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

Comparing libev/ev.pod (file contents):
Revision 1.46 by root, Mon Nov 26 10:20:43 2007 UTC vs.
Revision 1.60 by root, Wed Nov 28 18:29:30 2007 UTC

3libev - a high performance full-featured event loop written in C 3libev - a high performance full-featured event loop written in C
4 4
5=head1 SYNOPSIS 5=head1 SYNOPSIS
6 6
7 #include <ev.h> 7 #include <ev.h>
8
9=head1 EXAMPLE PROGRAM
10
11 #include <ev.h>
12
13 ev_io stdin_watcher;
14 ev_timer timeout_watcher;
15
16 /* called when data readable on stdin */
17 static void
18 stdin_cb (EV_P_ struct ev_io *w, int revents)
19 {
20 /* puts ("stdin ready"); */
21 ev_io_stop (EV_A_ w); /* just a syntax example */
22 ev_unloop (EV_A_ EVUNLOOP_ALL); /* leave all loop calls */
23 }
24
25 static void
26 timeout_cb (EV_P_ struct ev_timer *w, int revents)
27 {
28 /* puts ("timeout"); */
29 ev_unloop (EV_A_ EVUNLOOP_ONE); /* leave one loop call */
30 }
31
32 int
33 main (void)
34 {
35 struct ev_loop *loop = ev_default_loop (0);
36
37 /* initialise an io watcher, then start it */
38 ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
39 ev_io_start (loop, &stdin_watcher);
40
41 /* simple non-repeating 5.5 second timeout */
42 ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
43 ev_timer_start (loop, &timeout_watcher);
44
45 /* loop till timeout or data ready */
46 ev_loop (loop, 0);
47
48 return 0;
49 }
8 50
9=head1 DESCRIPTION 51=head1 DESCRIPTION
10 52
11Libev is an event loop: you register interest in certain events (such as a 53Libev is an event loop: you register interest in certain events (such as a
12file descriptor being readable or a timeout occuring), and it will manage 54file descriptor being readable or a timeout occuring), and it will manage
21details of the event, and then hand it over to libev by I<starting> the 63details of the event, and then hand it over to libev by I<starting> the
22watcher. 64watcher.
23 65
24=head1 FEATURES 66=head1 FEATURES
25 67
26Libev supports select, poll, the linux-specific epoll and the bsd-specific 68Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the
27kqueue mechanisms for file descriptor events, relative timers, absolute 69BSD-specific C<kqueue> and the Solaris-specific event port mechanisms
28timers with customised rescheduling, signal events, process status change 70for file descriptor events (C<ev_io>), the Linux C<inotify> interface
29events (related to SIGCHLD), and event watchers dealing with the event 71(for C<ev_stat>), relative timers (C<ev_timer>), absolute timers
30loop mechanism itself (idle, prepare and check watchers). It also is quite 72with customised rescheduling (C<ev_periodic>), synchronous signals
73(C<ev_signal>), process status change events (C<ev_child>), and event
74watchers dealing with the event loop mechanism itself (C<ev_idle>,
75C<ev_embed>, C<ev_prepare> and C<ev_check> watchers) as well as
76file watchers (C<ev_stat>) and even limited support for fork events
77(C<ev_fork>).
78
79It also is quite fast (see this
31fast (see this L<benchmark|http://libev.schmorp.de/bench.html> comparing 80L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent
32it to libevent for example). 81for example).
33 82
34=head1 CONVENTIONS 83=head1 CONVENTIONS
35 84
36Libev is very configurable. In this manual the default configuration 85Libev is very configurable. In this manual the default configuration will
37will be described, which supports multiple event loops. For more info 86be described, which supports multiple event loops. For more info about
38about various configuration options please have a look at the file 87various configuration options please have a look at B<EMBED> section in
39F<README.embed> in the libev distribution. If libev was configured without 88this manual. If libev was configured without support for multiple event
40support for multiple event loops, then all functions taking an initial 89loops, then all functions taking an initial argument of name C<loop>
41argument of name C<loop> (which is always of type C<struct ev_loop *>) 90(which is always of type C<struct ev_loop *>) will not have this argument.
42will not have this argument.
43 91
44=head1 TIME REPRESENTATION 92=head1 TIME REPRESENTATION
45 93
46Libev represents time as a single floating point number, representing the 94Libev represents time as a single floating point number, representing the
47(fractional) number of seconds since the (POSIX) epoch (somewhere near 95(fractional) number of seconds since the (POSIX) epoch (somewhere near
48the beginning of 1970, details are complicated, don't ask). This type is 96the beginning of 1970, details are complicated, don't ask). This type is
49called C<ev_tstamp>, which is what you should use too. It usually aliases 97called C<ev_tstamp>, which is what you should use too. It usually aliases
50to the C<double> type in C, and when you need to do any calculations on 98to the C<double> type in C, and when you need to do any calculations on
51it, you should treat it as such. 99it, you should treat it as such.
52 100
53
54=head1 GLOBAL FUNCTIONS 101=head1 GLOBAL FUNCTIONS
55 102
56These functions can be called anytime, even before initialising the 103These functions can be called anytime, even before initialising the
57library in any way. 104library in any way.
58 105
77Usually, it's a good idea to terminate if the major versions mismatch, 124Usually, it's a good idea to terminate if the major versions mismatch,
78as this indicates an incompatible change. Minor versions are usually 125as this indicates an incompatible change. Minor versions are usually
79compatible to older versions, so a larger minor version alone is usually 126compatible to older versions, so a larger minor version alone is usually
80not a problem. 127not a problem.
81 128
82Example: make sure we haven't accidentally been linked against the wrong 129Example: Make sure we haven't accidentally been linked against the wrong
83version: 130version.
84 131
85 assert (("libev version mismatch", 132 assert (("libev version mismatch",
86 ev_version_major () == EV_VERSION_MAJOR 133 ev_version_major () == EV_VERSION_MAJOR
87 && ev_version_minor () >= EV_VERSION_MINOR)); 134 && ev_version_minor () >= EV_VERSION_MINOR));
88 135
118 165
119See the description of C<ev_embed> watchers for more info. 166See the description of C<ev_embed> watchers for more info.
120 167
121=item ev_set_allocator (void *(*cb)(void *ptr, long size)) 168=item ev_set_allocator (void *(*cb)(void *ptr, long size))
122 169
123Sets the allocation function to use (the prototype is similar to the 170Sets the allocation function to use (the prototype is similar - the
124realloc C function, the semantics are identical). It is used to allocate 171semantics is identical - to the realloc C function). It is used to
125and free memory (no surprises here). If it returns zero when memory 172allocate and free memory (no surprises here). If it returns zero when
126needs to be allocated, the library might abort or take some potentially 173memory needs to be allocated, the library might abort or take some
127destructive action. The default is your system realloc function. 174potentially destructive action. The default is your system realloc
175function.
128 176
129You could override this function in high-availability programs to, say, 177You could override this function in high-availability programs to, say,
130free some memory if it cannot allocate memory, to use a special allocator, 178free some memory if it cannot allocate memory, to use a special allocator,
131or even to sleep a while and retry until some memory is available. 179or even to sleep a while and retry until some memory is available.
132 180
133Example: replace the libev allocator with one that waits a bit and then 181Example: Replace the libev allocator with one that waits a bit and then
134retries: better than mine). 182retries).
135 183
136 static void * 184 static void *
137 persistent_realloc (void *ptr, long size) 185 persistent_realloc (void *ptr, size_t size)
138 { 186 {
139 for (;;) 187 for (;;)
140 { 188 {
141 void *newptr = realloc (ptr, size); 189 void *newptr = realloc (ptr, size);
142 190
158callback is set, then libev will expect it to remedy the sitution, no 206callback is set, then libev will expect it to remedy the sitution, no
159matter what, when it returns. That is, libev will generally retry the 207matter what, when it returns. That is, libev will generally retry the
160requested operation, or, if the condition doesn't go away, do bad stuff 208requested operation, or, if the condition doesn't go away, do bad stuff
161(such as abort). 209(such as abort).
162 210
163Example: do the same thing as libev does internally: 211Example: This is basically the same thing that libev does internally, too.
164 212
165 static void 213 static void
166 fatal_error (const char *msg) 214 fatal_error (const char *msg)
167 { 215 {
168 perror (msg); 216 perror (msg);
314Similar to C<ev_default_loop>, but always creates a new event loop that is 362Similar to C<ev_default_loop>, but always creates a new event loop that is
315always distinct from the default loop. Unlike the default loop, it cannot 363always distinct from the default loop. Unlike the default loop, it cannot
316handle signal and child watchers, and attempts to do so will be greeted by 364handle signal and child watchers, and attempts to do so will be greeted by
317undefined behaviour (or a failed assertion if assertions are enabled). 365undefined behaviour (or a failed assertion if assertions are enabled).
318 366
319Example: try to create a event loop that uses epoll and nothing else. 367Example: Try to create a event loop that uses epoll and nothing else.
320 368
321 struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV); 369 struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
322 if (!epoller) 370 if (!epoller)
323 fatal ("no epoll found here, maybe it hides under your chair"); 371 fatal ("no epoll found here, maybe it hides under your chair");
324 372
423 Signals and child watchers are implemented as I/O watchers, and will 471 Signals and child watchers are implemented as I/O watchers, and will
424 be handled here by queueing them when their watcher gets executed. 472 be handled here by queueing them when their watcher gets executed.
425 - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK 473 - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
426 were used, return, otherwise continue with step *. 474 were used, return, otherwise continue with step *.
