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 or
14 references of user-installed signal handlers. These handlers will be called
15 with an argument which is the name of the signal that triggered it. A
16 signal may be generated intentionally from a particular keyboard sequence like
17 control-C or control-Z, sent to you from another process, or
18 triggered automatically by the kernel when special events transpire, like
19 a child process exiting, your process running out of stack space, or
20 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 perl5.7.3 it was necessary to do as little as you possibly can in your handler;
33 notice how all we do is set a global variable and then raise an exception. That's because on
34 most systems, libraries are not re-entrant; particularly, memory allocation and I/O
35 routines are not. That meant that doing nearly I<anything> in your handler could in theory
36 trigger a memory fault and subsequent core dump - see L<Deferred Signals> below.
38 The names of the signals are the ones listed out by C<kill -l> on your
39 system, or you can retrieve them from the Config module. Set up an
40 @signame list indexed by number to get the name and a %signo table
41 indexed by name to get the number:
44 defined $Config{sig_name} || die "No sigs?";
45 foreach $name (split(' ', $Config{sig_name})) {
51 So to check whether signal 17 and SIGALRM were the same, do just this:
53 print "signal #17 = $signame[17]\n";
55 print "SIGALRM is $signo{ALRM}\n";
58 You may also choose to assign the strings C<'IGNORE'> or C<'DEFAULT'> as
59 the handler, in which case Perl will try to discard the signal or do the
62 On most Unix platforms, the C<CHLD> (sometimes also known as C<CLD>) signal
63 has special behavior with respect to a value of C<'IGNORE'>.
64 Setting C<$SIG{CHLD}> to C<'IGNORE'> on such a platform has the effect of
65 not creating zombie processes when the parent process fails to C<wait()>
66 on its child processes (i.e. child processes are automatically reaped).
67 Calling C<wait()> with C<$SIG{CHLD}> set to C<'IGNORE'> usually returns
68 C<-1> on such platforms.
70 Some signals can be neither trapped nor ignored, such as
71 the KILL and STOP (but not the TSTP) signals. One strategy for
72 temporarily ignoring signals is to use a local() statement, which will be
73 automatically restored once your block is exited. (Remember that local()
74 values are "inherited" by functions called from within that block.)
77 local $SIG{INT} = 'IGNORE';
81 # interrupts still ignored, for now...
84 Sending a signal to a negative process ID means that you send the signal
85 to the entire Unix process-group. This code sends a hang-up signal to all
86 processes in the current process group (and sets $SIG{HUP} to IGNORE so
87 it doesn't kill itself):
90 local $SIG{HUP} = 'IGNORE';
92 # snazzy writing of: kill('HUP', -$$)
95 Another interesting signal to send is signal number zero. This doesn't
96 actually affect another process, but instead checks whether it's alive
97 or has changed its UID.
99 unless (kill 0 => $kid_pid) {
100 warn "something wicked happened to $kid_pid";
103 You might also want to employ anonymous functions for simple signal
106 $SIG{INT} = sub { die "\nOutta here!\n" };
108 But that will be problematic for the more complicated handlers that need
109 to reinstall themselves. Because Perl's signal mechanism is currently
110 based on the signal(3) function from the C library, you may sometimes be so
111 misfortunate as to run on systems where that function is "broken", that
112 is, it behaves in the old unreliable SysV way rather than the newer, more
113 reasonable BSD and POSIX fashion. So you'll see defensive people writing
114 signal handlers like this:
118 # loathe sysV: it makes us not only reinstate
119 # the handler, but place it after the wait
120 $SIG{CHLD} = \&REAPER;
122 $SIG{CHLD} = \&REAPER;
123 # now do something that forks...
127 use POSIX ":sys_wait_h";
130 # If a second child dies while in the signal handler caused by the
131 # first death, we won't get another signal. So must loop here else
132 # we will leave the unreaped child as a zombie. And the next time
133 # two children die we get another zombie. And so on.
