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 =item Signals as "faults"
345 Certain signals e.g. SEGV, ILL, BUS are generated as a result of
346 virtual memory or other "faults". These are normally fatal and there
347 is little a Perl-level handler can do with them. (In particular the
348 old signal scheme was particularly unsafe in such cases.) However if
349 a %SIG handler is set the new scheme simply sets a flag and returns as
350 described above. This may cause the operating system to try the
351 offending machine instruction again and - as nothing has changed - it
352 will generate the signal again. The result of this is a rather odd
353 "loop". In future Perl's signal mechanism may be changed to avoid this
354 - perhaps by simply disallowing %SIG handlers on signals of that
355 type. Until then the work-round is not to set a %SIG handler on those
356 signals. (Which signals they are is operating system dependant.)
358 =item Signals triggered by operating system state
360 On some operating systems certain signal handlers are supposed to "do
361 something" before returning. One example can be CHLD or CLD which
362 indicates a child process has completed. On some operating systems the
363 signal handler is expected to C<wait> for the completed child
364 process. On such systems the deferred signal scheme will not work for
365 those signals (it does not do the C<wait>). Again the failure will
366 look like a loop as the operating system will re-issue the signal as
367 there are un-waited-for completed child processes.
371 If you want the old signal behaviour back regardless of possible
372 memory corruption, set the environment variable C<PERL_SIGNALS> to
375 =head1 Using open() for IPC
377 Perl's basic open() statement can also be used for unidirectional
378 interprocess communication by either appending or prepending a pipe
379 symbol to the second argument to open(). Here's how to start
380 something up in a child process you intend to write to:
382 open(SPOOLER, "| cat -v | lpr -h 2>/dev/null")
383 || die "can't fork: $!";
384 local $SIG{PIPE} = sub { die "spooler pipe broke" };
385 print SPOOLER "stuff\n";
386 close SPOOLER || die "bad spool: $! $?";
388 And here's how to start up a child process you intend to read from:
390 open(STATUS, "netstat -an 2>&1 |")
391 || die "can't fork: $!";
393 next if /^(tcp|udp)/;
396 close STATUS || die "bad netstat: $! $?";
398 If one can be sure that a particular program is a Perl script that is
399 expecting filenames in @ARGV, the clever programmer can write something
402 % program f1 "cmd1|" - f2 "cmd2|" f3 < tmpfile
404 and irrespective of which shell it's called from, the Perl program will
405 read from the file F<f1>, the process F<cmd1>, standard input (F<tmpfile>
406 in this case), the F<f2> file, the F<cmd2> command, and finally the F<f3>
407 file. Pretty nifty, eh?
409 You might notice that you could use backticks for much the
410 same effect as opening a pipe for reading:
412 print grep { !/^(tcp|udp)/ } `netstat -an 2>&1`;
413 die "bad netstat" if $?;
415 While this is true on the surface, it's much more efficient to process the
416 file one line or record at a time because then you don't have to read the
417 whole thing into memory at once. It also gives you finer control of the
418 whole process, letting you to kill off the child process early if you'd
421 Be careful to check both the open() and the close() return values. If
422 you're I<writing> to a pipe, you should also trap SIGPIPE. Otherwise,
423 think of what happens when you start up a pipe to a command that doesn't
424 exist: the open() will in all likelihood succeed (it only reflects the
425 fork()'s success), but then your output will fail--spectacularly. Perl
426 can't know whether the command worked because your command is actually
427 running in a separate process whose exec() might have failed. Therefore,
428 while readers of bogus commands return just a quick end of file, writers
429 to bogus command will trigger a signal they'd better be prepared to
432 open(FH, "|bogus") or die "can't fork: $!";
433 print FH "bang\n" or die "can't write: $!";
434 close FH or die "can't close: $!";
436 That won't blow up until the close, and it will blow up with a SIGPIPE.
437 To catch it, you could use this:
439 $SIG{PIPE} = 'IGNORE';
440 open(FH, "|bogus") or die "can't fork: $!";
441 print FH "bang\n" or die "can't write: $!";
442 close FH or die "can't close: status=$?";
446 Both the main process and any child processes it forks share the same
447 STDIN, STDOUT, and STDERR filehandles. If both processes try to access
448 them at once, strange things can happen. You may also want to close
449 or reopen the filehandles for the child. You can get around this by
450 opening your pipe with open(), but on some systems this means that the
451 child process cannot outlive the parent.
453 =head2 Background Processes
455 You can run a command in the background with:
459 The command's STDOUT and STDERR (and possibly STDIN, depending on your
460 shell) will be the same as the parent's. You won't need to catch
461 SIGCHLD because of the double-fork taking place (see below for more
464 =head2 Complete Dissociation of Child from Parent
466 In some cases (starting server processes, for instance) you'll want to
467 completely dissociate the child process from the parent. This is
468 often called daemonization. A well behaved daemon will also chdir()
469 to the root directory (so it doesn't prevent unmounting the filesystem
470 containing the directory from which it was launched) and redirect its
471 standard file descriptors from and to F</dev/null> (so that random
472 output doesn't wind up on the user's terminal).
477 chdir '/' or die "Can't chdir to /: $!";
478 open STDIN, '/dev/null' or die "Can't read /dev/null: $!";
479 open STDOUT, '>/dev/null'
480 or die "Can't write to /dev/null: $!";
481 defined(my $pid = fork) or die "Can't fork: $!";
483 setsid or die "Can't start a new session: $!";
484 open STDERR, '>&STDOUT' or die "Can't dup stdout: $!";
487 The fork() has to come before the setsid() to ensure that you aren't a
488 process group leader (the setsid() will fail if you are). If your
489 system doesn't have the setsid() function, open F</dev/tty> and use the
490 C<TIOCNOTTY> ioctl() on it instead. See L<tty(4)> for details.
492 Non-Unix users should check their Your_OS::Process module for other
495 =head2 Safe Pipe Opens
497 Another interesting approach to IPC is making your single program go
498 multiprocess and communicate between (or even amongst) yourselves. The
499 open() function will accept a file argument of either C<"-|"> or C<"|-">
500 to do a very interesting thing: it forks a child connected to the
501 filehandle you've opened. The child is running the same program as the
502 parent. This is useful for safely opening a file when running under an
503 assumed UID or GID, for example. If you open a pipe I<to> minus, you can
504 write to the filehandle you opened and your kid will find it in his
505 STDIN. If you open a pipe I<from> minus, you can read from the filehandle
506 you opened whatever your kid writes to his STDOUT.
508 use English '-no_match_vars';
512 $pid = open(KID_TO_WRITE, "|-");
513 unless (defined $pid) {
514 warn "cannot fork: $!";
515 die "bailing out" if $sleep_count++ > 6;
518 } until defined $pid;
521 print KID_TO_WRITE @some_data;
522 close(KID_TO_WRITE) || warn "kid exited $?";
524 ($EUID, $EGID) = ($UID, $GID); # suid progs only
525 open (FILE, "> /safe/file")
526 || die "can't open /safe/file: $!";
528 print FILE; # child's STDIN is parent's KID
530 exit; # don't forget this
533 Another common use for this construct is when you need to execute
534 something without the shell's interference. With system(), it's
535 straightforward, but you can't use a pipe open or backticks safely.
536 That's because there's no way to stop the shell from getting its hands on
537 your arguments. Instead, use lower-level control to call exec() directly.
539 Here's a safe backtick or pipe open for read:
541 # add error processing as above
542 $pid = open(KID_TO_READ, "-|");
545 while (<KID_TO_READ>) {
546 # do something interesting
548 close(KID_TO_READ) || warn "kid exited $?";
551 ($EUID, $EGID) = ($UID, $GID); # suid only
552 exec($program, @options, @args)
553 || die "can't exec program: $!";
558 And here's a safe pipe open for writing:
560 # add error processing as above
561 $pid = open(KID_TO_WRITE, "|-");
562 $SIG{PIPE} = sub { die "whoops, $program pipe broke" };
568 close(KID_TO_WRITE) || warn "kid exited $?";
571 ($EUID, $EGID) = ($UID, $GID);
572 exec($program, @options, @args)
573 || die "can't exec program: $!";
577 Since Perl 5.8.0, you can also use the list form of C<open> for pipes :
580 open KID_PS, "-|", "ps", "aux" or die $!;
582 forks the ps(1) command (without spawning a shell, as there are more than
583 three arguments to open()), and reads its standard output via the
584 C<KID_PS> filehandle. The corresponding syntax to read from command
585 pipes (with C<"|-"> in place of C<"-|">) is also implemented.
587 Note that these operations are full Unix forks, which means they may not be
588 correctly implemented on alien systems. Additionally, these are not true
589 multithreading. If you'd like to learn more about threading, see the
590 F<modules> file mentioned below in the SEE ALSO section.
592 =head2 Bidirectional Communication with Another Process
594 While this works reasonably well for unidirectional communication, what
595 about bidirectional communication? The obvious thing you'd like to do
596 doesn't actually work:
598 open(PROG_FOR_READING_AND_WRITING, "| some program |")
600 and if you forget to use the C<use warnings> pragma or the B<-w> flag,
601 then you'll miss out entirely on the diagnostic message:
603 Can't do bidirectional pipe at -e line 1.
605 If you really want to, you can use the standard open2() library function
606 to catch both ends. There's also an open3() for tridirectional I/O so you
607 can also catch your child's STDERR, but doing so would then require an
608 awkward select() loop and wouldn't allow you to use normal Perl input
611 If you look at its source, you'll see that open2() uses low-level
612 primitives like Unix pipe() and exec() calls to create all the connections.
613 While it might have been slightly more efficient by using socketpair(), it
614 would have then been even less portable than it already is. The open2()
615 and open3() functions are unlikely to work anywhere except on a Unix
616 system or some other one purporting to be POSIX compliant.
618 Here's an example of using open2():
622 $pid = open2(*Reader, *Writer, "cat -u -n" );
623 print Writer "stuff\n";
626 The problem with this is that Unix buffering is really going to
627 ruin your day. Even though your C<Writer> filehandle is auto-flushed,
628 and the process on the other end will get your data in a timely manner,
629 you can't usually do anything to force it to give it back to you
630 in a similarly quick fashion. In this case, we could, because we
631 gave I<cat> a B<-u> flag to make it unbuffered. But very few Unix
632 commands are designed to operate over pipes, so this seldom works
633 unless you yourself wrote the program on the other end of the
636 A solution to this is the nonstandard F<Comm.pl> library. It uses
637 pseudo-ttys to make your program behave more reasonably:
640 $ph = open_proc('cat -n');
642 print $ph "a line\n";
643 print "got back ", scalar <$ph>;
646 This way you don't have to have control over the source code of the
647 program you're using. The F<Comm> library also has expect()
648 and interact() functions. Find the library (and we hope its
649 successor F<IPC::Chat>) at your nearest CPAN archive as detailed
650 in the SEE ALSO section below.
652 The newer Expect.pm module from CPAN also addresses this kind of thing.
653 This module requires two other modules from CPAN: IO::Pty and IO::Stty.
654 It sets up a pseudo-terminal to interact with programs that insist on
655 using talking to the terminal device driver. If your system is
656 amongst those supported, this may be your best bet.
658 =head2 Bidirectional Communication with Yourself
660 If you want, you may make low-level pipe() and fork()
661 to stitch this together by hand. This example only
662 talks to itself, but you could reopen the appropriate
663 handles to STDIN and STDOUT and call other processes.
666 # pipe1 - bidirectional communication using two pipe pairs
667 # designed for the socketpair-challenged
668 use IO::Handle; # thousands of lines just for autoflush :-(
669 pipe(PARENT_RDR, CHILD_WTR); # XXX: failure?
670 pipe(CHILD_RDR, PARENT_WTR); # XXX: failure?
671 CHILD_WTR->autoflush(1);
672 PARENT_WTR->autoflush(1);
675 close PARENT_RDR; close PARENT_WTR;
676 print CHILD_WTR "Parent Pid $$ is sending this\n";
677 chomp($line = <CHILD_RDR>);
678 print "Parent Pid $$ just read this: `$line'\n";
679 close CHILD_RDR; close CHILD_WTR;
682 die "cannot fork: $!" unless defined $pid;
683 close CHILD_RDR; close CHILD_WTR;
684 chomp($line = <PARENT_RDR>);
685 print "Child Pid $$ just read this: `$line'\n";
686 print PARENT_WTR "Child Pid $$ is sending this\n";
687 close PARENT_RDR; close PARENT_WTR;
691 But you don't actually have to make two pipe calls. If you
692 have the socketpair() system call, it will do this all for you.
695 # pipe2 - bidirectional communication using socketpair
696 # "the best ones always go both ways"
699 use IO::Handle; # thousands of lines just for autoflush :-(
700 # We say AF_UNIX because although *_LOCAL is the
701 # POSIX 1003.1g form of the constant, many machines
702 # still don't have it.
703 socketpair(CHILD, PARENT, AF_UNIX, SOCK_STREAM, PF_UNSPEC)
704 or die "socketpair: $!";
707 PARENT->autoflush(1);
711 print CHILD "Parent Pid $$ is sending this\n";
712 chomp($line = <CHILD>);
713 print "Parent Pid $$ just read this: `$line'\n";
717 die "cannot fork: $!" unless defined $pid;
719 chomp($line = <PARENT>);
720 print "Child Pid $$ just read this: `$line'\n";
721 print PARENT "Child Pid $$ is sending this\n";
726 =head1 Sockets: Client/Server Communication
728 While not limited to Unix-derived operating systems (e.g., WinSock on PCs
729 provides socket support, as do some VMS libraries), you may not have
730 sockets on your system, in which case this section probably isn't going to do
731 you much good. With sockets, you can do both virtual circuits (i.e., TCP
732 streams) and datagrams (i.e., UDP packets). You may be able to do even more
733 depending on your system.
735 The Perl function calls for dealing with sockets have the same names as
736 the corresponding system calls in C, but their arguments tend to differ
737 for two reasons: first, Perl filehandles work differently than C file
738 descriptors. Second, Perl already knows the length of its strings, so you
739 don't need to pass that information.
741 One of the major problems with old socket code in Perl was that it used
742 hard-coded values for some of the constants, which severely hurt
743 portability. If you ever see code that does anything like explicitly
744 setting C<$AF_INET = 2>, you know you're in for big trouble: An
745 immeasurably superior approach is to use the C<Socket> module, which more
746 reliably grants access to various constants and functions you'll need.
748 If you're not writing a server/client for an existing protocol like
749 NNTP or SMTP, you should give some thought to how your server will
750 know when the client has finished talking, and vice-versa. Most
751 protocols are based on one-line messages and responses (so one party
752 knows the other has finished when a "\n" is received) or multi-line
753 messages and responses that end with a period on an empty line
754 ("\n.\n" terminates a message/response).
756 =head2 Internet Line Terminators
758 The Internet line terminator is "\015\012". Under ASCII variants of
759 Unix, that could usually be written as "\r\n", but under other systems,
760 "\r\n" might at times be "\015\015\012", "\012\012\015", or something
761 completely different. The standards specify writing "\015\012" to be
762 conformant (be strict in what you provide), but they also recommend
763 accepting a lone "\012" on input (but be lenient in what you require).
764 We haven't always been very good about that in the code in this manpage,
765 but unless you're on a Mac, you'll probably be ok.
767 =head2 Internet TCP Clients and Servers
769 Use Internet-domain sockets when you want to do client-server
770 communication that might extend to machines outside of your own system.
772 Here's a sample TCP client using Internet-domain sockets:
777 my ($remote,$port, $iaddr, $paddr, $proto, $line);
779 $remote = shift || 'localhost';
780 $port = shift || 2345; # random port
781 if ($port =~ /\D/) { $port = getservbyname($port, 'tcp') }
782 die "No port" unless $port;
783 $iaddr = inet_aton($remote) || die "no host: $remote";
784 $paddr = sockaddr_in($port, $iaddr);
786 $proto = getprotobyname('tcp');
787 socket(SOCK, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
788 connect(SOCK, $paddr) || die "connect: $!";
789 while (defined($line = <SOCK>)) {
793 close (SOCK) || die "close: $!";
796 And here's a corresponding server to go along with it. We'll
797 leave the address as INADDR_ANY so that the kernel can choose
798 the appropriate interface on multihomed hosts. If you want sit
799 on a particular interface (like the external side of a gateway
800 or firewall machine), you should fill this in with your real address
805 BEGIN { $ENV{PATH} = '/usr/ucb:/bin' }
808 my $EOL = "\015\012";
810 sub logmsg { print "$0 $$: @_ at ", scalar localtime, "\n" }
812 my $port = shift || 2345;
813 my $proto = getprotobyname('tcp');
815 ($port) = $port =~ /^(\d+)$/ or die "invalid port";
817 socket(Server, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
818 setsockopt(Server, SOL_SOCKET, SO_REUSEADDR,
819 pack("l", 1)) || die "setsockopt: $!";
820 bind(Server, sockaddr_in($port, INADDR_ANY)) || die "bind: $!";
821 listen(Server,SOMAXCONN) || die "listen: $!";
823 logmsg "server started on port $port";
827 $SIG{CHLD} = \&REAPER;
829 for ( ; $paddr = accept(Client,Server); close Client) {
830 my($port,$iaddr) = sockaddr_in($paddr);
831 my $name = gethostbyaddr($iaddr,AF_INET);
833 logmsg "connection from $name [",
834 inet_ntoa($iaddr), "]
837 print Client "Hello there, $name, it's now ",
838 scalar localtime, $EOL;
841 And here's a multithreaded version. It's multithreaded in that
842 like most typical servers, it spawns (forks) a slave server to
843 handle the client request so that the master server can quickly
844 go back to service a new client.
848 BEGIN { $ENV{PATH} = '/usr/ucb:/bin' }
851 my $EOL = "\015\012";
853 sub spawn; # forward declaration
854 sub logmsg { print "$0 $$: @_ at ", scalar localtime, "\n" }
856 my $port = shift || 2345;
857 my $proto = getprotobyname('tcp');
859 ($port) = $port =~ /^(\d+)$/ or die "invalid port";
861 socket(Server, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
862 setsockopt(Server, SOL_SOCKET, SO_REUSEADDR,
863 pack("l", 1)) || die "setsockopt: $!";
864 bind(Server, sockaddr_in($port, INADDR_ANY)) || die "bind: $!";
865 listen(Server,SOMAXCONN) || die "listen: $!";
867 logmsg "server started on port $port";
872 use POSIX ":sys_wait_h";
875 while (($waitedpid = waitpid(-1,WNOHANG)) > 0) {
876 logmsg "reaped $waitedpid" . ($? ? " with exit $?" : '');
878 $SIG{CHLD} = \&REAPER; # loathe sysV
881 $SIG{CHLD} = \&REAPER;
883 for ( $waitedpid = 0;
884 ($paddr = accept(Client,Server)) || $waitedpid;
885 $waitedpid = 0, close Client)
887 next if $waitedpid and not $paddr;
888 my($port,$iaddr) = sockaddr_in($paddr);
889 my $name = gethostbyaddr($iaddr,AF_INET);
891 logmsg "connection from $name [",
892 inet_ntoa($iaddr), "]
897 print "Hello there, $name, it's now ", scalar localtime, $EOL;
898 exec '/usr/games/fortune' # XXX: `wrong' line terminators
899 or confess "can't exec fortune: $!";
907 unless (@_ == 0 && $coderef && ref($coderef) eq 'CODE') {
908 confess "usage: spawn CODEREF";
912 if (!defined($pid = fork)) {
913 logmsg "cannot fork: $!";
917 return; # I'm the parent
919 # else I'm the child -- go spawn
921 open(STDIN, "<&Client") || die "can't dup client to stdin";
922 open(STDOUT, ">&Client") || die "can't dup client to stdout";
923 ## open(STDERR, ">&STDOUT") || die "can't dup stdout to stderr";
927 This server takes the trouble to clone off a child version via fork() for
928 each incoming request. That way it can handle many requests at once,
929 which you might not always want. Even if you don't fork(), the listen()
930 will allow that many pending connections. Forking servers have to be
931 particularly careful about cleaning up their dead children (called
932 "zombies" in Unix parlance), because otherwise you'll quickly fill up your
935 We suggest that you use the B<-T> flag to use taint checking (see L<perlsec>)
936 even if we aren't running setuid or setgid. This is always a good idea
937 for servers and other programs run on behalf of someone else (like CGI
938 scripts), because it lessens the chances that people from the outside will
939 be able to compromise your system.
941 Let's look at another TCP client. This one connects to the TCP "time"
942 service on a number of different machines and shows how far their clocks
943 differ from the system on which it's being run:
949 my $SECS_of_70_YEARS = 2208988800;
950 sub ctime { scalar localtime(shift) }
952 my $iaddr = gethostbyname('localhost');
953 my $proto = getprotobyname('tcp');
954 my $port = getservbyname('time', 'tcp');
955 my $paddr = sockaddr_in(0, $iaddr);
959 printf "%-24s %8s %s\n", "localhost", 0, ctime(time());
961 foreach $host (@ARGV) {
962 printf "%-24s ", $host;
963 my $hisiaddr = inet_aton($host) || die "unknown host";
964 my $hispaddr = sockaddr_in($port, $hisiaddr);
965 socket(SOCKET, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
966 connect(SOCKET, $hispaddr) || die "bind: $!";
968 read(SOCKET, $rtime, 4);
970 my $histime = unpack("N", $rtime) - $SECS_of_70_YEARS ;
971 printf "%8d %s\n", $histime - time, ctime($histime);
974 =head2 Unix-Domain TCP Clients and Servers
976 That's fine for Internet-domain clients and servers, but what about local
977 communications? While you can use the same setup, sometimes you don't
978 want to. Unix-domain sockets are local to the current host, and are often
979 used internally to implement pipes. Unlike Internet domain sockets, Unix
980 domain sockets can show up in the file system with an ls(1) listing.
983 srw-rw-rw- 1 root 0 Oct 31 07:23 /dev/log
985 You can test for these with Perl's B<-S> file test:
987 unless ( -S '/dev/log' ) {
988 die "something's wicked with the log system";
991 Here's a sample Unix-domain client:
996 my ($rendezvous, $line);
998 $rendezvous = shift || '/tmp/catsock';
999 socket(SOCK, PF_UNIX, SOCK_STREAM, 0) || die "socket: $!";
1000 connect(SOCK, sockaddr_un($rendezvous)) || die "connect: $!";
1001 while (defined($line = <SOCK>)) {
1006 And here's a corresponding server. You don't have to worry about silly
1007 network terminators here because Unix domain sockets are guaranteed
1008 to be on the localhost, and thus everything works right.
1015 BEGIN { $ENV{PATH} = '/usr/ucb:/bin' }
1016 sub spawn; # forward declaration
1017 sub logmsg { print "$0 $$: @_ at ", scalar localtime, "\n" }
1019 my $NAME = '/tmp/catsock';
1020 my $uaddr = sockaddr_un($NAME);
1021 my $proto = getprotobyname('tcp');
1023 socket(Server,PF_UNIX,SOCK_STREAM,0) || die "socket: $!";
1025 bind (Server, $uaddr) || die "bind: $!";
1026 listen(Server,SOMAXCONN) || die "listen: $!";
1028 logmsg "server started on $NAME";
1032 use POSIX ":sys_wait_h";
1035 while (($waitedpid = waitpid(-1,WNOHANG)) > 0) {
1036 logmsg "reaped $waitedpid" . ($? ? " with exit $?" : '');
1038 $SIG{CHLD} = \&REAPER; # loathe sysV
1041 $SIG{CHLD} = \&REAPER;
1044 for ( $waitedpid = 0;
1045 accept(Client,Server) || $waitedpid;
1046 $waitedpid = 0, close Client)
1049 logmsg "connection on $NAME";
1051 print "Hello there, it's now ", scalar localtime, "\n";
1052 exec '/usr/games/fortune' or die "can't exec fortune: $!";
1057 my $coderef = shift;
1059 unless (@_ == 0 && $coderef && ref($coderef) eq 'CODE') {
1060 confess "usage: spawn CODEREF";
1064 if (!defined($pid = fork)) {
1065 logmsg "cannot fork: $!";
1068 logmsg "begat $pid";
1069 return; # I'm the parent
1071 # else I'm the child -- go spawn
1073 open(STDIN, "<&Client") || die "can't dup client to stdin";
1074 open(STDOUT, ">&Client") || die "can't dup client to stdout";
1075 ## open(STDERR, ">&STDOUT") || die "can't dup stdout to stderr";
1079 As you see, it's remarkably similar to the Internet domain TCP server, so
1080 much so, in fact, that we've omitted several duplicate functions--spawn(),
1081 logmsg(), ctime(), and REAPER()--which are exactly the same as in the
1084 So why would you ever want to use a Unix domain socket instead of a
1085 simpler named pipe? Because a named pipe doesn't give you sessions. You
1086 can't tell one process's data from another's. With socket programming,
1087 you get a separate session for each client: that's why accept() takes two
1090 For example, let's say that you have a long running database server daemon
1091 that you want folks from the World Wide Web to be able to access, but only
1092 if they go through a CGI interface. You'd have a small, simple CGI
1093 program that does whatever checks and logging you feel like, and then acts
1094 as a Unix-domain client and connects to your private server.
1096 =head1 TCP Clients with IO::Socket
1098 For those preferring a higher-level interface to socket programming, the
1099 IO::Socket module provides an object-oriented approach. IO::Socket is
1100 included as part of the standard Perl distribution as of the 5.004
1101 release. If you're running an earlier version of Perl, just fetch
1102 IO::Socket from CPAN, where you'll also find modules providing easy
1103 interfaces to the following systems: DNS, FTP, Ident (RFC 931), NIS and
1104 NISPlus, NNTP, Ping, POP3, SMTP, SNMP, SSLeay, Telnet, and Time--just
1107 =head2 A Simple Client
1109 Here's a client that creates a TCP connection to the "daytime"
1110 service at port 13 of the host name "localhost" and prints out everything
1111 that the server there cares to provide.
1115 $remote = IO::Socket::INET->new(
1117 PeerAddr => "localhost",
1118 PeerPort => "daytime(13)",
1120 or die "cannot connect to daytime port at localhost";
1121 while ( <$remote> ) { print }
1123 When you run this program, you should get something back that
1126 Wed May 14 08:40:46 MDT 1997
1128 Here are what those parameters to the C<new> constructor mean:
1134 This is which protocol to use. In this case, the socket handle returned
1135 will be connected to a TCP socket, because we want a stream-oriented
1136 connection, that is, one that acts pretty much like a plain old file.
1137 Not all sockets are this of this type. For example, the UDP protocol
1138 can be used to make a datagram socket, used for message-passing.
1142 This is the name or Internet address of the remote host the server is
1143 running on. We could have specified a longer name like C<"www.perl.com">,
1144 or an address like C<"204.148.40.9">. For demonstration purposes, we've
1145 used the special hostname C<"localhost">, which should always mean the
1146 current machine you're running on. The corresponding Internet address
1147 for localhost is C<"127.1">, if you'd rather use that.
1151 This is the service name or port number we'd like to connect to.
1152 We could have gotten away with using just C<"daytime"> on systems with a
1153 well-configured system services file,[FOOTNOTE: The system services file
1154 is in I</etc/services> under Unix] but just in case, we've specified the
1155 port number (13) in parentheses. Using just the number would also have
1156 worked, but constant numbers make careful programmers nervous.
1160 Notice how the return value from the C<new> constructor is used as
1161 a filehandle in the C<while> loop? That's what's called an indirect
1162 filehandle, a scalar variable containing a filehandle. You can use
1163 it the same way you would a normal filehandle. For example, you
1164 can read one line from it this way:
1168 all remaining lines from is this way:
1172 and send a line of data to it this way:
1174 print $handle "some data\n";
1176 =head2 A Webget Client
1178 Here's a simple client that takes a remote host to fetch a document
1179 from, and then a list of documents to get from that host. This is a
1180 more interesting client than the previous one because it first sends
1181 something to the server before fetching the server's response.
1185 unless (@ARGV > 1) { die "usage: $0 host document ..." }
1186 $host = shift(@ARGV);
1189 foreach $document ( @ARGV ) {
1190 $remote = IO::Socket::INET->new( Proto => "tcp",
1192 PeerPort => "http(80)",
1194 unless ($remote) { die "cannot connect to http daemon on $host" }
1195 $remote->autoflush(1);
1196 print $remote "GET $document HTTP/1.0" . $BLANK;
1197 while ( <$remote> ) { print }
1201 The web server handing the "http" service, which is assumed to be at
1202 its standard port, number 80. If the web server you're trying to
1203 connect to is at a different port (like 1080 or 8080), you should specify
1204 as the named-parameter pair, C<< PeerPort => 8080 >>. The C<autoflush>
1205 method is used on the socket because otherwise the system would buffer
1206 up the output we sent it. (If you're on a Mac, you'll also need to
1207 change every C<"\n"> in your code that sends data over the network to
1208 be a C<"\015\012"> instead.)
1210 Connecting to the server is only the first part of the process: once you
1211 have the connection, you have to use the server's language. Each server
1212 on the network has its own little command language that it expects as
1213 input. The string that we send to the server starting with "GET" is in
1214 HTTP syntax. In this case, we simply request each specified document.
1215 Yes, we really are making a new connection for each document, even though
1216 it's the same host. That's the way you always used to have to speak HTTP.
1217 Recent versions of web browsers may request that the remote server leave
1218 the connection open a little while, but the server doesn't have to honor
1221 Here's an example of running that program, which we'll call I<webget>:
1223 % webget www.perl.com /guanaco.html
1224 HTTP/1.1 404 File Not Found
1225 Date: Thu, 08 May 1997 18:02:32 GMT
1226 Server: Apache/1.2b6
1228 Content-type: text/html
1230 <HEAD><TITLE>404 File Not Found</TITLE></HEAD>
1231 <BODY><H1>File Not Found</H1>
1232 The requested URL /guanaco.html was not found on this server.<P>
1235 Ok, so that's not very interesting, because it didn't find that
1236 particular document. But a long response wouldn't have fit on this page.
1238 For a more fully-featured version of this program, you should look to
1239 the I<lwp-request> program included with the LWP modules from CPAN.
1241 =head2 Interactive Client with IO::Socket
1243 Well, that's all fine if you want to send one command and get one answer,
1244 but what about setting up something fully interactive, somewhat like
1245 the way I<telnet> works? That way you can type a line, get the answer,
1246 type a line, get the answer, etc.
1248 This client is more complicated than the two we've done so far, but if
1249 you're on a system that supports the powerful C<fork> call, the solution
1250 isn't that rough. Once you've made the connection to whatever service
1251 you'd like to chat with, call C<fork> to clone your process. Each of
1252 these two identical process has a very simple job to do: the parent
1253 copies everything from the socket to standard output, while the child
1254 simultaneously copies everything from standard input to the socket.
1255 To accomplish the same thing using just one process would be I<much>
1256 harder, because it's easier to code two processes to do one thing than it
1257 is to code one process to do two things. (This keep-it-simple principle
1258 a cornerstones of the Unix philosophy, and good software engineering as
1259 well, which is probably why it's spread to other systems.)
1266 my ($host, $port, $kidpid, $handle, $line);
1268 unless (@ARGV == 2) { die "usage: $0 host port" }
1269 ($host, $port) = @ARGV;
1271 # create a tcp connection to the specified host and port
1272 $handle = IO::Socket::INET->new(Proto => "tcp",
1275 or die "can't connect to port $port on $host: $!";
1277 $handle->autoflush(1); # so output gets there right away
1278 print STDERR "[Connected to $host:$port]\n";
1280 # split the program into two processes, identical twins
1281 die "can't fork: $!" unless defined($kidpid = fork());
1283 # the if{} block runs only in the parent process
1285 # copy the socket to standard output
1286 while (defined ($line = <$handle>)) {
1289 kill("TERM", $kidpid); # send SIGTERM to child
1291 # the else{} block runs only in the child process
1293 # copy standard input to the socket
1294 while (defined ($line = <STDIN>)) {
1295 print $handle $line;
1299 The C<kill> function in the parent's C<if> block is there to send a
1300 signal to our child process (current running in the C<else> block)
1301 as soon as the remote server has closed its end of the connection.
1303 If the remote server sends data a byte at time, and you need that
1304 data immediately without waiting for a newline (which might not happen),
1305 you may wish to replace the C<while> loop in the parent with the
1309 while (sysread($handle, $byte, 1) == 1) {
1313 Making a system call for each byte you want to read is not very efficient
1314 (to put it mildly) but is the simplest to explain and works reasonably
1317 =head1 TCP Servers with IO::Socket
1319 As always, setting up a server is little bit more involved than running a client.
1320 The model is that the server creates a special kind of socket that
1321 does nothing but listen on a particular port for incoming connections.
1322 It does this by calling the C<< IO::Socket::INET->new() >> method with
1323 slightly different arguments than the client did.
1329 This is which protocol to use. Like our clients, we'll
1330 still specify C<"tcp"> here.
1335 port in the C<LocalPort> argument, which we didn't do for the client.
1336 This is service name or port number for which you want to be the
1337 server. (Under Unix, ports under 1024 are restricted to the
1338 superuser.) In our sample, we'll use port 9000, but you can use
1339 any port that's not currently in use on your system. If you try
1340 to use one already in used, you'll get an "Address already in use"
1341 message. Under Unix, the C<netstat -a> command will show
1342 which services current have servers.
1346 The C<Listen> parameter is set to the maximum number of
1347 pending connections we can accept until we turn away incoming clients.
1348 Think of it as a call-waiting queue for your telephone.
1349 The low-level Socket module has a special symbol for the system maximum, which
1354 The C<Reuse> parameter is needed so that we restart our server
1355 manually without waiting a few minutes to allow system buffers to
1360 Once the generic server socket has been created using the parameters
1361 listed above, the server then waits for a new client to connect
1362 to it. The server blocks in the C<accept> method, which eventually accepts a
1363 bidirectional connection from the remote client. (Make sure to autoflush
1364 this handle to circumvent buffering.)
1366 To add to user-friendliness, our server prompts the user for commands.
1367 Most servers don't do this. Because of the prompt without a newline,
1368 you'll have to use the C<sysread> variant of the interactive client above.
1370 This server accepts one of five different commands, sending output
1371 back to the client. Note that unlike most network servers, this one
1372 only handles one incoming client at a time. Multithreaded servers are
1373 covered in Chapter 6 of the Camel.
1375 Here's the code. We'll
1379 use Net::hostent; # for OO version of gethostbyaddr
1381 $PORT = 9000; # pick something not in use
1383 $server = IO::Socket::INET->new( Proto => 'tcp',
1385 Listen => SOMAXCONN,
1388 die "can't setup server" unless $server;
1389 print "[Server $0 accepting clients]\n";
1391 while ($client = $server->accept()) {
1392 $client->autoflush(1);
1393 print $client "Welcome to $0; type help for command list.\n";
1394 $hostinfo = gethostbyaddr($client->peeraddr);
1395 printf "[Connect from %s]\n", $hostinfo ? $hostinfo->name : $client->peerhost;
1396 print $client "Command? ";
1397 while ( <$client>) {
1398 next unless /\S/; # blank line
1399 if (/quit|exit/i) { last; }
1400 elsif (/date|time/i) { printf $client "%s\n", scalar localtime; }
1401 elsif (/who/i ) { print $client `who 2>&1`; }
1402 elsif (/cookie/i ) { print $client `/usr/games/fortune 2>&1`; }
1403 elsif (/motd/i ) { print $client `cat /etc/motd 2>&1`; }
1405 print $client "Commands: quit date who cookie motd\n";
1408 print $client "Command? ";
1413 =head1 UDP: Message Passing
1415 Another kind of client-server setup is one that uses not connections, but
1416 messages. UDP communications involve much lower overhead but also provide
1417 less reliability, as there are no promises that messages will arrive at
1418 all, let alone in order and unmangled. Still, UDP offers some advantages
1419 over TCP, including being able to "broadcast" or "multicast" to a whole
1420 bunch of destination hosts at once (usually on your local subnet). If you
1421 find yourself overly concerned about reliability and start building checks
1422 into your message system, then you probably should use just TCP to start
1425 Note that UDP datagrams are I<not> a bytestream and should not be treated
1426 as such. This makes using I/O mechanisms with internal buffering
1427 like stdio (i.e. print() and friends) especially cumbersome. Use syswrite(),
1428 or better send(), like in the example below.
1430 Here's a UDP program similar to the sample Internet TCP client given
1431 earlier. However, instead of checking one host at a time, the UDP version
1432 will check many of them asynchronously by simulating a multicast and then
1433 using select() to do a timed-out wait for I/O. To do something similar
1434 with TCP, you'd have to use a different socket handle for each host.
1441 my ( $count, $hisiaddr, $hispaddr, $histime,
1442 $host, $iaddr, $paddr, $port, $proto,
1443 $rin, $rout, $rtime, $SECS_of_70_YEARS);
1445 $SECS_of_70_YEARS = 2208988800;
1447 $iaddr = gethostbyname(hostname());
1448 $proto = getprotobyname('udp');
1449 $port = getservbyname('time', 'udp');
1450 $paddr = sockaddr_in(0, $iaddr); # 0 means let kernel pick
1452 socket(SOCKET, PF_INET, SOCK_DGRAM, $proto) || die "socket: $!";
1453 bind(SOCKET, $paddr) || die "bind: $!";
1456 printf "%-12s %8s %s\n", "localhost", 0, scalar localtime time;
1460 $hisiaddr = inet_aton($host) || die "unknown host";
1461 $hispaddr = sockaddr_in($port, $hisiaddr);
1462 defined(send(SOCKET, 0, 0, $hispaddr)) || die "send $host: $!";
1466 vec($rin, fileno(SOCKET), 1) = 1;
1468 # timeout after 10.0 seconds
1469 while ($count && select($rout = $rin, undef, undef, 10.0)) {
1471 ($hispaddr = recv(SOCKET, $rtime, 4, 0)) || die "recv: $!";
1472 ($port, $hisiaddr) = sockaddr_in($hispaddr);
1473 $host = gethostbyaddr($hisiaddr, AF_INET);
1474 $histime = unpack("N", $rtime) - $SECS_of_70_YEARS ;
1475 printf "%-12s ", $host;
1476 printf "%8d %s\n", $histime - time, scalar localtime($histime);
1480 Note that this example does not include any retries and may consequently
1481 fail to contact a reachable host. The most prominent reason for this
1482 is congestion of the queues on the sending host if the number of
1483 list of hosts to contact is sufficiently large.
1487 While System V IPC isn't so widely used as sockets, it still has some
1488 interesting uses. You can't, however, effectively use SysV IPC or
1489 Berkeley mmap() to have shared memory so as to share a variable amongst
1490 several processes. That's because Perl would reallocate your string when
1491 you weren't wanting it to.
1493 Here's a small example showing shared memory usage.
1495 use IPC::SysV qw(IPC_PRIVATE IPC_RMID S_IRWXU);
1498 $id = shmget(IPC_PRIVATE, $size, S_IRWXU) || die "$!";
1499 print "shm key $id\n";
1501 $message = "Message #1";
1502 shmwrite($id, $message, 0, 60) || die "$!";
1503 print "wrote: '$message'\n";
1504 shmread($id, $buff, 0, 60) || die "$!";
1505 print "read : '$buff'\n";
1507 # the buffer of shmread is zero-character end-padded.
1508 substr($buff, index($buff, "\0")) = '';
1509 print "un" unless $buff eq $message;
1512 print "deleting shm $id\n";
1513 shmctl($id, IPC_RMID, 0) || die "$!";
1515 Here's an example of a semaphore:
1517 use IPC::SysV qw(IPC_CREAT);
1520 $id = semget($IPC_KEY, 10, 0666 | IPC_CREAT ) || die "$!";
1521 print "shm key $id\n";
1523 Put this code in a separate file to be run in more than one process.
1524 Call the file F<take>:
1526 # create a semaphore
1529 $id = semget($IPC_KEY, 0 , 0 );
1530 die if !defined($id);
1536 # wait for semaphore to be zero
1538 $opstring1 = pack("s!s!s!", $semnum, $semop, $semflag);
1540 # Increment the semaphore count
1542 $opstring2 = pack("s!s!s!", $semnum, $semop, $semflag);
1543 $opstring = $opstring1 . $opstring2;
1545 semop($id,$opstring) || die "$!";
1547 Put this code in a separate file to be run in more than one process.
1548 Call this file F<give>:
1550 # 'give' the semaphore
1551 # run this in the original process and you will see
1552 # that the second process continues
1555 $id = semget($IPC_KEY, 0, 0);
1556 die if !defined($id);
1561 # Decrement the semaphore count
1563 $opstring = pack("s!s!s!", $semnum, $semop, $semflag);
1565 semop($id,$opstring) || die "$!";
1567 The SysV IPC code above was written long ago, and it's definitely
1568 clunky looking. For a more modern look, see the IPC::SysV module
1569 which is included with Perl starting from Perl 5.005.
1571 A small example demonstrating SysV message queues:
1573 use IPC::SysV qw(IPC_PRIVATE IPC_RMID IPC_CREAT S_IRWXU);
1575 my $id = msgget(IPC_PRIVATE, IPC_CREAT | S_IRWXU);
1577 my $sent = "message";
1583 if (msgsnd($id, pack("l! a*", $type_sent, $sent), 0)) {
1584 if (msgrcv($id, $rcvd, 60, 0, 0)) {
1585 ($type_rcvd, $rcvd) = unpack("l! a*", $rcvd);
1586 if ($rcvd eq $sent) {
1592 die "# msgrcv failed\n";
1595 die "# msgsnd failed\n";
1597 msgctl($id, IPC_RMID, 0) || die "# msgctl failed: $!\n";
1599 die "# msgget failed\n";
1604 Most of these routines quietly but politely return C<undef> when they
1605 fail instead of causing your program to die right then and there due to
1606 an uncaught exception. (Actually, some of the new I<Socket> conversion
1607 functions croak() on bad arguments.) It is therefore essential to
1608 check return values from these functions. Always begin your socket
1609 programs this way for optimal success, and don't forget to add B<-T>
1610 taint checking flag to the #! line for servers:
1619 All these routines create system-specific portability problems. As noted
1620 elsewhere, Perl is at the mercy of your C libraries for much of its system
1621 behaviour. It's probably safest to assume broken SysV semantics for
1622 signals and to stick with simple TCP and UDP socket operations; e.g., don't
1623 try to pass open file descriptors over a local UDP datagram socket if you
1624 want your code to stand a chance of being portable.
1626 As mentioned in the signals section, because few vendors provide C
1627 libraries that are safely re-entrant, the prudent programmer will do
1628 little else within a handler beyond setting a numeric variable that
1629 already exists; or, if locked into a slow (restarting) system call,
1630 using die() to raise an exception and longjmp(3) out. In fact, even
1631 these may in some cases cause a core dump. It's probably best to avoid
1632 signals except where they are absolutely inevitable. This
1633 will be addressed in a future release of Perl.
1637 Tom Christiansen, with occasional vestiges of Larry Wall's original
1638 version and suggestions from the Perl Porters.
1642 There's a lot more to networking than this, but this should get you
1645 For intrepid programmers, the indispensable textbook is I<Unix
1646 Network Programming, 2nd Edition, Volume 1> by W. Richard Stevens
1647 (published by Prentice-Hall). Note that most books on networking
1648 address the subject from the perspective of a C programmer; translation
1649 to Perl is left as an exercise for the reader.
1651 The IO::Socket(3) manpage describes the object library, and the Socket(3)
1652 manpage describes the low-level interface to sockets. Besides the obvious
1653 functions in L<perlfunc>, you should also check out the F<modules> file
1654 at your nearest CPAN site. (See L<perlmodlib> or best yet, the F<Perl
1655 FAQ> for a description of what CPAN is and where to get it.)
1657 Section 5 of the F<modules> file is devoted to "Networking, Device Control
1658 (modems), and Interprocess Communication", and contains numerous unbundled
1659 modules numerous networking modules, Chat and Expect operations, CGI
1660 programming, DCE, FTP, IPC, NNTP, Proxy, Ptty, RPC, SNMP, SMTP, Telnet,
1661 Threads, and ToolTalk--just to name a few.