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

Comparing libev/ev.pod (file contents):
Revision 1.328 by root, Sun Oct 24 20:11:11 2010 UTC vs.
Revision 1.379 by root, Tue Jul 12 23:32:10 2011 UTC

58 ev_timer_start (loop, &timeout_watcher); 58 ev_timer_start (loop, &timeout_watcher);
59 59
60 // now wait for events to arrive 60 // now wait for events to arrive
61 ev_run (loop, 0); 61 ev_run (loop, 0);
62 62
63 // unloop was called, so exit 63 // break was called, so exit
64 return 0; 64 return 0;
65 } 65 }
66 66
67=head1 ABOUT THIS DOCUMENT 67=head1 ABOUT THIS DOCUMENT
68 68
77on event-based programming, nor will it introduce event-based programming 77on event-based programming, nor will it introduce event-based programming
78with libev. 78with libev.
79 79
80Familiarity with event based programming techniques in general is assumed 80Familiarity with event based programming techniques in general is assumed
81throughout this document. 81throughout this document.
82
83=head1 WHAT TO READ WHEN IN A HURRY
84
85This manual tries to be very detailed, but unfortunately, this also makes
86it very long. If you just want to know the basics of libev, I suggest
87reading L<ANATOMY OF A WATCHER>, then the L<EXAMPLE PROGRAM> above and
88look up the missing functions in L<GLOBAL FUNCTIONS> and the C<ev_io> and
89C<ev_timer> sections in L<WATCHER TYPES>.
82 90
83=head1 ABOUT LIBEV 91=head1 ABOUT LIBEV
84 92
85Libev is an event loop: you register interest in certain events (such as a 93Libev is an event loop: you register interest in certain events (such as a
86file descriptor being readable or a timeout occurring), and it will manage 94file descriptor being readable or a timeout occurring), and it will manage
170you actually want to know. Also interesting is the combination of 178you actually want to know. Also interesting is the combination of
171C<ev_update_now> and C<ev_now>. 179C<ev_update_now> and C<ev_now>.
172 180
173=item ev_sleep (ev_tstamp interval) 181=item ev_sleep (ev_tstamp interval)
174 182
175Sleep for the given interval: The current thread will be blocked until 183Sleep for the given interval: The current thread will be blocked
176either it is interrupted or the given time interval has passed. Basically 184until either it is interrupted or the given time interval has
185passed (approximately - it might return a bit earlier even if not
186interrupted). Returns immediately if C<< interval <= 0 >>.
187
177this is a sub-second-resolution C<sleep ()>. 188Basically this is a sub-second-resolution C<sleep ()>.
189
190The range of the C<interval> is limited - libev only guarantees to work
191with sleep times of up to one day (C<< interval <= 86400 >>).
178 192
179=item int ev_version_major () 193=item int ev_version_major ()
180 194
181=item int ev_version_minor () 195=item int ev_version_minor ()
182 196
233the current system, you would need to look at C<ev_embeddable_backends () 247the current system, you would need to look at C<ev_embeddable_backends ()
234& ev_supported_backends ()>, likewise for recommended ones. 248& ev_supported_backends ()>, likewise for recommended ones.
235 249
236See the description of C<ev_embed> watchers for more info. 250See the description of C<ev_embed> watchers for more info.
237 251
238=item ev_set_allocator (void *(*cb)(void *ptr, long size)) [NOT REENTRANT] 252=item ev_set_allocator (void *(*cb)(void *ptr, long size))
239 253
240Sets the allocation function to use (the prototype is similar - the 254Sets the allocation function to use (the prototype is similar - the
241semantics are identical to the C<realloc> C89/SuS/POSIX function). It is 255semantics are identical to the C<realloc> C89/SuS/POSIX function). It is
242used to allocate and free memory (no surprises here). If it returns zero 256used to allocate and free memory (no surprises here). If it returns zero
243when memory needs to be allocated (C<size != 0>), the library might abort 257when memory needs to be allocated (C<size != 0>), the library might abort
269 } 283 }
270 284
271 ... 285 ...
272 ev_set_allocator (persistent_realloc); 286 ev_set_allocator (persistent_realloc);
273 287
274=item ev_set_syserr_cb (void (*cb)(const char *msg)); [NOT REENTRANT] 288=item ev_set_syserr_cb (void (*cb)(const char *msg))
275 289
276Set the callback function to call on a retryable system call error (such 290Set the callback function to call on a retryable system call error (such
277as failed select, poll, epoll_wait). The message is a printable string 291as failed select, poll, epoll_wait). The message is a printable string
278indicating the system call or subsystem causing the problem. If this 292indicating the system call or subsystem causing the problem. If this
279callback is set, then libev will expect it to remedy the situation, no 293callback is set, then libev will expect it to remedy the situation, no
291 } 305 }
292 306
293 ... 307 ...
294 ev_set_syserr_cb (fatal_error); 308 ev_set_syserr_cb (fatal_error);
295 309
310=item ev_feed_signal (int signum)
311
312This function can be used to "simulate" a signal receive. It is completely
313safe to call this function at any time, from any context, including signal
314handlers or random threads.
315
316Its main use is to customise signal handling in your process, especially
317in the presence of threads. For example, you could block signals
318by default in all threads (and specifying C<EVFLAG_NOSIGMASK> when
319creating any loops), and in one thread, use C<sigwait> or any other
320mechanism to wait for signals, then "deliver" them to libev by calling
321C<ev_feed_signal>.
322
296=back 323=back
297 324
298=head1 FUNCTIONS CONTROLLING EVENT LOOPS 325=head1 FUNCTIONS CONTROLLING EVENT LOOPS
299 326
300An event loop is described by a C<struct ev_loop *> (the C<struct> is 327An event loop is described by a C<struct ev_loop *> (the C<struct> is
301I<not> optional in this case unless libev 3 compatibility is disabled, as 328I<not> optional in this case unless libev 3 compatibility is disabled, as
302libev 3 had an C<ev_loop> function colliding with the struct name). 329libev 3 had an C<ev_loop> function colliding with the struct name).
303 330
304The library knows two types of such loops, the I<default> loop, which 331The library knows two types of such loops, the I<default> loop, which
305supports signals and child events, and dynamically created event loops 332supports child process events, and dynamically created event loops which
306which do not. 333do not.
307 334
308=over 4 335=over 4
309 336
310=item struct ev_loop *ev_default_loop (unsigned int flags) 337=item struct ev_loop *ev_default_loop (unsigned int flags)
311 338
347=item struct ev_loop *ev_loop_new (unsigned int flags) 374=item struct ev_loop *ev_loop_new (unsigned int flags)
348 375
349This will create and initialise a new event loop object. If the loop 376This will create and initialise a new event loop object. If the loop
350could not be initialised, returns false. 377could not be initialised, returns false.
351 378
352Note that this function I<is> thread-safe, and one common way to use 379This function is thread-safe, and one common way to use libev with
353libev with threads is indeed to create one loop per thread, and using the 380threads is indeed to create one loop per thread, and using the default
354default loop in the "main" or "initial" thread. 381loop in the "main" or "initial" thread.
355 382
356The flags argument can be used to specify special behaviour or specific 383The flags argument can be used to specify special behaviour or specific
357backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>). 384backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
358 385
359The following flags are supported: 386The following flags are supported:
394environment variable. 421environment variable.
395 422
396=item C<EVFLAG_NOINOTIFY> 423=item C<EVFLAG_NOINOTIFY>
397 424
398When this flag is specified, then libev will not attempt to use the 425When this flag is specified, then libev will not attempt to use the
399I<inotify> API for it's C<ev_stat> watchers. Apart from debugging and 426I<inotify> API for its C<ev_stat> watchers. Apart from debugging and
400testing, this flag can be useful to conserve inotify file descriptors, as 427testing, this flag can be useful to conserve inotify file descriptors, as
401otherwise each loop using C<ev_stat> watchers consumes one inotify handle. 428otherwise each loop using C<ev_stat> watchers consumes one inotify handle.
402 429
403=item C<EVFLAG_SIGNALFD> 430=item C<EVFLAG_SIGNALFD>
404 431
405When this flag is specified, then libev will attempt to use the 432When this flag is specified, then libev will attempt to use the
406I<signalfd> API for it's C<ev_signal> (and C<ev_child>) watchers. This API 433I<signalfd> API for its C<ev_signal> (and C<ev_child>) watchers. This API
407delivers signals synchronously, which makes it both faster and might make 434delivers signals synchronously, which makes it both faster and might make
408it possible to get the queued signal data. It can also simplify signal 435it possible to get the queued signal data. It can also simplify signal
409handling with threads, as long as you properly block signals in your 436handling with threads, as long as you properly block signals in your
410threads that are not interested in handling them. 437threads that are not interested in handling them.
411 438
412Signalfd will not be used by default as this changes your signal mask, and 439Signalfd will not be used by default as this changes your signal mask, and
413there are a lot of shoddy libraries and programs (glib's threadpool for 440there are a lot of shoddy libraries and programs (glib's threadpool for
414example) that can't properly initialise their signal masks. 441example) that can't properly initialise their signal masks.
442
443=item C<EVFLAG_NOSIGMASK>
444
445When this flag is specified, then libev will avoid to modify the signal
446mask. Specifically, this means you have to make sure signals are unblocked
447when you want to receive them.
448
449This behaviour is useful when you want to do your own signal handling, or
450want to handle signals only in specific threads and want to avoid libev
451unblocking the signals.
452
453It's also required by POSIX in a threaded program, as libev calls
454C<sigprocmask>, whose behaviour is officially unspecified.
455
456This flag's behaviour will become the default in future versions of libev.
415 457
416=item C<EVBACKEND_SELECT> (value 1, portable select backend) 458=item C<EVBACKEND_SELECT> (value 1, portable select backend)
417 459
418This is your standard select(2) backend. Not I<completely> standard, as 460This is your standard select(2) backend. Not I<completely> standard, as
419libev tries to roll its own fd_set with no limits on the number of fds, 461libev tries to roll its own fd_set with no limits on the number of fds,
447=item C<EVBACKEND_EPOLL> (value 4, Linux) 489=item C<EVBACKEND_EPOLL> (value 4, Linux)
448 490
449Use the linux-specific epoll(7) interface (for both pre- and post-2.6.9 491Use the linux-specific epoll(7) interface (for both pre- and post-2.6.9
450kernels). 492kernels).
451 493
452For few fds, this backend is a bit little slower than poll and select, 494For few fds, this backend is a bit little slower than poll and select, but
453but it scales phenomenally better. While poll and select usually scale 495it scales phenomenally better. While poll and select usually scale like
454like O(total_fds) where n is the total number of fds (or the highest fd), 496O(total_fds) where total_fds is the total number of fds (or the highest
455epoll scales either O(1) or O(active_fds). 497fd), epoll scales either O(1) or O(active_fds).
456 498
457The epoll mechanism deserves honorable mention as the most misdesigned 499The epoll mechanism deserves honorable mention as the most misdesigned
458of the more advanced event mechanisms: mere annoyances include silently 500of the more advanced event mechanisms: mere annoyances include silently
459dropping file descriptors, requiring a system call per change per file 501dropping file descriptors, requiring a system call per change per file
460descriptor (and unnecessary guessing of parameters), problems with dup and 502descriptor (and unnecessary guessing of parameters), problems with dup,
503returning before the timeout value, resulting in additional iterations
504(and only giving 5ms accuracy while select on the same platform gives
461so on. The biggest issue is fork races, however - if a program forks then 5050.1ms) and so on. The biggest issue is fork races, however - if a program
462I<both> parent and child process have to recreate the epoll set, which can 506forks then I<both> parent and child process have to recreate the epoll
463take considerable time (one syscall per file descriptor) and is of course 507set, which can take considerable time (one syscall per file descriptor)
464hard to detect. 508and is of course hard to detect.