134 while (($child = waitpid(-1,WNOHANG)) > 0) {
135 $Kid_Status{$child} = $?;
137 $SIG{CHLD} = \&REAPER; # still loathe sysV
139 $SIG{CHLD} = \&REAPER;
140 # do something that forks...
142 Signal handling is also used for timeouts in Unix, While safely
143 protected within an C<eval{}> block, you set a signal handler to trap
144 alarm signals and then schedule to have one delivered to you in some
145 number of seconds. Then try your blocking operation, clearing the alarm
146 when it's done but not before you've exited your C<eval{}> block. If it
147 goes off, you'll use die() to jump out of the block, much as you might
148 using longjmp() or throw() in other languages.
153 local $SIG{ALRM} = sub { die "alarm clock restart" };
155 flock(FH, 2); # blocking write lock
158 if ($@ and $@ !~ /alarm clock restart/) { die }
160 If the operation being timed out is system() or qx(), this technique
161 is liable to generate zombies. If this matters to you, you'll
162 need to do your own fork() and exec(), and kill the errant child process.
164 For more complex signal handling, you might see the standard POSIX
165 module. Lamentably, this is almost entirely undocumented, but
166 the F<t/lib/posix.t> file from the Perl source distribution has some
171 A named pipe (often referred to as a FIFO) is an old Unix IPC
172 mechanism for processes communicating on the same machine. It works
173 just like a regular, connected anonymous pipes, except that the
174 processes rendezvous using a filename and don't have to be related.
176 To create a named pipe, use the Unix command mknod(1) or on some
177 systems, mkfifo(1). These may not be in your normal path.
179 # system return val is backwards, so && not ||
181 $ENV{PATH} .= ":/etc:/usr/etc";
182 if ( system('mknod', $path, 'p')
183 && system('mkfifo', $path) )
185 die "mk{nod,fifo} $path failed";
189 A fifo is convenient when you want to connect a process to an unrelated
190 one. When you open a fifo, the program will block until there's something
193 For example, let's say you'd like to have your F<.signature> file be a
194 named pipe that has a Perl program on the other end. Now every time any
195 program (like a mailer, news reader, finger program, etc.) tries to read
196 from that file, the reading program will block and your program will
197 supply the new signature. We'll use the pipe-checking file test B<-p>
198 to find out whether anyone (or anything) has accidentally removed our fifo.
201 $FIFO = '.signature';
202 $ENV{PATH} .= ":/etc:/usr/games";
207 system('mknod', $FIFO, 'p')
208 && die "can't mknod $FIFO: $!";
211 # next line blocks until there's a reader
212 open (FIFO, "> $FIFO") || die "can't write $FIFO: $!";
213 print FIFO "John Smith (smith\@host.org)\n", `fortune -s`;
215 sleep 2; # to avoid dup signals
218 =head2 Deferred Signals
220 In perls before perl5.7.3 by installing Perl code to deal with signals, you were exposing
221 yourself to danger from two things. First, few system library functions are
222 re-entrant. If the signal interrupts while Perl is executing one function
223 (like malloc(3) or printf(3)), and your signal handler then calls the
224 same function again, you could get unpredictable behavior--often, a
225 core dump. Second, Perl isn't itself re-entrant at the lowest levels.
226 If the signal interrupts Perl while Perl is changing its own internal
227 data structures, similarly unpredictable behaviour may result.
229 There were two things you could do, knowing this: be paranoid or be
230 pragmatic. The paranoid approach was to do as little as possible in your
231 signal handler. Set an existing integer variable that already has a
232 value, and return. This doesn't help you if you're in a slow system call,
233 which will just restart. That means you have to C<die> to longjump(3) out
234 of the handler. Even this is a little cavalier for the true paranoiac,
235 who avoids C<die> in a handler because the system I<is> out to get you.
236 The pragmatic approach was to say ``I know the risks, but prefer the
237 convenience'', and to do anything you wanted in your signal handler,
238 and be prepared to clean up core dumps now and again.
240 In perl5.7.3 and later to avoid these problems signals are "deferred" - that is when the
241 signal is delivered to the process by the system (to the C code that implements perl) a flag
242 is set, and the handler returns immediately. Then at strategic "safe" points in the perl
243 interpreter (e.g. when it is about to execute a new opcode) the flags are checked
244 and the perl level handler from %SIG is executed. The "deferred" scheme allows much more
245 flexibility in the coding of signal handler as we know perl interpreter is in a
246 safe state, and that we are not in a system library function when the handler is called.
247 However the implementation does differ from previous perls in the following ways:
251 =item Long running opcodes
253 As perl interpreter only looks at the signal flags when it about to execute a new
254 opcode if a signal arrives during a long running opcode (e.g. a regular expression
255 operation on a very large string) then signal will not be seen until operation completes.
257 =item Interrupting IO
259 When a signal is delivered (e.g. INT control-C) the operating system breaks into
260 IO operations like C<read> (used to implement perls E<lt>E<gt> operator). On
261 older perls the handler was called immediately (and as C<read> is not "unsafe" this
262 worked well). With the "deferred" scheme the handler is not called immediately,
263 and if perl is using system's C<stdio> library that library may re-start the C<read>
264 without returning to perl and giving it a chance to call the %SIG handler. If this
265 happens on your system the solution is to use C<:perlio> layer to do IO - at least
266 on those handles which you want to be able to break into with signals. (The C<:perlio>
267 layer checks the signal flags and calls %SIG handlers before resuming IO operation.)
269 =item Signals as "faults"
271 Certain signals e.g. SEGV, ILL, BUS are generated as a result of virtual memory or
272 other "faults". These are normally fatal and there is little a perl-level handler
273 can do with them. (In particular the old signal scheme was particularly unsafe
274 in such cases.) However if a %SIG handler is set the new scheme simply sets a flag
275 and returns as described above. This may cause the operating system to try the
276 offending machine instruction again and - as nothing has changed - it will generate
277 the signal again. The result of this is a rather odd "loop". In future perl's signal
278 mechanism may be changed to avoid this - perhaps by simply disallowing %SIG handlers
279 on signals of that type. Until then the work-round is not to set a %SIG handler on
280 those signals. (Which signals they are is operating system dependant.)
282 =item Signals triggered by operating system state
284 On some operating systems certain signal handlers are supposed to "do something"
285 before returning. One example can be CHLD or CLD which indicates a child process
286 has completed. On some operating systems the signal handler is expected to C<wait>
287 for the completed child process. On such systems the deferred signal scheme will
288 not work for those signals (it does not do the C<wait>). Again the failure will
289 look like a loop as the operating system will re-issue the signal as there are
290 un-waited-for completed child processes.
294 =head1 Using open() for IPC
296 Perl's basic open() statement can also be used for unidirectional interprocess
297 communication by either appending or prepending a pipe symbol to the second
298 argument to open(). Here's how to start something up in a child process you
301 open(SPOOLER, "| cat -v | lpr -h 2>/dev/null")
302 || die "can't fork: $!";
303 local $SIG{PIPE} = sub { die "spooler pipe broke" };
304 print SPOOLER "stuff\n";
305 close SPOOLER || die "bad spool: $! $?";
307 And here's how to start up a child process you intend to read from:
309 open(STATUS, "netstat -an 2>&1 |")
310 || die "can't fork: $!";
312 next if /^(tcp|udp)/;
315 close STATUS || die "bad netstat: $! $?";
317 If one can be sure that a particular program is a Perl script that is
318 expecting filenames in @ARGV, the clever programmer can write something
321 % program f1 "cmd1|" - f2 "cmd2|" f3 < tmpfile
323 and irrespective of which shell it's called from, the Perl program will
324 read from the file F<f1>, the process F<cmd1>, standard input (F<tmpfile>
325 in this case), the F<f2> file, the F<cmd2> command, and finally the F<f3>
326 file. Pretty nifty, eh?