427 475
428Example: queue some jobs and then loop until no events are outsanding 476Example: Queue some jobs and then loop until no events are outsanding
429anymore. 477anymore.
430 478
431 ... queue jobs here, make sure they register event watchers as long 479 ... queue jobs here, make sure they register event watchers as long
432 ... as they still have work to do (even an idle watcher will do..) 480 ... as they still have work to do (even an idle watcher will do..)
433 ev_loop (my_loop, 0); 481 ev_loop (my_loop, 0);
453visible to the libev user and should not keep C<ev_loop> from exiting if 501visible to the libev user and should not keep C<ev_loop> from exiting if
454no event watchers registered by it are active. It is also an excellent 502no event watchers registered by it are active. It is also an excellent
455way to do this for generic recurring timers or from within third-party 503way to do this for generic recurring timers or from within third-party
456libraries. Just remember to I<unref after start> and I<ref before stop>. 504libraries. Just remember to I<unref after start> and I<ref before stop>.
457 505
458Example: create a signal watcher, but keep it from keeping C<ev_loop> 506Example: Create a signal watcher, but keep it from keeping C<ev_loop>
459running when nothing else is active. 507running when nothing else is active.
460 508
461 struct dv_signal exitsig; 509 struct ev_signal exitsig;
462 ev_signal_init (&exitsig, sig_cb, SIGINT); 510 ev_signal_init (&exitsig, sig_cb, SIGINT);
463 ev_signal_start (myloop, &exitsig); 511 ev_signal_start (loop, &exitsig);
464 evf_unref (myloop); 512 evf_unref (loop);
465 513
466Example: for some weird reason, unregister the above signal handler again. 514Example: For some weird reason, unregister the above signal handler again.
467 515
468 ev_ref (myloop); 516 ev_ref (loop);
469 ev_signal_stop (myloop, &exitsig); 517 ev_signal_stop (loop, &exitsig);
470 518
471=back 519=back
472 520
473 521
474=head1 ANATOMY OF A WATCHER 522=head1 ANATOMY OF A WATCHER
544The signal specified in the C<ev_signal> watcher has been received by a thread. 592The signal specified in the C<ev_signal> watcher has been received by a thread.
545 593
546=item C<EV_CHILD> 594=item C<EV_CHILD>
547 595
548The pid specified in the C<ev_child> watcher has received a status change. 596The pid specified in the C<ev_child> watcher has received a status change.
597
598=item C<EV_STAT>
599
600The path specified in the C<ev_stat> watcher changed its attributes somehow.
549 601
550=item C<EV_IDLE> 602=item C<EV_IDLE>
551 603
552The C<ev_idle> watcher has determined that you have nothing better to do. 604The C<ev_idle> watcher has determined that you have nothing better to do.
553 605
561received events. Callbacks of both watcher types can start and stop as 613received events. Callbacks of both watcher types can start and stop as
562many watchers as they want, and all of them will be taken into account 614many watchers as they want, and all of them will be taken into account
563(for example, a C<ev_prepare> watcher might start an idle watcher to keep 615(for example, a C<ev_prepare> watcher might start an idle watcher to keep
564C<ev_loop> from blocking). 616C<ev_loop> from blocking).
565 617
618=item C<EV_EMBED>
619
620The embedded event loop specified in the C<ev_embed> watcher needs attention.
621
622=item C<EV_FORK>
623
624The event loop has been resumed in the child process after fork (see
625C<ev_fork>).
626
566=item C<EV_ERROR> 627=item C<EV_ERROR>
567 628
568An unspecified error has occured, the watcher has been stopped. This might 629An unspecified error has occured, the watcher has been stopped. This might
569happen because the watcher could not be properly started because libev 630happen because the watcher could not be properly started because libev
570ran out of memory, a file descriptor was found to be closed or any other 631ran out of memory, a file descriptor was found to be closed or any other
644events but its callback has not yet been invoked). As long as a watcher 705events but its callback has not yet been invoked). As long as a watcher
645is pending (but not active) you must not call an init function on it (but 706is pending (but not active) you must not call an init function on it (but
646C<ev_TYPE_set> is safe) and you must make sure the watcher is available to 707C<ev_TYPE_set> is safe) and you must make sure the watcher is available to
647libev (e.g. you cnanot C<free ()> it). 708libev (e.g. you cnanot C<free ()> it).
648 709
649=item callback = ev_cb (ev_TYPE *watcher) 710=item callback ev_cb (ev_TYPE *watcher)
650 711
651Returns the callback currently set on the watcher. 712Returns the callback currently set on the watcher.
652 713
653=item ev_cb_set (ev_TYPE *watcher, callback) 714=item ev_cb_set (ev_TYPE *watcher, callback)
654 715
682 { 743 {
683 struct my_io *w = (struct my_io *)w_; 744 struct my_io *w = (struct my_io *)w_;
684 ... 745 ...
