… | |
… | |
40 | points in your program, so locking and parallel access are rarely an |
40 | points in your program, so locking and parallel access are rarely an |
41 | issue, making thread programming much safer and easier than using other |
41 | issue, making thread programming much safer and easier than using other |
42 | thread models. |
42 | thread models. |
43 | |
43 | |
44 | Unlike the so-called "Perl threads" (which are not actually real threads |
44 | Unlike the so-called "Perl threads" (which are not actually real threads |
45 | but only the windows process emulation ported to unix), Coro provides a |
45 | but only the windows process emulation (see section of same name for more |
|
|
46 | details) ported to unix, and as such act as processes), Coro provides |
46 | full shared address space, which makes communication between threads |
47 | a full shared address space, which makes communication between threads |
47 | very easy. And threads are fast, too: disabling the Windows process |
48 | very easy. And Coro's threads are fast, too: disabling the Windows |
48 | emulation code in your perl and using Coro can easily result in a two to |
49 | process emulation code in your perl and using Coro can easily result in |
49 | four times speed increase for your programs. |
50 | a two to four times speed increase for your programs. A parallel matrix |
|
|
51 | multiplication benchmark runs over 300 times faster on a single core than |
|
|
52 | perl's pseudo-threads on a quad core using all four cores. |
50 | |
53 | |
51 | Coro achieves that by supporting multiple running interpreters that share |
54 | Coro achieves that by supporting multiple running interpreters that share |
52 | data, which is especially useful to code pseudo-parallel processes and |
55 | data, which is especially useful to code pseudo-parallel processes and |
53 | for event-based programming, such as multiple HTTP-GET requests running |
56 | for event-based programming, such as multiple HTTP-GET requests running |
54 | concurrently. See L<Coro::AnyEvent> to learn more on how to integrate Coro |
57 | concurrently. See L<Coro::AnyEvent> to learn more on how to integrate Coro |
55 | into an event-based environment. |
58 | into an event-based environment. |
56 | |
59 | |
57 | In this module, a thread is defined as "callchain + lexical variables + |
60 | In this module, a thread is defined as "callchain + lexical variables + |
58 | @_ + $_ + $@ + $/ + C stack), that is, a thread has its own callchain, |
61 | some package variables + C stack), that is, a thread has its own callchain, |
59 | its own set of lexicals and its own set of perls most important global |
62 | its own set of lexicals and its own set of perls most important global |
60 | variables (see L<Coro::State> for more configuration and background info). |
63 | variables (see L<Coro::State> for more configuration and background info). |
61 | |
64 | |
62 | See also the C<SEE ALSO> section at the end of this document - the Coro |
65 | See also the C<SEE ALSO> section at the end of this document - the Coro |
63 | module family is quite large. |
66 | module family is quite large. |
64 | |
67 | |
65 | =cut |
68 | =cut |
66 | |
69 | |
67 | package Coro; |
70 | package Coro; |
68 | |
71 | |
69 | use strict qw(vars subs); |
72 | use common::sense; |
70 | no warnings "uninitialized"; |
73 | |
|
|
74 | use Carp (); |
71 | |
75 | |
72 | use Guard (); |
76 | use Guard (); |
73 | |
77 | |
74 | use Coro::State; |
78 | use Coro::State; |
75 | |
79 | |
… | |
… | |
77 | |
81 | |
78 | our $idle; # idle handler |
82 | our $idle; # idle handler |
79 | our $main; # main coro |
83 | our $main; # main coro |
80 | our $current; # current coro |
84 | our $current; # current coro |
81 | |
85 | |
82 | our $VERSION = 5.13; |
86 | our $VERSION = 5.17; |
83 | |
87 | |
84 | our @EXPORT = qw(async async_pool cede schedule terminate current unblock_sub); |
88 | our @EXPORT = qw(async async_pool cede schedule terminate current unblock_sub); |
85 | our %EXPORT_TAGS = ( |
89 | our %EXPORT_TAGS = ( |
86 | prio => [qw(PRIO_MAX PRIO_HIGH PRIO_NORMAL PRIO_LOW PRIO_IDLE PRIO_MIN)], |
90 | prio => [qw(PRIO_MAX PRIO_HIGH PRIO_NORMAL PRIO_LOW PRIO_IDLE PRIO_MIN)], |
87 | ); |
91 | ); |
… | |
… | |
150 | handlers), then it must be prepared to be called recursively itself. |
154 | handlers), then it must be prepared to be called recursively itself. |
151 | |
155 | |
152 | =cut |
156 | =cut |
153 | |
157 | |
154 | $idle = sub { |
158 | $idle = sub { |
155 | require Carp; |
159 | warn "oi\n";#d# |
156 | Carp::croak ("FATAL: deadlock detected"); |
160 | Carp::confess ("FATAL: deadlock detected"); |
157 | }; |
161 | }; |
158 | |
162 | |
159 | # this coro is necessary because a coro |
163 | # this coro is necessary because a coro |
160 | # cannot destroy itself. |
164 | # cannot destroy itself. |
161 | our @destroy; |
165 | our @destroy; |
… | |
… | |
203 | Example: Create a new coro that just prints its arguments. |
207 | Example: Create a new coro that just prints its arguments. |
204 | |
208 | |
205 | async { |
209 | async { |
206 | print "@_\n"; |
210 | print "@_\n"; |
207 | } 1,2,3,4; |
211 | } 1,2,3,4; |
208 | |
|
|
209 | =cut |
|
|
210 | |
|
|
211 | sub async(&@) { |
|
|
212 | my $coro = new Coro @_; |
|
|
213 | $coro->ready; |
|
|
214 | $coro |
|
|
215 | } |
|
|
216 | |
212 | |
217 | =item async_pool { ... } [@args...] |
213 | =item async_pool { ... } [@args...] |
218 | |
214 | |
219 | Similar to C<async>, but uses a coro pool, so you should not call |
215 | Similar to C<async>, but uses a coro pool, so you should not call |
220 | terminate or join on it (although you are allowed to), and you get a |
216 | terminate or join on it (although you are allowed to), and you get a |
… | |
… | |
335 | |
331 | |
336 | These functions implement the same concept as C<dynamic-wind> in scheme |
332 | These functions implement the same concept as C<dynamic-wind> in scheme |
337 | does, and are useful when you want to localise some resource to a specific |
333 | does, and are useful when you want to localise some resource to a specific |
338 | coro. |
334 | coro. |
339 | |
335 | |
340 | They slow down coro switching considerably for coros that use |
336 | They slow down thread switching considerably for coros that use them |
341 | them (But coro switching is still reasonably fast if the handlers are |
337 | (about 40% for a BLOCK with a single assignment, so thread switching is |
342 | fast). |
338 | still reasonably fast if the handlers are fast). |
343 | |
339 | |
344 | These functions are best understood by an example: The following function |
340 | These functions are best understood by an example: The following function |
345 | will change the current timezone to "Antarctica/South_Pole", which |
341 | will change the current timezone to "Antarctica/South_Pole", which |
346 | requires a call to C<tzset>, but by using C<on_enter> and C<on_leave>, |
342 | requires a call to C<tzset>, but by using C<on_enter> and C<on_leave>, |
347 | which remember/change the current timezone and restore the previous |
343 | which remember/change the current timezone and restore the previous |
348 | value, respectively, the timezone is only changes for the coro that |
344 | value, respectively, the timezone is only changed for the coro that |
349 | installed those handlers. |
345 | installed those handlers. |
350 | |
346 | |
351 | use POSIX qw(tzset); |
347 | use POSIX qw(tzset); |
352 | |
348 | |
353 | async { |
349 | async { |
… | |
… | |
370 | }; |
366 | }; |
371 | |
367 | |
372 | This can be used to localise about any resource (locale, uid, current |
368 | This can be used to localise about any resource (locale, uid, current |
373 | working directory etc.) to a block, despite the existance of other |
369 | working directory etc.) to a block, despite the existance of other |
374 | coros. |
370 | coros. |
|
|
371 | |
|
|
372 | Another interesting example implements time-sliced multitasking using |
|
|
373 | interval timers (this could obviously be optimised, but does the job): |
|
|
374 | |
|
|
375 | # "timeslice" the given block |
|
|
376 | sub timeslice(&) { |
|
|
377 | use Time::HiRes (); |
|
|
378 | |
|
|
379 | Coro::on_enter { |
|
|
380 | # on entering the thread, we set an VTALRM handler to cede |
|
|
381 | $SIG{VTALRM} = sub { cede }; |
|
|
382 | # and then start the interval timer |
|
|
383 | Time::HiRes::setitimer &Time::HiRes::ITIMER_VIRTUAL, 0.01, 0.01; |
|
|
384 | }; |
|
|
385 | Coro::on_leave { |
|
|
386 | # on leaving the thread, we stop the interval timer again |
|
|
387 | Time::HiRes::setitimer &Time::HiRes::ITIMER_VIRTUAL, 0, 0; |
|
|
388 | }; |
|
|
389 | |
|
|
390 | &{+shift}; |
|
|
391 | } |
|
|
392 | |
|
|
393 | # use like this: |
|
|
394 | timeslice { |
|
|
395 | # The following is an endless loop that would normally |
|
|
396 | # monopolise the process. Since it runs in a timesliced |
|
|
397 | # environment, it will regularly cede to other threads. |
|
|
398 | while () { } |
|
|
399 | }; |
|
|
400 | |
375 | |
401 | |
376 | =item killall |
402 | =item killall |
377 | |
403 | |
378 | Kills/terminates/cancels all coros except the currently running one. |
404 | Kills/terminates/cancels all coros except the currently running one. |
379 | |
405 | |
… | |
… | |
423 | the ready queue, do nothing and return false. |
449 | the ready queue, do nothing and return false. |
424 | |
450 | |
425 | This ensures that the scheduler will resume this coro automatically |
451 | This ensures that the scheduler will resume this coro automatically |
426 | once all the coro of higher priority and all coro of the same |
452 | once all the coro of higher priority and all coro of the same |
427 | priority that were put into the ready queue earlier have been resumed. |
453 | priority that were put into the ready queue earlier have been resumed. |
|
|
454 | |
|
|
455 | =item $coro->suspend |
|
|
456 | |
|
|
457 | Suspends the specified coro. A suspended coro works just like any other |
|
|
458 | coro, except that the scheduler will not select a suspended coro for |
|
|
459 | execution. |
|
|
460 | |
|
|
461 | Suspending a coro can be useful when you want to keep the coro from |
|
|
462 | running, but you don't want to destroy it, or when you want to temporarily |
|
|
463 | freeze a coro (e.g. for debugging) to resume it later. |
|
|
464 | |
|
|
465 | A scenario for the former would be to suspend all (other) coros after a |
|
|
466 | fork and keep them alive, so their destructors aren't called, but new |
|
|
467 | coros can be created. |
|
|
468 | |
|
|
469 | =item $coro->resume |
|
|
470 | |
|
|
471 | If the specified coro was suspended, it will be resumed. Note that when |
|
|
472 | the coro was in the ready queue when it was suspended, it might have been |
|
|
473 | unreadied by the scheduler, so an activation might have been lost. |
|
|
474 | |
|
|
475 | To avoid this, it is best to put a suspended coro into the ready queue |
|
|
476 | unconditionally, as every synchronisation mechanism must protect itself |
|
|
477 | against spurious wakeups, and the one in the Coro family certainly do |
|
|
478 | that. |
428 | |
479 | |
429 | =item $is_ready = $coro->is_ready |
480 | =item $is_ready = $coro->is_ready |
430 | |
481 | |
431 | Returns true iff the Coro object is in the ready queue. Unless the Coro |
482 | Returns true iff the Coro object is in the ready queue. Unless the Coro |
432 | object gets destroyed, it will eventually be scheduled by the scheduler. |
483 | object gets destroyed, it will eventually be scheduled by the scheduler. |
… | |
… | |
693 | Wait for the specified rouse callback (or the last one that was created in |
744 | Wait for the specified rouse callback (or the last one that was created in |
694 | this coro). |
745 | this coro). |
695 | |
746 | |
696 | As soon as the callback is invoked (or when the callback was invoked |
747 | As soon as the callback is invoked (or when the callback was invoked |
697 | before C<rouse_wait>), it will return the arguments originally passed to |
748 | before C<rouse_wait>), it will return the arguments originally passed to |
698 | the rouse callback. |
749 | the rouse callback. In scalar context, that means you get the I<last> |
|
|
750 | argument, just as if C<rouse_wait> had a C<return ($a1, $a2, $a3...)> |
|
|
751 | statement at the end. |
699 | |
752 | |
700 | See the section B<HOW TO WAIT FOR A CALLBACK> for an actual usage example. |
753 | See the section B<HOW TO WAIT FOR A CALLBACK> for an actual usage example. |
701 | |
754 | |
702 | =back |
755 | =back |
703 | |
756 | |
… | |
… | |
802 | works. |
855 | works. |
803 | |
856 | |
804 | =back |
857 | =back |
805 | |
858 | |
806 | |
859 | |
|
|
860 | =head1 WINDOWS PROCESS EMULATION |
|
|
861 | |
|
|
862 | A great many people seem to be confused about ithreads (for example, Chip |
|
|
863 | Salzenberg called me unintelligent, incapable, stupid and gullible, |
|
|
864 | while in the same mail making rather confused statements about perl |
|
|
865 | ithreads (for example, that memory or files would be shared), showing his |
|
|
866 | lack of understanding of this area - if it is hard to understand for Chip, |
|
|
867 | it is probably not obvious to everybody). |
|
|
868 | |
|
|
869 | What follows is an ultra-condensed version of my talk about threads in |
|
|
870 | scripting languages given onthe perl workshop 2009: |
|
|
871 | |
|
|
872 | The so-called "ithreads" were originally implemented for two reasons: |
|
|
873 | first, to (badly) emulate unix processes on native win32 perls, and |
|
|
874 | secondly, to replace the older, real thread model ("5.005-threads"). |
|
|
875 | |
|
|
876 | It does that by using threads instead of OS processes. The difference |
|
|
877 | between processes and threads is that threads share memory (and other |
|
|
878 | state, such as files) between threads within a single process, while |
|
|
879 | processes do not share anything (at least not semantically). That |
|
|
880 | means that modifications done by one thread are seen by others, while |
|
|
881 | modifications by one process are not seen by other processes. |
|
|
882 | |
|
|
883 | The "ithreads" work exactly like that: when creating a new ithreads |
|
|
884 | process, all state is copied (memory is copied physically, files and code |
|
|
885 | is copied logically). Afterwards, it isolates all modifications. On UNIX, |
|
|
886 | the same behaviour can be achieved by using operating system processes, |
|
|
887 | except that UNIX typically uses hardware built into the system to do this |
|
|
888 | efficiently, while the windows process emulation emulates this hardware in |
|
|
889 | software (rather efficiently, but of course it is still much slower than |
|
|
890 | dedicated hardware). |
|
|
891 | |
|
|
892 | As mentioned before, loading code, modifying code, modifying data |
|
|
893 | structures and so on is only visible in the ithreads process doing the |
|
|
894 | modification, not in other ithread processes within the same OS process. |
|
|
895 | |
|
|
896 | This is why "ithreads" do not implement threads for perl at all, only |
|
|
897 | processes. What makes it so bad is that on non-windows platforms, you can |
|
|
898 | actually take advantage of custom hardware for this purpose (as evidenced |
|
|
899 | by the forks module, which gives you the (i-) threads API, just much |
|
|
900 | faster). |
|
|
901 | |
|
|
902 | Sharing data is in the i-threads model is done by transfering data |
|
|
903 | structures between threads using copying semantics, which is very slow - |
|
|
904 | shared data simply does not exist. Benchmarks using i-threads which are |
|
|
905 | communication-intensive show extremely bad behaviour with i-threads (in |
|
|
906 | fact, so bad that Coro, which cannot take direct advantage of multiple |
|
|
907 | CPUs, is often orders of magnitude faster because it shares data using |
|
|
908 | real threads, refer to my talk for details). |
|
|
909 | |
|
|
910 | As summary, i-threads *use* threads to implement processes, while |
|
|
911 | the compatible forks module *uses* processes to emulate, uhm, |
|
|
912 | processes. I-threads slow down every perl program when enabled, and |
|
|
913 | outside of windows, serve no (or little) practical purpose, but |
|
|
914 | disadvantages every single-threaded Perl program. |
|
|
915 | |
|
|
916 | This is the reason that I try to avoid the name "ithreads", as it is |
|
|
917 | misleading as it implies that it implements some kind of thread model for |
|
|
918 | perl, and prefer the name "windows process emulation", which describes the |
|
|
919 | actual use and behaviour of it much better. |
|
|
920 | |
807 | =head1 SEE ALSO |
921 | =head1 SEE ALSO |
808 | |
922 | |
809 | Event-Loop integration: L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>. |
923 | Event-Loop integration: L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>. |
810 | |
924 | |
811 | Debugging: L<Coro::Debug>. |
925 | Debugging: L<Coro::Debug>. |