328 You might notice that you could use backticks for much the
329 same effect as opening a pipe for reading:
331 print grep { !/^(tcp|udp)/ } `netstat -an 2>&1`;
332 die "bad netstat" if $?;
334 While this is true on the surface, it's much more efficient to process the
335 file one line or record at a time because then you don't have to read the
336 whole thing into memory at once. It also gives you finer control of the
337 whole process, letting you to kill off the child process early if you'd
340 Be careful to check both the open() and the close() return values. If
341 you're I<writing> to a pipe, you should also trap SIGPIPE. Otherwise,
342 think of what happens when you start up a pipe to a command that doesn't
343 exist: the open() will in all likelihood succeed (it only reflects the
344 fork()'s success), but then your output will fail--spectacularly. Perl
345 can't know whether the command worked because your command is actually
346 running in a separate process whose exec() might have failed. Therefore,
347 while readers of bogus commands return just a quick end of file, writers
348 to bogus command will trigger a signal they'd better be prepared to
351 open(FH, "|bogus") or die "can't fork: $!";
352 print FH "bang\n" or die "can't write: $!";
353 close FH or die "can't close: $!";
355 That won't blow up until the close, and it will blow up with a SIGPIPE.
356 To catch it, you could use this:
358 $SIG{PIPE} = 'IGNORE';
359 open(FH, "|bogus") or die "can't fork: $!";
360 print FH "bang\n" or die "can't write: $!";
361 close FH or die "can't close: status=$?";
365 Both the main process and any child processes it forks share the same
366 STDIN, STDOUT, and STDERR filehandles. If both processes try to access
367 them at once, strange things can happen. You may also want to close
368 or reopen the filehandles for the child. You can get around this by
369 opening your pipe with open(), but on some systems this means that the
370 child process cannot outlive the parent.
372 =head2 Background Processes
374 You can run a command in the background with:
378 The command's STDOUT and STDERR (and possibly STDIN, depending on your
379 shell) will be the same as the parent's. You won't need to catch
380 SIGCHLD because of the double-fork taking place (see below for more
383 =head2 Complete Dissociation of Child from Parent
385 In some cases (starting server processes, for instance) you'll want to
386 completely dissociate the child process from the parent. This is
387 often called daemonization. A well behaved daemon will also chdir()
388 to the root directory (so it doesn't prevent unmounting the filesystem
389 containing the directory from which it was launched) and redirect its
390 standard file descriptors from and to F</dev/null> (so that random
391 output doesn't wind up on the user's terminal).
396 chdir '/' or die "Can't chdir to /: $!";
397 open STDIN, '/dev/null' or die "Can't read /dev/null: $!";
398 open STDOUT, '>/dev/null'
399 or die "Can't write to /dev/null: $!";
400 defined(my $pid = fork) or die "Can't fork: $!";
402 setsid or die "Can't start a new session: $!";
403 open STDERR, '>&STDOUT' or die "Can't dup stdout: $!";
406 The fork() has to come before the setsid() to ensure that you aren't a
407 process group leader (the setsid() will fail if you are). If your
408 system doesn't have the setsid() function, open F</dev/tty> and use the
409 C<TIOCNOTTY> ioctl() on it instead. See L<tty(4)> for details.
411 Non-Unix users should check their Your_OS::Process module for other
414 =head2 Safe Pipe Opens
416 Another interesting approach to IPC is making your single program go
417 multiprocess and communicate between (or even amongst) yourselves. The
418 open() function will accept a file argument of either C<"-|"> or C<"|-">
419 to do a very interesting thing: it forks a child connected to the
420 filehandle you've opened. The child is running the same program as the
421 parent. This is useful for safely opening a file when running under an
422 assumed UID or GID, for example. If you open a pipe I<to> minus, you can
423 write to the filehandle you opened and your kid will find it in his
424 STDIN. If you open a pipe I<from> minus, you can read from the filehandle
425 you opened whatever your kid writes to his STDOUT.
427 use English '-no_match_vars';
431 $pid = open(KID_TO_WRITE, "|-");
432 unless (defined $pid) {
433 warn "cannot fork: $!";
434 die "bailing out" if $sleep_count++ > 6;
437 } until defined $pid;
440 print KID_TO_WRITE @some_data;
441 close(KID_TO_WRITE) || warn "kid exited $?";
443 ($EUID, $EGID) = ($UID, $GID); # suid progs only
444 open (FILE, "> /safe/file")
445 || die "can't open /safe/file: $!";
447 print FILE; # child's STDIN is parent's KID
449 exit; # don't forget this
452 Another common use for this construct is when you need to execute
453 something without the shell's interference. With system(), it's
454 straightforward, but you can't use a pipe open or backticks safely.
455 That's because there's no way to stop the shell from getting its hands on
456 your arguments. Instead, use lower-level control to call exec() directly.
458 Here's a safe backtick or pipe open for read:
460 # add error processing as above
461 $pid = open(KID_TO_READ, "-|");
464 while (<KID_TO_READ>) {
465 # do something interesting
467 close(KID_TO_READ) || warn "kid exited $?";
470 ($EUID, $EGID) = ($UID, $GID); # suid only
471 exec($program, @options, @args)
472 || die "can't exec program: $!";
477 And here's a safe pipe open for writing:
479 # add error processing as above
480 $pid = open(KID_TO_WRITE, "|-");
481 $SIG{ALRM} = sub { die "whoops, $program pipe broke" };
487 close(KID_TO_WRITE) || warn "kid exited $?";
490 ($EUID, $EGID) = ($UID, $GID);
491 exec($program, @options, @args)
492 || die "can't exec program: $!";
496 Note that these operations are full Unix forks, which means they may not be
497 correctly implemented on alien systems. Additionally, these are not true
498 multithreading. If you'd like to learn more about threading, see the
499 F<modules> file mentioned below in the SEE ALSO section.
501 =head2 Bidirectional Communication with Another Process
503 While this works reasonably well for unidirectional communication, what
504 about bidirectional communication? The obvious thing you'd like to do
505 doesn't actually work:
507 open(PROG_FOR_READING_AND_WRITING, "| some program |")
509 and if you forget to use the C<use warnings> pragma or the B<-w> flag,
510 then you'll miss out entirely on the diagnostic message:
512 Can't do bidirectional pipe at -e line 1.
514 If you really want to, you can use the standard open2() library function
515 to catch both ends. There's also an open3() for tridirectional I/O so you
516 can also catch your child's STDERR, but doing so would then require an
517 awkward select() loop and wouldn't allow you to use normal Perl input
520 If you look at its source, you'll see that open2() uses low-level
521 primitives like Unix pipe() and exec() calls to create all the connections.
522 While it might have been slightly more efficient by using socketpair(), it
523 would have then been even less portable than it already is. The open2()
524 and open3() functions are unlikely to work anywhere except on a Unix
525 system or some other one purporting to be POSIX compliant.
527 Here's an example of using open2():
531 $pid = open2(*Reader, *Writer, "cat -u -n" );
532 print Writer "stuff\n";
535 The problem with this is that Unix buffering is really going to
536 ruin your day. Even though your C<Writer> filehandle is auto-flushed,
537 and the process on the other end will get your data in a timely manner,
538 you can't usually do anything to force it to give it back to you
539 in a similarly quick fashion. In this case, we could, because we
540 gave I<cat> a B<-u> flag to make it unbuffered. But very few Unix
541 commands are designed to operate over pipes, so this seldom works
542 unless you yourself wrote the program on the other end of the
545 A solution to this is the nonstandard F<Comm.pl> library. It uses
546 pseudo-ttys to make your program behave more reasonably:
549 $ph = open_proc('cat -n');
551 print $ph "a line\n";
552 print "got back ", scalar <$ph>;
555 This way you don't have to have control over the source code of the
556 program you're using. The F<Comm> library also has expect()
557 and interact() functions. Find the library (and we hope its
558 successor F<IPC::Chat>) at your nearest CPAN archive as detailed
559 in the SEE ALSO section below.
561 The newer Expect.pm module from CPAN also addresses this kind of thing.
562 This module requires two other modules from CPAN: IO::Pty and IO::Stty.
563 It sets up a pseudo-terminal to interact with programs that insist on
564 using talking to the terminal device driver. If your system is
565 amongst those supported, this may be your best bet.
567 =head2 Bidirectional Communication with Yourself
569 If you want, you may make low-level pipe() and fork()
570 to stitch this together by hand. This example only
571 talks to itself, but you could reopen the appropriate
572 handles to STDIN and STDOUT and call other processes.
575 # pipe1 - bidirectional communication using two pipe pairs
576 # designed for the socketpair-challenged
577 use IO::Handle; # thousands of lines just for autoflush :-(
578 pipe(PARENT_RDR, CHILD_WTR); # XXX: failure?
579 pipe(CHILD_RDR, PARENT_WTR); # XXX: failure?
580 CHILD_WTR->autoflush(1);
581 PARENT_WTR->autoflush(1);
584 close PARENT_RDR; close PARENT_WTR;
585 print CHILD_WTR "Parent Pid $$ is sending this\n";
586 chomp($line = <CHILD_RDR>);
587 print "Parent Pid $$ just read this: `$line'\n";
588 close CHILD_RDR; close CHILD_WTR;
591 die "cannot fork: $!" unless defined $pid;
592 close CHILD_RDR; close CHILD_WTR;
593 chomp($line = <PARENT_RDR>);
594 print "Child Pid $$ just read this: `$line'\n";
595 print PARENT_WTR "Child Pid $$ is sending this\n";
596 close PARENT_RDR; close PARENT_WTR;
600 But you don't actually have to make two pipe calls. If you
601 have the socketpair() system call, it will do this all for you.
604 # pipe2 - bidirectional communication using socketpair
605 # "the best ones always go both ways"
608 use IO::Handle; # thousands of lines just for autoflush :-(
609 # We say AF_UNIX because although *_LOCAL is the
610 # POSIX 1003.1g form of the constant, many machines
611 # still don't have it.
612 socketpair(CHILD, PARENT, AF_UNIX, SOCK_STREAM, PF_UNSPEC)
613 or die "socketpair: $!";
616 PARENT->autoflush(1);
620 print CHILD "Parent Pid $$ is sending this\n";
621 chomp($line = <CHILD>);
622 print "Parent Pid $$ just read this: `$line'\n";
626 die "cannot fork: $!" unless defined $pid;
628 chomp($line = <PARENT>);
629 print "Child Pid $$ just read this: `$line'\n";
630 print PARENT "Child Pid $$ is sending this\n";
635 =head1 Sockets: Client/Server Communication
637 While not limited to Unix-derived operating systems (e.g., WinSock on PCs
638 provides socket support, as do some VMS libraries), you may not have
639 sockets on your system, in which case this section probably isn't going to do
640 you much good. With sockets, you can do both virtual circuits (i.e., TCP
641 streams) and datagrams (i.e., UDP packets). You may be able to do even more
642 depending on your system.
644 The Perl function calls for dealing with sockets have the same names as
645 the corresponding system calls in C, but their arguments tend to differ
646 for two reasons: first, Perl filehandles work differently than C file
647 descriptors. Second, Perl already knows the length of its strings, so you
648 don't need to pass that information.
650 One of the major problems with old socket code in Perl was that it used
651 hard-coded values for some of the constants, which severely hurt
652 portability. If you ever see code that does anything like explicitly
653 setting C<$AF_INET = 2>, you know you're in for big trouble: An
654 immeasurably superior approach is to use the C<Socket> module, which more
655 reliably grants access to various constants and functions you'll need.
