Commit | Line | Data |
---|---|---|
920dae64 AT |
1 | =head1 NAME |
2 | ||
3 | perlthrtut - tutorial on threads in Perl | |
4 | ||
5 | =head1 DESCRIPTION | |
6 | ||
7 | B<NOTE>: this tutorial describes the new Perl threading flavour | |
8 | introduced in Perl 5.6.0 called interpreter threads, or B<ithreads> | |
9 | for short. In this model each thread runs in its own Perl interpreter, | |
10 | and any data sharing between threads must be explicit. | |
11 | ||
12 | There is another older Perl threading flavour called the 5.005 model, | |
13 | unsurprisingly for 5.005 versions of Perl. The old model is known to | |
14 | have problems, deprecated, and will probably be removed around release | |
15 | 5.10. You are strongly encouraged to migrate any existing 5.005 | |
16 | threads code to the new model as soon as possible. | |
17 | ||
18 | You can see which (or neither) threading flavour you have by | |
19 | running C<perl -V> and looking at the C<Platform> section. | |
20 | If you have C<useithreads=define> you have ithreads, if you | |
21 | have C<use5005threads=define> you have 5.005 threads. | |
22 | If you have neither, you don't have any thread support built in. | |
23 | If you have both, you are in trouble. | |
24 | ||
25 | The user-level interface to the 5.005 threads was via the L<Threads> | |
26 | class, while ithreads uses the L<threads> class. Note the change in case. | |
27 | ||
28 | =head1 Status | |
29 | ||
30 | The ithreads code has been available since Perl 5.6.0, and is considered | |
31 | stable. The user-level interface to ithreads (the L<threads> classes) | |
32 | appeared in the 5.8.0 release, and as of this time is considered stable | |
33 | although it should be treated with caution as with all new features. | |
34 | ||
35 | =head1 What Is A Thread Anyway? | |
36 | ||
37 | A thread is a flow of control through a program with a single | |
38 | execution point. | |
39 | ||
40 | Sounds an awful lot like a process, doesn't it? Well, it should. | |
41 | Threads are one of the pieces of a process. Every process has at least | |
42 | one thread and, up until now, every process running Perl had only one | |
43 | thread. With 5.8, though, you can create extra threads. We're going | |
44 | to show you how, when, and why. | |
45 | ||
46 | =head1 Threaded Program Models | |
47 | ||
48 | There are three basic ways that you can structure a threaded | |
49 | program. Which model you choose depends on what you need your program | |
50 | to do. For many non-trivial threaded programs you'll need to choose | |
51 | different models for different pieces of your program. | |
52 | ||
53 | =head2 Boss/Worker | |
54 | ||
55 | The boss/worker model usually has one "boss" thread and one or more | |
56 | "worker" threads. The boss thread gathers or generates tasks that need | |
57 | to be done, then parcels those tasks out to the appropriate worker | |
58 | thread. | |
59 | ||
60 | This model is common in GUI and server programs, where a main thread | |
61 | waits for some event and then passes that event to the appropriate | |
62 | worker threads for processing. Once the event has been passed on, the | |
63 | boss thread goes back to waiting for another event. | |
64 | ||
65 | The boss thread does relatively little work. While tasks aren't | |
66 | necessarily performed faster than with any other method, it tends to | |
67 | have the best user-response times. | |
68 | ||
69 | =head2 Work Crew | |
70 | ||
71 | In the work crew model, several threads are created that do | |
72 | essentially the same thing to different pieces of data. It closely | |
73 | mirrors classical parallel processing and vector processors, where a | |
74 | large array of processors do the exact same thing to many pieces of | |
75 | data. | |
76 | ||
77 | This model is particularly useful if the system running the program | |
78 | will distribute multiple threads across different processors. It can | |
79 | also be useful in ray tracing or rendering engines, where the | |
80 | individual threads can pass on interim results to give the user visual | |
81 | feedback. | |
82 | ||
83 | =head2 Pipeline | |
84 | ||
85 | The pipeline model divides up a task into a series of steps, and | |
86 | passes the results of one step on to the thread processing the | |
87 | next. Each thread does one thing to each piece of data and passes the | |
88 | results to the next thread in line. | |
89 | ||
90 | This model makes the most sense if you have multiple processors so two | |
91 | or more threads will be executing in parallel, though it can often | |
92 | make sense in other contexts as well. It tends to keep the individual | |
93 | tasks small and simple, as well as allowing some parts of the pipeline | |
94 | to block (on I/O or system calls, for example) while other parts keep | |
95 | going. If you're running different parts of the pipeline on different | |
96 | processors you may also take advantage of the caches on each | |
97 | processor. | |
98 | ||
99 | This model is also handy for a form of recursive programming where, | |
100 | rather than having a subroutine call itself, it instead creates | |
101 | another thread. Prime and Fibonacci generators both map well to this | |
102 | form of the pipeline model. (A version of a prime number generator is | |
103 | presented later on.) | |
104 | ||
105 | =head1 What kind of threads are Perl threads? | |
106 | ||
107 | If you have experience with other thread implementations, you might | |
108 | find that things aren't quite what you expect. It's very important to | |
109 | remember when dealing with Perl threads that Perl Threads Are Not X | |
110 | Threads, for all values of X. They aren't POSIX threads, or | |
111 | DecThreads, or Java's Green threads, or Win32 threads. There are | |
112 | similarities, and the broad concepts are the same, but if you start | |
113 | looking for implementation details you're going to be either | |
114 | disappointed or confused. Possibly both. | |
115 | ||
116 | This is not to say that Perl threads are completely different from | |
117 | everything that's ever come before--they're not. Perl's threading | |
118 | model owes a lot to other thread models, especially POSIX. Just as | |
119 | Perl is not C, though, Perl threads are not POSIX threads. So if you | |
120 | find yourself looking for mutexes, or thread priorities, it's time to | |
121 | step back a bit and think about what you want to do and how Perl can | |
122 | do it. | |
123 | ||
124 | However it is important to remember that Perl threads cannot magically | |
125 | do things unless your operating systems threads allows it. So if your | |
126 | system blocks the entire process on sleep(), Perl usually will as well. | |
127 | ||
128 | Perl Threads Are Different. | |
129 | ||
130 | =head1 Thread-Safe Modules | |
131 | ||
132 | The addition of threads has changed Perl's internals | |
133 | substantially. There are implications for people who write | |
134 | modules with XS code or external libraries. However, since perl data is | |
135 | not shared among threads by default, Perl modules stand a high chance of | |
136 | being thread-safe or can be made thread-safe easily. Modules that are not | |
137 | tagged as thread-safe should be tested or code reviewed before being used | |
138 | in production code. | |
139 | ||
140 | Not all modules that you might use are thread-safe, and you should | |
141 | always assume a module is unsafe unless the documentation says | |
142 | otherwise. This includes modules that are distributed as part of the | |
143 | core. Threads are a new feature, and even some of the standard | |
144 | modules aren't thread-safe. | |
145 | ||
146 | Even if a module is thread-safe, it doesn't mean that the module is optimized | |
147 | to work well with threads. A module could possibly be rewritten to utilize | |
148 | the new features in threaded Perl to increase performance in a threaded | |
149 | environment. | |
150 | ||
151 | If you're using a module that's not thread-safe for some reason, you | |
152 | can protect yourself by using it from one, and only one thread at all. | |
153 | If you need multiple threads to access such a module, you can use semaphores and | |
154 | lots of programming discipline to control access to it. Semaphores | |
155 | are covered in L</"Basic semaphores">. | |
156 | ||
157 | See also L</"Thread-Safety of System Libraries">. | |
158 | ||
159 | =head1 Thread Basics | |
160 | ||
161 | The core L<threads> module provides the basic functions you need to write | |
162 | threaded programs. In the following sections we'll cover the basics, | |
163 | showing you what you need to do to create a threaded program. After | |
164 | that, we'll go over some of the features of the L<threads> module that | |
165 | make threaded programming easier. | |
166 | ||
167 | =head2 Basic Thread Support | |
168 | ||
169 | Thread support is a Perl compile-time option - it's something that's | |
170 | turned on or off when Perl is built at your site, rather than when | |
171 | your programs are compiled. If your Perl wasn't compiled with thread | |
172 | support enabled, then any attempt to use threads will fail. | |
173 | ||
174 | Your programs can use the Config module to check whether threads are | |
175 | enabled. If your program can't run without them, you can say something | |
176 | like: | |
177 | ||
178 | $Config{useithreads} or die "Recompile Perl with threads to run this program."; | |
179 | ||
180 | A possibly-threaded program using a possibly-threaded module might | |
181 | have code like this: | |
182 | ||
183 | use Config; | |
184 | use MyMod; | |
185 | ||
186 | BEGIN { | |
187 | if ($Config{useithreads}) { | |
188 | # We have threads | |
189 | require MyMod_threaded; | |
190 | import MyMod_threaded; | |
191 | } else { | |
192 | require MyMod_unthreaded; | |
193 | import MyMod_unthreaded; | |
194 | } | |
195 | } | |
196 | ||
197 | Since code that runs both with and without threads is usually pretty | |
198 | messy, it's best to isolate the thread-specific code in its own | |
199 | module. In our example above, that's what MyMod_threaded is, and it's | |
200 | only imported if we're running on a threaded Perl. | |
201 | ||
202 | =head2 A Note about the Examples | |
203 | ||
204 | Although thread support is considered to be stable, there are still a number | |
205 | of quirks that may startle you when you try out any of the examples below. | |
206 | In a real situation, care should be taken that all threads are finished | |
207 | executing before the program exits. That care has B<not> been taken in these | |
208 | examples in the interest of simplicity. Running these examples "as is" will | |
209 | produce error messages, usually caused by the fact that there are still | |
210 | threads running when the program exits. You should not be alarmed by this. | |
211 | Future versions of Perl may fix this problem. | |
212 | ||
213 | =head2 Creating Threads | |
214 | ||
215 | The L<threads> package provides the tools you need to create new | |
216 | threads. Like any other module, you need to tell Perl that you want to use | |
217 | it; C<use threads> imports all the pieces you need to create basic | |
218 | threads. | |
219 | ||
220 | The simplest, most straightforward way to create a thread is with new(): | |
221 | ||
222 | use threads; | |
223 | ||
224 | $thr = threads->new(\&sub1); | |
225 | ||
226 | sub sub1 { | |
227 | print "In the thread\n"; | |
228 | } | |
229 | ||
230 | The new() method takes a reference to a subroutine and creates a new | |
231 | thread, which starts executing in the referenced subroutine. Control | |
232 | then passes both to the subroutine and the caller. | |
233 | ||
234 | If you need to, your program can pass parameters to the subroutine as | |
235 | part of the thread startup. Just include the list of parameters as | |
236 | part of the C<threads::new> call, like this: | |
237 | ||
238 | use threads; | |
239 | ||
240 | $Param3 = "foo"; | |
241 | $thr = threads->new(\&sub1, "Param 1", "Param 2", $Param3); | |
242 | $thr = threads->new(\&sub1, @ParamList); | |
243 | $thr = threads->new(\&sub1, qw(Param1 Param2 Param3)); | |
244 | ||
245 | sub sub1 { | |
246 | my @InboundParameters = @_; | |
247 | print "In the thread\n"; | |
248 | print "got parameters >", join("<>", @InboundParameters), "<\n"; | |
249 | } | |
250 | ||
251 | ||
252 | The last example illustrates another feature of threads. You can spawn | |
253 | off several threads using the same subroutine. Each thread executes | |
254 | the same subroutine, but in a separate thread with a separate | |
255 | environment and potentially separate arguments. | |
256 | ||
257 | C<create()> is a synonym for C<new()>. | |
258 | ||
259 | =head2 Waiting For A Thread To Exit | |
260 | ||
261 | Since threads are also subroutines, they can return values. To wait | |
262 | for a thread to exit and extract any values it might return, you can | |
263 | use the join() method: | |
264 | ||
265 | use threads; | |
266 | ||
267 | $thr = threads->new(\&sub1); | |
268 | ||
269 | @ReturnData = $thr->join; | |
270 | print "Thread returned @ReturnData"; | |
271 | ||
272 | sub sub1 { return "Fifty-six", "foo", 2; } | |
273 | ||
274 | In the example above, the join() method returns as soon as the thread | |
275 | ends. In addition to waiting for a thread to finish and gathering up | |
276 | any values that the thread might have returned, join() also performs | |
277 | any OS cleanup necessary for the thread. That cleanup might be | |
278 | important, especially for long-running programs that spawn lots of | |
279 | threads. If you don't want the return values and don't want to wait | |
280 | for the thread to finish, you should call the detach() method | |
281 | instead, as described next. | |
282 | ||
283 | =head2 Ignoring A Thread | |
284 | ||
285 | join() does three things: it waits for a thread to exit, cleans up | |
286 | after it, and returns any data the thread may have produced. But what | |
287 | if you're not interested in the thread's return values, and you don't | |
288 | really care when the thread finishes? All you want is for the thread | |
289 | to get cleaned up after when it's done. | |
290 | ||
291 | In this case, you use the detach() method. Once a thread is detached, | |
292 | it'll run until it's finished, then Perl will clean up after it | |
293 | automatically. | |
294 | ||
295 | use threads; | |
296 | ||
297 | $thr = threads->new(\&sub1); # Spawn the thread | |
298 | ||
299 | $thr->detach; # Now we officially don't care any more | |
300 | ||
301 | sub sub1 { | |
302 | $a = 0; | |
303 | while (1) { | |
304 | $a++; | |
305 | print "\$a is $a\n"; | |
306 | sleep 1; | |
307 | } | |
308 | } | |
309 | ||
310 | Once a thread is detached, it may not be joined, and any return data | |
311 | that it might have produced (if it was done and waiting for a join) is | |
312 | lost. | |
313 | ||
314 | =head1 Threads And Data | |
315 | ||
316 | Now that we've covered the basics of threads, it's time for our next | |
317 | topic: data. Threading introduces a couple of complications to data | |
318 | access that non-threaded programs never need to worry about. | |
319 | ||
320 | =head2 Shared And Unshared Data | |
321 | ||
322 | The biggest difference between Perl ithreads and the old 5.005 style | |
323 | threading, or for that matter, to most other threading systems out there, | |
324 | is that by default, no data is shared. When a new perl thread is created, | |
325 | all the data associated with the current thread is copied to the new | |
326 | thread, and is subsequently private to that new thread! | |
327 | This is similar in feel to what happens when a UNIX process forks, | |
328 | except that in this case, the data is just copied to a different part of | |
329 | memory within the same process rather than a real fork taking place. | |
330 | ||
331 | To make use of threading however, one usually wants the threads to share | |
332 | at least some data between themselves. This is done with the | |
333 | L<threads::shared> module and the C< : shared> attribute: | |
334 | ||
335 | use threads; | |
336 | use threads::shared; | |
337 | ||
338 | my $foo : shared = 1; | |
339 | my $bar = 1; | |
340 | threads->new(sub { $foo++; $bar++ })->join; | |
341 | ||
342 | print "$foo\n"; #prints 2 since $foo is shared | |
343 | print "$bar\n"; #prints 1 since $bar is not shared | |
344 | ||
345 | In the case of a shared array, all the array's elements are shared, and for | |
346 | a shared hash, all the keys and values are shared. This places | |
347 | restrictions on what may be assigned to shared array and hash elements: only | |
348 | simple values or references to shared variables are allowed - this is | |
349 | so that a private variable can't accidentally become shared. A bad | |
350 | assignment will cause the thread to die. For example: | |
351 | ||
352 | use threads; | |
353 | use threads::shared; | |
354 | ||
355 | my $var = 1; | |
356 | my $svar : shared = 2; | |
357 | my %hash : shared; | |
358 | ||
359 | ... create some threads ... | |
360 | ||
361 | $hash{a} = 1; # all threads see exists($hash{a}) and $hash{a} == 1 | |
362 | $hash{a} = $var # okay - copy-by-value: same effect as previous | |
363 | $hash{a} = $svar # okay - copy-by-value: same effect as previous | |
364 | $hash{a} = \$svar # okay - a reference to a shared variable | |
365 | $hash{a} = \$var # This will die | |
366 | delete $hash{a} # okay - all threads will see !exists($hash{a}) | |
367 | ||
368 | Note that a shared variable guarantees that if two or more threads try to | |
369 | modify it at the same time, the internal state of the variable will not | |
370 | become corrupted. However, there are no guarantees beyond this, as | |
371 | explained in the next section. | |
372 | ||
373 | =head2 Thread Pitfalls: Races | |
374 | ||
375 | While threads bring a new set of useful tools, they also bring a | |
376 | number of pitfalls. One pitfall is the race condition: | |
377 | ||
378 | use threads; | |
379 | use threads::shared; | |
380 | ||
381 | my $a : shared = 1; | |
382 | $thr1 = threads->new(\&sub1); | |
383 | $thr2 = threads->new(\&sub2); | |
384 | ||
385 | $thr1->join; | |
386 | $thr2->join; | |
387 | print "$a\n"; | |
388 | ||
389 | sub sub1 { my $foo = $a; $a = $foo + 1; } | |
390 | sub sub2 { my $bar = $a; $a = $bar + 1; } | |
391 | ||
392 | What do you think $a will be? The answer, unfortunately, is "it | |
393 | depends." Both sub1() and sub2() access the global variable $a, once | |
394 | to read and once to write. Depending on factors ranging from your | |
395 | thread implementation's scheduling algorithm to the phase of the moon, | |
396 | $a can be 2 or 3. | |
397 | ||
398 | Race conditions are caused by unsynchronized access to shared | |
399 | data. Without explicit synchronization, there's no way to be sure that | |
400 | nothing has happened to the shared data between the time you access it | |
401 | and the time you update it. Even this simple code fragment has the | |
402 | possibility of error: | |
403 | ||
404 | use threads; | |
405 | my $a : shared = 2; | |
406 | my $b : shared; | |
407 | my $c : shared; | |
408 | my $thr1 = threads->create(sub { $b = $a; $a = $b + 1; }); | |
409 | my $thr2 = threads->create(sub { $c = $a; $a = $c + 1; }); | |
410 | $thr1->join; | |
411 | $thr2->join; | |
412 | ||
413 | Two threads both access $a. Each thread can potentially be interrupted | |
414 | at any point, or be executed in any order. At the end, $a could be 3 | |
415 | or 4, and both $b and $c could be 2 or 3. | |
416 | ||
417 | Even C<$a += 5> or C<$a++> are not guaranteed to be atomic. | |
418 | ||
419 | Whenever your program accesses data or resources that can be accessed | |
420 | by other threads, you must take steps to coordinate access or risk | |
421 | data inconsistency and race conditions. Note that Perl will protect its | |
422 | internals from your race conditions, but it won't protect you from you. | |
423 | ||
424 | =head1 Synchronization and control | |
425 | ||
426 | Perl provides a number of mechanisms to coordinate the interactions | |
427 | between themselves and their data, to avoid race conditions and the like. | |
428 | Some of these are designed to resemble the common techniques used in thread | |
429 | libraries such as C<pthreads>; others are Perl-specific. Often, the | |
430 | standard techniques are clumsy and difficult to get right (such as | |
431 | condition waits). Where possible, it is usually easier to use Perlish | |
432 | techniques such as queues, which remove some of the hard work involved. | |
433 | ||
434 | =head2 Controlling access: lock() | |
435 | ||
436 | The lock() function takes a shared variable and puts a lock on it. | |
437 | No other thread may lock the variable until the variable is unlocked | |
438 | by the thread holding the lock. Unlocking happens automatically | |
439 | when the locking thread exits the outermost block that contains | |
440 | C<lock()> function. Using lock() is straightforward: this example has | |
441 | several threads doing some calculations in parallel, and occasionally | |
442 | updating a running total: | |
443 | ||
444 | use threads; | |
445 | use threads::shared; | |
446 | ||
447 | my $total : shared = 0; | |
448 | ||
449 | sub calc { | |
450 | for (;;) { | |
451 | my $result; | |
452 | # (... do some calculations and set $result ...) | |
453 | { | |
454 | lock($total); # block until we obtain the lock | |
455 | $total += $result; | |
456 | } # lock implicitly released at end of scope | |
457 | last if $result == 0; | |
458 | } | |
459 | } | |
460 | ||
461 | my $thr1 = threads->new(\&calc); | |
462 | my $thr2 = threads->new(\&calc); | |
463 | my $thr3 = threads->new(\&calc); | |
464 | $thr1->join; | |
465 | $thr2->join; | |
466 | $thr3->join; | |
467 | print "total=$total\n"; | |
468 | ||
469 | ||
470 | lock() blocks the thread until the variable being locked is | |
471 | available. When lock() returns, your thread can be sure that no other | |
472 | thread can lock that variable until the outermost block containing the | |
473 | lock exits. | |
474 | ||
475 | It's important to note that locks don't prevent access to the variable | |
476 | in question, only lock attempts. This is in keeping with Perl's | |
477 | longstanding tradition of courteous programming, and the advisory file | |
478 | locking that flock() gives you. | |
479 | ||
480 | You may lock arrays and hashes as well as scalars. Locking an array, | |
481 | though, will not block subsequent locks on array elements, just lock | |
482 | attempts on the array itself. | |
483 | ||
484 | Locks are recursive, which means it's okay for a thread to | |
485 | lock a variable more than once. The lock will last until the outermost | |
486 | lock() on the variable goes out of scope. For example: | |
487 | ||
488 | my $x : shared; | |
489 | doit(); | |
490 | ||
491 | sub doit { | |
492 | { | |
493 | { | |
494 | lock($x); # wait for lock | |
495 | lock($x); # NOOP - we already have the lock | |
496 | { | |
497 | lock($x); # NOOP | |
498 | { | |
499 | lock($x); # NOOP | |
500 | lockit_some_more(); | |
501 | } | |
502 | } | |
503 | } # *** implicit unlock here *** | |
504 | } | |
505 | } | |
506 | ||
507 | sub lockit_some_more { | |
508 | lock($x); # NOOP | |
509 | } # nothing happens here | |
510 | ||
511 | Note that there is no unlock() function - the only way to unlock a | |
512 | variable is to allow it to go out of scope. | |
513 | ||
514 | A lock can either be used to guard the data contained within the variable | |
515 | being locked, or it can be used to guard something else, like a section | |
516 | of code. In this latter case, the variable in question does not hold any | |
517 | useful data, and exists only for the purpose of being locked. In this | |
518 | respect, the variable behaves like the mutexes and basic semaphores of | |
519 | traditional thread libraries. | |
520 | ||
521 | =head2 A Thread Pitfall: Deadlocks | |
522 | ||
523 | Locks are a handy tool to synchronize access to data, and using them | |
524 | properly is the key to safe shared data. Unfortunately, locks aren't | |
525 | without their dangers, especially when multiple locks are involved. | |
526 | Consider the following code: | |
527 | ||
528 | use threads; | |
529 | ||
530 | my $a : shared = 4; | |
531 | my $b : shared = "foo"; | |
532 | my $thr1 = threads->new(sub { | |
533 | lock($a); | |
534 | sleep 20; | |
535 | lock($b); | |
536 | }); | |
537 | my $thr2 = threads->new(sub { | |
538 | lock($b); | |
539 | sleep 20; | |
540 | lock($a); | |
541 | }); | |
542 | ||
543 | This program will probably hang until you kill it. The only way it | |
544 | won't hang is if one of the two threads acquires both locks | |
545 | first. A guaranteed-to-hang version is more complicated, but the | |
546 | principle is the same. | |
547 | ||
548 | The first thread will grab a lock on $a, then, after a pause during which | |
549 | the second thread has probably had time to do some work, try to grab a | |
550 | lock on $b. Meanwhile, the second thread grabs a lock on $b, then later | |
551 | tries to grab a lock on $a. The second lock attempt for both threads will | |
552 | block, each waiting for the other to release its lock. | |
553 | ||
554 | This condition is called a deadlock, and it occurs whenever two or | |
555 | more threads are trying to get locks on resources that the others | |
556 | own. Each thread will block, waiting for the other to release a lock | |
557 | on a resource. That never happens, though, since the thread with the | |
558 | resource is itself waiting for a lock to be released. | |
559 | ||
560 | There are a number of ways to handle this sort of problem. The best | |
561 | way is to always have all threads acquire locks in the exact same | |
562 | order. If, for example, you lock variables $a, $b, and $c, always lock | |
563 | $a before $b, and $b before $c. It's also best to hold on to locks for | |
564 | as short a period of time to minimize the risks of deadlock. | |
565 | ||
566 | The other synchronization primitives described below can suffer from | |
567 | similar problems. | |
568 | ||
569 | =head2 Queues: Passing Data Around | |
570 | ||
571 | A queue is a special thread-safe object that lets you put data in one | |
572 | end and take it out the other without having to worry about | |
573 | synchronization issues. They're pretty straightforward, and look like | |
574 | this: | |
575 | ||
576 | use threads; | |
577 | use Thread::Queue; | |
578 | ||
579 | my $DataQueue = Thread::Queue->new; | |
580 | $thr = threads->new(sub { | |
581 | while ($DataElement = $DataQueue->dequeue) { | |
582 | print "Popped $DataElement off the queue\n"; | |
583 | } | |
584 | }); | |
585 | ||
586 | $DataQueue->enqueue(12); | |
587 | $DataQueue->enqueue("A", "B", "C"); | |
588 | $DataQueue->enqueue(\$thr); | |
589 | sleep 10; | |
590 | $DataQueue->enqueue(undef); | |
591 | $thr->join; | |
592 | ||
593 | You create the queue with C<new Thread::Queue>. Then you can | |
594 | add lists of scalars onto the end with enqueue(), and pop scalars off | |
595 | the front of it with dequeue(). A queue has no fixed size, and can grow | |
596 | as needed to hold everything pushed on to it. | |
597 | ||
598 | If a queue is empty, dequeue() blocks until another thread enqueues | |
599 | something. This makes queues ideal for event loops and other | |
600 | communications between threads. | |
601 | ||
602 | =head2 Semaphores: Synchronizing Data Access | |
603 | ||
604 | Semaphores are a kind of generic locking mechanism. In their most basic | |
605 | form, they behave very much like lockable scalars, except that they | |
606 | can't hold data, and that they must be explicitly unlocked. In their | |
607 | advanced form, they act like a kind of counter, and can allow multiple | |
608 | threads to have the 'lock' at any one time. | |
609 | ||
610 | =head2 Basic semaphores | |
611 | ||
612 | Semaphores have two methods, down() and up(): down() decrements the resource | |
613 | count, while up increments it. Calls to down() will block if the | |
614 | semaphore's current count would decrement below zero. This program | |
615 | gives a quick demonstration: | |
616 | ||
617 | use threads; | |
618 | use Thread::Semaphore; | |
619 | ||
620 | my $semaphore = new Thread::Semaphore; | |
621 | my $GlobalVariable : shared = 0; | |
622 | ||
623 | $thr1 = new threads \&sample_sub, 1; | |
624 | $thr2 = new threads \&sample_sub, 2; | |
625 | $thr3 = new threads \&sample_sub, 3; | |
626 | ||
627 | sub sample_sub { | |
628 | my $SubNumber = shift @_; | |
629 | my $TryCount = 10; | |
630 | my $LocalCopy; | |
631 | sleep 1; | |
632 | while ($TryCount--) { | |
633 | $semaphore->down; | |
634 | $LocalCopy = $GlobalVariable; | |
635 | print "$TryCount tries left for sub $SubNumber (\$GlobalVariable is $GlobalVariable)\n"; | |
636 | sleep 2; | |
637 | $LocalCopy++; | |
638 | $GlobalVariable = $LocalCopy; | |
639 | $semaphore->up; | |
640 | } | |
641 | } | |
642 | ||
643 | $thr1->join; | |
644 | $thr2->join; | |
645 | $thr3->join; | |
646 | ||
647 | The three invocations of the subroutine all operate in sync. The | |
648 | semaphore, though, makes sure that only one thread is accessing the | |
649 | global variable at once. | |
650 | ||
651 | =head2 Advanced Semaphores | |
652 | ||
653 | By default, semaphores behave like locks, letting only one thread | |
654 | down() them at a time. However, there are other uses for semaphores. | |
655 | ||
656 | Each semaphore has a counter attached to it. By default, semaphores are | |
657 | created with the counter set to one, down() decrements the counter by | |
658 | one, and up() increments by one. However, we can override any or all | |
659 | of these defaults simply by passing in different values: | |
660 | ||
661 | use threads; | |
662 | use Thread::Semaphore; | |
663 | my $semaphore = Thread::Semaphore->new(5); | |
664 | # Creates a semaphore with the counter set to five | |
665 | ||
666 | $thr1 = threads->new(\&sub1); | |
667 | $thr2 = threads->new(\&sub1); | |
668 | ||
669 | sub sub1 { | |
670 | $semaphore->down(5); # Decrements the counter by five | |
671 | # Do stuff here | |
672 | $semaphore->up(5); # Increment the counter by five | |
673 | } | |
674 | ||
675 | $thr1->detach; | |
676 | $thr2->detach; | |
677 | ||
678 | If down() attempts to decrement the counter below zero, it blocks until | |
679 | the counter is large enough. Note that while a semaphore can be created | |
680 | with a starting count of zero, any up() or down() always changes the | |
681 | counter by at least one, and so $semaphore->down(0) is the same as | |
682 | $semaphore->down(1). | |
683 | ||
684 | The question, of course, is why would you do something like this? Why | |
685 | create a semaphore with a starting count that's not one, or why | |
686 | decrement/increment it by more than one? The answer is resource | |
687 | availability. Many resources that you want to manage access for can be | |
688 | safely used by more than one thread at once. | |
689 | ||
690 | For example, let's take a GUI driven program. It has a semaphore that | |
691 | it uses to synchronize access to the display, so only one thread is | |
692 | ever drawing at once. Handy, but of course you don't want any thread | |
693 | to start drawing until things are properly set up. In this case, you | |
694 | can create a semaphore with a counter set to zero, and up it when | |
695 | things are ready for drawing. | |
696 | ||
697 | Semaphores with counters greater than one are also useful for | |
698 | establishing quotas. Say, for example, that you have a number of | |
699 | threads that can do I/O at once. You don't want all the threads | |
700 | reading or writing at once though, since that can potentially swamp | |
701 | your I/O channels, or deplete your process' quota of filehandles. You | |
702 | can use a semaphore initialized to the number of concurrent I/O | |
703 | requests (or open files) that you want at any one time, and have your | |
704 | threads quietly block and unblock themselves. | |
705 | ||
706 | Larger increments or decrements are handy in those cases where a | |
707 | thread needs to check out or return a number of resources at once. | |
708 | ||
709 | =head2 cond_wait() and cond_signal() | |
710 | ||
711 | These two functions can be used in conjunction with locks to notify | |
712 | co-operating threads that a resource has become available. They are | |
713 | very similar in use to the functions found in C<pthreads>. However | |
714 | for most purposes, queues are simpler to use and more intuitive. See | |
715 | L<threads::shared> for more details. | |
716 | ||
717 | =head2 Giving up control | |
718 | ||
719 | There are times when you may find it useful to have a thread | |
720 | explicitly give up the CPU to another thread. You may be doing something | |
721 | processor-intensive and want to make sure that the user-interface thread | |
722 | gets called frequently. Regardless, there are times that you might want | |
723 | a thread to give up the processor. | |
724 | ||
725 | Perl's threading package provides the yield() function that does | |
726 | this. yield() is pretty straightforward, and works like this: | |
727 | ||
728 | use threads; | |
729 | ||
730 | sub loop { | |
731 | my $thread = shift; | |
732 | my $foo = 50; | |
733 | while($foo--) { print "in thread $thread\n" } | |
734 | threads->yield; | |
735 | $foo = 50; | |
736 | while($foo--) { print "in thread $thread\n" } | |
737 | } | |
738 | ||
739 | my $thread1 = threads->new(\&loop, 'first'); | |
740 | my $thread2 = threads->new(\&loop, 'second'); | |
741 | my $thread3 = threads->new(\&loop, 'third'); | |
742 | ||
743 | It is important to remember that yield() is only a hint to give up the CPU, | |
744 | it depends on your hardware, OS and threading libraries what actually happens. | |
745 | B<On many operating systems, yield() is a no-op.> Therefore it is important | |
746 | to note that one should not build the scheduling of the threads around | |
747 | yield() calls. It might work on your platform but it won't work on another | |
748 | platform. | |
749 | ||
750 | =head1 General Thread Utility Routines | |
751 | ||
752 | We've covered the workhorse parts of Perl's threading package, and | |
753 | with these tools you should be well on your way to writing threaded | |
754 | code and packages. There are a few useful little pieces that didn't | |
755 | really fit in anyplace else. | |
756 | ||
757 | =head2 What Thread Am I In? | |
758 | ||
759 | The C<< threads->self >> class method provides your program with a way to | |
760 | get an object representing the thread it's currently in. You can use this | |
761 | object in the same way as the ones returned from thread creation. | |
762 | ||
763 | =head2 Thread IDs | |
764 | ||
765 | tid() is a thread object method that returns the thread ID of the | |
766 | thread the object represents. Thread IDs are integers, with the main | |
767 | thread in a program being 0. Currently Perl assigns a unique tid to | |
768 | every thread ever created in your program, assigning the first thread | |
769 | to be created a tid of 1, and increasing the tid by 1 for each new | |
770 | thread that's created. | |
771 | ||
772 | =head2 Are These Threads The Same? | |
773 | ||
774 | The equal() method takes two thread objects and returns true | |
775 | if the objects represent the same thread, and false if they don't. | |
776 | ||
777 | Thread objects also have an overloaded == comparison so that you can do | |
778 | comparison on them as you would with normal objects. | |
779 | ||
780 | =head2 What Threads Are Running? | |
781 | ||
782 | C<< threads->list >> returns a list of thread objects, one for each thread | |
783 | that's currently running and not detached. Handy for a number of things, | |
784 | including cleaning up at the end of your program: | |
785 | ||
786 | # Loop through all the threads | |
787 | foreach $thr (threads->list) { | |
788 | # Don't join the main thread or ourselves | |
789 | if ($thr->tid && !threads::equal($thr, threads->self)) { | |
790 | $thr->join; | |
791 | } | |
792 | } | |
793 | ||
794 | If some threads have not finished running when the main Perl thread | |
795 | ends, Perl will warn you about it and die, since it is impossible for Perl | |
796 | to clean up itself while other threads are running | |
797 | ||
798 | =head1 A Complete Example | |
799 | ||
800 | Confused yet? It's time for an example program to show some of the | |
801 | things we've covered. This program finds prime numbers using threads. | |
802 | ||
803 | 1 #!/usr/bin/perl -w | |
804 | 2 # prime-pthread, courtesy of Tom Christiansen | |
805 | 3 | |
806 | 4 use strict; | |
807 | 5 | |
808 | 6 use threads; | |
809 | 7 use Thread::Queue; | |
810 | 8 | |
811 | 9 my $stream = new Thread::Queue; | |
812 | 10 my $kid = new threads(\&check_num, $stream, 2); | |
813 | 11 | |
814 | 12 for my $i ( 3 .. 1000 ) { | |
815 | 13 $stream->enqueue($i); | |
816 | 14 } | |
817 | 15 | |
818 | 16 $stream->enqueue(undef); | |
819 | 17 $kid->join; | |
820 | 18 | |
821 | 19 sub check_num { | |
822 | 20 my ($upstream, $cur_prime) = @_; | |
823 | 21 my $kid; | |
824 | 22 my $downstream = new Thread::Queue; | |
825 | 23 while (my $num = $upstream->dequeue) { | |
826 | 24 next unless $num % $cur_prime; | |
827 | 25 if ($kid) { | |
828 | 26 $downstream->enqueue($num); | |
829 | 27 } else { | |
830 | 28 print "Found prime $num\n"; | |
831 | 29 $kid = new threads(\&check_num, $downstream, $num); | |
832 | 30 } | |
833 | 31 } | |
834 | 32 $downstream->enqueue(undef) if $kid; | |
835 | 33 $kid->join if $kid; | |
836 | 34 } | |
837 | ||
838 | This program uses the pipeline model to generate prime numbers. Each | |
839 | thread in the pipeline has an input queue that feeds numbers to be | |
840 | checked, a prime number that it's responsible for, and an output queue | |
841 | into which it funnels numbers that have failed the check. If the thread | |
842 | has a number that's failed its check and there's no child thread, then | |
843 | the thread must have found a new prime number. In that case, a new | |
844 | child thread is created for that prime and stuck on the end of the | |
845 | pipeline. | |
846 | ||
847 | This probably sounds a bit more confusing than it really is, so let's | |
848 | go through this program piece by piece and see what it does. (For | |
849 | those of you who might be trying to remember exactly what a prime | |
850 | number is, it's a number that's only evenly divisible by itself and 1) | |
851 | ||
852 | The bulk of the work is done by the check_num() subroutine, which | |
853 | takes a reference to its input queue and a prime number that it's | |
854 | responsible for. After pulling in the input queue and the prime that | |
855 | the subroutine's checking (line 20), we create a new queue (line 22) | |
856 | and reserve a scalar for the thread that we're likely to create later | |
857 | (line 21). | |
858 | ||
859 | The while loop from lines 23 to line 31 grabs a scalar off the input | |
860 | queue and checks against the prime this thread is responsible | |
861 | for. Line 24 checks to see if there's a remainder when we modulo the | |
862 | number to be checked against our prime. If there is one, the number | |
863 | must not be evenly divisible by our prime, so we need to either pass | |
864 | it on to the next thread if we've created one (line 26) or create a | |
865 | new thread if we haven't. | |
866 | ||
867 | The new thread creation is line 29. We pass on to it a reference to | |
868 | the queue we've created, and the prime number we've found. | |
869 | ||
870 | Finally, once the loop terminates (because we got a 0 or undef in the | |
871 | queue, which serves as a note to die), we pass on the notice to our | |
872 | child and wait for it to exit if we've created a child (lines 32 and | |
873 | 37). | |
874 | ||
875 | Meanwhile, back in the main thread, we create a queue (line 9) and the | |
876 | initial child thread (line 10), and pre-seed it with the first prime: | |
877 | 2. Then we queue all the numbers from 3 to 1000 for checking (lines | |
878 | 12-14), then queue a die notice (line 16) and wait for the first child | |
879 | thread to terminate (line 17). Because a child won't die until its | |
880 | child has died, we know that we're done once we return from the join. | |
881 | ||
882 | That's how it works. It's pretty simple; as with many Perl programs, | |
883 | the explanation is much longer than the program. | |
884 | ||
885 | =head1 Different implementations of threads | |
886 | ||
887 | Some background on thread implementations from the operating system | |
888 | viewpoint. There are three basic categories of threads: user-mode threads, | |
889 | kernel threads, and multiprocessor kernel threads. | |
890 | ||
891 | User-mode threads are threads that live entirely within a program and | |
892 | its libraries. In this model, the OS knows nothing about threads. As | |
893 | far as it's concerned, your process is just a process. | |
894 | ||
895 | This is the easiest way to implement threads, and the way most OSes | |
896 | start. The big disadvantage is that, since the OS knows nothing about | |
897 | threads, if one thread blocks they all do. Typical blocking activities | |
898 | include most system calls, most I/O, and things like sleep(). | |
899 | ||
900 | Kernel threads are the next step in thread evolution. The OS knows | |
901 | about kernel threads, and makes allowances for them. The main | |
902 | difference between a kernel thread and a user-mode thread is | |
903 | blocking. With kernel threads, things that block a single thread don't | |
904 | block other threads. This is not the case with user-mode threads, | |
905 | where the kernel blocks at the process level and not the thread level. | |
906 | ||
907 | This is a big step forward, and can give a threaded program quite a | |
908 | performance boost over non-threaded programs. Threads that block | |
909 | performing I/O, for example, won't block threads that are doing other | |
910 | things. Each process still has only one thread running at once, | |
911 | though, regardless of how many CPUs a system might have. | |
912 | ||
913 | Since kernel threading can interrupt a thread at any time, they will | |
914 | uncover some of the implicit locking assumptions you may make in your | |
915 | program. For example, something as simple as C<$a = $a + 2> can behave | |
916 | unpredictably with kernel threads if $a is visible to other | |
917 | threads, as another thread may have changed $a between the time it | |
918 | was fetched on the right hand side and the time the new value is | |
919 | stored. | |
920 | ||
921 | Multiprocessor kernel threads are the final step in thread | |
922 | support. With multiprocessor kernel threads on a machine with multiple | |
923 | CPUs, the OS may schedule two or more threads to run simultaneously on | |
924 | different CPUs. | |
925 | ||
926 | This can give a serious performance boost to your threaded program, | |
927 | since more than one thread will be executing at the same time. As a | |
928 | tradeoff, though, any of those nagging synchronization issues that | |
929 | might not have shown with basic kernel threads will appear with a | |
930 | vengeance. | |
931 | ||
932 | In addition to the different levels of OS involvement in threads, | |
933 | different OSes (and different thread implementations for a particular | |
934 | OS) allocate CPU cycles to threads in different ways. | |
935 | ||
936 | Cooperative multitasking systems have running threads give up control | |
937 | if one of two things happen. If a thread calls a yield function, it | |
938 | gives up control. It also gives up control if the thread does | |
939 | something that would cause it to block, such as perform I/O. In a | |
940 | cooperative multitasking implementation, one thread can starve all the | |
941 | others for CPU time if it so chooses. | |
942 | ||
943 | Preemptive multitasking systems interrupt threads at regular intervals | |
944 | while the system decides which thread should run next. In a preemptive | |
945 | multitasking system, one thread usually won't monopolize the CPU. | |
946 | ||
947 | On some systems, there can be cooperative and preemptive threads | |
948 | running simultaneously. (Threads running with realtime priorities | |
949 | often behave cooperatively, for example, while threads running at | |
950 | normal priorities behave preemptively.) | |
951 | ||
952 | Most modern operating systems support preemptive multitasking nowadays. | |
953 | ||
954 | =head1 Performance considerations | |
955 | ||
956 | The main thing to bear in mind when comparing ithreads to other threading | |
957 | models is the fact that for each new thread created, a complete copy of | |
958 | all the variables and data of the parent thread has to be taken. Thus | |
959 | thread creation can be quite expensive, both in terms of memory usage and | |
960 | time spent in creation. The ideal way to reduce these costs is to have a | |
961 | relatively short number of long-lived threads, all created fairly early | |
962 | on - before the base thread has accumulated too much data. Of course, this | |
963 | may not always be possible, so compromises have to be made. However, after | |
964 | a thread has been created, its performance and extra memory usage should | |
965 | be little different than ordinary code. | |
966 | ||
967 | Also note that under the current implementation, shared variables | |
968 | use a little more memory and are a little slower than ordinary variables. | |
969 | ||
970 | =head1 Process-scope Changes | |
971 | ||
972 | Note that while threads themselves are separate execution threads and | |
973 | Perl data is thread-private unless explicitly shared, the threads can | |
974 | affect process-scope state, affecting all the threads. | |
975 | ||
976 | The most common example of this is changing the current working | |
977 | directory using chdir(). One thread calls chdir(), and the working | |
978 | directory of all the threads changes. | |
979 | ||
980 | Even more drastic example of a process-scope change is chroot(): | |
981 | the root directory of all the threads changes, and no thread can | |
982 | undo it (as opposed to chdir()). | |
983 | ||
984 | Further examples of process-scope changes include umask() and | |
985 | changing uids/gids. | |
986 | ||
987 | Thinking of mixing fork() and threads? Please lie down and wait | |
988 | until the feeling passes. Be aware that the semantics of fork() vary | |
989 | between platforms. For example, some UNIX systems copy all the current | |
990 | threads into the child process, while others only copy the thread that | |
991 | called fork(). You have been warned! | |
992 | ||
993 | Similarly, mixing signals and threads should not be attempted. | |
994 | Implementations are platform-dependent, and even the POSIX | |
995 | semantics may not be what you expect (and Perl doesn't even | |
996 | give you the full POSIX API). | |
997 | ||
998 | =head1 Thread-Safety of System Libraries | |
999 | ||
1000 | Whether various library calls are thread-safe is outside the control | |
1001 | of Perl. Calls often suffering from not being thread-safe include: | |
1002 | localtime(), gmtime(), get{gr,host,net,proto,serv,pw}*(), readdir(), | |
1003 | rand(), and srand() -- in general, calls that depend on some global | |
1004 | external state. | |
1005 | ||
1006 | If the system Perl is compiled in has thread-safe variants of such | |
1007 | calls, they will be used. Beyond that, Perl is at the mercy of | |
1008 | the thread-safety or -unsafety of the calls. Please consult your | |
1009 | C library call documentation. | |
1010 | ||
1011 | On some platforms the thread-safe library interfaces may fail if the | |
1012 | result buffer is too small (for example the user group databases may | |
1013 | be rather large, and the reentrant interfaces may have to carry around | |
1014 | a full snapshot of those databases). Perl will start with a small | |
1015 | buffer, but keep retrying and growing the result buffer | |
1016 | until the result fits. If this limitless growing sounds bad for | |
1017 | security or memory consumption reasons you can recompile Perl with | |
1018 | PERL_REENTRANT_MAXSIZE defined to the maximum number of bytes you will | |
1019 | allow. | |
1020 | ||
1021 | =head1 Conclusion | |
1022 | ||
1023 | A complete thread tutorial could fill a book (and has, many times), | |
1024 | but with what we've covered in this introduction, you should be well | |
1025 | on your way to becoming a threaded Perl expert. | |
1026 | ||
1027 | =head1 Bibliography | |
1028 | ||
1029 |