465 509
466Epoll is also notoriously buggy - embedding epoll fds I<should> work, but 510Epoll is also notoriously buggy - embedding epoll fds I<should> work,
467of course I<doesn't>, and epoll just loves to report events for totally 511but of course I<doesn't>, and epoll just loves to report events for
468I<different> file descriptors (even already closed ones, so one cannot 512totally I<different> file descriptors (even already closed ones, so
469even remove them from the set) than registered in the set (especially 513one cannot even remove them from the set) than registered in the set
470on SMP systems). Libev tries to counter these spurious notifications by 514(especially on SMP systems). Libev tries to counter these spurious
471employing an additional generation counter and comparing that against the 515notifications by employing an additional generation counter and comparing
472events to filter out spurious ones, recreating the set when required. Last 516that against the events to filter out spurious ones, recreating the set
517when required. Epoll also erroneously rounds down timeouts, but gives you
518no way to know when and by how much, so sometimes you have to busy-wait
519because epoll returns immediately despite a nonzero timeout. And last
473not least, it also refuses to work with some file descriptors which work 520not least, it also refuses to work with some file descriptors which work
474perfectly fine with C<select> (files, many character devices...). 521perfectly fine with C<select> (files, many character devices...).
522
523Epoll is truly the train wreck among event poll mechanisms, a frankenpoll,
524cobbled together in a hurry, no thought to design or interaction with
525others. Oh, the pain, will it ever stop...
475 526
476While stopping, setting and starting an I/O watcher in the same iteration 527While stopping, setting and starting an I/O watcher in the same iteration
477will result in some caching, there is still a system call per such 528will result in some caching, there is still a system call per such
478incident (because the same I<file descriptor> could point to a different 529incident (because the same I<file descriptor> could point to a different
479I<file description> now), so its best to avoid that. Also, C<dup ()>'ed 530I<file description> now), so its best to avoid that. Also, C<dup ()>'ed
545=item C<EVBACKEND_PORT> (value 32, Solaris 10) 596=item C<EVBACKEND_PORT> (value 32, Solaris 10)
546 597
547This uses the Solaris 10 event port mechanism. As with everything on Solaris, 598This uses the Solaris 10 event port mechanism. As with everything on Solaris,
548it's really slow, but it still scales very well (O(active_fds)). 599it's really slow, but it still scales very well (O(active_fds)).
549 600
550Please note that Solaris event ports can deliver a lot of spurious
551notifications, so you need to use non-blocking I/O or other means to avoid
552blocking when no data (or space) is available.
553
554While this backend scales well, it requires one system call per active 601While this backend scales well, it requires one system call per active
555file descriptor per loop iteration. For small and medium numbers of file 602file descriptor per loop iteration. For small and medium numbers of file
556descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend 603descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
557might perform better. 604might perform better.
558 605
559On the positive side, with the exception of the spurious readiness 606On the positive side, this backend actually performed fully to
560notifications, this backend actually performed fully to specification
561in all tests and is fully embeddable, which is a rare feat among the 607specification in all tests and is fully embeddable, which is a rare feat
562OS-specific backends (I vastly prefer correctness over speed hacks). 608among the OS-specific backends (I vastly prefer correctness over speed
609hacks).
610
611On the negative side, the interface is I<bizarre> - so bizarre that
612even sun itself gets it wrong in their code examples: The event polling
613function sometimes returns events to the caller even though an error
614occurred, but with no indication whether it has done so or not (yes, it's
615even documented that way) - deadly for edge-triggered interfaces where you
616absolutely have to know whether an event occurred or not because you have
617to re-arm the watcher.
618
619Fortunately libev seems to be able to work around these idiocies.
563 620
564This backend maps C<EV_READ> and C<EV_WRITE> in the same way as 621This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
565C<EVBACKEND_POLL>. 622C<EVBACKEND_POLL>.
566 623
567=item C<EVBACKEND_ALL> 624=item C<EVBACKEND_ALL>
568 625
569Try all backends (even potentially broken ones that wouldn't be tried 626Try all backends (even potentially broken ones that wouldn't be tried
570with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as 627with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
571C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>. 628C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
572 629
573It is definitely not recommended to use this flag. 630It is definitely not recommended to use this flag, use whatever
631C<ev_recommended_backends ()> returns, or simply do not specify a backend
632at all.
633
634=item C<EVBACKEND_MASK>
635
636Not a backend at all, but a mask to select all backend bits from a
637C<flags> value, in case you want to mask out any backends from a flags
638value (e.g. when modifying the C<LIBEV_FLAGS> environment variable).
574 639
575=back 640=back
576 641
577If one or more of the backend flags are or'ed into the flags value, 642If one or more of the backend flags are or'ed into the flags value,
578then only these backends will be tried (in the reverse order as listed 643then only these backends will be tried (in the reverse order as listed
607This function is normally used on loop objects allocated by 672This function is normally used on loop objects allocated by
608C<ev_loop_new>, but it can also be used on the default loop returned by 673C<ev_loop_new>, but it can also be used on the default loop returned by
609C<ev_default_loop>, in which case it is not thread-safe. 674C<ev_default_loop>, in which case it is not thread-safe.
610 675
611Note that it is not advisable to call this function on the default loop 676Note that it is not advisable to call this function on the default loop
612except in the rare occasion where you really need to free it's resources. 677except in the rare occasion where you really need to free its resources.
613If you need dynamically allocated loops it is better to use C<ev_loop_new> 678If you need dynamically allocated loops it is better to use C<ev_loop_new>
614and C<ev_loop_destroy>. 679and C<ev_loop_destroy>.
615 680
616=item ev_loop_fork (loop) 681=item ev_loop_fork (loop)
617 682
665prepare and check phases. 730prepare and check phases.
666 731
667=item unsigned int ev_depth (loop) 732=item unsigned int ev_depth (loop)
668 733
669Returns the number of times C<ev_run> was entered minus the number of 734Returns the number of times C<ev_run> was entered minus the number of
670times C<ev_run> was exited, in other words, the recursion depth. 735times C<ev_run> was exited normally, in other words, the recursion depth.
671 736
672Outside C<ev_run>, this number is zero. In a callback, this number is 737Outside C<ev_run>, this number is zero. In a callback, this number is
673C<1>, unless C<ev_run> was invoked recursively (or from another thread), 738C<1>, unless C<ev_run> was invoked recursively (or from another thread),
674in which case it is higher. 739in which case it is higher.
675 740
676Leaving C<ev_run> abnormally (setjmp/longjmp, cancelling the thread 741Leaving C<ev_run> abnormally (setjmp/longjmp, cancelling the thread,
677etc.), doesn't count as "exit" - consider this as a hint to avoid such 742throwing an exception etc.), doesn't count as "exit" - consider this
678ungentleman-like behaviour unless it's really convenient. 743as a hint to avoid such ungentleman-like behaviour unless it's really
744convenient, in which case it is fully supported.
679 745
680=item unsigned int ev_backend (loop) 746=item unsigned int ev_backend (loop)
681 747
682Returns one of the C<EVBACKEND_*> flags indicating the event backend in 748Returns one of the C<EVBACKEND_*> flags indicating the event backend in
683use. 749use.
745finished (especially in interactive programs), but having a program 811finished (especially in interactive programs), but having a program
746that automatically loops as long as it has to and no longer by virtue 812that automatically loops as long as it has to and no longer by virtue
747of relying on its watchers stopping correctly, that is truly a thing of 813of relying on its watchers stopping correctly, that is truly a thing of
748beauty. 814beauty.
749 815
816This function is also I<mostly> exception-safe - you can break out of
817a C<ev_run> call by calling C<longjmp> in a callback, throwing a C++
818exception and so on. This does not decrement the C<ev_depth> value, nor
819will it clear any outstanding C<EVBREAK_ONE> breaks.
820
750A flags value of C<EVRUN_NOWAIT> will look for new events, will handle 821A flags value of C<EVRUN_NOWAIT> will look for new events, will handle
751those events and any already outstanding ones, but will not wait and 822those events and any already outstanding ones, but will not wait and
752block your process in case there are no events and will return after one 823block your process in case there are no events and will return after one
753iteration of the loop. This is sometimes useful to poll and handle new 824iteration of the loop. This is sometimes useful to poll and handle new
754events while doing lengthy calculations, to keep the program responsive. 825events while doing lengthy calculations, to keep the program responsive.
763This is useful if you are waiting for some external event in conjunction 834This is useful if you are waiting for some external event in conjunction
764with something not expressible using other libev watchers (i.e. "roll your 835with something not expressible using other libev watchers (i.e. "roll your
765own C<ev_run>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is 836own C<ev_run>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
766usually a better approach for this kind of thing. 837usually a better approach for this kind of thing.
767 838
768Here are the gory details of what C<ev_run> does: 839Here are the gory details of what C<ev_run> does (this is for your
840understanding, not a guarantee that things will work exactly like this in
841future versions):
769 842
770 - Increment loop depth. 843 - Increment loop depth.
771 - Reset the ev_break status. 844 - Reset the ev_break status.
772 - Before the first iteration, call any pending watchers. 845 - Before the first iteration, call any pending watchers.
773 LOOP: 846 LOOP:
806anymore. 879anymore.
807 880
808 ... queue jobs here, make sure they register event watchers as long 881 ... queue jobs here, make sure they register event watchers as long
809 ... as they still have work to do (even an idle watcher will do..) 882 ... as they still have work to do (even an idle watcher will do..)
810 ev_run (my_loop, 0); 883 ev_run (my_loop, 0);
811 ... jobs done or somebody called unloop. yeah! 884 ... jobs done or somebody called break. yeah!
812 885
813=item ev_break (loop, how) 886=item ev_break (loop, how)
814 887
815Can be used to make a call to C<ev_run> return early (but only after it 888Can be used to make a call to C<ev_run> return early (but only after it
816has processed all outstanding events). The C<how> argument must be either 889has processed all outstanding events). The C<how> argument must be either
817C<EVBREAK_ONE>, which will make the innermost C<ev_run> call return, or 890C<EVBREAK_ONE>, which will make the innermost C<ev_run> call return, or
818C<EVBREAK_ALL>, which will make all nested C<ev_run> calls return. 891C<EVBREAK_ALL>, which will make all nested C<ev_run> calls return.
819 892
820This "unloop state" will be cleared when entering C<ev_run> again. 893This "break state" will be cleared on the next call to C<ev_run>.
821 894
822It is safe to call C<ev_break> from outside any C<ev_run> calls. ##TODO## 895It is safe to call C<ev_break> from outside any C<ev_run> calls, too, in
896which case it will have no effect.
823 897
824=item ev_ref (loop) 898=item ev_ref (loop)
825 899
826=item ev_unref (loop) 900=item ev_unref (loop)
827 901
848running when nothing else is active. 922running when nothing else is active.
849 923
850 ev_signal exitsig; 924 ev_signal exitsig;
851 ev_signal_init (&exitsig, sig_cb, SIGINT); 925 ev_signal_init (&exitsig, sig_cb, SIGINT);
852 ev_signal_start (loop, &exitsig); 926 ev_signal_start (loop, &exitsig);
853 evf_unref (loop); 927 ev_unref (loop);
854 928
855Example: For some weird reason, unregister the above signal handler again. 929Example: For some weird reason, unregister the above signal handler again.