685 } 746 }
686 747
687More interesting and less C-conformant ways of catsing your callback type 748More interesting and less C-conformant ways of casting your callback type
688have been omitted.... 749instead have been omitted.
750
751Another common scenario is having some data structure with multiple
752watchers:
753
754 struct my_biggy
755 {
756 int some_data;
757 ev_timer t1;
758 ev_timer t2;
759 }
760
761In this case getting the pointer to C<my_biggy> is a bit more complicated,
762you need to use C<offsetof>:
763
764 #include <stddef.h>
765
766 static void
767 t1_cb (EV_P_ struct ev_timer *w, int revents)
768 {
769 struct my_biggy big = (struct my_biggy *
770 (((char *)w) - offsetof (struct my_biggy, t1));
771 }
772
773 static void
774 t2_cb (EV_P_ struct ev_timer *w, int revents)
775 {
776 struct my_biggy big = (struct my_biggy *
777 (((char *)w) - offsetof (struct my_biggy, t2));
778 }
689 779
690 780
691=head1 WATCHER TYPES 781=head1 WATCHER TYPES
692 782
693This section describes each watcher in detail, but will not repeat 783This section describes each watcher in detail, but will not repeat
694information given in the last section. 784information given in the last section. Any initialisation/set macros,
785functions and members specific to the watcher type are explained.
786
787Members are additionally marked with either I<[read-only]>, meaning that,
788while the watcher is active, you can look at the member and expect some
789sensible content, but you must not modify it (you can modify it while the
790watcher is stopped to your hearts content), or I<[read-write]>, which
791means you can expect it to have some sensible content while the watcher
792is active, but you can also modify it. Modifying it may not do something
793sensible or take immediate effect (or do anything at all), but libev will
794not crash or malfunction in any way.
695 795
696 796
697=head2 C<ev_io> - is this file descriptor readable or writable? 797=head2 C<ev_io> - is this file descriptor readable or writable?
698 798
699I/O watchers check whether a file descriptor is readable or writable 799I/O watchers check whether a file descriptor is readable or writable
742 842
743Configures an C<ev_io> watcher. The C<fd> is the file descriptor to 843Configures an C<ev_io> watcher. The C<fd> is the file descriptor to
744rceeive events for and events is either C<EV_READ>, C<EV_WRITE> or 844rceeive events for and events is either C<EV_READ>, C<EV_WRITE> or
745C<EV_READ | EV_WRITE> to receive the given events. 845C<EV_READ | EV_WRITE> to receive the given events.
746 846
847=item int fd [read-only]
848
849The file descriptor being watched.
850
851=item int events [read-only]
852
853The events being watched.
854
747=back 855=back
748 856
749Example: call C<stdin_readable_cb> when STDIN_FILENO has become, well 857Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
750readable, but only once. Since it is likely line-buffered, you could 858readable, but only once. Since it is likely line-buffered, you could
751attempt to read a whole line in the callback: 859attempt to read a whole line in the callback.
752 860
753 static void 861 static void
754 stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents) 862 stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)
755 { 863 {
756 ev_io_stop (loop, w); 864 ev_io_stop (loop, w);
814 922
815If the timer is repeating, either start it if necessary (with the repeat 923If the timer is repeating, either start it if necessary (with the repeat
816value), or reset the running timer to the repeat value. 924value), or reset the running timer to the repeat value.
817 925
818This sounds a bit complicated, but here is a useful and typical 926This sounds a bit complicated, but here is a useful and typical
819example: Imagine you have a tcp connection and you want a so-called idle 927example: Imagine you have a tcp connection and you want a so-called
820timeout, that is, you want to be called when there have been, say, 60 928idle timeout, that is, you want to be called when there have been,
821seconds of inactivity on the socket. The easiest way to do this is to 929say, 60 seconds of inactivity on the socket. The easiest way to do
822configure an C<ev_timer> with after=repeat=60 and calling ev_timer_again each 930this is to configure an C<ev_timer> with C<after>=C<repeat>=C<60> and calling
823time you successfully read or write some data. If you go into an idle 931C<ev_timer_again> each time you successfully read or write some data. If
824state where you do not expect data to travel on the socket, you can stop 932you go into an idle state where you do not expect data to travel on the
825the timer, and again will automatically restart it if need be. 933socket, you can stop the timer, and again will automatically restart it if
934need be.
935
936You can also ignore the C<after> value and C<ev_timer_start> altogether
937and only ever use the C<repeat> value:
938
939 ev_timer_init (timer, callback, 0., 5.);
940 ev_timer_again (loop, timer);
941 ...
942 timer->again = 17.;
943 ev_timer_again (loop, timer);
944 ...
945 timer->again = 10.;
946 ev_timer_again (loop, timer);
947
948This is more efficient then stopping/starting the timer eahc time you want
949to modify its timeout value.