657 If you're not writing a server/client for an existing protocol like
658 NNTP or SMTP, you should give some thought to how your server will
659 know when the client has finished talking, and vice-versa. Most
660 protocols are based on one-line messages and responses (so one party
661 knows the other has finished when a "\n" is received) or multi-line
662 messages and responses that end with a period on an empty line
663 ("\n.\n" terminates a message/response).
665 =head2 Internet Line Terminators
667 The Internet line terminator is "\015\012". Under ASCII variants of
668 Unix, that could usually be written as "\r\n", but under other systems,
669 "\r\n" might at times be "\015\015\012", "\012\012\015", or something
670 completely different. The standards specify writing "\015\012" to be
671 conformant (be strict in what you provide), but they also recommend
672 accepting a lone "\012" on input (but be lenient in what you require).
673 We haven't always been very good about that in the code in this manpage,
674 but unless you're on a Mac, you'll probably be ok.
676 =head2 Internet TCP Clients and Servers
678 Use Internet-domain sockets when you want to do client-server
679 communication that might extend to machines outside of your own system.
681 Here's a sample TCP client using Internet-domain sockets:
686 my ($remote,$port, $iaddr, $paddr, $proto, $line);
688 $remote = shift || 'localhost';
689 $port = shift || 2345; # random port
690 if ($port =~ /\D/) { $port = getservbyname($port, 'tcp') }
691 die "No port" unless $port;
692 $iaddr = inet_aton($remote) || die "no host: $remote";
693 $paddr = sockaddr_in($port, $iaddr);
695 $proto = getprotobyname('tcp');
696 socket(SOCK, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
697 connect(SOCK, $paddr) || die "connect: $!";
698 while (defined($line = <SOCK>)) {
702 close (SOCK) || die "close: $!";
705 And here's a corresponding server to go along with it. We'll
706 leave the address as INADDR_ANY so that the kernel can choose
707 the appropriate interface on multihomed hosts. If you want sit
708 on a particular interface (like the external side of a gateway
709 or firewall machine), you should fill this in with your real address
714 BEGIN { $ENV{PATH} = '/usr/ucb:/bin' }
717 my $EOL = "\015\012";
719 sub logmsg { print "$0 $$: @_ at ", scalar localtime, "\n" }
721 my $port = shift || 2345;
722 my $proto = getprotobyname('tcp');
724 ($port) = $port =~ /^(\d+)$/ or die "invalid port";
726 socket(Server, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
727 setsockopt(Server, SOL_SOCKET, SO_REUSEADDR,
728 pack("l", 1)) || die "setsockopt: $!";
729 bind(Server, sockaddr_in($port, INADDR_ANY)) || die "bind: $!";
730 listen(Server,SOMAXCONN) || die "listen: $!";
732 logmsg "server started on port $port";
736 $SIG{CHLD} = \&REAPER;
738 for ( ; $paddr = accept(Client,Server); close Client) {
739 my($port,$iaddr) = sockaddr_in($paddr);
740 my $name = gethostbyaddr($iaddr,AF_INET);
742 logmsg "connection from $name [",
743 inet_ntoa($iaddr), "]
746 print Client "Hello there, $name, it's now ",
747 scalar localtime, $EOL;
750 And here's a multithreaded version. It's multithreaded in that
751 like most typical servers, it spawns (forks) a slave server to
752 handle the client request so that the master server can quickly
753 go back to service a new client.