856 930
857 ev_ref (loop); 931 ev_ref (loop);
858 ev_signal_stop (loop, &exitsig); 932 ev_signal_stop (loop, &exitsig);
878overhead for the actual polling but can deliver many events at once. 952overhead for the actual polling but can deliver many events at once.
879 953
880By setting a higher I<io collect interval> you allow libev to spend more 954By setting a higher I<io collect interval> you allow libev to spend more
881time collecting I/O events, so you can handle more events per iteration, 955time collecting I/O events, so you can handle more events per iteration,
882at the cost of increasing latency. Timeouts (both C<ev_periodic> and 956at the cost of increasing latency. Timeouts (both C<ev_periodic> and
883C<ev_timer>) will be not affected. Setting this to a non-null value will 957C<ev_timer>) will not be affected. Setting this to a non-null value will
884introduce an additional C<ev_sleep ()> call into most loop iterations. The 958introduce an additional C<ev_sleep ()> call into most loop iterations. The
885sleep time ensures that libev will not poll for I/O events more often then 959sleep time ensures that libev will not poll for I/O events more often then
886once per this interval, on average. 960once per this interval, on average (as long as the host time resolution is
961good enough).
887 962
888Likewise, by setting a higher I<timeout collect interval> you allow libev 963Likewise, by setting a higher I<timeout collect interval> you allow libev
889to spend more time collecting timeouts, at the expense of increased 964to spend more time collecting timeouts, at the expense of increased
890latency/jitter/inexactness (the watcher callback will be called 965latency/jitter/inexactness (the watcher callback will be called
891later). C<ev_io> watchers will not be affected. Setting this to a non-null 966later). C<ev_io> watchers will not be affected. Setting this to a non-null
970See also the locking example in the C<THREADS> section later in this 1045See also the locking example in the C<THREADS> section later in this
971document. 1046document.
972 1047
973=item ev_set_userdata (loop, void *data) 1048=item ev_set_userdata (loop, void *data)
974 1049
975=item ev_userdata (loop) 1050=item void *ev_userdata (loop)
976 1051
977Set and retrieve a single C<void *> associated with a loop. When 1052Set and retrieve a single C<void *> associated with a loop. When
978C<ev_set_userdata> has never been called, then C<ev_userdata> returns 1053C<ev_set_userdata> has never been called, then C<ev_userdata> returns
979C<0.> 1054C<0>.
980 1055
981These two functions can be used to associate arbitrary data with a loop, 1056These two functions can be used to associate arbitrary data with a loop,
982and are intended solely for the C<invoke_pending_cb>, C<release> and 1057and are intended solely for the C<invoke_pending_cb>, C<release> and
983C<acquire> callbacks described above, but of course can be (ab-)used for 1058C<acquire> callbacks described above, but of course can be (ab-)used for
984any other purpose as well. 1059any other purpose as well.
1114The event loop has been resumed in the child process after fork (see 1189The event loop has been resumed in the child process after fork (see
1115C<ev_fork>). 1190C<ev_fork>).
1116 1191
1117=item C<EV_CLEANUP> 1192=item C<EV_CLEANUP>
1118 1193
1119The event loop is abotu to be destroyed (see C<ev_cleanup>). 1194The event loop is about to be destroyed (see C<ev_cleanup>).
1120 1195
1121=item C<EV_ASYNC> 1196=item C<EV_ASYNC>
1122 1197
1123The given async watcher has been asynchronously notified (see C<ev_async>). 1198The given async watcher has been asynchronously notified (see C<ev_async>).
1124 1199
1146programs, though, as the fd could already be closed and reused for another 1221programs, though, as the fd could already be closed and reused for another
1147thing, so beware. 1222thing, so beware.
1148 1223
1149=back 1224=back
1150 1225
1226=head2 GENERIC WATCHER FUNCTIONS
1227
1228=over 4
1229
1230=item C<ev_init> (ev_TYPE *watcher, callback)
1231
1232This macro initialises the generic portion of a watcher. The contents
1233of the watcher object can be arbitrary (so C<malloc> will do). Only
1234the generic parts of the watcher are initialised, you I<need> to call
1235the type-specific C<ev_TYPE_set> macro afterwards to initialise the
1236type-specific parts. For each type there is also a C<ev_TYPE_init> macro
1237which rolls both calls into one.
1238
1239You can reinitialise a watcher at any time as long as it has been stopped
1240(or never started) and there are no pending events outstanding.
1241
1242The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
1243int revents)>.
1244
1245Example: Initialise an C<ev_io> watcher in two steps.
1246
1247 ev_io w;
1248 ev_init (&w, my_cb);
1249 ev_io_set (&w, STDIN_FILENO, EV_READ);
1250
1251=item C<ev_TYPE_set> (ev_TYPE *watcher, [args])
1252
1253This macro initialises the type-specific parts of a watcher. You need to
1254call C<ev_init> at least once before you call this macro, but you can
1255call C<ev_TYPE_set> any number of times. You must not, however, call this
1256macro on a watcher that is active (it can be pending, however, which is a
1257difference to the C<ev_init> macro).
1258
1259Although some watcher types do not have type-specific arguments
1260(e.g. C<ev_prepare>) you still need to call its C<set> macro.
1261
1262See C<ev_init>, above, for an example.
1263
1264=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
1265
1266This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
1267calls into a single call. This is the most convenient method to initialise
1268a watcher. The same limitations apply, of course.
1269
1270Example: Initialise and set an C<ev_io> watcher in one step.
1271
1272 ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
1273
1274=item C<ev_TYPE_start> (loop, ev_TYPE *watcher)
1275
1276Starts (activates) the given watcher. Only active watchers will receive
1277events. If the watcher is already active nothing will happen.
1278
1279Example: Start the C<ev_io> watcher that is being abused as example in this
1280whole section.
1281
1282 ev_io_start (EV_DEFAULT_UC, &w);
1283
1284=item C<ev_TYPE_stop> (loop, ev_TYPE *watcher)
1285
1286Stops the given watcher if active, and clears the pending status (whether
1287the watcher was active or not).
1288
1289It is possible that stopped watchers are pending - for example,
1290non-repeating timers are being stopped when they become pending - but
1291calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
1292pending. If you want to free or reuse the memory used by the watcher it is
1293therefore a good idea to always call its C<ev_TYPE_stop> function.
1294
1295=item bool ev_is_active (ev_TYPE *watcher)
1296
1297Returns a true value iff the watcher is active (i.e. it has been started
1298and not yet been stopped). As long as a watcher is active you must not modify
1299it.
1300
1301=item bool ev_is_pending (ev_TYPE *watcher)
1302
1303Returns a true value iff the watcher is pending, (i.e. it has outstanding
1304events but its callback has not yet been invoked). As long as a watcher
1305is pending (but not active) you must not call an init function on it (but
1306C<ev_TYPE_set> is safe), you must not change its priority, and you must
1307make sure the watcher is available to libev (e.g. you cannot C<free ()>
1308it).
1309
1310=item callback ev_cb (ev_TYPE *watcher)
1311
1312Returns the callback currently set on the watcher.
1313
1314=item ev_cb_set (ev_TYPE *watcher, callback)
1315
1316Change the callback. You can change the callback at virtually any time
1317(modulo threads).
1318
1319=item ev_set_priority (ev_TYPE *watcher, int priority)
1320
1321=item int ev_priority (ev_TYPE *watcher)
1322
1323Set and query the priority of the watcher. The priority is a small
1324integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
1325(default: C<-2>). Pending watchers with higher priority will be invoked
1326before watchers with lower priority, but priority will not keep watchers
1327from being executed (except for C<ev_idle> watchers).
1328
1329If you need to suppress invocation when higher priority events are pending
1330you need to look at C<ev_idle> watchers, which provide this functionality.
1331
1332You I<must not> change the priority of a watcher as long as it is active or
1333pending.
1334
1335Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
1336fine, as long as you do not mind that the priority value you query might
1337or might not have been clamped to the valid range.
1338
1339The default priority used by watchers when no priority has been set is
1340always C<0>, which is supposed to not be too high and not be too low :).
1341
1342See L<WATCHER PRIORITY MODELS>, below, for a more thorough treatment of
1343priorities.
1344
1345=item ev_invoke (loop, ev_TYPE *watcher, int revents)
1346
1347Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
1348C<loop> nor C<revents> need to be valid as long as the watcher callback
1349can deal with that fact, as both are simply passed through to the
1350callback.
1351
1352=item int ev_clear_pending (loop, ev_TYPE *watcher)
1353
1354If the watcher is pending, this function clears its pending status and
1355returns its C<revents> bitset (as if its callback was invoked). If the
1356watcher isn't pending it does nothing and returns C<0>.
1357
1358Sometimes it can be useful to "poll" a watcher instead of waiting for its
1359callback to be invoked, which can be accomplished with this function.
1360
1361=item ev_feed_event (loop, ev_TYPE *watcher, int revents)
1362
1363Feeds the given event set into the event loop, as if the specified event
1364had happened for the specified watcher (which must be a pointer to an
1365initialised but not necessarily started event watcher). Obviously you must
1366not free the watcher as long as it has pending events.
1367
1368Stopping the watcher, letting libev invoke it, or calling
1369C<ev_clear_pending> will clear the pending event, even if the watcher was
1370not started in the first place.
1371
1372See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related
1373functions that do not need a watcher.
1374
1375=back
1376
1377See also the L<ASSOCIATING CUSTOM DATA WITH A WATCHER> and L<BUILDING YOUR
1378OWN COMPOSITE WATCHERS> idioms.
1379
1151=head2 WATCHER STATES 1380=head2 WATCHER STATES
1152 1381
1153There are various watcher states mentioned throughout this manual - 1382There are various watcher states mentioned throughout this manual -
1154active, pending and so on. In this section these states and the rules to 1383active, pending and so on. In this section these states and the rules to
1155transition between them will be described in more detail - and while these 1384transition between them will be described in more detail - and while these
1157 1386
1158=over 4 1387=over 4
1159 1388
1160=item initialiased 1389=item initialiased
1161 1390
1162Before a watcher can be registered with the event looop it has to be 1391Before a watcher can be registered with the event loop it has to be
1163initialised. This can be done with a call to C<ev_TYPE_init>, or calls to 1392initialised. This can be done with a call to C<ev_TYPE_init>, or calls to
1164C<ev_init> followed by the watcher-specific C<ev_TYPE_set> function. 1393C<ev_init> followed by the watcher-specific C<ev_TYPE_set> function.
1165 1394
1166In this state it is simply some block of memory that is suitable for use 1395In this state it is simply some block of memory that is suitable for
1167in an event loop. It can be moved around, freed, reused etc. at will. 1396use in an event loop. It can be moved around, freed, reused etc. at
1397will - as long as you either keep the memory contents intact, or call
1398C<ev_TYPE_init> again.
1168 1399
1169=item started/running/active 1400=item started/running/active
1170 1401
1171Once a watcher has been started with a call to C<ev_TYPE_start> it becomes 1402Once a watcher has been started with a call to C<ev_TYPE_start> it becomes
1172property of the event loop, and is actively waiting for events. While in 1403property of the event loop, and is actively waiting for events. While in
1200latter will clear any pending state the watcher might be in, regardless 1431latter will clear any pending state the watcher might be in, regardless
1201of whether it was active or not, so stopping a watcher explicitly before 1432of whether it was active or not, so stopping a watcher explicitly before
1202freeing it is often a good idea. 1433freeing it is often a good idea.