950
951=item ev_tstamp repeat [read-write]
952
953The current C<repeat> value. Will be used each time the watcher times out
954or C<ev_timer_again> is called and determines the next timeout (if any),
955which is also when any modifications are taken into account.
826 956
827=back 957=back
828 958
829Example: create a timer that fires after 60 seconds. 959Example: Create a timer that fires after 60 seconds.
830 960
831 static void 961 static void
832 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents) 962 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
833 { 963 {
834 .. one minute over, w is actually stopped right here 964 .. one minute over, w is actually stopped right here
836 966
837 struct ev_timer mytimer; 967 struct ev_timer mytimer;
838 ev_timer_init (&mytimer, one_minute_cb, 60., 0.); 968 ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
839 ev_timer_start (loop, &mytimer); 969 ev_timer_start (loop, &mytimer);
840 970
841Example: create a timeout timer that times out after 10 seconds of 971Example: Create a timeout timer that times out after 10 seconds of
842inactivity. 972inactivity.
843 973
844 static void 974 static void
845 timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents) 975 timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
846 { 976 {
957Simply stops and restarts the periodic watcher again. This is only useful 1087Simply stops and restarts the periodic watcher again. This is only useful
958when you changed some parameters or the reschedule callback would return 1088when you changed some parameters or the reschedule callback would return
959a different time than the last time it was called (e.g. in a crond like 1089a different time than the last time it was called (e.g. in a crond like
960program when the crontabs have changed). 1090program when the crontabs have changed).
961 1091
1092=item ev_tstamp interval [read-write]
1093
1094The current interval value. Can be modified any time, but changes only
1095take effect when the periodic timer fires or C<ev_periodic_again> is being
1096called.
1097
1098=item ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]
1099
1100The current reschedule callback, or C<0>, if this functionality is
1101switched off. Can be changed any time, but changes only take effect when
1102the periodic timer fires or C<ev_periodic_again> is being called.
1103
962=back 1104=back
963 1105
964Example: call a callback every hour, or, more precisely, whenever the 1106Example: Call a callback every hour, or, more precisely, whenever the
965system clock is divisible by 3600. The callback invocation times have 1107system clock is divisible by 3600. The callback invocation times have
966potentially a lot of jittering, but good long-term stability. 1108potentially a lot of jittering, but good long-term stability.
967 1109
968 static void 1110 static void
969 clock_cb (struct ev_loop *loop, struct ev_io *w, int revents) 1111 clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)
973 1115
974 struct ev_periodic hourly_tick; 1116 struct ev_periodic hourly_tick;
975 ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0); 1117 ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
976 ev_periodic_start (loop, &hourly_tick); 1118 ev_periodic_start (loop, &hourly_tick);
977 1119
978Example: the same as above, but use a reschedule callback to do it: 1120Example: The same as above, but use a reschedule callback to do it:
979 1121
980 #include <math.h> 1122 #include <math.h>
981 1123
982 static ev_tstamp 1124 static ev_tstamp
983 my_scheduler_cb (struct ev_periodic *w, ev_tstamp now) 1125 my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)
985 return fmod (now, 3600.) + 3600.; 1127 return fmod (now, 3600.) + 3600.;
986 } 1128 }
987 1129
988 ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb); 1130 ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
989 1131
990Example: call a callback every hour, starting now: 1132Example: Call a callback every hour, starting now:
991 1133
992 struct ev_periodic hourly_tick; 1134 struct ev_periodic hourly_tick;
993 ev_periodic_init (&hourly_tick, clock_cb, 1135 ev_periodic_init (&hourly_tick, clock_cb,
994 fmod (ev_now (loop), 3600.), 3600., 0); 1136 fmod (ev_now (loop), 3600.), 3600., 0);
995 ev_periodic_start (loop, &hourly_tick); 1137 ev_periodic_start (loop, &hourly_tick);
1016=item ev_signal_set (ev_signal *, int signum) 1158=item ev_signal_set (ev_signal *, int signum)
1017 1159
1018Configures the watcher to trigger on the given signal number (usually one 1160Configures the watcher to trigger on the given signal number (usually one
1019of the C<SIGxxx> constants). 1161of the C<SIGxxx> constants).
1020 1162
1163=item int signum [read-only]
1164
1165The signal the watcher watches out for.
1166
1021=back 1167=back
1022 1168
1023 1169
1024=head2 C<ev_child> - watch out for process status changes 1170=head2 C<ev_child> - watch out for process status changes
1025 1171
1037at the C<rstatus> member of the C<ev_child> watcher structure to see 1183at the C<rstatus> member of the C<ev_child> watcher structure to see
1038the status word (use the macros from C<sys/wait.h> and see your systems 1184the status word (use the macros from C<sys/wait.h> and see your systems
1039C<waitpid> documentation). The C<rpid> member contains the pid of the 1185C<waitpid> documentation). The C<rpid> member contains the pid of the
1040process causing the status change. 1186process causing the status change.
