1PERLOTHRTUT(1) Perl Programmers Reference Guide PERLOTHRTUT(1)
2
6 perlothrtut - old tutorial on threads in Perl
7
9 WARNING: This tutorial describes the old-style thread model that was
10 introduced in release 5.005. This model is now deprecated, and will be
11 removed, probably in version 5.10. The interfaces described here were
12 considered experimental, and are likely to be buggy.
13 For information about the new interpreter threads ("ithreads") model,
15 see the perlthrtut tutorial, and the threads and threads::shared mod‐
16 ules.
17 You are strongly encouraged to migrate any existing threads code to the
19 new model as soon as possible.
20
22 A thread is a flow of control through a program with a single execution
23 point.
24 Sounds an awful lot like a process, doesn't it? Well, it should.
26 Threads are one of the pieces of a process. Every process has at least
27 one thread and, up until now, every process running Perl had only one
28 thread. With 5.005, though, you can create extra threads. We're going
29 to show you how, when, and why.
30
32 There are three basic ways that you can structure a threaded program.
33 Which model you choose depends on what you need your program to do.
34 For many non-trivial threaded programs you'll need to choose different
35 models for different pieces of your program.
36 Boss/Worker
38 The boss/worker model usually has one `boss' thread and one or more
40 `worker' threads. The boss thread gathers or generates tasks that need
41 to be done, then parcels those tasks out to the appropriate worker
42 thread.
43 This model is common in GUI and server programs, where a main thread
45 waits for some event and then passes that event to the appropriate
46 worker threads for processing. Once the event has been passed on, the
47 boss thread goes back to waiting for another event.
48 The boss thread does relatively little work. While tasks aren't neces‐
50 sarily performed faster than with any other method, it tends to have
51 the best user-response times.
52 Work Crew
54 In the work crew model, several threads are created that do essentially
56 the same thing to different pieces of data. It closely mirrors classi‐
57 cal parallel processing and vector processors, where a large array of
58 processors do the exact same thing to many pieces of data.
59 This model is particularly useful if the system running the program
61 will distribute multiple threads across different processors. It can
62 also be useful in ray tracing or rendering engines, where the individ‐
63 ual threads can pass on interim results to give the user visual feed‐
64 back.
65 Pipeline
67 The pipeline model divides up a task into a series of steps, and passes
69 the results of one step on to the thread processing the next. Each
70 thread does one thing to each piece of data and passes the results to
71 the next thread in line.
72 This model makes the most sense if you have multiple processors so two
74 or more threads will be executing in parallel, though it can often make
75 sense in other contexts as well. It tends to keep the individual tasks
76 small and simple, as well as allowing some parts of the pipeline to
77 block (on I/O or system calls, for example) while other parts keep
78 going. If you're running different parts of the pipeline on different
79 processors you may also take advantage of the caches on each processor.
80 This model is also handy for a form of recursive programming where,
82 rather than having a subroutine call itself, it instead creates another
83 thread. Prime and Fibonacci generators both map well to this form of
84 the pipeline model. (A version of a prime number generator is presented
85 later on.)
86
88 There are several different ways to implement threads on a system. How
89 threads are implemented depends both on the vendor and, in some cases,
90 the version of the operating system. Often the first implementation
91 will be relatively simple, but later versions of the OS will be more
92 sophisticated.
93 While the information in this section is useful, it's not necessary, so
95 you can skip it if you don't feel up to it.
96 There are three basic categories of threads-user-mode threads, kernel
98 threads, and multiprocessor kernel threads.
99 User-mode threads are threads that live entirely within a program and
101 its libraries. In this model, the OS knows nothing about threads. As
102 far as it's concerned, your process is just a process.
103 This is the easiest way to implement threads, and the way most OSes
105 start. The big disadvantage is that, since the OS knows nothing about
106 threads, if one thread blocks they all do. Typical blocking activities
107 include most system calls, most I/O, and things like sleep().
108 Kernel threads are the next step in thread evolution. The OS knows
110 about kernel threads, and makes allowances for them. The main differ‐
111 ence between a kernel thread and a user-mode thread is blocking. With
112 kernel threads, things that block a single thread don't block other
113 threads. This is not the case with user-mode threads, where the kernel
114 blocks at the process level and not the thread level.
115 This is a big step forward, and can give a threaded program quite a
117 performance boost over non-threaded programs. Threads that block per‐
118 forming I/O, for example, won't block threads that are doing other
119 things. Each process still has only one thread running at once,
120 though, regardless of how many CPUs a system might have.
121 Since kernel threading can interrupt a thread at any time, they will
123 uncover some of the implicit locking assumptions you may make in your
124 program. For example, something as simple as "$a = $a + 2" can behave
125 unpredictably with kernel threads if $a is visible to other threads, as
126 another thread may have changed $a between the time it was fetched on
127 the right hand side and the time the new value is stored.
128 Multiprocessor Kernel Threads are the final step in thread support.
130 With multiprocessor kernel threads on a machine with multiple CPUs, the
131 OS may schedule two or more threads to run simultaneously on different
132 CPUs.
133 This can give a serious performance boost to your threaded program,
135 since more than one thread will be executing at the same time. As a
136 tradeoff, though, any of those nagging synchronization issues that
137 might not have shown with basic kernel threads will appear with a
138 vengeance.
139 In addition to the different levels of OS involvement in threads, dif‐
141 ferent OSes (and different thread implementations for a particular OS)
142 allocate CPU cycles to threads in different ways.
143 Cooperative multitasking systems have running threads give up control
145 if one of two things happen. If a thread calls a yield function, it
146 gives up control. It also gives up control if the thread does some‐
147 thing that would cause it to block, such as perform I/O. In a coopera‐
148 tive multitasking implementation, one thread can starve all the others
149 for CPU time if it so chooses.
150 Preemptive multitasking systems interrupt threads at regular intervals
152 while the system decides which thread should run next. In a preemptive
153 multitasking system, one thread usually won't monopolize the CPU.
154 On some systems, there can be cooperative and preemptive threads run‐
156 ning simultaneously. (Threads running with realtime priorities often
157 behave cooperatively, for example, while threads running at normal pri‐
158 orities behave preemptively.)
159
161 If you have experience with other thread implementations, you might
162 find that things aren't quite what you expect. It's very important to
163 remember when dealing with Perl threads that Perl Threads Are Not X
164 Threads, for all values of X. They aren't POSIX threads, or Dec‐
165 Threads, or Java's Green threads, or Win32 threads. There are similar‐
166 ities, and the broad concepts are the same, but if you start looking
167 for implementation details you're going to be either disappointed or
168 confused. Possibly both.
169 This is not to say that Perl threads are completely different from
171 everything that's ever come before--they're not. Perl's threading
172 model owes a lot to other thread models, especially POSIX. Just as
173 Perl is not C, though, Perl threads are not POSIX threads. So if you
174 find yourself looking for mutexes, or thread priorities, it's time to
175 step back a bit and think about what you want to do and how Perl can do
176 it.
177
179 The addition of threads has changed Perl's internals substantially.
180 There are implications for people who write modules--especially modules
181 with XS code or external libraries. While most modules won't encounter
182 any problems, modules that aren't explicitly tagged as thread-safe
183 should be tested before being used in production code.
184 Not all modules that you might use are thread-safe, and you should
186 always assume a module is unsafe unless the documentation says other‐
187 wise. This includes modules that are distributed as part of the core.
188 Threads are a beta feature, and even some of the standard modules
189 aren't thread-safe.
190 If you're using a module that's not thread-safe for some reason, you
192 can protect yourself by using semaphores and lots of programming disci‐
193 pline to control access to the module. Semaphores are covered later in
194 the article. Perl Threads Are Different
195
197 The core Thread module provides the basic functions you need to write
198 threaded programs. In the following sections we'll cover the basics,
199 showing you what you need to do to create a threaded program. After
200 that, we'll go over some of the features of the Thread module that make
201 threaded programming easier.
202 Basic Thread Support
204 Thread support is a Perl compile-time option-it's something that's
206 turned on or off when Perl is built at your site, rather than when your
207 programs are compiled. If your Perl wasn't compiled with thread support
208 enabled, then any attempt to use threads will fail.
209 Remember that the threading support in 5.005 is in beta release, and
211 should be treated as such. You should expect that it may not function
212 entirely properly, and the thread interface may well change some before
213 it is a fully supported, production release. The beta version
214 shouldn't be used for mission-critical projects. Having said that,
215 threaded Perl is pretty nifty, and worth a look.
216 Your programs can use the Config module to check whether threads are
218 enabled. If your program can't run without them, you can say something
219 like:
220 $Config{usethreads} or die "Recompile Perl with threads to run this program.";
222 A possibly-threaded program using a possibly-threaded module might have
224 code like this:
225 use Config;
227 use MyMod;
228 if ($Config{usethreads}) {
230 # We have threads
231 require MyMod_threaded;
232 import MyMod_threaded;
233 } else {
234 require MyMod_unthreaded;
235 import MyMod_unthreaded;
236 }
237 Since code that runs both with and without threads is usually pretty
239 messy, it's best to isolate the thread-specific code in its own module.
240 In our example above, that's what MyMod_threaded is, and it's only
241 imported if we're running on a threaded Perl.
242 Creating Threads
244 The Thread package provides the tools you need to create new threads.
246 Like any other module, you need to tell Perl you want to use it; use
247 Thread imports all the pieces you need to create basic threads.
248 The simplest, straightforward way to create a thread is with new():
250 use Thread;
252 $thr = new Thread \&sub1;
254 sub sub1 {
256 print "In the thread\n";
257 }
258 The new() method takes a reference to a subroutine and creates a new
260 thread, which starts executing in the referenced subroutine. Control
261 then passes both to the subroutine and the caller.
262 If you need to, your program can pass parameters to the subroutine as
264 part of the thread startup. Just include the list of parameters as
265 part of the "Thread::new" call, like this:
266 use Thread;
268 $Param3 = "foo";
269 $thr = new Thread \&sub1, "Param 1", "Param 2", $Param3;
270 $thr = new Thread \&sub1, @ParamList;
271 $thr = new Thread \&sub1, qw(Param1 Param2 $Param3);
272 sub sub1 {
274 my @InboundParameters = @_;
275 print "In the thread\n";
276 print "got parameters >", join("<>", @InboundParameters), "<\n";
277 }
278 The subroutine runs like a normal Perl subroutine, and the call to new
280 Thread returns whatever the subroutine returns.
281 The last example illustrates another feature of threads. You can spawn
283 off several threads using the same subroutine. Each thread executes
284 the same subroutine, but in a separate thread with a separate environ‐
285 ment and potentially separate arguments.
286 The other way to spawn a new thread is with async(), which is a way to
288 spin off a chunk of code like eval(), but into its own thread:
289 use Thread qw(async);
291 $LineCount = 0;
293 $thr = async {
295 while(<>) {$LineCount++}
296 print "Got $LineCount lines\n";
297 };
298 print "Waiting for the linecount to end\n";
300 $thr->join;
301 print "All done\n";
302 You'll notice we did a use Thread qw(async) in that example. async is
304 not exported by default, so if you want it, you'll either need to
305 import it before you use it or fully qualify it as Thread::async.
306 You'll also note that there's a semicolon after the closing brace.
307 That's because async() treats the following block as an anonymous sub‐
308 routine, so the semicolon is necessary.
309 Like eval(), the code executes in the same context as it would if it
311 weren't spun off. Since both the code inside and after the async start
312 executing, you need to be careful with any shared resources. Locking
313 and other synchronization techniques are covered later.
314 Giving up control
316 There are times when you may find it useful to have a thread explicitly
318 give up the CPU to another thread. Your threading package might not
319 support preemptive multitasking for threads, for example, or you may be
320 doing something compute-intensive and want to make sure that the user-
321 interface thread gets called frequently. Regardless, there are times
322 that you might want a thread to give up the processor.
323 Perl's threading package provides the yield() function that does this.
325 yield() is pretty straightforward, and works like this:
326 use Thread qw(yield async);
328 async {
329 my $foo = 50;
330 while ($foo--) { print "first async\n" }
331 yield;
332 $foo = 50;
333 while ($foo--) { print "first async\n" }
334 };
335 async {
336 my $foo = 50;
337 while ($foo--) { print "second async\n" }
338 yield;
339 $foo = 50;
340 while ($foo--) { print "second async\n" }
341 };
342 Waiting For A Thread To Exit
344 Since threads are also subroutines, they can return values. To wait
346 for a thread to exit and extract any scalars it might return, you can
347 use the join() method.
348 use Thread;
350 $thr = new Thread \&sub1;
351 @ReturnData = $thr->join;
353 print "Thread returned @ReturnData";
354 sub sub1 { return "Fifty-six", "foo", 2; }
356 In the example above, the join() method returns as soon as the thread
358 ends. In addition to waiting for a thread to finish and gathering up
359 any values that the thread might have returned, join() also performs
360 any OS cleanup necessary for the thread. That cleanup might be impor‐
361 tant, especially for long-running programs that spawn lots of threads.
362 If you don't want the return values and don't want to wait for the
363 thread to finish, you should call the detach() method instead. detach()
364 is covered later in the article.
365 Errors In Threads
367 So what happens when an error occurs in a thread? Any errors that could
369 be caught with eval() are postponed until the thread is joined. If
370 your program never joins, the errors appear when your program exits.
371 Errors deferred until a join() can be caught with eval():
373 use Thread qw(async);
375 $thr = async {$b = 3/0}; # Divide by zero error
376 $foo = eval {$thr->join};
377 if ($@) {
378 print "died with error $@\n";
379 } else {
380 print "Hey, why aren't you dead?\n";
381 }
382 eval() passes any results from the joined thread back unmodified, so if
384 you want the return value of the thread, this is your only chance to
385 get them.
386 Ignoring A Thread
388 join() does three things: it waits for a thread to exit, cleans up
390 after it, and returns any data the thread may have produced. But what
391 if you're not interested in the thread's return values, and you don't
392 really care when the thread finishes? All you want is for the thread to
393 get cleaned up after when it's done.
394 In this case, you use the detach() method. Once a thread is detached,
396 it'll run until it's finished, then Perl will clean up after it auto‐
397 matically.
398 use Thread;
400 $thr = new Thread \&sub1; # Spawn the thread
401 $thr->detach; # Now we officially don't care any more
403 sub sub1 {
405 $a = 0;
406 while (1) {
407 $a++;
408 print "\$a is $a\n";
409 sleep 1;
410 }
411 }
412 Once a thread is detached, it may not be joined, and any output that it
414 might have produced (if it was done and waiting for a join) is lost.
415
417 Now that we've covered the basics of threads, it's time for our next
418 topic: data. Threading introduces a couple of complications to data
419 access that non-threaded programs never need to worry about.
420 Shared And Unshared Data
422 The single most important thing to remember when using threads is that
424 all threads potentially have access to all the data anywhere in your
425 program. While this is true with a nonthreaded Perl program as well,
426 it's especially important to remember with a threaded program, since
427 more than one thread can be accessing this data at once.
428 Perl's scoping rules don't change because you're using threads. If a
430 subroutine (or block, in the case of async()) could see a variable if
431 you weren't running with threads, it can see it if you are. This is
432 especially important for the subroutines that create, and makes "my"
433 variables even more important. Remember--if your variables aren't lex‐
434 ically scoped (declared with "my") you're probably sharing them between
435 threads.
436 Thread Pitfall: Races
438 While threads bring a new set of useful tools, they also bring a number
440 of pitfalls. One pitfall is the race condition:
441 use Thread;
443 $a = 1;
444 $thr1 = Thread->new(\&sub1);
445 $thr2 = Thread->new(\&sub2);
446 sleep 10;
448 print "$a\n";
449 sub sub1 { $foo = $a; $a = $foo + 1; }
451 sub sub2 { $bar = $a; $a = $bar + 1; }
452 What do you think $a will be? The answer, unfortunately, is "it
454 depends." Both sub1() and sub2() access the global variable $a, once to
455 read and once to write. Depending on factors ranging from your thread
456 implementation's scheduling algorithm to the phase of the moon, $a can
457 be 2 or 3.
458 Race conditions are caused by unsynchronized access to shared data.
460 Without explicit synchronization, there's no way to be sure that noth‐
461 ing has happened to the shared data between the time you access it and
462 the time you update it. Even this simple code fragment has the possi‐
463 bility of error:
464 use Thread qw(async);
466 $a = 2;
467 async{ $b = $a; $a = $b + 1; };
468 async{ $c = $a; $a = $c + 1; };
469 Two threads both access $a. Each thread can potentially be interrupted
471 at any point, or be executed in any order. At the end, $a could be 3
472 or 4, and both $b and $c could be 2 or 3.
473 Whenever your program accesses data or resources that can be accessed
475 by other threads, you must take steps to coordinate access or risk data
476 corruption and race conditions.
477 Controlling access: lock()
479 The lock() function takes a variable (or subroutine, but we'll get to
481 that later) and puts a lock on it. No other thread may lock the vari‐
482 able until the locking thread exits the innermost block containing the
483 lock. Using lock() is straightforward:
484 use Thread qw(async);
486 $a = 4;
487 $thr1 = async {
488 $foo = 12;
489 {
490 lock ($a); # Block until we get access to $a
491 $b = $a;
492 $a = $b * $foo;
493 }
494 print "\$foo was $foo\n";
495 };
496 $thr2 = async {
497 $bar = 7;
498 {
499 lock ($a); # Block until we can get access to $a
500 $c = $a;
501 $a = $c * $bar;
502 }
503 print "\$bar was $bar\n";
504 };
505 $thr1->join;
506 $thr2->join;
507 print "\$a is $a\n";
508 lock() blocks the thread until the variable being locked is available.
510 When lock() returns, your thread can be sure that no other thread can
511 lock that variable until the innermost block containing the lock exits.
512 It's important to note that locks don't prevent access to the variable
514 in question, only lock attempts. This is in keeping with Perl's long‐
515 standing tradition of courteous programming, and the advisory file
516 locking that flock() gives you. Locked subroutines behave differently,
517 however. We'll cover that later in the article.
518 You may lock arrays and hashes as well as scalars. Locking an array,
520 though, will not block subsequent locks on array elements, just lock
521 attempts on the array itself.
522 Finally, locks are recursive, which means it's okay for a thread to
524 lock a variable more than once. The lock will last until the outermost
525 lock() on the variable goes out of scope.
526 Thread Pitfall: Deadlocks
528 Locks are a handy tool to synchronize access to data. Using them prop‐
530 erly is the key to safe shared data. Unfortunately, locks aren't with‐
531 out their dangers. Consider the following code:
532 use Thread qw(async yield);
534 $a = 4;
535 $b = "foo";
536 async {
537 lock($a);
538 yield;
539 sleep 20;
540 lock ($b);
541 };
542 async {
543 lock($b);
544 yield;
545 sleep 20;
546 lock ($a);
547 };
548 This program will probably hang until you kill it. The only way it
550 won't hang is if one of the two async() routines acquires both locks
551 first. A guaranteed-to-hang version is more complicated, but the prin‐
552 ciple is the same.
553 The first thread spawned by async() will grab a lock on $a then, a sec‐
555 ond or two later, try to grab a lock on $b. Meanwhile, the second
556 thread grabs a lock on $b, then later tries to grab a lock on $a. The
557 second lock attempt for both threads will block, each waiting for the
558 other to release its lock.
559 This condition is called a deadlock, and it occurs whenever two or more
561 threads are trying to get locks on resources that the others own. Each
562 thread will block, waiting for the other to release a lock on a
563 resource. That never happens, though, since the thread with the
564 resource is itself waiting for a lock to be released.
565 There are a number of ways to handle this sort of problem. The best
567 way is to always have all threads acquire locks in the exact same
568 order. If, for example, you lock variables $a, $b, and $c, always lock
569 $a before $b, and $b before $c. It's also best to hold on to locks for
570 as short a period of time to minimize the risks of deadlock.
571 Queues: Passing Data Around
573 A queue is a special thread-safe object that lets you put data in one
575 end and take it out the other without having to worry about synchro‐
576 nization issues. They're pretty straightforward, and look like this:
577 use Thread qw(async);
579 use Thread::Queue;
580 my $DataQueue = new Thread::Queue;
582 $thr = async {
583 while ($DataElement = $DataQueue->dequeue) {
584 print "Popped $DataElement off the queue\n";
585 }
586 };
587 $DataQueue->enqueue(12);
589 $DataQueue->enqueue("A", "B", "C");
590 $DataQueue->enqueue(\$thr);
591 sleep 10;
592 $DataQueue->enqueue(undef);
593 You create the queue with new Thread::Queue. Then you can add lists of
595 scalars onto the end with enqueue(), and pop scalars off the front of
596 it with dequeue(). A queue has no fixed size, and can grow as needed
597 to hold everything pushed on to it.
598 If a queue is empty, dequeue() blocks until another thread enqueues
600 something. This makes queues ideal for event loops and other communi‐
601 cations between threads.
602
604 In addition to providing thread-safe access to data via locks and
605 queues, threaded Perl also provides general-purpose semaphores for
606 coarser synchronization than locks provide and thread-safe access to
607 entire subroutines.
608 Semaphores: Synchronizing Data Access
610 Semaphores are a kind of generic locking mechanism. Unlike lock, which
612 gets a lock on a particular scalar, Perl doesn't associate any particu‐
613 lar thing with a semaphore so you can use them to control access to
614 anything you like. In addition, semaphores can allow more than one
615 thread to access a resource at once, though by default semaphores only
616 allow one thread access at a time.
617 Basic semaphores
619 Semaphores have two methods, down and up. down decrements the
620 resource count, while up increments it. down calls will block if
621 the semaphore's current count would decrement below zero. This
622 program gives a quick demonstration:
623 use Thread qw(yield);
625 use Thread::Semaphore;
626 my $semaphore = new Thread::Semaphore;
627 $GlobalVariable = 0;
628 $thr1 = new Thread \&sample_sub, 1;
630 $thr2 = new Thread \&sample_sub, 2;
631 $thr3 = new Thread \&sample_sub, 3;
632 sub sample_sub {
634 my $SubNumber = shift @_;
635 my $TryCount = 10;
636 my $LocalCopy;
637 sleep 1;
638 while ($TryCount--) {
639 $semaphore->down;
640 $LocalCopy = $GlobalVariable;
641 print "$TryCount tries left for sub $SubNumber (\$GlobalVariable is $GlobalVariable)\n";
642 yield;
643 sleep 2;
644 $LocalCopy++;
645 $GlobalVariable = $LocalCopy;
646 $semaphore->up;
647 }
648 }
649 The three invocations of the subroutine all operate in sync. The
651 semaphore, though, makes sure that only one thread is accessing the
652 global variable at once.
653 Advanced Semaphores
655 By default, semaphores behave like locks, letting only one thread
656 down() them at a time. However, there are other uses for sema‐
657 phores.
658 Each semaphore has a counter attached to it. down() decrements the
660 counter and up() increments the counter. By default, semaphores
661 are created with the counter set to one, down() decrements by one,
662 and up() increments by one. If down() attempts to decrement the
663 counter below zero, it blocks until the counter is large enough.
664 Note that while a semaphore can be created with a starting count of
665 zero, any up() or down() always changes the counter by at least
666 one. $semaphore->down(0) is the same as $semaphore->down(1).
667 The question, of course, is why would you do something like this?
669 Why create a semaphore with a starting count that's not one, or why
670 decrement/increment it by more than one? The answer is resource
671 availability. Many resources that you want to manage access for
672 can be safely used by more than one thread at once.
673 For example, let's take a GUI driven program. It has a semaphore
675 that it uses to synchronize access to the display, so only one
676 thread is ever drawing at once. Handy, but of course you don't
677 want any thread to start drawing until things are properly set up.
678 In this case, you can create a semaphore with a counter set to
679 zero, and up it when things are ready for drawing.
680 Semaphores with counters greater than one are also useful for
682 establishing quotas. Say, for example, that you have a number of
683 threads that can do I/O at once. You don't want all the threads
684 reading or writing at once though, since that can potentially swamp
685 your I/O channels, or deplete your process' quota of filehandles.
686 You can use a semaphore initialized to the number of concurrent I/O
687 requests (or open files) that you want at any one time, and have
688 your threads quietly block and unblock themselves.
689 Larger increments or decrements are handy in those cases where a
691 thread needs to check out or return a number of resources at once.
692 Attributes: Restricting Access To Subroutines
694 In addition to synchronizing access to data or resources, you might
696 find it useful to synchronize access to subroutines. You may be
697 accessing a singular machine resource (perhaps a vector processor), or
698 find it easier to serialize calls to a particular subroutine than to
699 have a set of locks and semaphores.
700 One of the additions to Perl 5.005 is subroutine attributes. The
702 Thread package uses these to provide several flavors of serialization.
703 It's important to remember that these attributes are used in the compi‐
704 lation phase of your program so you can't change a subroutine's behav‐
705 ior while your program is actually running.
706 Subroutine Locks
708 The basic subroutine lock looks like this:
710 sub test_sub :locked {
712 }
713 This ensures that only one thread will be executing this subroutine at
715 any one time. Once a thread calls this subroutine, any other thread
716 that calls it will block until the thread in the subroutine exits it.
717 A more elaborate example looks like this:
718 use Thread qw(yield);
720 new Thread \&thread_sub, 1;
722 new Thread \&thread_sub, 2;
723 new Thread \&thread_sub, 3;
724 new Thread \&thread_sub, 4;
725 sub sync_sub :locked {
727 my $CallingThread = shift @_;
728 print "In sync_sub for thread $CallingThread\n";
729 yield;
730 sleep 3;
731 print "Leaving sync_sub for thread $CallingThread\n";
732 }
733 sub thread_sub {
735 my $ThreadID = shift @_;
736 print "Thread $ThreadID calling sync_sub\n";
737 sync_sub($ThreadID);
738 print "$ThreadID is done with sync_sub\n";
739 }
740 The "locked" attribute tells perl to lock sync_sub(), and if you run
742 this, you can see that only one thread is in it at any one time.
743 Methods
745 Locking an entire subroutine can sometimes be overkill, especially when
747 dealing with Perl objects. When calling a method for an object, for
748 example, you want to serialize calls to a method, so that only one
749 thread will be in the subroutine for a particular object, but threads
750 calling that subroutine for a different object aren't blocked. The
751 method attribute indicates whether the subroutine is really a method.
752 use Thread;
754 sub tester {
756 my $thrnum = shift @_;
757 my $bar = new Foo;
758 foreach (1..10) {
759 print "$thrnum calling per_object\n";
760 $bar->per_object($thrnum);
761 print "$thrnum out of per_object\n";
762 yield;
763 print "$thrnum calling one_at_a_time\n";
764 $bar->one_at_a_time($thrnum);
765 print "$thrnum out of one_at_a_time\n";
766 yield;
767 }
768 }
769 foreach my $thrnum (1..10) {
771 new Thread \&tester, $thrnum;
772 }
773 package Foo;
775 sub new {
776 my $class = shift @_;
777 return bless [@_], $class;
778 }
779 sub per_object :locked :method {
781 my ($class, $thrnum) = @_;
782 print "In per_object for thread $thrnum\n";
783 yield;
784 sleep 2;
785 print "Exiting per_object for thread $thrnum\n";
786 }
787 sub one_at_a_time :locked {
789 my ($class, $thrnum) = @_;
790 print "In one_at_a_time for thread $thrnum\n";
791 yield;
792 sleep 2;
793 print "Exiting one_at_a_time for thread $thrnum\n";
794 }
795 As you can see from the output (omitted for brevity; it's 800 lines)
797 all the threads can be in per_object() simultaneously, but only one
798 thread is ever in one_at_a_time() at once.
799 Locking A Subroutine
801 You can lock a subroutine as you would lock a variable. Subroutine
803 locks work the same as specifying a "locked" attribute for the subrou‐
804 tine, and block all access to the subroutine for other threads until
805 the lock goes out of scope. When the subroutine isn't locked, any num‐
806 ber of threads can be in it at once, and getting a lock on a subroutine
807 doesn't affect threads already in the subroutine. Getting a lock on a
808 subroutine looks like this:
809 lock(\&sub_to_lock);
811 Simple enough. Unlike the "locked" attribute, which is a compile time
813 option, locking and unlocking a subroutine can be done at runtime at
814 your discretion. There is some runtime penalty to using lock(\&sub)
815 instead of the "locked" attribute, so make sure you're choosing the
816 proper method to do the locking.
817 You'd choose lock(\&sub) when writing modules and code to run on both
819 threaded and unthreaded Perl, especially for code that will run on
820 5.004 or earlier Perls. In that case, it's useful to have subroutines
821 that should be serialized lock themselves if they're running threaded,
822 like so:
823 package Foo;
825 use Config;
826 $Running_Threaded = 0;
827 BEGIN { $Running_Threaded = $Config{'usethreads'} }
829 sub sub1 { lock(\&sub1) if $Running_Threaded }
831 This way you can ensure single-threadedness regardless of which version
833 of Perl you're running.
834
836 We've covered the workhorse parts of Perl's threading package, and with
837 these tools you should be well on your way to writing threaded code and
838 packages. There are a few useful little pieces that didn't really fit
839 in anyplace else.
840 What Thread Am I In?
842 The Thread->self method provides your program with a way to get an
844 object representing the thread it's currently in. You can use this
845 object in the same way as the ones returned from the thread creation.
846 Thread IDs
848 tid() is a thread object method that returns the thread ID of the
850 thread the object represents. Thread IDs are integers, with the main
851 thread in a program being 0. Currently Perl assigns a unique tid to
852 every thread ever created in your program, assigning the first thread
853 to be created a tid of 1, and increasing the tid by 1 for each new
854 thread that's created.
855 Are These Threads The Same?
857 The equal() method takes two thread objects and returns true if the
859 objects represent the same thread, and false if they don't.
860 What Threads Are Running?
862 Thread->list returns a list of thread objects, one for each thread
864 that's currently running. Handy for a number of things, including
865 cleaning up at the end of your program:
866 # Loop through all the threads
868 foreach $thr (Thread->list) {
869 # Don't join the main thread or ourselves
870 if ($thr->tid && !Thread::equal($thr, Thread->self)) {
871 $thr->join;
872 }
873 }
874 The example above is just for illustration. It isn't strictly neces‐
876 sary to join all the threads you create, since Perl detaches all the
877 threads before it exits.
878
880 Confused yet? It's time for an example program to show some of the
881 things we've covered. This program finds prime numbers using threads.
882 1 #!/usr/bin/perl -w
884 2 # prime-pthread, courtesy of Tom Christiansen
885 3
886 4 use strict;
887 5
888 6 use Thread;
889 7 use Thread::Queue;
890 8
891 9 my $stream = new Thread::Queue;
892 10 my $kid = new Thread(\&check_num, $stream, 2);
893 11
894 12 for my $i ( 3 .. 1000 ) {
895 13 $stream->enqueue($i);
896 14 }
897 15
898 16 $stream->enqueue(undef);
899 17 $kid->join();
900 18
901 19 sub check_num {
902 20 my ($upstream, $cur_prime) = @_;
903 21 my $kid;
904 22 my $downstream = new Thread::Queue;
905 23 while (my $num = $upstream->dequeue) {
906 24 next unless $num % $cur_prime;
907 25 if ($kid) {
908 26 $downstream->enqueue($num);
909 27 } else {
910 28 print "Found prime $num\n";
911 29 $kid = new Thread(\&check_num, $downstream, $num);
912 30 }
913 31 }
914 32 $downstream->enqueue(undef) if $kid;
915 33 $kid->join() if $kid;
916 34 }
917 This program uses the pipeline model to generate prime numbers. Each
919 thread in the pipeline has an input queue that feeds numbers to be
920 checked, a prime number that it's responsible for, and an output queue
921 that it funnels numbers that have failed the check into. If the thread
922 has a number that's failed its check and there's no child thread, then
923 the thread must have found a new prime number. In that case, a new
924 child thread is created for that prime and stuck on the end of the
925 pipeline.
926 This probably sounds a bit more confusing than it really is, so lets go
928 through this program piece by piece and see what it does. (For those
929 of you who might be trying to remember exactly what a prime number is,
930 it's a number that's only evenly divisible by itself and 1)
931 The bulk of the work is done by the check_num() subroutine, which takes
933 a reference to its input queue and a prime number that it's responsible
934 for. After pulling in the input queue and the prime that the subrou‐
935 tine's checking (line 20), we create a new queue (line 22) and reserve
936 a scalar for the thread that we're likely to create later (line 21).
937 The while loop from lines 23 to line 31 grabs a scalar off the input
939 queue and checks against the prime this thread is responsible for.
940 Line 24 checks to see if there's a remainder when we modulo the number
941 to be checked against our prime. If there is one, the number must not
942 be evenly divisible by our prime, so we need to either pass it on to
943 the next thread if we've created one (line 26) or create a new thread
944 if we haven't.
945 The new thread creation is line 29. We pass on to it a reference to
947 the queue we've created, and the prime number we've found.
948 Finally, once the loop terminates (because we got a 0 or undef in the
950 queue, which serves as a note to die), we pass on the notice to our
951 child and wait for it to exit if we've created a child (Lines 32 and
952 37).
953 Meanwhile, back in the main thread, we create a queue (line 9) and the
955 initial child thread (line 10), and pre-seed it with the first prime:
956 2. Then we queue all the numbers from 3 to 1000 for checking (lines
957 12-14), then queue a die notice (line 16) and wait for the first child
958 thread to terminate (line 17). Because a child won't die until its
959 child has died, we know that we're done once we return from the join.
960 That's how it works. It's pretty simple; as with many Perl programs,
962 the explanation is much longer than the program.
963
965 A complete thread tutorial could fill a book (and has, many times), but
966 this should get you well on your way. The final authority on how
967 Perl's threads behave is the documentation bundled with the Perl dis‐
968 tribution, but with what we've covered in this article, you should be
969 well on your way to becoming a threaded Perl expert.
970
972 Here's a short bibliography courtesy of Jürgen Christoffel:
973 Introductory Texts
975 Birrell, Andrew D. An Introduction to Programming with Threads. Digital
977 Equipment Corporation, 1989, DEC-SRC Research Report #35 online as
978 http://www.research.digital.com/SRC/staff/birrell/bib.html (highly rec‐
979 ommended)
980 Robbins, Kay. A., and Steven Robbins. Practical Unix Programming: A
982 Guide to Concurrency, Communication, and Multithreading. Prentice-Hall,
983 1996.
984 Lewis, Bill, and Daniel J. Berg. Multithreaded Programming with
986 Pthreads. Prentice Hall, 1997, ISBN 0-13-443698-9 (a well-written
987 introduction to threads).
988 Nelson, Greg (editor). Systems Programming with Modula-3. Prentice
990 Hall, 1991, ISBN 0-13-590464-1.
991 Nichols, Bradford, Dick Buttlar, and Jacqueline Proulx Farrell.
993 Pthreads Programming. O'Reilly & Associates, 1996, ISBN 156592-115-1
994 (covers POSIX threads).
995 OS-Related References
997 Boykin, Joseph, David Kirschen, Alan Langerman, and Susan LoVerso. Pro‐
999 gramming under Mach. Addison-Wesley, 1994, ISBN 0-201-52739-1.
1000 Tanenbaum, Andrew S. Distributed Operating Systems. Prentice Hall,
1002 1995, ISBN 0-13-219908-4 (great textbook).
1003 Silberschatz, Abraham, and Peter B. Galvin. Operating System Concepts,
1005 4th ed. Addison-Wesley, 1995, ISBN 0-201-59292-4
1006 Other References
1008 Arnold, Ken and James Gosling. The Java Programming Language, 2nd ed.
1010 Addison-Wesley, 1998, ISBN 0-201-31006-6.
1011 Le Sergent, T. and B. Berthomieu. "Incremental MultiThreaded Garbage
1013 Collection on Virtually Shared Memory Architectures" in Memory Manage‐
1014 ment: Proc. of the International Workshop IWMM 92, St. Malo, France,
1015 September 1992, Yves Bekkers and Jacques Cohen, eds. Springer, 1992,
1016 ISBN 3540-55940-X (real-life thread applications).
1017
1019 Thanks (in no particular order) to Chaim Frenkel, Steve Fink, Gurusamy
1020 Sarathy, Ilya Zakharevich, Benjamin Sugars, Jürgen Christoffel, Joshua
1021 Pritikin, and Alan Burlison, for their help in reality-checking and
1022 polishing this article. Big thanks to Tom Christiansen for his rewrite
1023 of the prime number generator.
1024
1026 Dan Sugalski <sugalskd@ous.edu>
1027
1029 This article originally appeared in The Perl Journal #10, and is copy‐
1030 right 1998 The Perl Journal. It appears courtesy of Jon Orwant and The
1031 Perl Journal. This document may be distributed under the same terms as
1032 Perl itself.
1033perl v5.8.8 2006-01-07 PERLOTHRTUT(1)