1203 1434
1204While stopped (and not pending) the watcher is essentially in the 1435While stopped (and not pending) the watcher is essentially in the
1205initialised state, that is it can be reused, moved, modified in any way 1436initialised state, that is, it can be reused, moved, modified in any way
1206you wish. 1437you wish (but when you trash the memory block, you need to C<ev_TYPE_init>
1438it again).
1207 1439
1208=back 1440=back
1209
1210=head2 GENERIC WATCHER FUNCTIONS
1211
1212=over 4
1213
1214=item C<ev_init> (ev_TYPE *watcher, callback)
1215
1216This macro initialises the generic portion of a watcher. The contents
1217of the watcher object can be arbitrary (so C<malloc> will do). Only
1218the generic parts of the watcher are initialised, you I<need> to call
1219the type-specific C<ev_TYPE_set> macro afterwards to initialise the
1220type-specific parts. For each type there is also a C<ev_TYPE_init> macro
1221which rolls both calls into one.
1222
1223You can reinitialise a watcher at any time as long as it has been stopped
1224(or never started) and there are no pending events outstanding.
1225
1226The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
1227int revents)>.
1228
1229Example: Initialise an C<ev_io> watcher in two steps.
1230
1231 ev_io w;
1232 ev_init (&w, my_cb);
1233 ev_io_set (&w, STDIN_FILENO, EV_READ);
1234
1235=item C<ev_TYPE_set> (ev_TYPE *watcher, [args])
1236
1237This macro initialises the type-specific parts of a watcher. You need to
1238call C<ev_init> at least once before you call this macro, but you can
1239call C<ev_TYPE_set> any number of times. You must not, however, call this
1240macro on a watcher that is active (it can be pending, however, which is a
1241difference to the C<ev_init> macro).
1242
1243Although some watcher types do not have type-specific arguments
1244(e.g. C<ev_prepare>) you still need to call its C<set> macro.
1245
1246See C<ev_init>, above, for an example.
1247
1248=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
1249
1250This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
1251calls into a single call. This is the most convenient method to initialise
1252a watcher. The same limitations apply, of course.
1253
1254Example: Initialise and set an C<ev_io> watcher in one step.
1255
1256 ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
1257
1258=item C<ev_TYPE_start> (loop, ev_TYPE *watcher)
1259
1260Starts (activates) the given watcher. Only active watchers will receive
1261events. If the watcher is already active nothing will happen.
1262
1263Example: Start the C<ev_io> watcher that is being abused as example in this
1264whole section.
1265
1266 ev_io_start (EV_DEFAULT_UC, &w);
1267
1268=item C<ev_TYPE_stop> (loop, ev_TYPE *watcher)
1269
1270Stops the given watcher if active, and clears the pending status (whether
1271the watcher was active or not).
1272
1273It is possible that stopped watchers are pending - for example,
1274non-repeating timers are being stopped when they become pending - but
1275calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
1276pending. If you want to free or reuse the memory used by the watcher it is
1277therefore a good idea to always call its C<ev_TYPE_stop> function.
1278
1279=item bool ev_is_active (ev_TYPE *watcher)
1280
1281Returns a true value iff the watcher is active (i.e. it has been started
1282and not yet been stopped). As long as a watcher is active you must not modify
1283it.
1284
1285=item bool ev_is_pending (ev_TYPE *watcher)
1286
1287Returns a true value iff the watcher is pending, (i.e. it has outstanding
1288events but its callback has not yet been invoked). As long as a watcher
1289is pending (but not active) you must not call an init function on it (but
1290C<ev_TYPE_set> is safe), you must not change its priority, and you must
1291make sure the watcher is available to libev (e.g. you cannot C<free ()>
1292it).
1293
1294=item callback ev_cb (ev_TYPE *watcher)
1295
1296Returns the callback currently set on the watcher.
1297
1298=item ev_cb_set (ev_TYPE *watcher, callback)
1299
1300Change the callback. You can change the callback at virtually any time
1301(modulo threads).
1302
1303=item ev_set_priority (ev_TYPE *watcher, int priority)
1304
1305=item int ev_priority (ev_TYPE *watcher)
1306
1307Set and query the priority of the watcher. The priority is a small
1308integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
1309(default: C<-2>). Pending watchers with higher priority will be invoked
1310before watchers with lower priority, but priority will not keep watchers
1311from being executed (except for C<ev_idle> watchers).
1312
1313If you need to suppress invocation when higher priority events are pending
1314you need to look at C<ev_idle> watchers, which provide this functionality.
1315
1316You I<must not> change the priority of a watcher as long as it is active or
1317pending.
1318
1319Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
1320fine, as long as you do not mind that the priority value you query might
1321or might not have been clamped to the valid range.
1322
1323The default priority used by watchers when no priority has been set is
1324always C<0>, which is supposed to not be too high and not be too low :).
1325
1326See L<WATCHER PRIORITY MODELS>, below, for a more thorough treatment of
1327priorities.
1328
1329=item ev_invoke (loop, ev_TYPE *watcher, int revents)
1330
1331Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
1332C<loop> nor C<revents> need to be valid as long as the watcher callback
1333can deal with that fact, as both are simply passed through to the
1334callback.
1335
1336=item int ev_clear_pending (loop, ev_TYPE *watcher)
1337
1338If the watcher is pending, this function clears its pending status and
1339returns its C<revents> bitset (as if its callback was invoked). If the
1340watcher isn't pending it does nothing and returns C<0>.
1341
1342Sometimes it can be useful to "poll" a watcher instead of waiting for its
1343callback to be invoked, which can be accomplished with this function.
1344
1345=item ev_feed_event (loop, ev_TYPE *watcher, int revents)
1346
1347Feeds the given event set into the event loop, as if the specified event
1348had happened for the specified watcher (which must be a pointer to an
1349initialised but not necessarily started event watcher). Obviously you must
1350not free the watcher as long as it has pending events.
1351
1352Stopping the watcher, letting libev invoke it, or calling
1353C<ev_clear_pending> will clear the pending event, even if the watcher was
1354not started in the first place.
1355
1356See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related
1357functions that do not need a watcher.
1358
1359=back
1360
1361
1362=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
1363
1364Each watcher has, by default, a member C<void *data> that you can change
1365and read at any time: libev will completely ignore it. This can be used
1366to associate arbitrary data with your watcher. If you need more data and
1367don't want to allocate memory and store a pointer to it in that data
1368member, you can also "subclass" the watcher type and provide your own
1369data:
1370
1371 struct my_io
1372 {
1373 ev_io io;
1374 int otherfd;
1375 void *somedata;
1376 struct whatever *mostinteresting;
1377 };
1378
1379 ...
1380 struct my_io w;
1381 ev_io_init (&w.io, my_cb, fd, EV_READ);
1382
1383And since your callback will be called with a pointer to the watcher, you
1384can cast it back to your own type:
1385
1386 static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
1387 {
1388 struct my_io *w = (struct my_io *)w_;
1389 ...
1390 }
1391
1392More interesting and less C-conformant ways of casting your callback type
1393instead have been omitted.
1394
1395Another common scenario is to use some data structure with multiple
1396embedded watchers:
1397
1398 struct my_biggy
1399 {
1400 int some_data;
1401 ev_timer t1;
1402 ev_timer t2;
1403 }
1404
1405In this case getting the pointer to C<my_biggy> is a bit more
1406complicated: Either you store the address of your C<my_biggy> struct
1407in the C<data> member of the watcher (for woozies), or you need to use
1408some pointer arithmetic using C<offsetof> inside your watchers (for real
1409programmers):
1410
1411 #include <stddef.h>
1412
1413 static void
1414 t1_cb (EV_P_ ev_timer *w, int revents)
1415 {
1416 struct my_biggy big = (struct my_biggy *)
1417 (((char *)w) - offsetof (struct my_biggy, t1));
1418 }
1419
1420 static void
1421 t2_cb (EV_P_ ev_timer *w, int revents)
1422 {
1423 struct my_biggy big = (struct my_biggy *)
1424 (((char *)w) - offsetof (struct my_biggy, t2));
1425 }
1426 1441
1427=head2 WATCHER PRIORITY MODELS 1442=head2 WATCHER PRIORITY MODELS
1428 1443
1429Many event loops support I<watcher priorities>, which are usually small 1444Many event loops support I<watcher priorities>, which are usually small
1430integers that influence the ordering of event callback invocation 1445integers that influence the ordering of event callback invocation
1557In general you can register as many read and/or write event watchers per 1572In general you can register as many read and/or write event watchers per
1558fd as you want (as long as you don't confuse yourself). Setting all file 1573fd as you want (as long as you don't confuse yourself). Setting all file
1559descriptors to non-blocking mode is also usually a good idea (but not 1574descriptors to non-blocking mode is also usually a good idea (but not
1560required if you know what you are doing). 1575required if you know what you are doing).
1561 1576
1562If you cannot use non-blocking mode, then force the use of a
1563known-to-be-good backend (at the time of this writing, this includes only
1564C<EVBACKEND_SELECT> and C<EVBACKEND_POLL>). The same applies to file
1565descriptors for which non-blocking operation makes no sense (such as
1566files) - libev doesn't guarantee any specific behaviour in that case.
1567
1568Another thing you have to watch out for is that it is quite easy to 1577Another thing you have to watch out for is that it is quite easy to
1569receive "spurious" readiness notifications, that is your callback might 1578receive "spurious" readiness notifications, that is, your callback might
1570be called with C<EV_READ> but a subsequent C<read>(2) will actually block 1579be called with C<EV_READ> but a subsequent C<read>(2) will actually block
1571because there is no data. Not only are some backends known to create a 1580because there is no data. It is very easy to get into this situation even
1572lot of those (for example Solaris ports), it is very easy to get into 1581with a relatively standard program structure. Thus it is best to always
1573this situation even with a relatively standard program structure. Thus 1582use non-blocking I/O: An extra C<read>(2) returning C<EAGAIN> is far
1574it is best to always use non-blocking I/O: An extra C<read>(2) returning
1575C<EAGAIN> is far preferable to a program hanging until some data arrives. 1583preferable to a program hanging until some data arrives.
1576 1584
1577If you cannot run the fd in non-blocking mode (for example you should 1585If you cannot run the fd in non-blocking mode (for example you should
1578not play around with an Xlib connection), then you have to separately 1586not play around with an Xlib connection), then you have to separately
1579re-test whether a file descriptor is really ready with a known-to-be good 1587re-test whether a file descriptor is really ready with a known-to-be good
1580interface such as poll (fortunately in our Xlib example, Xlib already 1588interface such as poll (fortunately in the case of Xlib, it already does
1581does this on its own, so its quite safe to use). Some people additionally 1589this on its own, so its quite safe to use). Some people additionally
1582use C<SIGALRM> and an interval timer, just to be sure you won't block 1590use C<SIGALRM> and an interval timer, just to be sure you won't block
1583indefinitely. 1591indefinitely.
1584 1592
1585But really, best use non-blocking mode. 1593But really, best use non-blocking mode.
1586 1594
1614 1622
1615There is no workaround possible except not registering events 1623There is no workaround possible except not registering events
1616for potentially C<dup ()>'ed file descriptors, or to resort to 1624for potentially C<dup ()>'ed file descriptors, or to resort to
1617C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>. 1625C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
1618 1626
1627=head3 The special problem of files
1628
1629Many people try to use C<select> (or libev) on file descriptors
1630representing files, and expect it to become ready when their program
1631doesn't block on disk accesses (which can take a long time on their own).
1632
1633However, this cannot ever work in the "expected" way - you get a readiness
1634notification as soon as the kernel knows whether and how much data is
1635there, and in the case of open files, that's always the case, so you
1636always get a readiness notification instantly, and your read (or possibly
1637write) will still block on the disk I/O.
1638
1639Another way to view it is that in the case of sockets, pipes, character
1640devices and so on, there is another party (the sender) that delivers data
1641on its own, but in the case of files, there is no such thing: the disk
1642will not send data on its own, simply because it doesn't know what you
1643wish to read - you would first have to request some data.
1644
1645Since files are typically not-so-well supported by advanced notification
1646mechanism, libev tries hard to emulate POSIX behaviour with respect
1647to files, even though you should not use it. The reason for this is
1648convenience: sometimes you want to watch STDIN or STDOUT, which is
1649usually a tty, often a pipe, but also sometimes files or special devices
1650(for example, C<epoll> on Linux works with F</dev/random> but not with
1651F</dev/urandom>), and even though the file might better be served with
1652asynchronous I/O instead of with non-blocking I/O, it is still useful when
1653it "just works" instead of freezing.
1654
1655So avoid file descriptors pointing to files when you know it (e.g. use
1656libeio), but use them when it is convenient, e.g. for STDIN/STDOUT, or
1657when you rarely read from a file instead of from a socket, and want to
1658reuse the same code path.
1659
1619=head3 The special problem of fork 1660=head3 The special problem of fork
1620 1661
1621Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit 1662Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
1622useless behaviour. Libev fully supports fork, but needs to be told about 1663useless behaviour. Libev fully supports fork, but needs to be told about
1623it in the child. 1664it in the child if you want to continue to use it in the child.
1624 1665
1625To support fork in your programs, you either have to call 1666To support fork in your child processes, you have to call C<ev_loop_fork
1626C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child, 1667()> after a fork in the child, enable C<EVFLAG_FORKCHECK>, or resort to
1627enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or 1668C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
1628C<EVBACKEND_POLL>.
1629 1669
1630=head3 The special problem of SIGPIPE 1670=head3 The special problem of SIGPIPE
1631 1671
1632While not really specific to libev, it is easy to forget about C<SIGPIPE>: 1672While not really specific to libev, it is easy to forget about C<SIGPIPE>:
1633when writing to a pipe whose other end has been closed, your program gets 1673when writing to a pipe whose other end has been closed, your program gets
1983keep up with the timer (because it takes longer than those 10 seconds to 2023keep up with the timer (because it takes longer than those 10 seconds to
1984do stuff) the timer will not fire more than once per event loop iteration. 2024do stuff) the timer will not fire more than once per event loop iteration.
1985 2025
1986=item ev_timer_again (loop, ev_timer *) 2026=item ev_timer_again (loop, ev_timer *)
1987 2027
1988This will act as if the timer timed out and restart it again if it is 2028This will act as if the timer timed out and restarts it again if it is
1989repeating. The exact semantics are: 2029repeating. The exact semantics are:
1990 2030
1991If the timer is pending, its pending status is cleared. 2031If the timer is pending, its pending status is cleared.
1992 2032
1993If the timer is started but non-repeating, stop it (as if it timed out). 2033If the timer is started but non-repeating, stop it (as if it timed out).
2123 2163
2124Another way to think about it (for the mathematically inclined) is that 2164Another way to think about it (for the mathematically inclined) is that
2125C<ev_periodic> will try to run the callback in this mode at the next possible 2165C<ev_periodic> will try to run the callback in this mode at the next possible
2126time where C<time = offset (mod interval)>, regardless of any time jumps. 2166time where C<time = offset (mod interval)>, regardless of any time jumps.
2127 2167
2128For numerical stability it is preferable that the C<offset> value is near 2168The C<interval> I<MUST> be positive, and for numerical stability, the
2129C<ev_now ()> (the current time), but there is no range requirement for 2169interval value should be higher than C<1/8192> (which is around 100
2130this value, and in fact is often specified as zero. 2170microseconds) and C<offset> should be higher than C<0> and should have
2171at most a similar magnitude as the current time (say, within a factor of
2172ten). Typical values for offset are, in fact, C<0> or something between
2173C<0> and C<interval>, which is also the recommended range.
2131 2174
2132Note also that there is an upper limit to how often a timer can fire (CPU 2175Note also that there is an upper limit to how often a timer can fire (CPU
2133speed for example), so if C<interval> is very small then timing stability 2176speed for example), so if C<interval> is very small then timing stability
2134will of course deteriorate. Libev itself tries to be exact to be about one 2177will of course deteriorate. Libev itself tries to be exact to be about one
2135millisecond (if the OS supports it and the machine is fast enough). 2178millisecond (if the OS supports it and the machine is fast enough).
2249 2292
2250=head2 C<ev_signal> - signal me when a signal gets signalled! 2293=head2 C<ev_signal> - signal me when a signal gets signalled!
2251 2294
2252Signal watchers will trigger an event when the process receives a specific 2295Signal watchers will trigger an event when the process receives a specific
2253signal one or more times. Even though signals are very asynchronous, libev 2296signal one or more times. Even though signals are very asynchronous, libev
2254will try it's best to deliver signals synchronously, i.e. as part of the 2297will try its best to deliver signals synchronously, i.e. as part of the
2255normal event processing, like any other event. 2298normal event processing, like any other event.
2256 2299
2257If you want signals to be delivered truly asynchronously, just use 2300If you want signals to be delivered truly asynchronously, just use
2258C<sigaction> as you would do without libev and forget about sharing 2301C<sigaction> as you would do without libev and forget about sharing
2259the signal. You can even use C<ev_async> from a signal handler to 2302the signal. You can even use C<ev_async> from a signal handler to
2278=head3 The special problem of inheritance over fork/execve/pthread_create 2321=head3 The special problem of inheritance over fork/execve/pthread_create
2279 2322
2280Both the signal mask (C<sigprocmask>) and the signal disposition 2323Both the signal mask (C<sigprocmask>) and the signal disposition
2281(C<sigaction>) are unspecified after starting a signal watcher (and after 2324(C<sigaction>) are unspecified after starting a signal watcher (and after
2282stopping it again), that is, libev might or might not block the signal, 2325stopping it again), that is, libev might or might not block the signal,
2283and might or might not set or restore the installed signal handler. 2326and might or might not set or restore the installed signal handler (but
2327see C<EVFLAG_NOSIGMASK>).
2284 2328
2285While this does not matter for the signal disposition (libev never 2329While this does not matter for the signal disposition (libev never
2286sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on 2330sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on
2287C<execve>), this matters for the signal mask: many programs do not expect 2331C<execve>), this matters for the signal mask: many programs do not expect
2288certain signals to be blocked. 2332certain signals to be blocked.
2301I<has> to modify the signal mask, at least temporarily. 2345I<has> to modify the signal mask, at least temporarily.
2302 2346
2303So I can't stress this enough: I<If you do not reset your signal mask when 2347So I can't stress this enough: I<If you do not reset your signal mask when
2304you expect it to be empty, you have a race condition in your code>. This 2348you expect it to be empty, you have a race condition in your code>. This
2305is not a libev-specific thing, this is true for most event libraries. 2349is not a libev-specific thing, this is true for most event libraries.
2350
2351=head3 The special problem of threads signal handling
2352
2353POSIX threads has problematic signal handling semantics, specifically,
2354a lot of functionality (sigfd, sigwait etc.) only really works if all
2355threads in a process block signals, which is hard to achieve.
2356
2357When you want to use sigwait (or mix libev signal handling with your own
2358for the same signals), you can tackle this problem by globally blocking
2359all signals before creating any threads (or creating them with a fully set
2360sigprocmask) and also specifying the C<EVFLAG_NOSIGMASK> when creating
2361loops. Then designate one thread as "signal receiver thread" which handles
2362these signals. You can pass on any signals that libev might be interested
2363in by calling C<ev_feed_signal>.
2306 2364
2307=head3 Watcher-Specific Functions and Data Members 2365=head3 Watcher-Specific Functions and Data Members
2308 2366
2309=over 4 2367=over 4
2310 2368
3098 3156
3099=item ev_fork_init (ev_fork *, callback) 3157=item ev_fork_init (ev_fork *, callback)
3100 3158
3101Initialises and configures the fork watcher - it has no parameters of any 3159Initialises and configures the fork watcher - it has no parameters of any
3102kind. There is a C<ev_fork_set> macro, but using it is utterly pointless, 3160kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
3103believe me. 3161really.
3104 3162
3105=back 3163=back
3106 3164
3107 3165
3108=head2 C<ev_cleanup> - even the best things end 3166=head2 C<ev_cleanup> - even the best things end
3126 3184
3127=item ev_cleanup_init (ev_cleanup *, callback) 3185=item ev_cleanup_init (ev_cleanup *, callback)
3128 3186
3129Initialises and configures the cleanup watcher - it has no parameters of 3187Initialises and configures the cleanup watcher - it has no parameters of
3130any kind. There is a C<ev_cleanup_set> macro, but using it is utterly 3188any kind. There is a C<ev_cleanup_set> macro, but using it is utterly
3131pointless, believe me. 3189pointless, I assure you.
3132 3190
3133=back 3191=back
3134 3192
3135Example: Register an atexit handler to destroy the default loop, so any 3193Example: Register an atexit handler to destroy the default loop, so any
3136cleanup functions are called. 3194cleanup functions are called.
3145 atexit (program_exits); 3203 atexit (program_exits);
3146 3204
3147 3205
3148=head2 C<ev_async> - how to wake up an event loop 3206=head2 C<ev_async> - how to wake up an event loop
3149 3207
3150In general, you cannot use an C<ev_run> from multiple threads or other 3208In general, you cannot use an C<ev_loop> from multiple threads or other
3151asynchronous sources such as signal handlers (as opposed to multiple event 3209asynchronous sources such as signal handlers (as opposed to multiple event
3152loops - those are of course safe to use in different threads). 3210loops - those are of course safe to use in different threads).
3153 3211
3154Sometimes, however, you need to wake up an event loop you do not control, 3212Sometimes, however, you need to wake up an event loop you do not control,
3155for example because it belongs to another thread. This is what C<ev_async> 3213for example because it belongs to another thread. This is what C<ev_async>
3157it by calling C<ev_async_send>, which is thread- and signal safe. 3215it by calling C<ev_async_send>, which is thread- and signal safe.
3158 3216
3159This functionality is very similar to C<ev_signal> watchers, as signals, 3217This functionality is very similar to C<ev_signal> watchers, as signals,
3160too, are asynchronous in nature, and signals, too, will be compressed 3218too, are asynchronous in nature, and signals, too, will be compressed
3161(i.e. the number of callback invocations may be less than the number of 3219(i.e. the number of callback invocations may be less than the number of
3162C<ev_async_sent> calls). 3220C<ev_async_sent> calls). In fact, you could use signal watchers as a kind
3163 3221of "global async watchers" by using a watcher on an otherwise unused
3164Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not 3222signal, and C<ev_feed_signal> to signal this watcher from another thread,
3165just the default loop. 3223even without knowing which loop owns the signal.
3166 3224
3167=head3 Queueing 3225=head3 Queueing
3168 3226
3169C<ev_async> does not support queueing of data in any way. The reason 3227C<ev_async> does not support queueing of data in any way. The reason
3170is that the author does not know of a simple (or any) algorithm for a 3228is that the author does not know of a simple (or any) algorithm for a
3262trust me. 3320trust me.
3263 3321
3264=item ev_async_send (loop, ev_async *) 3322=item ev_async_send (loop, ev_async *)
3265 3323
3266Sends/signals/activates the given C<ev_async> watcher, that is, feeds 3324Sends/signals/activates the given C<ev_async> watcher, that is, feeds
3267an C<EV_ASYNC> event on the watcher into the event loop. Unlike 3325an C<EV_ASYNC> event on the watcher into the event loop, and instantly
3326returns.
3327
3268C<ev_feed_event>, this call is safe to do from other threads, signal or 3328Unlike C<ev_feed_event>, this call is safe to do from other threads,
3269similar contexts (see the discussion of C<EV_ATOMIC_T> in the embedding 3329signal or similar contexts (see the discussion of C<EV_ATOMIC_T> in the
3270section below on what exactly this means). 3330embedding section below on what exactly this means).
3271 3331
3272Note that, as with other watchers in libev, multiple events might get 3332Note that, as with other watchers in libev, multiple events might get
3273compressed into a single callback invocation (another way to look at this 3333compressed into a single callback invocation (another way to look at
3274is that C<ev_async> watchers are level-triggered, set on C<ev_async_send>, 3334this is that C<ev_async> watchers are level-triggered: they are set on
3275reset when the event loop detects that). 3335C<ev_async_send>, reset when the event loop detects that).
3276 3336
3277This call incurs the overhead of a system call only once per event loop 3337This call incurs the overhead of at most one extra system call per event
3278iteration, so while the overhead might be noticeable, it doesn't apply to 3338loop iteration, if the event loop is blocked, and no syscall at all if
3279repeated calls to C<ev_async_send> for the same event loop. 3339the event loop (or your program) is processing events. That means that
3340repeated calls are basically free (there is no need to avoid calls for
3341performance reasons) and that the overhead becomes smaller (typically
3342zero) under load.
3280 3343
3281=item bool = ev_async_pending (ev_async *) 3344=item bool = ev_async_pending (ev_async *)
3282 3345
3283Returns a non-zero value when C<ev_async_send> has been called on the 3346Returns a non-zero value when C<ev_async_send> has been called on the
3284watcher but the event has not yet been processed (or even noted) by the 3347watcher but the event has not yet been processed (or even noted) by the
3343Feed an event on the given fd, as if a file descriptor backend detected 3406Feed an event on the given fd, as if a file descriptor backend detected
3344the given events it. 3407the given events it.
3345 3408
3346=item ev_feed_signal_event (loop, int signum) 3409=item ev_feed_signal_event (loop, int signum)
3347 3410
3348Feed an event as if the given signal occurred (C<loop> must be the default 3411Feed an event as if the given signal occurred. See also C<ev_feed_signal>,
3349loop!). 3412which is async-safe.
3350 3413
3351=back 3414=back
3415
3416
3417=head1 COMMON OR USEFUL IDIOMS (OR BOTH)
3418
3419This section explains some common idioms that are not immediately
3420obvious. Note that examples are sprinkled over the whole manual, and this
3421section only contains stuff that wouldn't fit anywhere else.
3422
3423=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
3424
3425Each watcher has, by default, a C<void *data> member that you can read
3426or modify at any time: libev will completely ignore it. This can be used
3427to associate arbitrary data with your watcher. If you need more data and
3428don't want to allocate memory separately and store a pointer to it in that
3429data member, you can also "subclass" the watcher type and provide your own
3430data:
3431
3432 struct my_io
3433 {
3434 ev_io io;
3435 int otherfd;
3436 void *somedata;
3437 struct whatever *mostinteresting;
3438 };
3439
3440 ...
3441 struct my_io w;
3442 ev_io_init (&w.io, my_cb, fd, EV_READ);
3443
3444And since your callback will be called with a pointer to the watcher, you
3445can cast it back to your own type:
3446
3447 static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
3448 {
3449 struct my_io *w = (struct my_io *)w_;
3450 ...
3451 }
3452
3453More interesting and less C-conformant ways of casting your callback
3454function type instead have been omitted.
3455
3456=head2 BUILDING YOUR OWN COMPOSITE WATCHERS
3457
3458Another common scenario is to use some data structure with multiple
3459embedded watchers, in effect creating your own watcher that combines
3460multiple libev event sources into one "super-watcher":
3461
3462 struct my_biggy
3463 {
3464 int some_data;
3465 ev_timer t1;
3466 ev_timer t2;
3467 }
3468
3469In this case getting the pointer to C<my_biggy> is a bit more
3470complicated: Either you store the address of your C<my_biggy> struct in
3471the C<data> member of the watcher (for woozies or C++ coders), or you need
3472to use some pointer arithmetic using C<offsetof> inside your watchers (for
3473real programmers):
3474
3475 #include <stddef.h>
3476
3477 static void
3478 t1_cb (EV_P_ ev_timer *w, int revents)
3479 {
3480 struct my_biggy big = (struct my_biggy *)
3481 (((char *)w) - offsetof (struct my_biggy, t1));
3482 }
3483
3484 static void
3485 t2_cb (EV_P_ ev_timer *w, int revents)
3486 {
3487 struct my_biggy big = (struct my_biggy *)
3488 (((char *)w) - offsetof (struct my_biggy, t2));
3489 }
3490
3491=head2 MODEL/NESTED EVENT LOOP INVOCATIONS AND EXIT CONDITIONS
3492
3493Often (especially in GUI toolkits) there are places where you have
3494I<modal> interaction, which is most easily implemented by recursively
3495invoking C<ev_run>.
3496
3497This brings the problem of exiting - a callback might want to finish the
3498main C<ev_run> call, but not the nested one (e.g. user clicked "Quit", but
3499a modal "Are you sure?" dialog is still waiting), or just the nested one
3500and not the main one (e.g. user clocked "Ok" in a modal dialog), or some
3501other combination: In these cases, C<ev_break> will not work alone.
3502
3503The solution is to maintain "break this loop" variable for each C<ev_run>
3504invocation, and use a loop around C<ev_run> until the condition is
3505triggered, using C<EVRUN_ONCE>:
3506
3507 // main loop
3508 int exit_main_loop = 0;
3509
3510 while (!exit_main_loop)
3511 ev_run (EV_DEFAULT_ EVRUN_ONCE);
3512
3513 // in a model watcher
3514 int exit_nested_loop = 0;
3515
3516 while (!exit_nested_loop)
3517 ev_run (EV_A_ EVRUN_ONCE);
3518
3519To exit from any of these loops, just set the corresponding exit variable:
3520
3521 // exit modal loop
3522 exit_nested_loop = 1;
3523
3524 // exit main program, after modal loop is finished
3525 exit_main_loop = 1;
3526
3527 // exit both
3528 exit_main_loop = exit_nested_loop = 1;
3529
3530=head2 THREAD LOCKING EXAMPLE
3531
3532Here is a fictitious example of how to run an event loop in a different
3533thread from where callbacks are being invoked and watchers are
3534created/added/removed.
3535
3536For a real-world example, see the C<EV::Loop::Async> perl module,
3537which uses exactly this technique (which is suited for many high-level
3538languages).
3539
3540The example uses a pthread mutex to protect the loop data, a condition
3541variable to wait for callback invocations, an async watcher to notify the
3542event loop thread and an unspecified mechanism to wake up the main thread.
3543
3544First, you need to associate some data with the event loop:
3545
3546 typedef struct {
3547 mutex_t lock; /* global loop lock */
3548 ev_async async_w;
3549 thread_t tid;
3550 cond_t invoke_cv;
3551 } userdata;
3552
3553 void prepare_loop (EV_P)
3554 {
3555 // for simplicity, we use a static userdata struct.
3556 static userdata u;
3557
3558 ev_async_init (&u->async_w, async_cb);
3559 ev_async_start (EV_A_ &u->async_w);
3560
3561 pthread_mutex_init (&u->lock, 0);
3562 pthread_cond_init (&u->invoke_cv, 0);
3563
3564 // now associate this with the loop
3565 ev_set_userdata (EV_A_ u);
3566 ev_set_invoke_pending_cb (EV_A_ l_invoke);
3567 ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
3568
3569 // then create the thread running ev_run
3570 pthread_create (&u->tid, 0, l_run, EV_A);
3571 }
3572
3573The callback for the C<ev_async> watcher does nothing: the watcher is used
3574solely to wake up the event loop so it takes notice of any new watchers
3575that might have been added:
3576
3577 static void
3578 async_cb (EV_P_ ev_async *w, int revents)
3579 {
3580 // just used for the side effects
3581 }
3582
3583The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
3584protecting the loop data, respectively.
3585
3586 static void
3587 l_release (EV_P)
3588 {
3589 userdata *u = ev_userdata (EV_A);
3590 pthread_mutex_unlock (&u->lock);
3591 }
3592
3593 static void
3594 l_acquire (EV_P)
3595 {
3596 userdata *u = ev_userdata (EV_A);
3597 pthread_mutex_lock (&u->lock);
3598 }
3599
3600The event loop thread first acquires the mutex, and then jumps straight
3601into C<ev_run>:
3602
3603 void *
3604 l_run (void *thr_arg)
3605 {
3606 struct ev_loop *loop = (struct ev_loop *)thr_arg;
3607
3608 l_acquire (EV_A);
3609 pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
3610 ev_run (EV_A_ 0);
3611 l_release (EV_A);
3612
3613 return 0;
3614 }
3615
3616Instead of invoking all pending watchers, the C<l_invoke> callback will
3617signal the main thread via some unspecified mechanism (signals? pipe
3618writes? C<Async::Interrupt>?) and then waits until all pending watchers
3619have been called (in a while loop because a) spurious wakeups are possible
3620and b) skipping inter-thread-communication when there are no pending
3621watchers is very beneficial):
3622
3623 static void
3624 l_invoke (EV_P)
3625 {
3626 userdata *u = ev_userdata (EV_A);
3627
3628 while (ev_pending_count (EV_A))
3629 {
3630 wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
3631 pthread_cond_wait (&u->invoke_cv, &u->lock);
3632 }
3633 }
3634
3635Now, whenever the main thread gets told to invoke pending watchers, it
3636will grab the lock, call C<ev_invoke_pending> and then signal the loop
3637thread to continue:
3638
3639 static void
3640 real_invoke_pending (EV_P)
3641 {
3642 userdata *u = ev_userdata (EV_A);
3643
3644 pthread_mutex_lock (&u->lock);
3645 ev_invoke_pending (EV_A);
3646 pthread_cond_signal (&u->invoke_cv);
3647 pthread_mutex_unlock (&u->lock);
3648 }
3649
3650Whenever you want to start/stop a watcher or do other modifications to an
3651event loop, you will now have to lock:
3652
3653 ev_timer timeout_watcher;
3654 userdata *u = ev_userdata (EV_A);
3655
3656 ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
3657
3658 pthread_mutex_lock (&u->lock);
3659 ev_timer_start (EV_A_ &timeout_watcher);
3660 ev_async_send (EV_A_ &u->async_w);
3661 pthread_mutex_unlock (&u->lock);
3662
3663Note that sending the C<ev_async> watcher is required because otherwise
3664an event loop currently blocking in the kernel will have no knowledge
3665about the newly added timer. By waking up the loop it will pick up any new
3666watchers in the next event loop iteration.
3667
3668=head2 THREADS, COROUTINES, CONTINUATIONS, QUEUES... INSTEAD OF CALLBACKS
3669
3670While the overhead of a callback that e.g. schedules a thread is small, it
3671is still an overhead. If you embed libev, and your main usage is with some
3672kind of threads or coroutines, you might want to customise libev so that
3673doesn't need callbacks anymore.
3674
3675Imagine you have coroutines that you can switch to using a function
3676C<switch_to (coro)>, that libev runs in a coroutine called C<libev_coro>
3677and that due to some magic, the currently active coroutine is stored in a
3678global called C<current_coro>. Then you can build your own "wait for libev
3679event" primitive by changing C<EV_CB_DECLARE> and C<EV_CB_INVOKE> (note
3680the differing C<;> conventions):
3681
3682 #define EV_CB_DECLARE(type) struct my_coro *cb;
3683 #define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb)
3684
3685That means instead of having a C callback function, you store the
3686coroutine to switch to in each watcher, and instead of having libev call
3687your callback, you instead have it switch to that coroutine.
3688
3689A coroutine might now wait for an event with a function called
3690C<wait_for_event>. (the watcher needs to be started, as always, but it doesn't
3691matter when, or whether the watcher is active or not when this function is
3692called):
3693
3694 void
3695 wait_for_event (ev_watcher *w)
3696 {
3697 ev_cb_set (w) = current_coro;
3698 switch_to (libev_coro);
3699 }
3700
3701That basically suspends the coroutine inside C<wait_for_event> and
3702continues the libev coroutine, which, when appropriate, switches back to
3703this or any other coroutine. I am sure if you sue this your own :)
3704
3705You can do similar tricks if you have, say, threads with an event queue -
3706instead of storing a coroutine, you store the queue object and instead of
3707switching to a coroutine, you push the watcher onto the queue and notify
3708any waiters.
3709
3710To embed libev, see L<EMBEDDING>, but in short, it's easiest to create two
3711files, F<my_ev.h> and F<my_ev.c> that include the respective libev files:
3712
3713 // my_ev.h
3714 #define EV_CB_DECLARE(type) struct my_coro *cb;
3715 #define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb);
3716 #include "../libev/ev.h"
3717
3718 // my_ev.c
3719 #define EV_H "my_ev.h"
3720 #include "../libev/ev.c"
3721
3722And then use F<my_ev.h> when you would normally use F<ev.h>, and compile
3723F<my_ev.c> into your project. When properly specifying include paths, you
3724can even use F<ev.h> as header file name directly.
3352 3725
3353 3726
3354=head1 LIBEVENT EMULATION 3727=head1 LIBEVENT EMULATION
3355 3728
3356Libev offers a compatibility emulation layer for libevent. It cannot 3729Libev offers a compatibility emulation layer for libevent. It cannot
3357emulate the internals of libevent, so here are some usage hints: 3730emulate the internals of libevent, so here are some usage hints:
3358 3731
3359=over 4 3732=over 4
3733
3734=item * Only the libevent-1.4.1-beta API is being emulated.
3735
3736This was the newest libevent version available when libev was implemented,
3737and is still mostly unchanged in 2010.
3360 3738
3361=item * Use it by including <event.h>, as usual. 3739=item * Use it by including <event.h>, as usual.
3362 3740
3363=item * The following members are fully supported: ev_base, ev_callback, 3741=item * The following members are fully supported: ev_base, ev_callback,
3364ev_arg, ev_fd, ev_res, ev_events. 3742ev_arg, ev_fd, ev_res, ev_events.
3370=item * Priorities are not currently supported. Initialising priorities 3748=item * Priorities are not currently supported. Initialising priorities
3371will fail and all watchers will have the same priority, even though there 3749will fail and all watchers will have the same priority, even though there
3372is an ev_pri field. 3750is an ev_pri field.
3373 3751
3374=item * In libevent, the last base created gets the signals, in libev, the 3752=item * In libevent, the last base created gets the signals, in libev, the
3375first base created (== the default loop) gets the signals. 3753base that registered the signal gets the signals.
3376 3754
3377=item * Other members are not supported. 3755=item * Other members are not supported.
3378 3756
3379=item * The libev emulation is I<not> ABI compatible to libevent, you need 3757=item * The libev emulation is I<not> ABI compatible to libevent, you need
3380to use the libev header file and library. 3758to use the libev header file and library.
3399Care has been taken to keep the overhead low. The only data member the C++ 3777Care has been taken to keep the overhead low. The only data member the C++
3400classes add (compared to plain C-style watchers) is the event loop pointer 3778classes add (compared to plain C-style watchers) is the event loop pointer
3401that the watcher is associated with (or no additional members at all if 3779that the watcher is associated with (or no additional members at all if
3402you disable C<EV_MULTIPLICITY> when embedding libev). 3780you disable C<EV_MULTIPLICITY> when embedding libev).
3403 3781
3404Currently, functions, and static and non-static member functions can be 3782Currently, functions, static and non-static member functions and classes
3405used as callbacks. Other types should be easy to add as long as they only 3783with C<operator ()> can be used as callbacks. Other types should be easy
3406need one additional pointer for context. If you need support for other 3784to add as long as they only need one additional pointer for context. If
3407types of functors please contact the author (preferably after implementing 3785you need support for other types of functors please contact the author
3408it). 3786(preferably after implementing it).
3409 3787
3410Here is a list of things available in the C<ev> namespace: 3788Here is a list of things available in the C<ev> namespace:
3411 3789
3412=over 4 3790=over 4
3413 3791
3566watchers in the constructor. 3944watchers in the constructor.
3567 3945
3568 class myclass 3946 class myclass
3569 { 3947 {
3570 ev::io io ; void io_cb (ev::io &w, int revents); 3948 ev::io io ; void io_cb (ev::io &w, int revents);
3571 ev::io2 io2 ; void io2_cb (ev::io &w, int revents); 3949 ev::io io2 ; void io2_cb (ev::io &w, int revents);
3572 ev::idle idle; void idle_cb (ev::idle &w, int revents); 3950 ev::idle idle; void idle_cb (ev::idle &w, int revents);
3573 3951
3574 myclass (int fd) 3952 myclass (int fd)
3575 { 3953 {
3576 io .set <myclass, &myclass::io_cb > (this); 3954 io .set <myclass, &myclass::io_cb > (this);
3627L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>. 4005L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.
3628 4006
3629=item D 4007=item D
3630 4008
3631Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to 4009Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
3632be found at L<http://proj.llucax.com.ar/wiki/evd>. 4010be found at L<http://www.llucax.com.ar/proj/ev.d/index.html>.
3633 4011
3634=item Ocaml 4012=item Ocaml
3635 4013
3636Erkki Seppala has written Ocaml bindings for libev, to be found at 4014Erkki Seppala has written Ocaml bindings for libev, to be found at
3637L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>. 4015L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
3840supported). It will also not define any of the structs usually found in 4218supported). It will also not define any of the structs usually found in
3841F<event.h> that are not directly supported by the libev core alone. 4219F<event.h> that are not directly supported by the libev core alone.
3842 4220
3843In standalone mode, libev will still try to automatically deduce the 4221In standalone mode, libev will still try to automatically deduce the
3844configuration, but has to be more conservative. 4222configuration, but has to be more conservative.
4223
4224=item EV_USE_FLOOR
4225
4226If defined to be C<1>, libev will use the C<floor ()> function for its
4227periodic reschedule calculations, otherwise libev will fall back on a
4228portable (slower) implementation. If you enable this, you usually have to
4229link against libm or something equivalent. Enabling this when the C<floor>
4230function is not available will fail, so the safe default is to not enable
4231this.
3845 4232
3846=item EV_USE_MONOTONIC 4233=item EV_USE_MONOTONIC
3847 4234
3848If defined to be C<1>, libev will try to detect the availability of the 4235If defined to be C<1>, libev will try to detect the availability of the
3849monotonic clock option at both compile time and runtime. Otherwise no 4236monotonic clock option at both compile time and runtime. Otherwise no
3982indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled. 4369indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
3983 4370
3984=item EV_ATOMIC_T 4371=item EV_ATOMIC_T
3985 4372
3986Libev requires an integer type (suitable for storing C<0> or C<1>) whose 4373Libev requires an integer type (suitable for storing C<0> or C<1>) whose
3987access is atomic with respect to other threads or signal contexts. No such 4374access is atomic and serialised with respect to other threads or signal
3988type is easily found in the C language, so you can provide your own type 4375contexts. No such type is easily found in the C language, so you can
3989that you know is safe for your purposes. It is used both for signal handler "locking" 4376provide your own type that you know is safe for your purposes. It is used
3990as well as for signal and thread safety in C<ev_async> watchers. 4377both for signal handler "locking" as well as for signal and thread safety
4378in C<ev_async> watchers.
3991 4379
3992In the absence of this define, libev will use C<sig_atomic_t volatile> 4380In the absence of this define, libev will use C<sig_atomic_t volatile>
3993(from F<signal.h>), which is usually good enough on most platforms. 4381(from F<signal.h>), which is usually good enough on most platforms,
4382although strictly speaking using a type that also implies a memory fence
4383is required.
3994 4384
3995=item EV_H (h) 4385=item EV_H (h)
3996 4386
3997The name of the F<ev.h> header file used to include it. The default if 4387The name of the F<ev.h> header file used to include it. The default if
3998undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be 4388undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
4281And a F<ev_cpp.C> implementation file that contains libev proper and is compiled: 4671And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
4282 4672
4283 #include "ev_cpp.h" 4673 #include "ev_cpp.h"
4284 #include "ev.c" 4674 #include "ev.c"
4285 4675
4286=head1 INTERACTION WITH OTHER PROGRAMS OR LIBRARIES 4676=head1 INTERACTION WITH OTHER PROGRAMS, LIBRARIES OR THE ENVIRONMENT
4287 4677
4288=head2 THREADS AND COROUTINES 4678=head2 THREADS AND COROUTINES
4289 4679
4290=head3 THREADS 4680=head3 THREADS
4291 4681
4342default loop and triggering an C<ev_async> watcher from the default loop 4732default loop and triggering an C<ev_async> watcher from the default loop
4343watcher callback into the event loop interested in the signal. 4733watcher callback into the event loop interested in the signal.
4344 4734
4345=back 4735=back
4346 4736
4347=head4 THREAD LOCKING EXAMPLE 4737See also L<THREAD LOCKING EXAMPLE>.
4348
4349Here is a fictitious example of how to run an event loop in a different
4350thread than where callbacks are being invoked and watchers are
4351created/added/removed.
4352
4353For a real-world example, see the C<EV::Loop::Async> perl module,
4354which uses exactly this technique (which is suited for many high-level
4355languages).
4356
4357The example uses a pthread mutex to protect the loop data, a condition
4358variable to wait for callback invocations, an async watcher to notify the
4359event loop thread and an unspecified mechanism to wake up the main thread.
4360
4361First, you need to associate some data with the event loop:
4362
4363 typedef struct {
4364 mutex_t lock; /* global loop lock */
4365 ev_async async_w;
4366 thread_t tid;
4367 cond_t invoke_cv;
4368 } userdata;
4369
4370 void prepare_loop (EV_P)
4371 {
4372 // for simplicity, we use a static userdata struct.
4373 static userdata u;
4374
4375 ev_async_init (&u->async_w, async_cb);
4376 ev_async_start (EV_A_ &u->async_w);
4377
4378 pthread_mutex_init (&u->lock, 0);
4379 pthread_cond_init (&u->invoke_cv, 0);
4380
4381 // now associate this with the loop
4382 ev_set_userdata (EV_A_ u);
4383 ev_set_invoke_pending_cb (EV_A_ l_invoke);
4384 ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
4385
4386 // then create the thread running ev_loop
4387 pthread_create (&u->tid, 0, l_run, EV_A);
4388 }
4389
4390The callback for the C<ev_async> watcher does nothing: the watcher is used
4391solely to wake up the event loop so it takes notice of any new watchers
4392that might have been added:
4393
4394 static void
4395 async_cb (EV_P_ ev_async *w, int revents)
4396 {
4397 // just used for the side effects
4398 }
4399
4400The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
4401protecting the loop data, respectively.
4402
4403 static void
4404 l_release (EV_P)
4405 {
4406 userdata *u = ev_userdata (EV_A);
4407 pthread_mutex_unlock (&u->lock);
4408 }
4409
4410 static void
4411 l_acquire (EV_P)
4412 {
4413 userdata *u = ev_userdata (EV_A);
4414 pthread_mutex_lock (&u->lock);
4415 }
4416
4417The event loop thread first acquires the mutex, and then jumps straight
4418into C<ev_run>:
4419
4420 void *
4421 l_run (void *thr_arg)
4422 {
4423 struct ev_loop *loop = (struct ev_loop *)thr_arg;
4424
4425 l_acquire (EV_A);
4426 pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
4427 ev_run (EV_A_ 0);
4428 l_release (EV_A);
4429
4430 return 0;
4431 }
4432
4433Instead of invoking all pending watchers, the C<l_invoke> callback will
4434signal the main thread via some unspecified mechanism (signals? pipe
4435writes? C<Async::Interrupt>?) and then waits until all pending watchers
4436have been called (in a while loop because a) spurious wakeups are possible
4437and b) skipping inter-thread-communication when there are no pending
4438watchers is very beneficial):
4439
4440 static void
4441 l_invoke (EV_P)
4442 {
4443 userdata *u = ev_userdata (EV_A);
4444
4445 while (ev_pending_count (EV_A))
4446 {
4447 wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
4448 pthread_cond_wait (&u->invoke_cv, &u->lock);
4449 }
4450 }
4451
4452Now, whenever the main thread gets told to invoke pending watchers, it
4453will grab the lock, call C<ev_invoke_pending> and then signal the loop
4454thread to continue:
4455
4456 static void
4457 real_invoke_pending (EV_P)
4458 {
4459 userdata *u = ev_userdata (EV_A);
4460
4461 pthread_mutex_lock (&u->lock);
4462 ev_invoke_pending (EV_A);
4463 pthread_cond_signal (&u->invoke_cv);
4464 pthread_mutex_unlock (&u->lock);
4465 }
4466
4467Whenever you want to start/stop a watcher or do other modifications to an
4468event loop, you will now have to lock:
4469
4470 ev_timer timeout_watcher;
4471 userdata *u = ev_userdata (EV_A);
4472
4473 ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
4474
4475 pthread_mutex_lock (&u->lock);
4476 ev_timer_start (EV_A_ &timeout_watcher);
4477 ev_async_send (EV_A_ &u->async_w);
4478 pthread_mutex_unlock (&u->lock);
4479
4480Note that sending the C<ev_async> watcher is required because otherwise
4481an event loop currently blocking in the kernel will have no knowledge
4482about the newly added timer. By waking up the loop it will pick up any new
4483watchers in the next event loop iteration.
4484 4738
4485=head3 COROUTINES 4739=head3 COROUTINES
4486 4740
4487Libev is very accommodating to coroutines ("cooperative threads"): 4741Libev is very accommodating to coroutines ("cooperative threads"):
4488libev fully supports nesting calls to its functions from different 4742libev fully supports nesting calls to its functions from different
4653requires, and its I/O model is fundamentally incompatible with the POSIX 4907requires, and its I/O model is fundamentally incompatible with the POSIX
4654model. Libev still offers limited functionality on this platform in 4908model. Libev still offers limited functionality on this platform in
4655the form of the C<EVBACKEND_SELECT> backend, and only supports socket 4909the form of the C<EVBACKEND_SELECT> backend, and only supports socket
4656descriptors. This only applies when using Win32 natively, not when using 4910descriptors. This only applies when using Win32 natively, not when using
4657e.g. cygwin. Actually, it only applies to the microsofts own compilers, 4911e.g. cygwin. Actually, it only applies to the microsofts own compilers,
4658as every compielr comes with a slightly differently broken/incompatible 4912as every compiler comes with a slightly differently broken/incompatible
4659environment. 4913environment.
4660 4914
4661Lifting these limitations would basically require the full 4915Lifting these limitations would basically require the full
4662re-implementation of the I/O system. If you are into this kind of thing, 4916re-implementation of the I/O system. If you are into this kind of thing,
4663then note that glib does exactly that for you in a very portable way (note 4917then note that glib does exactly that for you in a very portable way (note
4757structure (guaranteed by POSIX but not by ISO C for example), but it also 5011structure (guaranteed by POSIX but not by ISO C for example), but it also
4758assumes that the same (machine) code can be used to call any watcher 5012assumes that the same (machine) code can be used to call any watcher
4759callback: The watcher callbacks have different type signatures, but libev 5013callback: The watcher callbacks have different type signatures, but libev
4760calls them using an C<ev_watcher *> internally. 5014calls them using an C<ev_watcher *> internally.
4761 5015
5016=item pointer accesses must be thread-atomic
5017
5018Accessing a pointer value must be atomic, it must both be readable and
5019writable in one piece - this is the case on all current architectures.
5020
4762=item C<sig_atomic_t volatile> must be thread-atomic as well 5021=item C<sig_atomic_t volatile> must be thread-atomic as well
4763 5022
4764The type C<sig_atomic_t volatile> (or whatever is defined as 5023The type C<sig_atomic_t volatile> (or whatever is defined as
4765C<EV_ATOMIC_T>) must be atomic with respect to accesses from different 5024C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
4766threads. This is not part of the specification for C<sig_atomic_t>, but is 5025threads. This is not part of the specification for C<sig_atomic_t>, but is
4791 5050
4792The type C<double> is used to represent timestamps. It is required to 5051The type C<double> is used to represent timestamps. It is required to
4793have at least 51 bits of mantissa (and 9 bits of exponent), which is 5052have at least 51 bits of mantissa (and 9 bits of exponent), which is
4794good enough for at least into the year 4000 with millisecond accuracy 5053good enough for at least into the year 4000 with millisecond accuracy
4795(the design goal for libev). This requirement is overfulfilled by 5054(the design goal for libev). This requirement is overfulfilled by
4796implementations using IEEE 754, which is basically all existing ones. With 5055implementations using IEEE 754, which is basically all existing ones.
5056
4797IEEE 754 doubles, you get microsecond accuracy until at least 2200. 5057With IEEE 754 doubles, you get microsecond accuracy until at least the
5058year 2255 (and millisecond accuray till the year 287396 - by then, libev
5059is either obsolete or somebody patched it to use C<long double> or
5060something like that, just kidding).
4798 5061
4799=back 5062=back
4800 5063
4801If you know of other additional requirements drop me a note. 5064If you know of other additional requirements drop me a note.
4802 5065
4864=item Processing ev_async_send: O(number_of_async_watchers) 5127=item Processing ev_async_send: O(number_of_async_watchers)
4865 5128
4866=item Processing signals: O(max_signal_number) 5129=item Processing signals: O(max_signal_number)
4867 5130
4868Sending involves a system call I<iff> there were no other C<ev_async_send> 5131Sending involves a system call I<iff> there were no other C<ev_async_send>
4869calls in the current loop iteration. Checking for async and signal events 5132calls in the current loop iteration and the loop is currently
5133blocked. Checking for async and signal events involves iterating over all
4870involves iterating over all running async watchers or all signal numbers. 5134running async watchers or all signal numbers.
4871 5135
4872=back 5136=back
4873 5137
4874 5138
4875=head1 PORTING FROM LIBEV 3.X TO 4.X 5139=head1 PORTING FROM LIBEV 3.X TO 4.X
4876 5140
4877The major version 4 introduced some minor incompatible changes to the API. 5141The major version 4 introduced some incompatible changes to the API.
4878 5142
4879At the moment, the C<ev.h> header file tries to implement superficial 5143At the moment, the C<ev.h> header file provides compatibility definitions
4880compatibility, so most programs should still compile. Those might be 5144for all changes, so most programs should still compile. The compatibility
4881removed in later versions of libev, so better update early than late. 5145layer might be removed in later versions of libev, so better update to the
5146new API early than late.
4882 5147
4883=over 4 5148=over 4
5149
5150=item C<EV_COMPAT3> backwards compatibility mechanism
5151
5152The backward compatibility mechanism can be controlled by
5153C<EV_COMPAT3>. See L<PREPROCESSOR SYMBOLS/MACROS> in the L<EMBEDDING>
5154section.
4884 5155
4885=item C<ev_default_destroy> and C<ev_default_fork> have been removed 5156=item C<ev_default_destroy> and C<ev_default_fork> have been removed
4886 5157
4887These calls can be replaced easily by their C<ev_loop_xxx> counterparts: 5158These calls can be replaced easily by their C<ev_loop_xxx> counterparts:
4888 5159
4914ev_loop> anymore and C<EV_TIMER> now follows the same naming scheme 5185ev_loop> anymore and C<EV_TIMER> now follows the same naming scheme
4915as all other watcher types. Note that C<ev_loop_fork> is still called 5186as all other watcher types. Note that C<ev_loop_fork> is still called
4916C<ev_loop_fork> because it would otherwise clash with the C<ev_fork> 5187C<ev_loop_fork> because it would otherwise clash with the C<ev_fork>
4917typedef. 5188typedef.
4918 5189
4919=item C<EV_COMPAT3> backwards compatibility mechanism
4920
4921The backward compatibility mechanism can be controlled by
4922C<EV_COMPAT3>. See L<PREPROCESSOR SYMBOLS/MACROS> in the L<EMBEDDING>
4923section.
4924
4925=item C<EV_MINIMAL> mechanism replaced by C<EV_FEATURES> 5190=item C<EV_MINIMAL> mechanism replaced by C<EV_FEATURES>
4926 5191
4927The preprocessor symbol C<EV_MINIMAL> has been replaced by a different 5192The preprocessor symbol C<EV_MINIMAL> has been replaced by a different
4928mechanism, C<EV_FEATURES>. Programs using C<EV_MINIMAL> usually compile 5193mechanism, C<EV_FEATURES>. Programs using C<EV_MINIMAL> usually compile
4929and work, but the library code will of course be larger. 5194and work, but the library code will of course be larger.
4991The physical time that is observed. It is apparently strictly monotonic :) 5256The physical time that is observed. It is apparently strictly monotonic :)
4992 5257
4993=item wall-clock time 5258=item wall-clock time
4994 5259
4995The time and date as shown on clocks. Unlike real time, it can actually 5260The time and date as shown on clocks. Unlike real time, it can actually
4996be wrong and jump forwards and backwards, e.g. when the you adjust your 5261be wrong and jump forwards and backwards, e.g. when you adjust your
4997clock. 5262clock.
4998 5263
4999=item watcher 5264=item watcher
5000 5265
5001A data structure that describes interest in certain events. Watchers need 5266A data structure that describes interest in certain events. Watchers need
5003 5268
5004=back 5269=back
5005 5270
5006=head1 AUTHOR 5271=head1 AUTHOR
5007 5272
5008Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael Magnusson. 5273Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael
5274Magnusson and Emanuele Giaquinta, and minor corrections by many others.
5009 5275

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines