3 perlipc - Perl interprocess communication (signals, fifos, pipes, safe subprocesses, sockets, and semaphores)
7 The basic IPC facilities of Perl are built out of the good old Unix
8 signals, named pipes, pipe opens, the Berkeley socket routines, and SysV
9 IPC calls. Each is used in slightly different situations.
13 Perl uses a simple signal handling model: the %SIG hash contains names
14 or references of user-installed signal handlers. These handlers will
15 be called with an argument which is the name of the signal that
16 triggered it. A signal may be generated intentionally from a
17 particular keyboard sequence like control-C or control-Z, sent to you
18 from another process, or triggered automatically by the kernel when
19 special events transpire, like a child process exiting, your process
20 running out of stack space, or hitting file size limit.
22 For example, to trap an interrupt signal, set up a handler like this:
27 die "Somebody sent me a SIG$signame";
29 $SIG{INT} = 'catch_zap'; # could fail in modules
30 $SIG{INT} = \&catch_zap; # best strategy
32 Prior to Perl 5.7.3 it was necessary to do as little as you possibly
33 could in your handler; notice how all we do is set a global variable
34 and then raise an exception. That's because on most systems,
35 libraries are not re-entrant; particularly, memory allocation and I/O
36 routines are not. That meant that doing nearly I<anything> in your
37 handler could in theory trigger a memory fault and subsequent core
38 dump - see L<Deferred Signals> below.
40 The names of the signals are the ones listed out by C<kill -l> on your
41 system, or you can retrieve them from the Config module. Set up an
42 @signame list indexed by number to get the name and a %signo table
43 indexed by name to get the number:
46 defined $Config{sig_name} || die "No sigs?";
47 foreach $name (split(' ', $Config{sig_name})) {
53 So to check whether signal 17 and SIGALRM were the same, do just this:
55 print "signal #17 = $signame[17]\n";
57 print "SIGALRM is $signo{ALRM}\n";
60 You may also choose to assign the strings C<'IGNORE'> or C<'DEFAULT'> as
61 the handler, in which case Perl will try to discard the signal or do the
64 On most Unix platforms, the C<CHLD> (sometimes also known as C<CLD>) signal
65 has special behavior with respect to a value of C<'IGNORE'>.
66 Setting C<$SIG{CHLD}> to C<'IGNORE'> on such a platform has the effect of
67 not creating zombie processes when the parent process fails to C<wait()>
68 on its child processes (i.e. child processes are automatically reaped).
69 Calling C<wait()> with C<$SIG{CHLD}> set to C<'IGNORE'> usually returns
70 C<-1> on such platforms.
72 Some signals can be neither trapped nor ignored, such as
73 the KILL and STOP (but not the TSTP) signals. One strategy for
74 temporarily ignoring signals is to use a local() statement, which will be
75 automatically restored once your block is exited. (Remember that local()
76 values are "inherited" by functions called from within that block.)
79 local $SIG{INT} = 'IGNORE';
83 # interrupts still ignored, for now...
86 Sending a signal to a negative process ID means that you send the signal
87 to the entire Unix process-group. This code sends a hang-up signal to all
88 processes in the current process group (and sets $SIG{HUP} to IGNORE so
89 it doesn't kill itself):
92 local $SIG{HUP} = 'IGNORE';
94 # snazzy writing of: kill('HUP', -$$)
97 Another interesting signal to send is signal number zero. This doesn't
98 actually affect another process, but instead checks whether it's alive
99 or has changed its UID.
101 unless (kill 0 => $kid_pid) {
102 warn "something wicked happened to $kid_pid";
105 You might also want to employ anonymous functions for simple signal
108 $SIG{INT} = sub { die "\nOutta here!\n" };
110 But that will be problematic for the more complicated handlers that need
111 to reinstall themselves. Because Perl's signal mechanism is currently
112 based on the signal(3) function from the C library, you may sometimes be so
113 misfortunate as to run on systems where that function is "broken", that
114 is, it behaves in the old unreliable SysV way rather than the newer, more
115 reasonable BSD and POSIX fashion. So you'll see defensive people writing
116 signal handlers like this:
120 # loathe sysV: it makes us not only reinstate
121 # the handler, but place it after the wait
122 $SIG{CHLD} = \&REAPER;
124 $SIG{CHLD} = \&REAPER;
125 # now do something that forks...
129 use POSIX ":sys_wait_h";
132 # If a second child dies while in the signal handler caused by the
133 # first death, we won't get another signal. So must loop here else
134 # we will leave the unreaped child as a zombie. And the next time
135 # two children die we get another zombie. And so on.
136 while (($child = waitpid(-1,WNOHANG)) > 0) {
137 $Kid_Status{$child} = $?;
139 $SIG{CHLD} = \&REAPER; # still loathe sysV
141 $SIG{CHLD} = \&REAPER;
142 # do something that forks...
144 Signal handling is also used for timeouts in Unix, While safely
145 protected within an C<eval{}> block, you set a signal handler to trap
146 alarm signals and then schedule to have one delivered to you in some
147 number of seconds. Then try your blocking operation, clearing the alarm
148 when it's done but not before you've exited your C<eval{}> block. If it
149 goes off, you'll use die() to jump out of the block, much as you might
150 using longjmp() or throw() in other languages.
155 local $SIG{ALRM} = sub { die "alarm clock restart" };
157 flock(FH, 2); # blocking write lock
160 if ($@ and $@ !~ /alarm clock restart/) { die }
162 If the operation being timed out is system() or qx(), this technique
163 is liable to generate zombies. If this matters to you, you'll
164 need to do your own fork() and exec(), and kill the errant child process.
166 For more complex signal handling, you might see the standard POSIX
167 module. Lamentably, this is almost entirely undocumented, but
168 the F<t/lib/posix.t> file from the Perl source distribution has some
171 =head2 Handling the SIGHUP Signal in Daemons
173 A process that usually starts when the system boots and shuts down
174 when the system is shut down is called a daemon (Disk And Execution
175 MONitor). If a daemon process has a configuration file which is
176 modified after the process has been started, there should be a way to
177 tell that process to re-read its configuration file, without stopping
178 the process. Many daemons provide this mechanism using the C<SIGHUP>
179 signal handler. When you want to tell the daemon to re-read the file
180 you simply send it the C<SIGHUP> signal.
182 Not all platforms automatically reinstall their (native) signal
183 handlers after a signal delivery. This means that the handler works
184 only the first time the signal is sent. The solution to this problem
185 is to use C<POSIX> signal handlers if available, their behaviour
188 The following example implements a simple daemon, which restarts
189 itself every time the C<SIGHUP> signal is received. The actual code is
190 located in the subroutine C<code()>, which simply prints some debug
191 info to show that it works and should be replaced with the real code.
197 use File::Basename ();
198 use File::Spec::Functions;
202 # make the daemon cross-platform, so exec always calls the script
203 # itself with the right path, no matter how the script was invoked.