1041 1187
1188=item int pid [read-only]
1189
1190The process id this watcher watches out for, or C<0>, meaning any process id.
1191
1192=item int rpid [read-write]
1193
1194The process id that detected a status change.
1195
1196=item int rstatus [read-write]
1197
1198The process exit/trace status caused by C<rpid> (see your systems
1199C<waitpid> and C<sys/wait.h> documentation for details).
1200
1042=back 1201=back
1043 1202
1044Example: try to exit cleanly on SIGINT and SIGTERM. 1203Example: Try to exit cleanly on SIGINT and SIGTERM.
1045 1204
1046 static void 1205 static void
1047 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents) 1206 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
1048 { 1207 {
1049 ev_unloop (loop, EVUNLOOP_ALL); 1208 ev_unloop (loop, EVUNLOOP_ALL);
1050 } 1209 }
1051 1210
1052 struct ev_signal signal_watcher; 1211 struct ev_signal signal_watcher;
1053 ev_signal_init (&signal_watcher, sigint_cb, SIGINT); 1212 ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
1054 ev_signal_start (loop, &sigint_cb); 1213 ev_signal_start (loop, &sigint_cb);
1214
1215
1216=head2 C<ev_stat> - did the file attributes just change?
1217
1218This watches a filesystem path for attribute changes. That is, it calls
1219C<stat> regularly (or when the OS says it changed) and sees if it changed
1220compared to the last time, invoking the callback if it did.
1221
1222The path does not need to exist: changing from "path exists" to "path does
1223not exist" is a status change like any other. The condition "path does
1224not exist" is signified by the C<st_nlink> field being zero (which is
1225otherwise always forced to be at least one) and all the other fields of
1226the stat buffer having unspecified contents.
1227
1228The path I<should> be absolute and I<must not> end in a slash. If it is
1229relative and your working directory changes, the behaviour is undefined.
1230
1231Since there is no standard to do this, the portable implementation simply
1232calls C<stat (2)> regularly on the path to see if it changed somehow. You
1233can specify a recommended polling interval for this case. If you specify
1234a polling interval of C<0> (highly recommended!) then a I<suitable,
1235unspecified default> value will be used (which you can expect to be around
1236five seconds, although this might change dynamically). Libev will also
1237impose a minimum interval which is currently around C<0.1>, but thats
1238usually overkill.
1239
1240This watcher type is not meant for massive numbers of stat watchers,
1241as even with OS-supported change notifications, this can be
1242resource-intensive.
1243
1244At the time of this writing, only the Linux inotify interface is
1245implemented (implementing kqueue support is left as an exercise for the
1246reader). Inotify will be used to give hints only and should not change the
1247semantics of C<ev_stat> watchers, which means that libev sometimes needs
1248to fall back to regular polling again even with inotify, but changes are
1249usually detected immediately, and if the file exists there will be no
1250polling.
1251
1252=over 4
1253
1254=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
1255
1256=item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)
1257
1258Configures the watcher to wait for status changes of the given
1259C<path>. The C<interval> is a hint on how quickly a change is expected to
1260be detected and should normally be specified as C<0> to let libev choose
1261a suitable value. The memory pointed to by C<path> must point to the same
1262path for as long as the watcher is active.
1263
1264The callback will be receive C<EV_STAT> when a change was detected,
1265relative to the attributes at the time the watcher was started (or the
1266last change was detected).
1267
1268=item ev_stat_stat (ev_stat *)
1269
1270Updates the stat buffer immediately with new values. If you change the
1271watched path in your callback, you could call this fucntion to avoid
1272detecting this change (while introducing a race condition). Can also be
1273useful simply to find out the new values.
1274
1275=item ev_statdata attr [read-only]
1276
1277The most-recently detected attributes of the file. Although the type is of
1278C<ev_statdata>, this is usually the (or one of the) C<struct stat> types
1279suitable for your system. If the C<st_nlink> member is C<0>, then there
1280was some error while C<stat>ing the file.
1281
1282=item ev_statdata prev [read-only]
1283
1284The previous attributes of the file. The callback gets invoked whenever
1285C<prev> != C<attr>.
1286
1287=item ev_tstamp interval [read-only]
1288
1289The specified interval.
1290
1291=item const char *path [read-only]
1292
1293The filesystem path that is being watched.
1294
1295=back
1296
1297Example: Watch C</etc/passwd> for attribute changes.
1298
1299 static void
1300 passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1301 {
1302 /* /etc/passwd changed in some way */
1303 if (w->attr.st_nlink)
1304 {
1305 printf ("passwd current size %ld\n", (long)w->attr.st_size);
1306 printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
1307 printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
1308 }
1309 else
1310 /* you shalt not abuse printf for puts */
1311 puts ("wow, /etc/passwd is not there, expect problems. "
1312 "if this is windows, they already arrived\n");
1313 }
1314
1315 ...
1316 ev_stat passwd;
1317
1318 ev_stat_init (&passwd, passwd_cb, "/etc/passwd");
1319 ev_stat_start (loop, &passwd);
1055 1320
1056 1321
1057=head2 C<ev_idle> - when you've got nothing better to do... 1322=head2 C<ev_idle> - when you've got nothing better to do...
1058 1323
1059Idle watchers trigger events when there are no other events are pending 1324Idle watchers trigger events when there are no other events are pending
1080kind. There is a C<ev_idle_set> macro, but using it is utterly pointless, 1345kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
1081believe me. 1346believe me.
1082 1347
1083=back 1348=back
1084 1349
1085Example: dynamically allocate an C<ev_idle>, start it, and in the 1350Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
1086callback, free it. Alos, use no error checking, as usual. 1351callback, free it. Also, use no error checking, as usual.
1087 1352
1088 static void 1353 static void
1089 idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents) 1354 idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
1090 { 1355 {
1091 free (w); 1356 free (w);
1292 1557
1293Make a single, non-blocking sweep over the embedded loop. This works 1558Make a single, non-blocking sweep over the embedded loop. This works
1294similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most 1559similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
1295apropriate way for embedded loops. 1560apropriate way for embedded loops.
1296 1561
1562=item struct ev_loop *loop [read-only]
1563
1564The embedded event loop.
1565
1566=back
1567
1568
1569=head2 C<ev_fork> - the audacity to resume the event loop after a fork
1570
1571Fork watchers are called when a C<fork ()> was detected (usually because
1572whoever is a good citizen cared to tell libev about it by calling
1573C<ev_default_fork> or C<ev_loop_fork>). The invocation is done before the
1574event loop blocks next and before C<ev_check> watchers are being called,
1575and only in the child after the fork. If whoever good citizen calling
1576C<ev_default_fork> cheats and calls it in the wrong process, the fork
1577handlers will be invoked, too, of course.
1578
1579=over 4
1580
1581=item ev_fork_init (ev_signal *, callback)
1582
1583Initialises and configures the fork watcher - it has no parameters of any
1584kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
1585believe me.
1586
1297=back 1587=back
1298 1588
1299 1589
1300=head1 OTHER FUNCTIONS 1590=head1 OTHER FUNCTIONS
1301 1591
1463 1753
1464=item w->sweep () C<ev::embed> only 1754=item w->sweep () C<ev::embed> only
1465 1755
1466Invokes C<ev_embed_sweep>. 1756Invokes C<ev_embed_sweep>.
1467 1757
1758=item w->update () C<ev::stat> only
1759
1760Invokes C<ev_stat_stat>.
1761
1468=back 1762=back
1469 1763
1470=back 1764=back
1471 1765
1472Example: Define a class with an IO and idle watcher, start one of them in 1766Example: Define a class with an IO and idle watcher, start one of them in
1484 : io (this, &myclass::io_cb), 1778 : io (this, &myclass::io_cb),
1485 idle (this, &myclass::idle_cb) 1779 idle (this, &myclass::idle_cb)
1486 { 1780 {
1487 io.start (fd, ev::READ); 1781 io.start (fd, ev::READ);
1488 } 1782 }
1783
1784
1785=head1 MACRO MAGIC
1786
1787Libev can be compiled with a variety of options, the most fundemantal is
1788C<EV_MULTIPLICITY>. This option determines wether (most) functions and
1789callbacks have an initial C<struct ev_loop *> argument.
1790
1791To make it easier to write programs that cope with either variant, the
1792following macros are defined:
1793
1794=over 4
1795
1796=item C<EV_A>, C<EV_A_>
1797
1798This provides the loop I<argument> for functions, if one is required ("ev
1799loop argument"). The C<EV_A> form is used when this is the sole argument,
1800C<EV_A_> is used when other arguments are following. Example:
1801
1802 ev_unref (EV_A);
1803 ev_timer_add (EV_A_ watcher);
1804 ev_loop (EV_A_ 0);
1805
1806It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
1807which is often provided by the following macro.
1808
1809=item C<EV_P>, C<EV_P_>
1810
1811This provides the loop I<parameter> for functions, if one is required ("ev
1812loop parameter"). The C<EV_P> form is used when this is the sole parameter,
1813C<EV_P_> is used when other parameters are following. Example:
1814
1815 // this is how ev_unref is being declared
1816 static void ev_unref (EV_P);
1817
1818 // this is how you can declare your typical callback
1819 static void cb (EV_P_ ev_timer *w, int revents)
1820
1821It declares a parameter C<loop> of type C<struct ev_loop *>, quite
1822suitable for use with C<EV_A>.
1823
1824=item C<EV_DEFAULT>, C<EV_DEFAULT_>
1825
1826Similar to the other two macros, this gives you the value of the default
1827loop, if multiple loops are supported ("ev loop default").
1828
1829=back
1830
1831Example: Declare and initialise a check watcher, working regardless of
1832wether multiple loops are supported or not.
1833
1834 static void
1835 check_cb (EV_P_ ev_timer *w, int revents)
1836 {
1837 ev_check_stop (EV_A_ w);
1838 }
1839
1840 ev_check check;
1841 ev_check_init (&check, check_cb);
1842 ev_check_start (EV_DEFAULT_ &check);
1843 ev_loop (EV_DEFAULT_ 0);
1844
1489 1845
1490=head1 EMBEDDING 1846=head1 EMBEDDING
1491 1847
1492Libev can (and often is) directly embedded into host 1848Libev can (and often is) directly embedded into host
1493applications. Examples of applications that embed it include the Deliantra 1849applications. Examples of applications that embed it include the Deliantra
1668 2024
1669=item EV_USE_DEVPOLL 2025=item EV_USE_DEVPOLL
1670 2026
1671reserved for future expansion, works like the USE symbols above. 2027reserved for future expansion, works like the USE symbols above.
1672 2028
2029=item EV_USE_INOTIFY
2030
2031If defined to be C<1>, libev will compile in support for the Linux inotify
2032interface to speed up C<ev_stat> watchers. Its actual availability will
2033be detected at runtime.
2034
1673=item EV_H 2035=item EV_H
1674 2036
1675The name of the F<ev.h> header file used to include it. The default if 2037The name of the F<ev.h> header file used to include it. The default if
1676undefined is C<< <ev.h> >> in F<event.h> and C<"ev.h"> in F<ev.c>. This 2038undefined is C<< <ev.h> >> in F<event.h> and C<"ev.h"> in F<ev.c>. This
1677can be used to virtually rename the F<ev.h> header file in case of conflicts. 2039can be used to virtually rename the F<ev.h> header file in case of conflicts.
1700will have the C<struct ev_loop *> as first argument, and you can create 2062will have the C<struct ev_loop *> as first argument, and you can create
1701additional independent event loops. Otherwise there will be no support 2063additional independent event loops. Otherwise there will be no support
1702for multiple event loops and there is no first event loop pointer 2064for multiple event loops and there is no first event loop pointer
1703argument. Instead, all functions act on the single default loop. 2065argument. Instead, all functions act on the single default loop.
1704 2066
1705=item EV_PERIODICS 2067=item EV_PERIODIC_ENABLE
1706 2068
1707If undefined or defined to be C<1>, then periodic timers are supported, 2069If undefined or defined to be C<1>, then periodic timers are supported. If
1708otherwise not. This saves a few kb of code. 2070defined to be C<0>, then they are not. Disabling them saves a few kB of
2071code.
2072
2073=item EV_EMBED_ENABLE
2074
2075If undefined or defined to be C<1>, then embed watchers are supported. If
2076defined to be C<0>, then they are not.
2077
2078=item EV_STAT_ENABLE
2079
2080If undefined or defined to be C<1>, then stat watchers are supported. If
2081defined to be C<0>, then they are not.
2082
2083=item EV_FORK_ENABLE
2084
2085If undefined or defined to be C<1>, then fork watchers are supported. If
2086defined to be C<0>, then they are not.
2087
2088=item EV_MINIMAL
2089
2090If you need to shave off some kilobytes of code at the expense of some
2091speed, define this symbol to C<1>. Currently only used for gcc to override
2092some inlining decisions, saves roughly 30% codesize of amd64.
2093
2094=item EV_PID_HASHSIZE
2095
2096C<ev_child> watchers use a small hash table to distribute workload by
2097pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more
2098than enough. If you need to manage thousands of children you might want to
2099increase this value (I<must> be a power of two).
2100
2101=item EV_INOTIFY_HASHSIZE
2102
2103C<ev_staz> watchers use a small hash table to distribute workload by
2104inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),
2105usually more than enough. If you need to manage thousands of C<ev_stat>
2106watchers you might want to increase this value (I<must> be a power of
2107two).
1709 2108
1710=item EV_COMMON 2109=item EV_COMMON
1711 2110
1712By default, all watchers have a C<void *data> member. By redefining 2111By default, all watchers have a C<void *data> member. By redefining
1713this macro to a something else you can include more and other types of 2112this macro to a something else you can include more and other types of
1773 2172
1774=item Starting io/check/prepare/idle/signal/child watchers: O(1) 2173=item Starting io/check/prepare/idle/signal/child watchers: O(1)
1775 2174
1776=item Stopping check/prepare/idle watchers: O(1) 2175=item Stopping check/prepare/idle watchers: O(1)
1777 2176
1778=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % 16)) 2177=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
1779 2178
1780=item Finding the next timer per loop iteration: O(1) 2179=item Finding the next timer per loop iteration: O(1)
1781 2180
1782=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd) 2181=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
1783 2182

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines