: the same as above, but use a reschedule callback to do it:</p>
892<pre> #include &lt;math.h&gt;
893
894 static ev_tstamp
895 my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)
896 {
897 return fmod (now, 3600.) + 3600.;
898 }
899
900 ev_periodic_init (&amp;hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
901
902</pre>
903<p>Example: call a callback every hour, starting now:</p>
904<pre> struct ev_periodic hourly_tick;
905 ev_periodic_init (&amp;hourly_tick, clock_cb,
906 fmod (ev_now (loop), 3600.), 3600., 0);
907 ev_periodic_start (loop, &amp;hourly_tick);
908
909
910
911
912</pre>
681 913
682</div> 914</div>
683<h2 id="code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled</h2> 915<h2 id="code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled</h2>
684<div id="code_ev_signal_code_signal_me_when_a-2"> 916<div id="code_ev_signal_code_signal_me_when_a-2">
685<p>Signal watchers will trigger an event when the process receives a specific 917<p>Signal watchers will trigger an event when the process receives a specific
… …
716the status word (use the macros from <code>sys/wait.h</code> and see your systems 948the status word (use the macros from <code>sys/wait.h</code> and see your systems
717<code>waitpid</code> documentation). The <code>rpid</code> member contains the pid of the 949<code>waitpid</code> documentation). The <code>rpid</code> member contains the pid of the
718process causing the status change.</p> 950process causing the status change.</p>
719 </dd> 951 </dd>
720</dl> 952</dl>
953<p>Example: try to exit cleanly on SIGINT and SIGTERM.</p>
954<pre> static void
955 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
956 {
957 ev_unloop (loop, EVUNLOOP_ALL);
958 }
959
960 struct ev_signal signal_watcher;
961 ev_signal_init (&amp;signal_watcher, sigint_cb, SIGINT);
962 ev_signal_start (loop, &amp;sigint_cb);
963
964
965
966
967</pre>
721 968
722</div> 969</div>
723<h2 id="code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do</h2> 970<h2 id="code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do</h2>
724<div id="code_ev_idle_code_when_you_ve_got_no-2"> 971<div id="code_ev_idle_code_when_you_ve_got_no-2">
725<p>Idle watchers trigger events when there are no other events are pending 972<p>Idle watchers trigger events when there are no other events are pending
… …
741 <p>Initialises and configures the idle watcher - it has no parameters of any 988 <p>Initialises and configures the idle watcher - it has no parameters of any
742kind. There is a <code>ev_idle_set</code> macro, but using it is utterly pointless, 989kind. There is a <code>ev_idle_set</code> macro, but using it is utterly pointless,
743believe me.</p> 990believe me.</p>
744 </dd> 991 </dd>
745</dl> 992</dl>
993<p>Example: dynamically allocate an <code>ev_idle</code>, start it, and in the
994callback, free it. Alos, use no error checking, as usual.</p>
995<pre> static void
996 idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
997 {
998 free (w);
999 // now do something you wanted to do when the program has
1000 // no longer asnything immediate to do.
1001 }
1002
1003 struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
1004 ev_idle_init (idle_watcher, idle_cb);
1005 ev_idle_start (loop, idle_cb);
1006
1007
1008
1009
1010</pre>
746 1011
747</div> 1012</div>
748<h2 id="code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop</h2> 1013<h2 id="code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop</h2>
749<div id="code_ev_prepare_code_and_code_ev_che-2"> 1014<div id="code_ev_prepare_code_and_code_ev_che-2">
750<p>Prepare and check watchers are usually (but not always) used in tandem: 1015<p>Prepare and check watchers are usually (but not always) used in tandem:
… …
776 <p>Initialises and configures the prepare or check watcher - they have no 1041 <p>Initialises and configures the prepare or check watcher - they have no
777parameters of any kind. There are <code>ev_prepare_set</code> and <code>ev_check_set</code> 1042parameters of any kind. There are <code>ev_prepare_set</code> and <code>ev_check_set</code>
778macros, but using them is utterly, utterly and completely pointless.</p> 1043macros, but using them is utterly, utterly and completely pointless.</p>
779 </dd> 1044 </dd>
780</dl> 1045</dl>
1046<p>Example: *TODO*.</p>
1047
1048
1049
1050
781 1051
782</div> 1052</div>
783<h1 id="OTHER_FUNCTIONS">OTHER FUNCTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p> 1053<h1 id="OTHER_FUNCTIONS">OTHER FUNCTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p>
784<div id="OTHER_FUNCTIONS_CONTENT"> 1054<div id="OTHER_FUNCTIONS_CONTENT">
785<p>There are some other functions of possible interest. Described. Here. Now.</p> 1055<p>There are some other functions of possible interest. Described. Here. Now.</p>
… …
829 <dd> 1099 <dd>
830 <p>Feed an event as if the given signal occured (loop must be the default loop!).</p> 1100 <p>Feed an event as if the given signal occured (loop must be the default loop!).</p>
831 </dd> 1101 </dd>
832</dl> 1102</dl>
833 1103
1104
1105
1106
1107
834</div> 1108</div>
835<h1 id="LIBEVENT_EMULATION">LIBEVENT EMULATION</h1><p><a href="#TOP" class="toplink">Top</a></p> 1109<h1 id="LIBEVENT_EMULATION">LIBEVENT EMULATION</h1><p><a href="#TOP" class="toplink">Top</a></p>
836<div id="LIBEVENT_EMULATION_CONTENT"> 1110<div id="LIBEVENT_EMULATION_CONTENT">
837<p>Libev offers a compatibility emulation layer for libevent. It cannot 1111<p>Libev offers a compatibility emulation layer for libevent. It cannot
838emulate the internals of libevent, so here are some usage hints:</p> 1112emulate the internals of libevent, so here are some usage hints:</p>

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines