This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
g++/Solaris: force d_dlopen=define
[perl5.git] / pod / perlthrtut.pod
CommitLineData
2605996a
JH
1=head1 NAME
2
3perlthrtut - tutorial on threads in Perl
4
5=head1 DESCRIPTION
6
53d7eaa8 7B<NOTE>: this tutorial describes the new Perl threading flavour
9316ed2f
JH
8introduced in Perl 5.6.0 called interpreter threads, or B<ithreads>
9for short. In this model each thread runs in its own Perl interpreter,
10and any data sharing between threads must be explicit.
11
12There is another older Perl threading flavour called the 5.005 model,
13unsurprisingly for 5.005 versions of Perl. The old model is known to
0b390a82 14have problems, is deprecated, and support for it will be removed in release
9316ed2f
JH
155.10. You are strongly encouraged to migrate any existing 5.005
16threads code to the new model as soon as possible.
2a4bf773 17
53d7eaa8 18You can see which (or neither) threading flavour you have by
6eded8f3 19running C<perl -V> and looking at the C<Platform> section.
53d7eaa8
JH
20If you have C<useithreads=define> you have ithreads, if you
21have C<use5005threads=define> you have 5.005 threads.
22If you have neither, you don't have any thread support built in.
23If you have both, you are in trouble.
2605996a 24
bfce6503
DM
25The user-level interface to the 5.005 threads was via the L<Threads>
26class, while ithreads uses the L<threads> class. Note the change in case.
27
28=head1 Status
29
30The ithreads code has been available since Perl 5.6.0, and is considered
31stable. The user-level interface to ithreads (the L<threads> classes)
9e75ef81
IT
32appeared in the 5.8.0 release, and as of this time is considered stable
33although it should be treated with caution as with all new features.
2605996a 34
c975c451
AB
35=head1 What Is A Thread Anyway?
36
37A thread is a flow of control through a program with a single
38execution point.
39
40Sounds an awful lot like a process, doesn't it? Well, it should.
41Threads are one of the pieces of a process. Every process has at least
42one thread and, up until now, every process running Perl had only one
43thread. With 5.8, though, you can create extra threads. We're going
44to show you how, when, and why.
45
46=head1 Threaded Program Models
47
48There are three basic ways that you can structure a threaded
49program. Which model you choose depends on what you need your program
50to do. For many non-trivial threaded programs you'll need to choose
51different models for different pieces of your program.
52
53=head2 Boss/Worker
54
b432a672
AL
55The boss/worker model usually has one "boss" thread and one or more
56"worker" threads. The boss thread gathers or generates tasks that need
c975c451
AB
57to be done, then parcels those tasks out to the appropriate worker
58thread.
59
60This model is common in GUI and server programs, where a main thread
61waits for some event and then passes that event to the appropriate
62worker threads for processing. Once the event has been passed on, the
63boss thread goes back to waiting for another event.
64
65The boss thread does relatively little work. While tasks aren't
66necessarily performed faster than with any other method, it tends to
67have the best user-response times.
68
69=head2 Work Crew
70
71In the work crew model, several threads are created that do
72essentially the same thing to different pieces of data. It closely
73mirrors classical parallel processing and vector processors, where a
74large array of processors do the exact same thing to many pieces of
75data.
76
77This model is particularly useful if the system running the program
78will distribute multiple threads across different processors. It can
79also be useful in ray tracing or rendering engines, where the
80individual threads can pass on interim results to give the user visual
81feedback.
82
83=head2 Pipeline
84
85The pipeline model divides up a task into a series of steps, and
86passes the results of one step on to the thread processing the
87next. Each thread does one thing to each piece of data and passes the
88results to the next thread in line.
89
90This model makes the most sense if you have multiple processors so two
91or more threads will be executing in parallel, though it can often
92make sense in other contexts as well. It tends to keep the individual
93tasks small and simple, as well as allowing some parts of the pipeline
94to block (on I/O or system calls, for example) while other parts keep
95going. If you're running different parts of the pipeline on different
96processors you may also take advantage of the caches on each
97processor.
98
99This model is also handy for a form of recursive programming where,
100rather than having a subroutine call itself, it instead creates
101another thread. Prime and Fibonacci generators both map well to this
102form of the pipeline model. (A version of a prime number generator is
103presented later on.)
104
bfce6503 105=head1 What kind of threads are Perl threads?
c975c451
AB
106
107If you have experience with other thread implementations, you might
108find that things aren't quite what you expect. It's very important to
109remember when dealing with Perl threads that Perl Threads Are Not X
110Threads, for all values of X. They aren't POSIX threads, or
111DecThreads, or Java's Green threads, or Win32 threads. There are
112similarities, and the broad concepts are the same, but if you start
113looking for implementation details you're going to be either
114disappointed or confused. Possibly both.
115
116This is not to say that Perl threads are completely different from
117everything that's ever come before--they're not. Perl's threading
118model owes a lot to other thread models, especially POSIX. Just as
119Perl is not C, though, Perl threads are not POSIX threads. So if you
120find yourself looking for mutexes, or thread priorities, it's time to
121step back a bit and think about what you want to do and how Perl can
122do it.
123
6eded8f3 124However it is important to remember that Perl threads cannot magically
c975c451 125do things unless your operating systems threads allows it. So if your
bfce6503 126system blocks the entire process on sleep(), Perl usually will as well.
c975c451 127
9316ed2f
JH
128Perl Threads Are Different.
129
cf5baa48 130=head1 Thread-Safe Modules
c975c451 131
cf5baa48 132The addition of threads has changed Perl's internals
c975c451 133substantially. There are implications for people who write
cf5baa48
JH
134modules with XS code or external libraries. However, since perl data is
135not shared among threads by default, Perl modules stand a high chance of
136being thread-safe or can be made thread-safe easily. Modules that are not
137tagged as thread-safe should be tested or code reviewed before being used
138in production code.
c975c451
AB
139
140Not all modules that you might use are thread-safe, and you should
141always assume a module is unsafe unless the documentation says
142otherwise. This includes modules that are distributed as part of the
143core. Threads are a new feature, and even some of the standard
bfce6503 144modules aren't thread-safe.
c975c451 145
cf5baa48 146Even if a module is thread-safe, it doesn't mean that the module is optimized
6eded8f3
SG
147to work well with threads. A module could possibly be rewritten to utilize
148the new features in threaded Perl to increase performance in a threaded
149environment.
c975c451
AB
150
151If you're using a module that's not thread-safe for some reason, you
cf5baa48
JH
152can protect yourself by using it from one, and only one thread at all.
153If you need multiple threads to access such a module, you can use semaphores and
154lots of programming discipline to control access to it. Semaphores
155are covered in L</"Basic semaphores">.
9316ed2f 156
cf5baa48 157See also L</"Thread-Safety of System Libraries">.
c975c451
AB
158
159=head1 Thread Basics
160
161The core L<threads> module provides the basic functions you need to write
162threaded programs. In the following sections we'll cover the basics,
163showing you what you need to do to create a threaded program. After
164that, we'll go over some of the features of the L<threads> module that
165make threaded programming easier.
166
167=head2 Basic Thread Support
168
6eded8f3 169Thread support is a Perl compile-time option - it's something that's
c975c451
AB
170turned on or off when Perl is built at your site, rather than when
171your programs are compiled. If your Perl wasn't compiled with thread
172support enabled, then any attempt to use threads will fail.
173
c975c451
AB
174Your programs can use the Config module to check whether threads are
175enabled. If your program can't run without them, you can say something
176like:
177
9316ed2f 178 $Config{useithreads} or die "Recompile Perl with threads to run this program.";
c975c451
AB
179
180A possibly-threaded program using a possibly-threaded module might
181have code like this:
182
cf5baa48
JH
183 use Config;
184 use MyMod;
c975c451 185
9316ed2f 186 BEGIN {
cf5baa48
JH
187 if ($Config{useithreads}) {
188 # We have threads
189 require MyMod_threaded;
190 import MyMod_threaded;
191 } else {
192 require MyMod_unthreaded;
193 import MyMod_unthreaded;
9316ed2f 194 }
cf5baa48 195 }
c975c451
AB
196
197Since code that runs both with and without threads is usually pretty
198messy, it's best to isolate the thread-specific code in its own
199module. In our example above, that's what MyMod_threaded is, and it's
200only imported if we're running on a threaded Perl.
201
8f95bfb9
EM
202=head2 A Note about the Examples
203
204Although thread support is considered to be stable, there are still a number
205of quirks that may startle you when you try out any of the examples below.
206In a real situation, care should be taken that all threads are finished
207executing before the program exits. That care has B<not> been taken in these
208examples in the interest of simplicity. Running these examples "as is" will
209produce error messages, usually caused by the fact that there are still
210threads running when the program exits. You should not be alarmed by this.
211Future versions of Perl may fix this problem.
212
c975c451
AB
213=head2 Creating Threads
214
215The L<threads> package provides the tools you need to create new
9e75ef81 216threads. Like any other module, you need to tell Perl that you want to use
c975c451
AB
217it; C<use threads> imports all the pieces you need to create basic
218threads.
219
9e75ef81 220The simplest, most straightforward way to create a thread is with new():
c975c451 221
0b390a82 222 use threads;
c975c451
AB
223
224 $thr = threads->new(\&sub1);
225
0b390a82
RGS
226 sub sub1 {
227 print "In the thread\n";
c975c451
AB
228 }
229
230The new() method takes a reference to a subroutine and creates a new
231thread, which starts executing in the referenced subroutine. Control
232then passes both to the subroutine and the caller.
233
234If you need to, your program can pass parameters to the subroutine as
235part of the thread startup. Just include the list of parameters as
236part of the C<threads::new> call, like this:
237
0b390a82 238 use threads;
bfce6503 239
0b390a82
RGS
240 $Param3 = "foo";
241 $thr = threads->new(\&sub1, "Param 1", "Param 2", $Param3);
242 $thr = threads->new(\&sub1, @ParamList);
8f95bfb9 243 $thr = threads->new(\&sub1, qw(Param1 Param2 Param3));
c975c451 244
0b390a82
RGS
245 sub sub1 {
246 my @InboundParameters = @_;
247 print "In the thread\n";
248 print "got parameters >", join("<>", @InboundParameters), "<\n";
c975c451
AB
249 }
250
251
252The last example illustrates another feature of threads. You can spawn
253off several threads using the same subroutine. Each thread executes
254the same subroutine, but in a separate thread with a separate
255environment and potentially separate arguments.
256
9316ed2f 257C<create()> is a synonym for C<new()>.
bfce6503 258
c975c451
AB
259=head2 Waiting For A Thread To Exit
260
261Since threads are also subroutines, they can return values. To wait
6eded8f3
SG
262for a thread to exit and extract any values it might return, you can
263use the join() method:
c975c451 264
0b390a82 265 use threads;
bfce6503 266
c975c451
AB
267 $thr = threads->new(\&sub1);
268
0b390a82
RGS
269 @ReturnData = $thr->join;
270 print "Thread returned @ReturnData";
c975c451
AB
271
272 sub sub1 { return "Fifty-six", "foo", 2; }
273
274In the example above, the join() method returns as soon as the thread
275ends. In addition to waiting for a thread to finish and gathering up
276any values that the thread might have returned, join() also performs
277any OS cleanup necessary for the thread. That cleanup might be
278important, especially for long-running programs that spawn lots of
279threads. If you don't want the return values and don't want to wait
280for the thread to finish, you should call the detach() method
bfce6503 281instead, as described next.
c975c451
AB
282
283=head2 Ignoring A Thread
284
285join() does three things: it waits for a thread to exit, cleans up
286after it, and returns any data the thread may have produced. But what
287if you're not interested in the thread's return values, and you don't
288really care when the thread finishes? All you want is for the thread
289to get cleaned up after when it's done.
290
291In this case, you use the detach() method. Once a thread is detached,
292it'll run until it's finished, then Perl will clean up after it
293automatically.
294
0b390a82 295 use threads;
bfce6503 296
6eded8f3 297 $thr = threads->new(\&sub1); # Spawn the thread
c975c451
AB
298
299 $thr->detach; # Now we officially don't care any more
300
cf5baa48 301 sub sub1 {
0b390a82
RGS
302 $a = 0;
303 while (1) {
304 $a++;
305 print "\$a is $a\n";
306 sleep 1;
307 }
c975c451
AB
308 }
309
bfce6503
DM
310Once a thread is detached, it may not be joined, and any return data
311that it might have produced (if it was done and waiting for a join) is
c975c451
AB
312lost.
313
314=head1 Threads And Data
315
316Now that we've covered the basics of threads, it's time for our next
317topic: data. Threading introduces a couple of complications to data
318access that non-threaded programs never need to worry about.
319
320=head2 Shared And Unshared Data
321
bfce6503
DM
322The biggest difference between Perl ithreads and the old 5.005 style
323threading, or for that matter, to most other threading systems out there,
324is that by default, no data is shared. When a new perl thread is created,
325all the data associated with the current thread is copied to the new
326thread, and is subsequently private to that new thread!
327This is similar in feel to what happens when a UNIX process forks,
328except that in this case, the data is just copied to a different part of
329memory within the same process rather than a real fork taking place.
c975c451 330
9e75ef81 331To make use of threading however, one usually wants the threads to share
bfce6503
DM
332at least some data between themselves. This is done with the
333L<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++; $bar++ })->join;
818c4caa 341
bfce6503
DM
342 print "$foo\n"; #prints 2 since $foo is shared
343 print "$bar\n"; #prints 1 since $bar is not shared
344
345In the case of a shared array, all the array's elements are shared, and for
346a shared hash, all the keys and values are shared. This places
347restrictions on what may be assigned to shared array and hash elements: only
348simple values or references to shared variables are allowed - this is
f3278b06 349so that a private variable can't accidentally become shared. A bad
bfce6503
DM
350assignment 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
56ca1794
IT
362 $hash{a} = $var # okay - copy-by-value: same effect as previous
363 $hash{a} = $svar # okay - copy-by-value: same effect as previous
bfce6503
DM
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
368Note that a shared variable guarantees that if two or more threads try to
369modify it at the same time, the internal state of the variable will not
370become corrupted. However, there are no guarantees beyond this, as
371explained in the next section.
c975c451 372
6eded8f3 373=head2 Thread Pitfalls: Races
c975c451
AB
374
375While threads bring a new set of useful tools, they also bring a
376number of pitfalls. One pitfall is the race condition:
377
0b390a82 378 use threads;
c975c451 379 use threads::shared;
bfce6503 380
0b390a82
RGS
381 my $a : shared = 1;
382 $thr1 = threads->new(\&sub1);
383 $thr2 = threads->new(\&sub2);
c975c451
AB
384
385 $thr1->join;
386 $thr2->join;
387 print "$a\n";
388
bfce6503
DM
389 sub sub1 { my $foo = $a; $a = $foo + 1; }
390 sub sub2 { my $bar = $a; $a = $bar + 1; }
c975c451
AB
391
392What do you think $a will be? The answer, unfortunately, is "it
393depends." Both sub1() and sub2() access the global variable $a, once
394to read and once to write. Depending on factors ranging from your
395thread implementation's scheduling algorithm to the phase of the moon,
396$a can be 2 or 3.
397
398Race conditions are caused by unsynchronized access to shared
399data. Without explicit synchronization, there's no way to be sure that
400nothing has happened to the shared data between the time you access it
401and the time you update it. Even this simple code fragment has the
402possibility of error:
403
0b390a82 404 use threads;
c975c451
AB
405 my $a : shared = 2;
406 my $b : shared;
407 my $c : shared;
0b390a82 408 my $thr1 = threads->create(sub { $b = $a; $a = $b + 1; });
c975c451 409 my $thr2 = threads->create(sub { $c = $a; $a = $c + 1; });
8f95bfb9
EM
410 $thr1->join;
411 $thr2->join;
c975c451
AB
412
413Two threads both access $a. Each thread can potentially be interrupted
414at any point, or be executed in any order. At the end, $a could be 3
415or 4, and both $b and $c could be 2 or 3.
416
bfce6503
DM
417Even C<$a += 5> or C<$a++> are not guaranteed to be atomic.
418
c975c451
AB
419Whenever your program accesses data or resources that can be accessed
420by other threads, you must take steps to coordinate access or risk
bfce6503
DM
421data inconsistency and race conditions. Note that Perl will protect its
422internals from your race conditions, but it won't protect you from you.
423
f3278b06 424=head1 Synchronization and control
bfce6503
DM
425
426Perl provides a number of mechanisms to coordinate the interactions
427between themselves and their data, to avoid race conditions and the like.
428Some of these are designed to resemble the common techniques used in thread
429libraries such as C<pthreads>; others are Perl-specific. Often, the
9e75ef81 430standard techniques are clumsy and difficult to get right (such as
bfce6503
DM
431condition waits). Where possible, it is usually easier to use Perlish
432techniques such as queues, which remove some of the hard work involved.
c975c451
AB
433
434=head2 Controlling access: lock()
435
0b390a82 436The lock() function takes a shared variable and puts a lock on it.
a6d05634 437No other thread may lock the variable until the variable is unlocked
bfce6503 438by the thread holding the lock. Unlocking happens automatically
0b390a82 439when the locking thread exits the block that contains the call to the
bfce6503 440C<lock()> function. Using lock() is straightforward: this example has
f3278b06 441several threads doing some calculations in parallel, and occasionally
bfce6503
DM
442updating 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
8f95bfb9 455 $total += $result;
f3278b06 456 } # lock implicitly released at end of scope
bfce6503
DM
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";
c975c451 468
c975c451
AB
469lock() blocks the thread until the variable being locked is
470available. When lock() returns, your thread can be sure that no other
0b390a82 471thread can lock that variable until the block containing the
c975c451
AB
472lock exits.
473
474It's important to note that locks don't prevent access to the variable
475in question, only lock attempts. This is in keeping with Perl's
476longstanding tradition of courteous programming, and the advisory file
0b390a82 477locking that flock() gives you.
c975c451
AB
478
479You may lock arrays and hashes as well as scalars. Locking an array,
480though, will not block subsequent locks on array elements, just lock
481attempts on the array itself.
482
bfce6503 483Locks are recursive, which means it's okay for a thread to
c975c451 484lock a variable more than once. The lock will last until the outermost
bfce6503
DM
485lock() on the variable goes out of scope. For example:
486
487 my $x : shared;
488 doit();
489
490 sub doit {
491 {
492 {
493 lock($x); # wait for lock
8f95bfb9 494 lock($x); # NOOP - we already have the lock
bfce6503
DM
495 {
496 lock($x); # NOOP
497 {
498 lock($x); # NOOP
499 lockit_some_more();
500 }
501 }
502 } # *** implicit unlock here ***
503 }
504 }
505
506 sub lockit_some_more {
507 lock($x); # NOOP
508 } # nothing happens here
509
510Note that there is no unlock() function - the only way to unlock a
0b390a82 511variable is to allow it to go out of scope.
bfce6503
DM
512
513A lock can either be used to guard the data contained within the variable
514being locked, or it can be used to guard something else, like a section
515of code. In this latter case, the variable in question does not hold any
516useful data, and exists only for the purpose of being locked. In this
517respect, the variable behaves like the mutexes and basic semaphores of
518traditional thread libraries.
c975c451 519
bfce6503 520=head2 A Thread Pitfall: Deadlocks
c975c451 521
bfce6503 522Locks are a handy tool to synchronize access to data, and using them
c975c451 523properly is the key to safe shared data. Unfortunately, locks aren't
f3278b06 524without their dangers, especially when multiple locks are involved.
bfce6503 525Consider the following code:
c975c451 526
0b390a82
RGS
527 use threads;
528
529 my $a : shared = 4;
530 my $b : shared = "foo";
531 my $thr1 = threads->new(sub {
532 lock($a);
533 sleep 20;
534 lock($b);
535 });
536 my $thr2 = threads->new(sub {
537 lock($b);
538 sleep 20;
539 lock($a);
c975c451
AB
540 });
541
542This program will probably hang until you kill it. The only way it
bfce6503 543won't hang is if one of the two threads acquires both locks
c975c451
AB
544first. A guaranteed-to-hang version is more complicated, but the
545principle is the same.
546
bfce6503
DM
547The first thread will grab a lock on $a, then, after a pause during which
548the second thread has probably had time to do some work, try to grab a
549lock on $b. Meanwhile, the second thread grabs a lock on $b, then later
550tries to grab a lock on $a. The second lock attempt for both threads will
551block, each waiting for the other to release its lock.
c975c451
AB
552
553This condition is called a deadlock, and it occurs whenever two or
554more threads are trying to get locks on resources that the others
555own. Each thread will block, waiting for the other to release a lock
556on a resource. That never happens, though, since the thread with the
557resource is itself waiting for a lock to be released.
558
559There are a number of ways to handle this sort of problem. The best
560way is to always have all threads acquire locks in the exact same
561order. If, for example, you lock variables $a, $b, and $c, always lock
562$a before $b, and $b before $c. It's also best to hold on to locks for
563as short a period of time to minimize the risks of deadlock.
564
48b96218 565The other synchronization primitives described below can suffer from
bfce6503
DM
566similar problems.
567
c975c451
AB
568=head2 Queues: Passing Data Around
569
570A queue is a special thread-safe object that lets you put data in one
571end and take it out the other without having to worry about
572synchronization issues. They're pretty straightforward, and look like
573this:
574
0b390a82 575 use threads;
83272a45 576 use Thread::Queue;
c975c451 577
0b390a82
RGS
578 my $DataQueue = Thread::Queue->new;
579 $thr = threads->new(sub {
580 while ($DataElement = $DataQueue->dequeue) {
c975c451 581 print "Popped $DataElement off the queue\n";
0b390a82
RGS
582 }
583 });
c975c451 584
0b390a82
RGS
585 $DataQueue->enqueue(12);
586 $DataQueue->enqueue("A", "B", "C");
587 $DataQueue->enqueue(\$thr);
588 sleep 10;
c975c451 589 $DataQueue->enqueue(undef);
8f95bfb9 590 $thr->join;
c975c451 591
83272a45 592You create the queue with C<new Thread::Queue>. Then you can
6eded8f3
SG
593add lists of scalars onto the end with enqueue(), and pop scalars off
594the front of it with dequeue(). A queue has no fixed size, and can grow
595as needed to hold everything pushed on to it.
c975c451
AB
596
597If a queue is empty, dequeue() blocks until another thread enqueues
598something. This makes queues ideal for event loops and other
599communications between threads.
600
c975c451
AB
601=head2 Semaphores: Synchronizing Data Access
602
bfce6503 603Semaphores are a kind of generic locking mechanism. In their most basic
fa11829f 604form, they behave very much like lockable scalars, except that they
bfce6503
DM
605can't hold data, and that they must be explicitly unlocked. In their
606advanced form, they act like a kind of counter, and can allow multiple
607threads to have the 'lock' at any one time.
2605996a 608
bfce6503 609=head2 Basic semaphores
2605996a 610
bfce6503
DM
611Semaphores have two methods, down() and up(): down() decrements the resource
612count, while up increments it. Calls to down() will block if the
c975c451
AB
613semaphore's current count would decrement below zero. This program
614gives a quick demonstration:
615
536bca94 616 use threads;
0b390a82 617 use Thread::Semaphore;
bfce6503 618
0b390a82 619 my $semaphore = new Thread::Semaphore;
bfce6503 620 my $GlobalVariable : shared = 0;
2605996a 621
0b390a82
RGS
622 $thr1 = new threads \&sample_sub, 1;
623 $thr2 = new threads \&sample_sub, 2;
c975c451 624 $thr3 = new threads \&sample_sub, 3;
2605996a 625
0b390a82
RGS
626 sub sample_sub {
627 my $SubNumber = shift @_;
628 my $TryCount = 10;
629 my $LocalCopy;
630 sleep 1;
631 while ($TryCount--) {
632 $semaphore->down;
633 $LocalCopy = $GlobalVariable;
634 print "$TryCount tries left for sub $SubNumber (\$GlobalVariable is $GlobalVariable)\n";
635 sleep 2;
636 $LocalCopy++;
637 $GlobalVariable = $LocalCopy;
638 $semaphore->up;
639 }
c975c451 640 }
6eded8f3 641
8f95bfb9
EM
642 $thr1->join;
643 $thr2->join;
644 $thr3->join;
2605996a 645
c975c451
AB
646The three invocations of the subroutine all operate in sync. The
647semaphore, though, makes sure that only one thread is accessing the
648global variable at once.
2605996a 649
bfce6503 650=head2 Advanced Semaphores
2605996a 651
c975c451
AB
652By default, semaphores behave like locks, letting only one thread
653down() them at a time. However, there are other uses for semaphores.
2605996a 654
6eded8f3
SG
655Each semaphore has a counter attached to it. By default, semaphores are
656created with the counter set to one, down() decrements the counter by
657one, and up() increments by one. However, we can override any or all
658of these defaults simply by passing in different values:
659
660 use threads;
83272a45
JH
661 use Thread::Semaphore;
662 my $semaphore = Thread::Semaphore->new(5);
6eded8f3
SG
663 # Creates a semaphore with the counter set to five
664
665 $thr1 = threads->new(\&sub1);
666 $thr2 = threads->new(\&sub1);
667
668 sub sub1 {
669 $semaphore->down(5); # Decrements the counter by five
670 # Do stuff here
671 $semaphore->up(5); # Increment the counter by five
672 }
673
8f95bfb9
EM
674 $thr1->detach;
675 $thr2->detach;
6eded8f3
SG
676
677If down() attempts to decrement the counter below zero, it blocks until
678the counter is large enough. Note that while a semaphore can be created
679with a starting count of zero, any up() or down() always changes the
680counter by at least one, and so $semaphore->down(0) is the same as
681$semaphore->down(1).
2605996a 682
c975c451
AB
683The question, of course, is why would you do something like this? Why
684create a semaphore with a starting count that's not one, or why
685decrement/increment it by more than one? The answer is resource
686availability. Many resources that you want to manage access for can be
687safely used by more than one thread at once.
2605996a 688
c975c451
AB
689For example, let's take a GUI driven program. It has a semaphore that
690it uses to synchronize access to the display, so only one thread is
691ever drawing at once. Handy, but of course you don't want any thread
692to start drawing until things are properly set up. In this case, you
693can create a semaphore with a counter set to zero, and up it when
694things are ready for drawing.
2605996a 695
c975c451
AB
696Semaphores with counters greater than one are also useful for
697establishing quotas. Say, for example, that you have a number of
698threads that can do I/O at once. You don't want all the threads
699reading or writing at once though, since that can potentially swamp
700your I/O channels, or deplete your process' quota of filehandles. You
701can use a semaphore initialized to the number of concurrent I/O
702requests (or open files) that you want at any one time, and have your
703threads quietly block and unblock themselves.
2605996a 704
c975c451
AB
705Larger increments or decrements are handy in those cases where a
706thread needs to check out or return a number of resources at once.
2605996a 707
bfce6503
DM
708=head2 cond_wait() and cond_signal()
709
710These two functions can be used in conjunction with locks to notify
711co-operating threads that a resource has become available. They are
712very similar in use to the functions found in C<pthreads>. However
713for most purposes, queues are simpler to use and more intuitive. See
714L<threads::shared> for more details.
2605996a 715
536bca94
EM
716=head2 Giving up control
717
718There are times when you may find it useful to have a thread
719explicitly give up the CPU to another thread. You may be doing something
720processor-intensive and want to make sure that the user-interface thread
721gets called frequently. Regardless, there are times that you might want
722a thread to give up the processor.
723
724Perl's threading package provides the yield() function that does
725this. yield() is pretty straightforward, and works like this:
726
0b390a82 727 use threads;
536bca94
EM
728
729 sub loop {
0b390a82
RGS
730 my $thread = shift;
731 my $foo = 50;
732 while($foo--) { print "in thread $thread\n" }
733 threads->yield;
734 $foo = 50;
735 while($foo--) { print "in thread $thread\n" }
536bca94
EM
736 }
737
738 my $thread1 = threads->new(\&loop, 'first');
739 my $thread2 = threads->new(\&loop, 'second');
740 my $thread3 = threads->new(\&loop, 'third');
741
742It is important to remember that yield() is only a hint to give up the CPU,
743it depends on your hardware, OS and threading libraries what actually happens.
744B<On many operating systems, yield() is a no-op.> Therefore it is important
745to note that one should not build the scheduling of the threads around
746yield() calls. It might work on your platform but it won't work on another
747platform.
748
c975c451
AB
749=head1 General Thread Utility Routines
750
751We've covered the workhorse parts of Perl's threading package, and
752with these tools you should be well on your way to writing threaded
753code and packages. There are a few useful little pieces that didn't
754really fit in anyplace else.
755
756=head2 What Thread Am I In?
757
bfce6503
DM
758The C<< threads->self >> class method provides your program with a way to
759get an object representing the thread it's currently in. You can use this
6eded8f3 760object in the same way as the ones returned from thread creation.
c975c451
AB
761
762=head2 Thread IDs
763
764tid() is a thread object method that returns the thread ID of the
765thread the object represents. Thread IDs are integers, with the main
766thread in a program being 0. Currently Perl assigns a unique tid to
767every thread ever created in your program, assigning the first thread
768to be created a tid of 1, and increasing the tid by 1 for each new
769thread that's created.
770
771=head2 Are These Threads The Same?
772
0b390a82 773The equal() method takes two thread objects and returns true
c975c451
AB
774if the objects represent the same thread, and false if they don't.
775
776Thread objects also have an overloaded == comparison so that you can do
777comparison on them as you would with normal objects.
778
779=head2 What Threads Are Running?
780
bfce6503 781C<< threads->list >> returns a list of thread objects, one for each thread
c975c451
AB
782that's currently running and not detached. Handy for a number of things,
783including cleaning up at the end of your program:
784
0b390a82
RGS
785 # Loop through all the threads
786 foreach $thr (threads->list) {
787 # Don't join the main thread or ourselves
788 if ($thr->tid && !threads::equal($thr, threads->self)) {
789 $thr->join;
790 }
c975c451
AB
791 }
792
bfce6503
DM
793If some threads have not finished running when the main Perl thread
794ends, Perl will warn you about it and die, since it is impossible for Perl
6eded8f3 795to clean up itself while other threads are running
c975c451
AB
796
797=head1 A Complete Example
798
799Confused yet? It's time for an example program to show some of the
800things we've covered. This program finds prime numbers using threads.
801
802 1 #!/usr/bin/perl -w
803 2 # prime-pthread, courtesy of Tom Christiansen
804 3
805 4 use strict;
806 5
807 6 use threads;
83272a45 808 7 use Thread::Queue;
c975c451 809 8
83272a45 810 9 my $stream = new Thread::Queue;
c975c451
AB
811 10 my $kid = new threads(\&check_num, $stream, 2);
812 11
813 12 for my $i ( 3 .. 1000 ) {
814 13 $stream->enqueue($i);
0b390a82 815 14 }
c975c451
AB
816 15
817 16 $stream->enqueue(undef);
8f95bfb9 818 17 $kid->join;
c975c451
AB
819 18
820 19 sub check_num {
821 20 my ($upstream, $cur_prime) = @_;
822 21 my $kid;
83272a45 823 22 my $downstream = new Thread::Queue;
c975c451
AB
824 23 while (my $num = $upstream->dequeue) {
825 24 next unless $num % $cur_prime;
826 25 if ($kid) {
827 26 $downstream->enqueue($num);
828 27 } else {
829 28 print "Found prime $num\n";
830 29 $kid = new threads(\&check_num, $downstream, $num);
831 30 }
0b390a82 832 31 }
c975c451 833 32 $downstream->enqueue(undef) if $kid;
8f95bfb9 834 33 $kid->join if $kid;
c975c451
AB
835 34 }
836
837This program uses the pipeline model to generate prime numbers. Each
838thread in the pipeline has an input queue that feeds numbers to be
839checked, a prime number that it's responsible for, and an output queue
9e75ef81 840into which it funnels numbers that have failed the check. If the thread
c975c451
AB
841has a number that's failed its check and there's no child thread, then
842the thread must have found a new prime number. In that case, a new
843child thread is created for that prime and stuck on the end of the
844pipeline.
845
6eded8f3 846This probably sounds a bit more confusing than it really is, so let's
c975c451
AB
847go through this program piece by piece and see what it does. (For
848those of you who might be trying to remember exactly what a prime
849number is, it's a number that's only evenly divisible by itself and 1)
850
851The bulk of the work is done by the check_num() subroutine, which
852takes a reference to its input queue and a prime number that it's
853responsible for. After pulling in the input queue and the prime that
854the subroutine's checking (line 20), we create a new queue (line 22)
855and reserve a scalar for the thread that we're likely to create later
856(line 21).
857
858The while loop from lines 23 to line 31 grabs a scalar off the input
859queue and checks against the prime this thread is responsible
860for. Line 24 checks to see if there's a remainder when we modulo the
861number to be checked against our prime. If there is one, the number
862must not be evenly divisible by our prime, so we need to either pass
863it on to the next thread if we've created one (line 26) or create a
864new thread if we haven't.
865
866The new thread creation is line 29. We pass on to it a reference to
867the queue we've created, and the prime number we've found.
868
869Finally, once the loop terminates (because we got a 0 or undef in the
870queue, which serves as a note to die), we pass on the notice to our
6eded8f3 871child and wait for it to exit if we've created a child (lines 32 and
c975c451
AB
87237).
873
874Meanwhile, back in the main thread, we create a queue (line 9) and the
875initial child thread (line 10), and pre-seed it with the first prime:
8762. Then we queue all the numbers from 3 to 1000 for checking (lines
87712-14), then queue a die notice (line 16) and wait for the first child
878thread to terminate (line 17). Because a child won't die until its
879child has died, we know that we're done once we return from the join.
880
881That's how it works. It's pretty simple; as with many Perl programs,
882the explanation is much longer than the program.
883
536bca94
EM
884=head1 Different implementations of threads
885
886Some background on thread implementations from the operating system
887viewpoint. There are three basic categories of threads: user-mode threads,
888kernel threads, and multiprocessor kernel threads.
889
890User-mode threads are threads that live entirely within a program and
891its libraries. In this model, the OS knows nothing about threads. As
892far as it's concerned, your process is just a process.
893
894This is the easiest way to implement threads, and the way most OSes
895start. The big disadvantage is that, since the OS knows nothing about
896threads, if one thread blocks they all do. Typical blocking activities
897include most system calls, most I/O, and things like sleep().
898
899Kernel threads are the next step in thread evolution. The OS knows
900about kernel threads, and makes allowances for them. The main
901difference between a kernel thread and a user-mode thread is
902blocking. With kernel threads, things that block a single thread don't
903block other threads. This is not the case with user-mode threads,
904where the kernel blocks at the process level and not the thread level.
905
906This is a big step forward, and can give a threaded program quite a
907performance boost over non-threaded programs. Threads that block
908performing I/O, for example, won't block threads that are doing other
909things. Each process still has only one thread running at once,
910though, regardless of how many CPUs a system might have.
911
912Since kernel threading can interrupt a thread at any time, they will
913uncover some of the implicit locking assumptions you may make in your
914program. For example, something as simple as C<$a = $a + 2> can behave
915unpredictably with kernel threads if $a is visible to other
916threads, as another thread may have changed $a between the time it
917was fetched on the right hand side and the time the new value is
918stored.
919
920Multiprocessor kernel threads are the final step in thread
921support. With multiprocessor kernel threads on a machine with multiple
922CPUs, the OS may schedule two or more threads to run simultaneously on
923different CPUs.
924
925This can give a serious performance boost to your threaded program,
926since more than one thread will be executing at the same time. As a
927tradeoff, though, any of those nagging synchronization issues that
928might not have shown with basic kernel threads will appear with a
929vengeance.
930
931In addition to the different levels of OS involvement in threads,
932different OSes (and different thread implementations for a particular
933OS) allocate CPU cycles to threads in different ways.
934
935Cooperative multitasking systems have running threads give up control
936if one of two things happen. If a thread calls a yield function, it
937gives up control. It also gives up control if the thread does
938something that would cause it to block, such as perform I/O. In a
939cooperative multitasking implementation, one thread can starve all the
940others for CPU time if it so chooses.
941
942Preemptive multitasking systems interrupt threads at regular intervals
943while the system decides which thread should run next. In a preemptive
944multitasking system, one thread usually won't monopolize the CPU.
945
946On some systems, there can be cooperative and preemptive threads
947running simultaneously. (Threads running with realtime priorities
948often behave cooperatively, for example, while threads running at
949normal priorities behave preemptively.)
950
951Most modern operating systems support preemptive multitasking nowadays.
952
bfce6503
DM
953=head1 Performance considerations
954
955The main thing to bear in mind when comparing ithreads to other threading
956models is the fact that for each new thread created, a complete copy of
957all the variables and data of the parent thread has to be taken. Thus
958thread creation can be quite expensive, both in terms of memory usage and
959time spent in creation. The ideal way to reduce these costs is to have a
960relatively short number of long-lived threads, all created fairly early
961on - before the base thread has accumulated too much data. Of course, this
962may not always be possible, so compromises have to be made. However, after
963a thread has been created, its performance and extra memory usage should
964be little different than ordinary code.
965
966Also note that under the current implementation, shared variables
967use a little more memory and are a little slower than ordinary variables.
968
cf5baa48
JH
969=head1 Process-scope Changes
970
971Note that while threads themselves are separate execution threads and
972Perl data is thread-private unless explicitly shared, the threads can
973affect process-scope state, affecting all the threads.
974
975The most common example of this is changing the current working
976directory using chdir(). One thread calls chdir(), and the working
977directory of all the threads changes.
bdcfa4c7 978
cf5baa48
JH
979Even more drastic example of a process-scope change is chroot():
980the root directory of all the threads changes, and no thread can
981undo it (as opposed to chdir()).
982
983Further examples of process-scope changes include umask() and
984changing uids/gids.
985
986Thinking of mixing fork() and threads? Please lie down and wait
a95a5f75
DM
987until the feeling passes. Be aware that the semantics of fork() vary
988between platforms. For example, some UNIX systems copy all the current
989threads into the child process, while others only copy the thread that
990called fork(). You have been warned!
cf5baa48 991
b03ad8f6
JH
992Similarly, mixing signals and threads should not be attempted.
993Implementations are platform-dependent, and even the POSIX
994semantics may not be what you expect (and Perl doesn't even
995give you the full POSIX API).
996
cf5baa48
JH
997=head1 Thread-Safety of System Libraries
998
999Whether various library calls are thread-safe is outside the control
1000of Perl. Calls often suffering from not being thread-safe include:
bdcfa4c7 1001localtime(), gmtime(), get{gr,host,net,proto,serv,pw}*(), readdir(),
cf5baa48
JH
1002rand(), and srand() -- in general, calls that depend on some global
1003external state.
80bbcbc4 1004
cf5baa48 1005If the system Perl is compiled in has thread-safe variants of such
80bbcbc4 1006calls, they will be used. Beyond that, Perl is at the mercy of
cf5baa48 1007the thread-safety or -unsafety of the calls. Please consult your
80bbcbc4
JH
1008C library call documentation.
1009
af685957
JH
1010On some platforms the thread-safe library interfaces may fail if the
1011result buffer is too small (for example the user group databases may
1012be rather large, and the reentrant interfaces may have to carry around
1013a full snapshot of those databases). Perl will start with a small
1014buffer, but keep retrying and growing the result buffer
1015until the result fits. If this limitless growing sounds bad for
1016security or memory consumption reasons you can recompile Perl with
1017PERL_REENTRANT_MAXSIZE defined to the maximum number of bytes you will
1018allow.
bdcfa4c7 1019
c975c451
AB
1020=head1 Conclusion
1021
1022A complete thread tutorial could fill a book (and has, many times),
6eded8f3
SG
1023but with what we've covered in this introduction, you should be well
1024on your way to becoming a threaded Perl expert.
c975c451
AB
1025
1026=head1 Bibliography
1027
1028Here's a short bibliography courtesy of Jürgen Christoffel:
1029
1030=head2 Introductory Texts
1031
1032Birrell, Andrew D. An Introduction to Programming with
1033Threads. Digital Equipment Corporation, 1989, DEC-SRC Research Report
1034#35 online as
6eded8f3
SG
1035http://gatekeeper.dec.com/pub/DEC/SRC/research-reports/abstracts/src-rr-035.html
1036(highly recommended)
c975c451
AB
1037
1038Robbins, Kay. A., and Steven Robbins. Practical Unix Programming: A
1039Guide to Concurrency, Communication, and
1040Multithreading. Prentice-Hall, 1996.
1041
1042Lewis, Bill, and Daniel J. Berg. Multithreaded Programming with
1043Pthreads. Prentice Hall, 1997, ISBN 0-13-443698-9 (a well-written
1044introduction to threads).
1045
1046Nelson, Greg (editor). Systems Programming with Modula-3. Prentice
1047Hall, 1991, ISBN 0-13-590464-1.
1048
1049Nichols, Bradford, Dick Buttlar, and Jacqueline Proulx Farrell.
1050Pthreads Programming. O'Reilly & Associates, 1996, ISBN 156592-115-1
1051(covers POSIX threads).
1052
1053=head2 OS-Related References
1054
1055Boykin, Joseph, David Kirschen, Alan Langerman, and Susan
1056LoVerso. Programming under Mach. Addison-Wesley, 1994, ISBN
10570-201-52739-1.
1058
1059Tanenbaum, Andrew S. Distributed Operating Systems. Prentice Hall,
10601995, ISBN 0-13-219908-4 (great textbook).
1061
1062Silberschatz, Abraham, and Peter B. Galvin. Operating System Concepts,
10634th ed. Addison-Wesley, 1995, ISBN 0-201-59292-4
1064
1065=head2 Other References
1066
1067Arnold, Ken and James Gosling. The Java Programming Language, 2nd
1068ed. Addison-Wesley, 1998, ISBN 0-201-31006-6.
1069
b03ad8f6
JH
1070comp.programming.threads FAQ,
1071L<http://www.serpentine.com/~bos/threads-faq/>
1072
c975c451
AB
1073Le Sergent, T. and B. Berthomieu. "Incremental MultiThreaded Garbage
1074Collection on Virtually Shared Memory Architectures" in Memory
1075Management: Proc. of the International Workshop IWMM 92, St. Malo,
1076France, September 1992, Yves Bekkers and Jacques Cohen, eds. Springer,
10771992, ISBN 3540-55940-X (real-life thread applications).
1078
5e549d84
JH
1079Artur Bergman, "Where Wizards Fear To Tread", June 11, 2002,
1080L<http://www.perl.com/pub/a/2002/06/11/threads.html>
1081
c975c451
AB
1082=head1 Acknowledgements
1083
1084Thanks (in no particular order) to Chaim Frenkel, Steve Fink, Gurusamy
1085Sarathy, Ilya Zakharevich, Benjamin Sugars, Jürgen Christoffel, Joshua
1086Pritikin, and Alan Burlison, for their help in reality-checking and
1087polishing this article. Big thanks to Tom Christiansen for his rewrite
1088of the prime number generator.
1089
1090=head1 AUTHOR
1091
9316ed2f 1092Dan Sugalski E<lt>dan@sidhe.org<gt>
c975c451
AB
1093
1094Slightly modified by Arthur Bergman to fit the new thread model/module.
1095
cf5baa48
JH
1096Reworked slightly by Jörg Walter E<lt>jwalt@cpan.org<gt> to be more concise
1097about thread-safety of perl code.
1098
536bca94
EM
1099Rearranged slightly by Elizabeth Mattijsen E<lt>liz@dijkmat.nl<gt> to put
1100less emphasis on yield().
1101
c975c451
AB
1102=head1 Copyrights
1103
bfce6503
DM
1104The original version of this article originally appeared in The Perl
1105Journal #10, and is copyright 1998 The Perl Journal. It appears courtesy
1106of Jon Orwant and The Perl Journal. This document may be distributed
1107under the same terms as Perl itself.
2605996a 1108
53d7eaa8 1109For more information please see L<threads> and L<threads::shared>.