204 my $script = File::Basename::basename($0);
205 my $SELF = catfile $FindBin::Bin, $script;
207 # POSIX unmasks the sigprocmask properly
208 my $sigset = POSIX::SigSet->new();
209 my $action = POSIX::SigAction->new('sigHUP_handler',
212 POSIX::sigaction(&POSIX::SIGHUP, $action);
215 print "got SIGHUP\n";
216 exec($SELF, @ARGV) or die "Couldn't restart: $!\n";
223 print "ARGV: @ARGV\n";
235 A named pipe (often referred to as a FIFO) is an old Unix IPC
236 mechanism for processes communicating on the same machine. It works
237 just like a regular, connected anonymous pipes, except that the
238 processes rendezvous using a filename and don't have to be related.
240 To create a named pipe, use the Unix command mknod(1) or on some
241 systems, mkfifo(1). These may not be in your normal path.
243 # system return val is backwards, so && not ||
245 $ENV{PATH} .= ":/etc:/usr/etc";
246 if ( system('mknod', $path, 'p')
247 && system('mkfifo', $path) )
249 die "mk{nod,fifo} $path failed";
253 A fifo is convenient when you want to connect a process to an unrelated
254 one. When you open a fifo, the program will block until there's something
257 For example, let's say you'd like to have your F<.signature> file be a
258 named pipe that has a Perl program on the other end. Now every time any
259 program (like a mailer, news reader, finger program, etc.) tries to read
260 from that file, the reading program will block and your program will
261 supply the new signature. We'll use the pipe-checking file test B<-p>
262 to find out whether anyone (or anything) has accidentally removed our fifo.
265 $FIFO = '.signature';
266 $ENV{PATH} .= ":/etc:/usr/games";
271 system('mknod', $FIFO, 'p')
272 && die "can't mknod $FIFO: $!";
275 # next line blocks until there's a reader
276 open (FIFO, "> $FIFO") || die "can't write $FIFO: $!";
277 print FIFO "John Smith (smith\@host.org)\n", `fortune -s`;
279 sleep 2; # to avoid dup signals
282 =head2 Deferred Signals (Safe signals)
284 In Perls before Perl 5.7.3 by installing Perl code to deal with
285 signals, you were exposing yourself to danger from two things. First,
286 few system library functions are re-entrant. If the signal interrupts
287 while Perl is executing one function (like malloc(3) or printf(3)),
288 and your signal handler then calls the same function again, you could
289 get unpredictable behavior--often, a core dump. Second, Perl isn't
290 itself re-entrant at the lowest levels. If the signal interrupts Perl
291 while Perl is changing its own internal data structures, similarly
292 unpredictable behaviour may result.
294 There were two things you could do, knowing this: be paranoid or be
295 pragmatic. The paranoid approach was to do as little as possible in your
296 signal handler. Set an existing integer variable that already has a
297 value, and return. This doesn't help you if you're in a slow system call,
298 which will just restart. That means you have to C<die> to longjump(3) out
299 of the handler. Even this is a little cavalier for the true paranoiac,
300 who avoids C<die> in a handler because the system I<is> out to get you.
301 The pragmatic approach was to say ``I know the risks, but prefer the
302 convenience'', and to do anything you wanted in your signal handler,
303 and be prepared to clean up core dumps now and again.
305 In Perl 5.7.3 and later to avoid these problems signals are
306 "deferred"-- that is when the signal is delivered to the process by
307 the system (to the C code that implements Perl) a flag is set, and the
308 handler returns immediately. Then at strategic "safe" points in the
309 Perl interpreter (e.g. when it is about to execute a new opcode) the
310 flags are checked and the Perl level handler from %SIG is
311 executed. The "deferred" scheme allows much more flexibility in the
312 coding of signal handler as we know Perl interpreter is in a safe
313 state, and that we are not in a system library function when the
314 handler is called. However the implementation does differ from
315 previous Perls in the following ways:
319 =item Long running opcodes
321 As Perl interpreter only looks at the signal flags when it about to
322 execute a new opcode if a signal arrives during a long running opcode
323 (e.g. a regular expression operation on a very large string) then
324 signal will not be seen until operation completes.
326 =item Interrupting IO
328 When a signal is delivered (e.g. INT control-C) the operating system
329 breaks into IO operations like C<read> (used to implement Perls
330 E<lt>E<gt> operator). On older Perls the handler was called
331 immediately (and as C<read> is not "unsafe" this worked well). With
332 the "deferred" scheme the handler is not called immediately, and if
333 Perl is using system's C<stdio> library that library may re-start the
334 C<read> without returning to Perl and giving it a chance to call the
335 %SIG handler. If this happens on your system the solution is to use
336 C<:perlio> layer to do IO - at least on those handles which you want
337 to be able to break into with signals. (The C<:perlio> layer checks
338 the signal flags and calls %SIG handlers before resuming IO operation.)
340 Note that the default in Perl 5.7.3 and later is to automatically use
341 the C<:perlio> layer.
343 Note that some networking library functions like gethostbyname() are
344 known to have their own implementations of timeouts which may conflict
345 with your timeouts. If you are having problems with such functions,
346 you can try using the POSIX sigaction() function, which bypasses the
347 Perl safe signals (note that this means subjecting yourself to
348 possible memory corruption, as described above). Instead of setting
349 C<$SIG{ALRM}> try something like the following:
352 sigaction SIGALRM, new POSIX::SigAction sub { die "alarm\n" }
353 or die "Error setting SIGALRM handler: $!\n";
355 =item Signals as "faults"
357 Certain signals e.g. SEGV, ILL, BUS are generated as a result of
358 virtual memory or other "faults". These are normally fatal and there
359 is little a Perl-level handler can do with them. (In particular the
360 old signal scheme was particularly unsafe in such cases.) However if
361 a %SIG handler is set the new scheme simply sets a flag and returns as
362 described above. This may cause the operating system to try the
363 offending machine instruction again and - as nothing has changed - it
364 will generate the signal again. The result of this is a rather odd
365 "loop". In future Perl's signal mechanism may be changed to avoid this
366 - perhaps by simply disallowing %SIG handlers on signals of that
367 type. Until then the work-round is not to set a %SIG handler on those
368 signals. (Which signals they are is operating system dependant.)
370 =item Signals triggered by operating system state
372 On some operating systems certain signal handlers are supposed to "do
373 something" before returning. One example can be CHLD or CLD which
374 indicates a child process has completed. On some operating systems the
375 signal handler is expected to C<wait> for the completed child
376 process. On such systems the deferred signal scheme will not work for
377 those signals (it does not do the C<wait>). Again the failure will
378 look like a loop as the operating system will re-issue the signal as
379 there are un-waited-for completed child processes.
383 If you want the old signal behaviour back regardless of possible
384 memory corruption, set the environment variable C<PERL_SIGNALS> to
385 C<"unsafe"> (a new feature since Perl 5.8.1).
387 =head1 Using open() for IPC
389 Perl's basic open() statement can also be used for unidirectional
390 interprocess communication by either appending or prepending a pipe
391 symbol to the second argument to open(). Here's how to start
392 something up in a child process you intend to write to:
394 open(SPOOLER, "| cat -v | lpr -h 2>/dev/null")
395 || die "can't fork: $!";
396 local $SIG{PIPE} = sub { die "spooler pipe broke" };
397 print SPOOLER "stuff\n";
398 close SPOOLER || die "bad spool: $! $?";
400 And here's how to start up a child process you intend to read from:
402 open(STATUS, "netstat -an 2>&1 |")
403 || die "can't fork: $!";
405 next if /^(tcp|udp)/;
408 close STATUS || die "bad netstat: $! $?";
410 If one can be sure that a particular program is a Perl script that is
411 expecting filenames in @ARGV, the clever programmer can write something
414 % program f1 "cmd1|" - f2 "cmd2|" f3 < tmpfile
416 and irrespective of which shell it's called from, the Perl program will
417 read from the file F<f1>, the process F<cmd1>, standard input (F<tmpfile>
418 in this case), the F<f2> file, the F<cmd2> command, and finally the F<f3>
419 file. Pretty nifty, eh?
421 You might notice that you could use backticks for much the
422 same effect as opening a pipe for reading:
424 print grep { !/^(tcp|udp)/ } `netstat -an 2>&1`;
425 die "bad netstat" if $?;
427 While this is true on the surface, it's much more efficient to process the
428 file one line or record at a time because then you don't have to read the
429 whole thing into memory at once. It also gives you finer control of the
430 whole process, letting you to kill off the child process early if you'd
433 Be careful to check both the open() and the close() return values. If
434 you're I<writing> to a pipe, you should also trap SIGPIPE. Otherwise,
435 think of what happens when you start up a pipe to a command that doesn't
436 exist: the open() will in all likelihood succeed (it only reflects the
437 fork()'s success), but then your output will fail--spectacularly. Perl
438 can't know whether the command worked because your command is actually
439 running in a separate process whose exec() might have failed. Therefore,
440 while readers of bogus commands return just a quick end of file, writers
441 to bogus command will trigger a signal they'd better be prepared to
444 open(FH, "|bogus") or die "can't fork: $!";
445 print FH "bang\n" or die "can't write: $!";
446 close FH or die "can't close: $!";
448 That won't blow up until the close, and it will blow up with a SIGPIPE.
449 To catch it, you could use this:
451 $SIG{PIPE} = 'IGNORE';
452 open(FH, "|bogus") or die "can't fork: $!";
453 print FH "bang\n" or die "can't write: $!";
454 close FH or die "can't close: status=$?";
458 Both the main process and any child processes it forks share the same
459 STDIN, STDOUT, and STDERR filehandles. If both processes try to access
460 them at once, strange things can happen. You may also want to close
461 or reopen the filehandles for the child. You can get around this by
462 opening your pipe with open(), but on some systems this means that the
463 child process cannot outlive the parent.
465 =head2 Background Processes
467 You can run a command in the background with:
471 The command's STDOUT and STDERR (and possibly STDIN, depending on your
472 shell) will be the same as the parent's. You won't need to catch
473 SIGCHLD because of the double-fork taking place (see below for more
476 =head2 Complete Dissociation of Child from Parent
478 In some cases (starting server processes, for instance) you'll want to
479 completely dissociate the child process from the parent. This is
480 often called daemonization. A well behaved daemon will also chdir()
481 to the root directory (so it doesn't prevent unmounting the filesystem
482 containing the directory from which it was launched) and redirect its
483 standard file descriptors from and to F</dev/null> (so that random
484 output doesn't wind up on the user's terminal).
489 chdir '/' or die "Can't chdir to /: $!";
490 open STDIN, '/dev/null' or die "Can't read /dev/null: $!";
491 open STDOUT, '>/dev/null'
492 or die "Can't write to /dev/null: $!";
493 defined(my $pid = fork) or die "Can't fork: $!";
495 setsid or die "Can't start a new session: $!";
496 open STDERR, '>&STDOUT' or die "Can't dup stdout: $!";
499 The fork() has to come before the setsid() to ensure that you aren't a
500 process group leader (the setsid() will fail if you are). If your
501 system doesn't have the setsid() function, open F</dev/tty> and use the
502 C<TIOCNOTTY> ioctl() on it instead. See L<tty(4)> for details.
504 Non-Unix users should check their Your_OS::Process module for other
507 =head2 Safe Pipe Opens
509 Another interesting approach to IPC is making your single program go
510 multiprocess and communicate between (or even amongst) yourselves. The
511 open() function will accept a file argument of either C<"-|"> or C<"|-">
512 to do a very interesting thing: it forks a child connected to the
513 filehandle you've opened. The child is running the same program as the
514 parent. This is useful for safely opening a file when running under an
515 assumed UID or GID, for example. If you open a pipe I<to> minus, you can
516 write to the filehandle you opened and your kid will find it in his
517 STDIN. If you open a pipe I<from> minus, you can read from the filehandle
518 you opened whatever your kid writes to his STDOUT.
520 use English '-no_match_vars';
524 $pid = open(KID_TO_WRITE, "|-");
525 unless (defined $pid) {
526 warn "cannot fork: $!";
527 die "bailing out" if $sleep_count++ > 6;
530 } until defined $pid;
533 print KID_TO_WRITE @some_data;
534 close(KID_TO_WRITE) || warn "kid exited $?";
536 ($EUID, $EGID) = ($UID, $GID); # suid progs only
537 open (FILE, "> /safe/file")
538 || die "can't open /safe/file: $!";
540 print FILE; # child's STDIN is parent's KID
542 exit; # don't forget this
545 Another common use for this construct is when you need to execute
546 something without the shell's interference. With system(), it's
547 straightforward, but you can't use a pipe open or backticks safely.
548 That's because there's no way to stop the shell from getting its hands on
549 your arguments. Instead, use lower-level control to call exec() directly.
551 Here's a safe backtick or pipe open for read:
553 # add error processing as above
554 $pid = open(KID_TO_READ, "-|");
557 while (<KID_TO_READ>) {
558 # do something interesting
560 close(KID_TO_READ) || warn "kid exited $?";
563 ($EUID, $EGID) = ($UID, $GID); # suid only
564 exec($program, @options, @args)
565 || die "can't exec program: $!";
570 And here's a safe pipe open for writing:
572 # add error processing as above
573 $pid = open(KID_TO_WRITE, "|-");
574 $SIG{PIPE} = sub { die "whoops, $program pipe broke" };
580 close(KID_TO_WRITE) || warn "kid exited $?";
583 ($EUID, $EGID) = ($UID, $GID);
584 exec($program, @options, @args)
585 || die "can't exec program: $!";
589 Since Perl 5.8.0, you can also use the list form of C<open> for pipes :
592 open KID_PS, "-|", "ps", "aux" or die $!;
594 forks the ps(1) command (without spawning a shell, as there are more than
595 three arguments to open()), and reads its standard output via the
596 C<KID_PS> filehandle. The corresponding syntax to read from command
597 pipes (with C<"|-"> in place of C<"-|">) is also implemented.
599 Note that these operations are full Unix forks, which means they may not be
600 correctly implemented on alien systems. Additionally, these are not true
601 multithreading. If you'd like to learn more about threading, see the
602 F<modules> file mentioned below in the SEE ALSO section.
604 =head2 Bidirectional Communication with Another Process
606 While this works reasonably well for unidirectional communication, what
607 about bidirectional communication? The obvious thing you'd like to do
608 doesn't actually work:
610 open(PROG_FOR_READING_AND_WRITING, "| some program |")
612 and if you forget to use the C<use warnings> pragma or the B<-w> flag,
613 then you'll miss out entirely on the diagnostic message:
615 Can't do bidirectional pipe at -e line 1.
617 If you really want to, you can use the standard open2() library function
618 to catch both ends. There's also an open3() for tridirectional I/O so you
619 can also catch your child's STDERR, but doing so would then require an
620 awkward select() loop and wouldn't allow you to use normal Perl input
623 If you look at its source, you'll see that open2() uses low-level
624 primitives like Unix pipe() and exec() calls to create all the connections.
625 While it might have been slightly more efficient by using socketpair(), it
626 would have then been even less portable than it already is. The open2()
627 and open3() functions are unlikely to work anywhere except on a Unix
628 system or some other one purporting to be POSIX compliant.
630 Here's an example of using open2():
634 $pid = open2(*Reader, *Writer, "cat -u -n" );
635 print Writer "stuff\n";
638 The problem with this is that Unix buffering is really going to
639 ruin your day. Even though your C<Writer> filehandle is auto-flushed,
640 and the process on the other end will get your data in a timely manner,
641 you can't usually do anything to force it to give it back to you
642 in a similarly quick fashion. In this case, we could, because we
643 gave I<cat> a B<-u> flag to make it unbuffered. But very few Unix
644 commands are designed to operate over pipes, so this seldom works
645 unless you yourself wrote the program on the other end of the
648 A solution to this is the nonstandard F<Comm.pl> library. It uses
649 pseudo-ttys to make your program behave more reasonably:
652 $ph = open_proc('cat -n');
654 print $ph "a line\n";
655 print "got back ", scalar <$ph>;
658 This way you don't have to have control over the source code of the
659 program you're using. The F<Comm> library also has expect()
660 and interact() functions. Find the library (and we hope its
661 successor F<IPC::Chat>) at your nearest CPAN archive as detailed
662 in the SEE ALSO section below.
664 The newer Expect.pm module from CPAN also addresses this kind of thing.
665 This module requires two other modules from CPAN: IO::Pty and IO::Stty.
666 It sets up a pseudo-terminal to interact with programs that insist on
667 using talking to the terminal device driver. If your system is
668 amongst those supported, this may be your best bet.
670 =head2 Bidirectional Communication with Yourself
672 If you want, you may make low-level pipe() and fork()
673 to stitch this together by hand. This example only
674 talks to itself, but you could reopen the appropriate
675 handles to STDIN and STDOUT and call other processes.
678 # pipe1 - bidirectional communication using two pipe pairs
679 # designed for the socketpair-challenged
680 use IO::Handle; # thousands of lines just for autoflush :-(
681 pipe(PARENT_RDR, CHILD_WTR); # XXX: failure?
682 pipe(CHILD_RDR, PARENT_WTR); # XXX: failure?
683 CHILD_WTR->autoflush(1);
684 PARENT_WTR->autoflush(1);
687 close PARENT_RDR; close PARENT_WTR;
688 print CHILD_WTR "Parent Pid $$ is sending this\n";
689 chomp($line = <CHILD_RDR>);
690 print "Parent Pid $$ just read this: `$line'\n";
691 close CHILD_RDR; close CHILD_WTR;
694 die "cannot fork: $!" unless defined $pid;
695 close CHILD_RDR; close CHILD_WTR;
696 chomp($line = <PARENT_RDR>);
697 print "Child Pid $$ just read this: `$line'\n";
698 print PARENT_WTR "Child Pid $$ is sending this\n";
699 close PARENT_RDR; close PARENT_WTR;
703 But you don't actually have to make two pipe calls. If you
704 have the socketpair() system call, it will do this all for you.
707 # pipe2 - bidirectional communication using socketpair
708 # "the best ones always go both ways"
711 use IO::Handle; # thousands of lines just for autoflush :-(
712 # We say AF_UNIX because although *_LOCAL is the
713 # POSIX 1003.1g form of the constant, many machines
714 # still don't have it.
715 socketpair(CHILD, PARENT, AF_UNIX, SOCK_STREAM, PF_UNSPEC)
716 or die "socketpair: $!";
719 PARENT->autoflush(1);
723 print CHILD "Parent Pid $$ is sending this\n";
724 chomp($line = <CHILD>);
725 print "Parent Pid $$ just read this: `$line'\n";
729 die "cannot fork: $!" unless defined $pid;
731 chomp($line = <PARENT>);
732 print "Child Pid $$ just read this: `$line'\n";
733 print PARENT "Child Pid $$ is sending this\n";
738 =head1 Sockets: Client/Server Communication
740 While not limited to Unix-derived operating systems (e.g., WinSock on PCs
741 provides socket support, as do some VMS libraries), you may not have
742 sockets on your system, in which case this section probably isn't going to do
743 you much good. With sockets, you can do both virtual circuits (i.e., TCP
744 streams) and datagrams (i.e., UDP packets). You may be able to do even more
745 depending on your system.
747 The Perl function calls for dealing with sockets have the same names as
748 the corresponding system calls in C, but their arguments tend to differ
749 for two reasons: first, Perl filehandles work differently than C file
750 descriptors. Second, Perl already knows the length of its strings, so you
751 don't need to pass that information.
753 One of the major problems with old socket code in Perl was that it used
754 hard-coded values for some of the constants, which severely hurt
755 portability. If you ever see code that does anything like explicitly
756 setting C<$AF_INET = 2>, you know you're in for big trouble: An
757 immeasurably superior approach is to use the C<Socket> module, which more
758 reliably grants access to various constants and functions you'll need.
760 If you're not writing a server/client for an existing protocol like
761 NNTP or SMTP, you should give some thought to how your server will
762 know when the client has finished talking, and vice-versa. Most
763 protocols are based on one-line messages and responses (so one party
764 knows the other has finished when a "\n" is received) or multi-line
765 messages and responses that end with a period on an empty line
766 ("\n.\n" terminates a message/response).
768 =head2 Internet Line Terminators
770 The Internet line terminator is "\015\012". Under ASCII variants of
771 Unix, that could usually be written as "\r\n", but under other systems,
772 "\r\n" might at times be "\015\015\012", "\012\012\015", or something
773 completely different. The standards specify writing "\015\012" to be
774 conformant (be strict in what you provide), but they also recommend
775 accepting a lone "\012" on input (but be lenient in what you require).
776 We haven't always been very good about that in the code in this manpage,
777 but unless you're on a Mac, you'll probably be ok.
779 =head2 Internet TCP Clients and Servers
781 Use Internet-domain sockets when you want to do client-server
782 communication that might extend to machines outside of your own system.
784 Here's a sample TCP client using Internet-domain sockets:
789 my ($remote,$port, $iaddr, $paddr, $proto, $line);
791 $remote = shift || 'localhost';
792 $port = shift || 2345; # random port
793 if ($port =~ /\D/) { $port = getservbyname($port, 'tcp') }
794 die "No port" unless $port;
795 $iaddr = inet_aton($remote) || die "no host: $remote";
796 $paddr = sockaddr_in($port, $iaddr);
798 $proto = getprotobyname('tcp');
799 socket(SOCK, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
800 connect(SOCK, $paddr) || die "connect: $!";
801 while (defined($line = <SOCK>)) {
805 close (SOCK) || die "close: $!";
808 And here's a corresponding server to go along with it. We'll
809 leave the address as INADDR_ANY so that the kernel can choose
810 the appropriate interface on multihomed hosts. If you want sit
811 on a particular interface (like the external side of a gateway
812 or firewall machine), you should fill this in with your real address
817 BEGIN { $ENV{PATH} = '/usr/ucb:/bin' }
820 my $EOL = "\015\012";
822 sub logmsg { print "$0 $$: @_ at ", scalar localtime, "\n" }
824 my $port = shift || 2345;
825 my $proto = getprotobyname('tcp');
827 ($port) = $port =~ /^(\d+)$/ or die "invalid port";
829 socket(Server, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
830 setsockopt(Server, SOL_SOCKET, SO_REUSEADDR,
831 pack("l", 1)) || die "setsockopt: $!";
832 bind(Server, sockaddr_in($port, INADDR_ANY)) || die "bind: $!";
833 listen(Server,SOMAXCONN) || die "listen: $!";
835 logmsg "server started on port $port";
839 $SIG{CHLD} = \&REAPER;
841 for ( ; $paddr = accept(Client,Server); close Client) {
842 my($port,$iaddr) = sockaddr_in($paddr);
843 my $name = gethostbyaddr($iaddr,AF_INET);
845 logmsg "connection from $name [",
846 inet_ntoa($iaddr), "]
849 print Client "Hello there, $name, it's now ",
850 scalar localtime, $EOL;
853 And here's a multithreaded version. It's multithreaded in that
854 like most typical servers, it spawns (forks) a slave server to
855 handle the client request so that the master server can quickly
856 go back to service a new client.
860 BEGIN { $ENV{PATH} = '/usr/ucb:/bin' }
863 my $EOL = "\015\012";
865 sub spawn; # forward declaration
866 sub logmsg { print "$0 $$: @_ at ", scalar localtime, "\n" }
868 my $port = shift || 2345;
869 my $proto = getprotobyname('tcp');
871 ($port) = $port =~ /^(\d+)$/ or die "invalid port";
873 socket(Server, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
874 setsockopt(Server, SOL_SOCKET, SO_REUSEADDR,
875 pack("l", 1)) || die "setsockopt: $!";
876 bind(Server, sockaddr_in($port, INADDR_ANY)) || die "bind: $!";
877 listen(Server,SOMAXCONN) || die "listen: $!";
879 logmsg "server started on port $port";
884 use POSIX ":sys_wait_h";
887 while (($waitedpid = waitpid(-1,WNOHANG)) > 0) {
888 logmsg "reaped $waitedpid" . ($? ? " with exit $?" : '');
890 $SIG{CHLD} = \&REAPER; # loathe sysV
893 $SIG{CHLD} = \&REAPER;
895 for ( $waitedpid = 0;
896 ($paddr = accept(Client,Server)) || $waitedpid;
897 $waitedpid = 0, close Client)
899 next if $waitedpid and not $paddr;
900 my($port,$iaddr) = sockaddr_in($paddr);
901 my $name = gethostbyaddr($iaddr,AF_INET);
903 logmsg "connection from $name [",
904 inet_ntoa($iaddr), "]
909 print "Hello there, $name, it's now ", scalar localtime, $EOL;
910 exec '/usr/games/fortune' # XXX: `wrong' line terminators
911 or confess "can't exec fortune: $!";
919 unless (@_ == 0 && $coderef && ref($coderef) eq 'CODE') {
920 confess "usage: spawn CODEREF";
924 if (!defined($pid = fork)) {
925 logmsg "cannot fork: $!";
929 return; # I'm the parent
931 # else I'm the child -- go spawn
933 open(STDIN, "<&Client") || die "can't dup client to stdin";
934 open(STDOUT, ">&Client") || die "can't dup client to stdout";
935 ## open(STDERR, ">&STDOUT") || die "can't dup stdout to stderr";
939 This server takes the trouble to clone off a child version via fork() for
940 each incoming request. That way it can handle many requests at once,
941 which you might not always want. Even if you don't fork(), the listen()
942 will allow that many pending connections. Forking servers have to be
943 particularly careful about cleaning up their dead children (called
944 "zombies" in Unix parlance), because otherwise you'll quickly fill up your
947 We suggest that you use the B<-T> flag to use taint checking (see L<perlsec>)
948 even if we aren't running setuid or setgid. This is always a good idea
949 for servers and other programs run on behalf of someone else (like CGI
950 scripts), because it lessens the chances that people from the outside will
951 be able to compromise your system.
953 Let's look at another TCP client. This one connects to the TCP "time"
954 service on a number of different machines and shows how far their clocks
955 differ from the system on which it's being run:
961 my $SECS_of_70_YEARS = 2208988800;
962 sub ctime { scalar localtime(shift) }
964 my $iaddr = gethostbyname('localhost');
965 my $proto = getprotobyname('tcp');
966 my $port = getservbyname('time', 'tcp');
967 my $paddr = sockaddr_in(0, $iaddr);
971 printf "%-24s %8s %s\n", "localhost", 0, ctime(time());
973 foreach $host (@ARGV) {
974 printf "%-24s ", $host;
975 my $hisiaddr = inet_aton($host) || die "unknown host";
976 my $hispaddr = sockaddr_in($port, $hisiaddr);
977 socket(SOCKET, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
978 connect(SOCKET, $hispaddr) || die "bind: $!";
980 read(SOCKET, $rtime, 4);
982 my $histime = unpack("N", $rtime) - $SECS_of_70_YEARS ;
983 printf "%8d %s\n", $histime - time, ctime($histime);
986 =head2 Unix-Domain TCP Clients and Servers
988 That's fine for Internet-domain clients and servers, but what about local
989 communications? While you can use the same setup, sometimes you don't
990 want to. Unix-domain sockets are local to the current host, and are often
991 used internally to implement pipes. Unlike Internet domain sockets, Unix
992 domain sockets can show up in the file system with an ls(1) listing.
995 srw-rw-rw- 1 root 0 Oct 31 07:23 /dev/log
997 You can test for these with Perl's B<-S> file test:
999 unless ( -S '/dev/log' ) {
1000 die "something's wicked with the log system";
1003 Here's a sample Unix-domain client:
1008 my ($rendezvous, $line);
1010 $rendezvous = shift || '/tmp/catsock';
1011 socket(SOCK, PF_UNIX, SOCK_STREAM, 0) || die "socket: $!";
1012 connect(SOCK, sockaddr_un($rendezvous)) || die "connect: $!";
1013 while (defined($line = <SOCK>)) {
1018 And here's a corresponding server. You don't have to worry about silly
1019 network terminators here because Unix domain sockets are guaranteed
1020 to be on the localhost, and thus everything works right.
1027 BEGIN { $ENV{PATH} = '/usr/ucb:/bin' }
1028 sub spawn; # forward declaration
1029 sub logmsg { print "$0 $$: @_ at ", scalar localtime, "\n" }
1031 my $NAME = '/tmp/catsock';
1032 my $uaddr = sockaddr_un($NAME);
1033 my $proto = getprotobyname('tcp');
1035 socket(Server,PF_UNIX,SOCK_STREAM,0) || die "socket: $!";
1037 bind (Server, $uaddr) || die "bind: $!";
1038 listen(Server,SOMAXCONN) || die "listen: $!";
1040 logmsg "server started on $NAME";
1044 use POSIX ":sys_wait_h";
1047 while (($waitedpid = waitpid(-1,WNOHANG)) > 0) {
1048 logmsg "reaped $waitedpid" . ($? ? " with exit $?" : '');
1050 $SIG{CHLD} = \&REAPER; # loathe sysV
1053 $SIG{CHLD} = \&REAPER;
1056 for ( $waitedpid = 0;
1057 accept(Client,Server) || $waitedpid;
1058 $waitedpid = 0, close Client)
1061 logmsg "connection on $NAME";
1063 print "Hello there, it's now ", scalar localtime, "\n";
1064 exec '/usr/games/fortune' or die "can't exec fortune: $!";
1069 my $coderef = shift;
1071 unless (@_ == 0 && $coderef && ref($coderef) eq 'CODE') {
1072 confess "usage: spawn CODEREF";
1076 if (!defined($pid = fork)) {
1077 logmsg "cannot fork: $!";
1080 logmsg "begat $pid";
1081 return; # I'm the parent
1083 # else I'm the child -- go spawn
1085 open(STDIN, "<&Client") || die "can't dup client to stdin";
1086 open(STDOUT, ">&Client") || die "can't dup client to stdout";
1087 ## open(STDERR, ">&STDOUT") || die "can't dup stdout to stderr";
1091 As you see, it's remarkably similar to the Internet domain TCP server, so
1092 much so, in fact, that we've omitted several duplicate functions--spawn(),
1093 logmsg(), ctime(), and REAPER()--which are exactly the same as in the
1096 So why would you ever want to use a Unix domain socket instead of a
1097 simpler named pipe? Because a named pipe doesn't give you sessions. You
1098 can't tell one process's data from another's. With socket programming,
1099 you get a separate session for each client: that's why accept() takes two
1102 For example, let's say that you have a long running database server daemon
1103 that you want folks from the World Wide Web to be able to access, but only
1104 if they go through a CGI interface. You'd have a small, simple CGI
1105 program that does whatever checks and logging you feel like, and then acts
1106 as a Unix-domain client and connects to your private server.
1108 =head1 TCP Clients with IO::Socket
1110 For those preferring a higher-level interface to socket programming, the
1111 IO::Socket module provides an object-oriented approach. IO::Socket is
1112 included as part of the standard Perl distribution as of the 5.004
1113 release. If you're running an earlier version of Perl, just fetch
1114 IO::Socket from CPAN, where you'll also find modules providing easy
1115 interfaces to the following systems: DNS, FTP, Ident (RFC 931), NIS and
1116 NISPlus, NNTP, Ping, POP3, SMTP, SNMP, SSLeay, Telnet, and Time--just
1119 =head2 A Simple Client
1121 Here's a client that creates a TCP connection to the "daytime"
1122 service at port 13 of the host name "localhost" and prints out everything
1123 that the server there cares to provide.
1127 $remote = IO::Socket::INET->new(
1129 PeerAddr => "localhost",
1130 PeerPort => "daytime(13)",
1132 or die "cannot connect to daytime port at localhost";
1133 while ( <$remote> ) { print }
1135 When you run this program, you should get something back that
1138 Wed May 14 08:40:46 MDT 1997
1140 Here are what those parameters to the C<new> constructor mean:
1146 This is which protocol to use. In this case, the socket handle returned
1147 will be connected to a TCP socket, because we want a stream-oriented
1148 connection, that is, one that acts pretty much like a plain old file.
1149 Not all sockets are this of this type. For example, the UDP protocol
1150 can be used to make a datagram socket, used for message-passing.
1154 This is the name or Internet address of the remote host the server is
1155 running on. We could have specified a longer name like C<"www.perl.com">,
1156 or an address like C<"204.148.40.9">. For demonstration purposes, we've
1157 used the special hostname C<"localhost">, which should always mean the
1158 current machine you're running on. The corresponding Internet address
1159 for localhost is C<"127.1">, if you'd rather use that.
1163 This is the service name or port number we'd like to connect to.
1164 We could have gotten away with using just C<"daytime"> on systems with a
1165 well-configured system services file,[FOOTNOTE: The system services file
1166 is in I</etc/services> under Unix] but just in case, we've specified the
1167 port number (13) in parentheses. Using just the number would also have
1168 worked, but constant numbers make careful programmers nervous.
1172 Notice how the return value from the C<new> constructor is used as
1173 a filehandle in the C<while> loop? That's what's called an indirect
1174 filehandle, a scalar variable containing a filehandle. You can use
1175 it the same way you would a normal filehandle. For example, you
1176 can read one line from it this way:
1180 all remaining lines from is this way:
1184 and send a line of data to it this way:
1186 print $handle "some data\n";
1188 =head2 A Webget Client
1190 Here's a simple client that takes a remote host to fetch a document
1191 from, and then a list of documents to get from that host. This is a
1192 more interesting client than the previous one because it first sends
1193 something to the server before fetching the server's response.
1197 unless (@ARGV > 1) { die "usage: $0 host document ..." }
1198 $host = shift(@ARGV);
1201 foreach $document ( @ARGV ) {
1202 $remote = IO::Socket::INET->new( Proto => "tcp",
1204 PeerPort => "http(80)",
1206 unless ($remote) { die "cannot connect to http daemon on $host" }
1207 $remote->autoflush(1);
1208 print $remote "GET $document HTTP/1.0" . $BLANK;
1209 while ( <$remote> ) { print }
1213 The web server handing the "http" service, which is assumed to be at
1214 its standard port, number 80. If the web server you're trying to
1215 connect to is at a different port (like 1080 or 8080), you should specify
1216 as the named-parameter pair, C<< PeerPort => 8080 >>. The C<autoflush>
1217 method is used on the socket because otherwise the system would buffer
1218 up the output we sent it. (If you're on a Mac, you'll also need to
1219 change every C<"\n"> in your code that sends data over the network to
1220 be a C<"\015\012"> instead.)
1222 Connecting to the server is only the first part of the process: once you
1223 have the connection, you have to use the server's language. Each server
1224 on the network has its own little command language that it expects as
1225 input. The string that we send to the server starting with "GET" is in
1226 HTTP syntax. In this case, we simply request each specified document.
1227 Yes, we really are making a new connection for each document, even though
1228 it's the same host. That's the way you always used to have to speak HTTP.
1229 Recent versions of web browsers may request that the remote server leave
1230 the connection open a little while, but the server doesn't have to honor
1233 Here's an example of running that program, which we'll call I<webget>:
1235 % webget www.perl.com /guanaco.html
1236 HTTP/1.1 404 File Not Found
1237 Date: Thu, 08 May 1997 18:02:32 GMT
1238 Server: Apache/1.2b6
1240 Content-type: text/html
1242 <HEAD><TITLE>404 File Not Found</TITLE></HEAD>
1243 <BODY><H1>File Not Found</H1>
1244 The requested URL /guanaco.html was not found on this server.<P>
1247 Ok, so that's not very interesting, because it didn't find that
1248 particular document. But a long response wouldn't have fit on this page.
1250 For a more fully-featured version of this program, you should look to
1251 the I<lwp-request> program included with the LWP modules from CPAN.
1253 =head2 Interactive Client with IO::Socket
1255 Well, that's all fine if you want to send one command and get one answer,
1256 but what about setting up something fully interactive, somewhat like
1257 the way I<telnet> works? That way you can type a line, get the answer,
1258 type a line, get the answer, etc.
1260 This client is more complicated than the two we've done so far, but if
1261 you're on a system that supports the powerful C<fork> call, the solution
1262 isn't that rough. Once you've made the connection to whatever service
1263 you'd like to chat with, call C<fork> to clone your process. Each of
1264 these two identical process has a very simple job to do: the parent
1265 copies everything from the socket to standard output, while the child
1266 simultaneously copies everything from standard input to the socket.
1267 To accomplish the same thing using just one process would be I<much>
1268 harder, because it's easier to code two processes to do one thing than it
1269 is to code one process to do two things. (This keep-it-simple principle
1270 a cornerstones of the Unix philosophy, and good software engineering as
1271 well, which is probably why it's spread to other systems.)
1278 my ($host, $port, $kidpid, $handle, $line);
1280 unless (@ARGV == 2) { die "usage: $0 host port" }
1281 ($host, $port) = @ARGV;
1283 # create a tcp connection to the specified host and port
1284 $handle = IO::Socket::INET->new(Proto => "tcp",
1287 or die "can't connect to port $port on $host: $!";
1289 $handle->autoflush(1); # so output gets there right away
1290 print STDERR "[Connected to $host:$port]\n";
1292 # split the program into two processes, identical twins
1293 die "can't fork: $!" unless defined($kidpid = fork());
1295 # the if{} block runs only in the parent process
1297 # copy the socket to standard output
1298 while (defined ($line = <$handle>)) {
1301 kill("TERM", $kidpid); # send SIGTERM to child
1303 # the else{} block runs only in the child process
1305 # copy standard input to the socket
1306 while (defined ($line = <STDIN>)) {
1307 print $handle $line;
1311 The C<kill> function in the parent's C<if> block is there to send a
1312 signal to our child process (current running in the C<else> block)
1313 as soon as the remote server has closed its end of the connection.
1315 If the remote server sends data a byte at time, and you need that
1316 data immediately without waiting for a newline (which might not happen),
1317 you may wish to replace the C<while> loop in the parent with the
1321 while (sysread($handle, $byte, 1) == 1) {
1325 Making a system call for each byte you want to read is not very efficient
1326 (to put it mildly) but is the simplest to explain and works reasonably
1329 =head1 TCP Servers with IO::Socket
1331 As always, setting up a server is little bit more involved than running a client.
1332 The model is that the server creates a special kind of socket that
1333 does nothing but listen on a particular port for incoming connections.
1334 It does this by calling the C<< IO::Socket::INET->new() >> method with
1335 slightly different arguments than the client did.
1341 This is which protocol to use. Like our clients, we'll
1342 still specify C<"tcp"> here.
1347 port in the C<LocalPort> argument, which we didn't do for the client.
1348 This is service name or port number for which you want to be the
1349 server. (Under Unix, ports under 1024 are restricted to the
1350 superuser.) In our sample, we'll use port 9000, but you can use
1351 any port that's not currently in use on your system. If you try
1352 to use one already in used, you'll get an "Address already in use"
1353 message. Under Unix, the C<netstat -a> command will show
1354 which services current have servers.
1358 The C<Listen> parameter is set to the maximum number of
1359 pending connections we can accept until we turn away incoming clients.
1360 Think of it as a call-waiting queue for your telephone.
1361 The low-level Socket module has a special symbol for the system maximum, which
1366 The C<Reuse> parameter is needed so that we restart our server
1367 manually without waiting a few minutes to allow system buffers to
1372 Once the generic server socket has been created using the parameters
1373 listed above, the server then waits for a new client to connect
1374 to it. The server blocks in the C<accept> method, which eventually accepts a
1375 bidirectional connection from the remote client. (Make sure to autoflush
1376 this handle to circumvent buffering.)
1378 To add to user-friendliness, our server prompts the user for commands.
1379 Most servers don't do this. Because of the prompt without a newline,
1380 you'll have to use the C<sysread> variant of the interactive client above.
1382 This server accepts one of five different commands, sending output
1383 back to the client. Note that unlike most network servers, this one
1384 only handles one incoming client at a time. Multithreaded servers are
1385 covered in Chapter 6 of the Camel.
1387 Here's the code. We'll
1391 use Net::hostent; # for OO version of gethostbyaddr
1393 $PORT = 9000; # pick something not in use
1395 $server = IO::Socket::INET->new( Proto => 'tcp',
1397 Listen => SOMAXCONN,
1400 die "can't setup server" unless $server;
1401 print "[Server $0 accepting clients]\n";
1403 while ($client = $server->accept()) {
1404 $client->autoflush(1);
1405 print $client "Welcome to $0; type help for command list.\n";
1406 $hostinfo = gethostbyaddr($client->peeraddr);
1407 printf "[Connect from %s]\n", $hostinfo ? $hostinfo->name : $client->peerhost;
1408 print $client "Command? ";
1409 while ( <$client>) {
1410 next unless /\S/; # blank line
1411 if (/quit|exit/i) { last; }
1412 elsif (/date|time/i) { printf $client "%s\n", scalar localtime; }
1413 elsif (/who/i ) { print $client `who 2>&1`; }
1414 elsif (/cookie/i ) { print $client `/usr/games/fortune 2>&1`; }
1415 elsif (/motd/i ) { print $client `cat /etc/motd 2>&1`; }
1417 print $client "Commands: quit date who cookie motd\n";
1420 print $client "Command? ";
1425 =head1 UDP: Message Passing
1427 Another kind of client-server setup is one that uses not connections, but
1428 messages. UDP communications involve much lower overhead but also provide
1429 less reliability, as there are no promises that messages will arrive at
1430 all, let alone in order and unmangled. Still, UDP offers some advantages
1431 over TCP, including being able to "broadcast" or "multicast" to a whole
1432 bunch of destination hosts at once (usually on your local subnet). If you
1433 find yourself overly concerned about reliability and start building checks
1434 into your message system, then you probably should use just TCP to start
1437 Note that UDP datagrams are I<not> a bytestream and should not be treated
1438 as such. This makes using I/O mechanisms with internal buffering
1439 like stdio (i.e. print() and friends) especially cumbersome. Use syswrite(),
1440 or better send(), like in the example below.
1442 Here's a UDP program similar to the sample Internet TCP client given
1443 earlier. However, instead of checking one host at a time, the UDP version
1444 will check many of them asynchronously by simulating a multicast and then
1445 using select() to do a timed-out wait for I/O. To do something similar
1446 with TCP, you'd have to use a different socket handle for each host.
1453 my ( $count, $hisiaddr, $hispaddr, $histime,
1454 $host, $iaddr, $paddr, $port, $proto,
1455 $rin, $rout, $rtime, $SECS_of_70_YEARS);
1457 $SECS_of_70_YEARS = 2208988800;
1459 $iaddr = gethostbyname(hostname());
1460 $proto = getprotobyname('udp');
1461 $port = getservbyname('time', 'udp');
1462 $paddr = sockaddr_in(0, $iaddr); # 0 means let kernel pick
1464 socket(SOCKET, PF_INET, SOCK_DGRAM, $proto) || die "socket: $!";
1465 bind(SOCKET, $paddr) || die "bind: $!";
1468 printf "%-12s %8s %s\n", "localhost", 0, scalar localtime time;
1472 $hisiaddr = inet_aton($host) || die "unknown host";
1473 $hispaddr = sockaddr_in($port, $hisiaddr);
1474 defined(send(SOCKET, 0, 0, $hispaddr)) || die "send $host: $!";
1478 vec($rin, fileno(SOCKET), 1) = 1;
1480 # timeout after 10.0 seconds
1481 while ($count && select($rout = $rin, undef, undef, 10.0)) {
1483 ($hispaddr = recv(SOCKET, $rtime, 4, 0)) || die "recv: $!";
1484 ($port, $hisiaddr) = sockaddr_in($hispaddr);
1485 $host = gethostbyaddr($hisiaddr, AF_INET);
1486 $histime = unpack("N", $rtime) - $SECS_of_70_YEARS ;
1487 printf "%-12s ", $host;
1488 printf "%8d %s\n", $histime - time, scalar localtime($histime);
1492 Note that this example does not include any retries and may consequently
1493 fail to contact a reachable host. The most prominent reason for this
1494 is congestion of the queues on the sending host if the number of
1495 list of hosts to contact is sufficiently large.
1499 While System V IPC isn't so widely used as sockets, it still has some
1500 interesting uses. You can't, however, effectively use SysV IPC or
1501 Berkeley mmap() to have shared memory so as to share a variable amongst
1502 several processes. That's because Perl would reallocate your string when
1503 you weren't wanting it to.
1505 Here's a small example showing shared memory usage.
1507 use IPC::SysV qw(IPC_PRIVATE IPC_RMID S_IRWXU);
1510 $id = shmget(IPC_PRIVATE, $size, S_IRWXU) || die "$!";
1511 print "shm key $id\n";
1513 $message = "Message #1";
1514 shmwrite($id, $message, 0, 60) || die "$!";
1515 print "wrote: '$message'\n";
1516 shmread($id, $buff, 0, 60) || die "$!";
1517 print "read : '$buff'\n";
1519 # the buffer of shmread is zero-character end-padded.
1520 substr($buff, index($buff, "\0")) = '';
1521 print "un" unless $buff eq $message;
1524 print "deleting shm $id\n";
1525 shmctl($id, IPC_RMID, 0) || die "$!";
1527 Here's an example of a semaphore:
1529 use IPC::SysV qw(IPC_CREAT);
1532 $id = semget($IPC_KEY, 10, 0666 | IPC_CREAT ) || die "$!";
1533 print "shm key $id\n";
1535 Put this code in a separate file to be run in more than one process.
1536 Call the file F<take>:
1538 # create a semaphore
1541 $id = semget($IPC_KEY, 0 , 0 );
1542 die if !defined($id);
1548 # wait for semaphore to be zero
1550 $opstring1 = pack("s!s!s!", $semnum, $semop, $semflag);
1552 # Increment the semaphore count
1554 $opstring2 = pack("s!s!s!", $semnum, $semop, $semflag);
1555 $opstring = $opstring1 . $opstring2;
1557 semop($id,$opstring) || die "$!";
1559 Put this code in a separate file to be run in more than one process.
1560 Call this file F<give>:
1562 # 'give' the semaphore
1563 # run this in the original process and you will see
1564 # that the second process continues
1567 $id = semget($IPC_KEY, 0, 0);
1568 die if !defined($id);
1573 # Decrement the semaphore count
1575 $opstring = pack("s!s!s!", $semnum, $semop, $semflag);
1577 semop($id,$opstring) || die "$!";
1579 The SysV IPC code above was written long ago, and it's definitely
1580 clunky looking. For a more modern look, see the IPC::SysV module
1581 which is included with Perl starting from Perl 5.005.
1583 A small example demonstrating SysV message queues:
1585 use IPC::SysV qw(IPC_PRIVATE IPC_RMID IPC_CREAT S_IRWXU);
1587 my $id = msgget(IPC_PRIVATE, IPC_CREAT | S_IRWXU);
1589 my $sent = "message";
1595 if (msgsnd($id, pack("l! a*", $type_sent, $sent), 0)) {
1596 if (msgrcv($id, $rcvd, 60, 0, 0)) {
1597 ($type_rcvd, $rcvd) = unpack("l! a*", $rcvd);
1598 if ($rcvd eq $sent) {
1604 die "# msgrcv failed\n";
1607 die "# msgsnd failed\n";
1609 msgctl($id, IPC_RMID, 0) || die "# msgctl failed: $!\n";
1611 die "# msgget failed\n";
1616 Most of these routines quietly but politely return C<undef> when they
1617 fail instead of causing your program to die right then and there due to
1618 an uncaught exception. (Actually, some of the new I<Socket> conversion
1619 functions croak() on bad arguments.) It is therefore essential to
1620 check return values from these functions. Always begin your socket
1621 programs this way for optimal success, and don't forget to add B<-T>
1622 taint checking flag to the #! line for servers:
1631 All these routines create system-specific portability problems. As noted
1632 elsewhere, Perl is at the mercy of your C libraries for much of its system
1633 behaviour. It's probably safest to assume broken SysV semantics for
1634 signals and to stick with simple TCP and UDP socket operations; e.g., don't
1635 try to pass open file descriptors over a local UDP datagram socket if you
1636 want your code to stand a chance of being portable.
1638 As mentioned in the signals section, because few vendors provide C
1639 libraries that are safely re-entrant, the prudent programmer will do
1640 little else within a handler beyond setting a numeric variable that
1641 already exists; or, if locked into a slow (restarting) system call,
1642 using die() to raise an exception and longjmp(3) out. In fact, even
1643 these may in some cases cause a core dump. It's probably best to avoid
1644 signals except where they are absolutely inevitable. This
1645 will be addressed in a future release of Perl.
1649 Tom Christiansen, with occasional vestiges of Larry Wall's original
1650 version and suggestions from the Perl Porters.
1654 There's a lot more to networking than this, but this should get you
1657 For intrepid programmers, the indispensable textbook is I<Unix
1658 Network Programming, 2nd Edition, Volume 1> by W. Richard Stevens
1659 (published by Prentice-Hall). Note that most books on networking
1660 address the subject from the perspective of a C programmer; translation
1661 to Perl is left as an exercise for the reader.
1663 The IO::Socket(3) manpage describes the object library, and the Socket(3)
1664 manpage describes the low-level interface to sockets. Besides the obvious
1665 functions in L<perlfunc>, you should also check out the F<modules> file
1666 at your nearest CPAN site. (See L<perlmodlib> or best yet, the F<Perl
1667 FAQ> for a description of what CPAN is and where to get it.)
1669 Section 5 of the F<modules> file is devoted to "Networking, Device Control
1670 (modems), and Interprocess Communication", and contains numerous unbundled
1671 modules numerous networking modules, Chat and Expect operations, CGI
1672 programming, DCE, FTP, IPC, NNTP, Proxy, Ptty, RPC, SNMP, SMTP, Telnet,
1673 Threads, and ToolTalk--just to name a few.