757 BEGIN { $ENV{PATH} = '/usr/ucb:/bin' }
760 my $EOL = "\015\012";
762 sub spawn; # forward declaration
763 sub logmsg { print "$0 $$: @_ at ", scalar localtime, "\n" }
765 my $port = shift || 2345;
766 my $proto = getprotobyname('tcp');
768 ($port) = $port =~ /^(\d+)$/ or die "invalid port";
770 socket(Server, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
771 setsockopt(Server, SOL_SOCKET, SO_REUSEADDR,
772 pack("l", 1)) || die "setsockopt: $!";
773 bind(Server, sockaddr_in($port, INADDR_ANY)) || die "bind: $!";
774 listen(Server,SOMAXCONN) || die "listen: $!";
776 logmsg "server started on port $port";
781 use POSIX ":sys_wait_h";
784 while (($waitedpid = waitpid(-1,WNOHANG)) > 0) {
785 logmsg "reaped $waitedpid" . ($? ? " with exit $?" : '');
787 $SIG{CHLD} = \&REAPER; # loathe sysV
790 $SIG{CHLD} = \&REAPER;
792 for ( $waitedpid = 0;
793 ($paddr = accept(Client,Server)) || $waitedpid;
794 $waitedpid = 0, close Client)
796 next if $waitedpid and not $paddr;
797 my($port,$iaddr) = sockaddr_in($paddr);
798 my $name = gethostbyaddr($iaddr,AF_INET);
800 logmsg "connection from $name [",
801 inet_ntoa($iaddr), "]
806 print "Hello there, $name, it's now ", scalar localtime, $EOL;
807 exec '/usr/games/fortune' # XXX: `wrong' line terminators
808 or confess "can't exec fortune: $!";
816 unless (@_ == 0 && $coderef && ref($coderef) eq 'CODE') {
817 confess "usage: spawn CODEREF";
821 if (!defined($pid = fork)) {
822 logmsg "cannot fork: $!";
826 return; # I'm the parent
828 # else I'm the child -- go spawn
830 open(STDIN, "<&Client") || die "can't dup client to stdin";
831 open(STDOUT, ">&Client") || die "can't dup client to stdout";
832 ## open(STDERR, ">&STDOUT") || die "can't dup stdout to stderr";
836 This server takes the trouble to clone off a child version via fork() for
837 each incoming request. That way it can handle many requests at once,
838 which you might not always want. Even if you don't fork(), the listen()
839 will allow that many pending connections. Forking servers have to be
840 particularly careful about cleaning up their dead children (called
841 "zombies" in Unix parlance), because otherwise you'll quickly fill up your
844 We suggest that you use the B<-T> flag to use taint checking (see L<perlsec>)
845 even if we aren't running setuid or setgid. This is always a good idea
846 for servers and other programs run on behalf of someone else (like CGI
847 scripts), because it lessens the chances that people from the outside will
848 be able to compromise your system.
850 Let's look at another TCP client. This one connects to the TCP "time"
851 service on a number of different machines and shows how far their clocks
852 differ from the system on which it's being run:
858 my $SECS_of_70_YEARS = 2208988800;
859 sub ctime { scalar localtime(shift) }
861 my $iaddr = gethostbyname('localhost');
862 my $proto = getprotobyname('tcp');
863 my $port = getservbyname('time', 'tcp');
864 my $paddr = sockaddr_in(0, $iaddr);
868 printf "%-24s %8s %s\n", "localhost", 0, ctime(time());
870 foreach $host (@ARGV) {
871 printf "%-24s ", $host;
872 my $hisiaddr = inet_aton($host) || die "unknown host";
873 my $hispaddr = sockaddr_in($port, $hisiaddr);
874 socket(SOCKET, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
875 connect(SOCKET, $hispaddr) || die "bind: $!";
877 read(SOCKET, $rtime, 4);
879 my $histime = unpack("N", $rtime) - $SECS_of_70_YEARS ;
880 printf "%8d %s\n", $histime - time, ctime($histime);
883 =head2 Unix-Domain TCP Clients and Servers
885 That's fine for Internet-domain clients and servers, but what about local
886 communications? While you can use the same setup, sometimes you don't
887 want to. Unix-domain sockets are local to the current host, and are often
888 used internally to implement pipes. Unlike Internet domain sockets, Unix
889 domain sockets can show up in the file system with an ls(1) listing.
892 srw-rw-rw- 1 root 0 Oct 31 07:23 /dev/log
894 You can test for these with Perl's B<-S> file test:
896 unless ( -S '/dev/log' ) {
897 die "something's wicked with the log system";
900 Here's a sample Unix-domain client:
905 my ($rendezvous, $line);
907 $rendezvous = shift || '/tmp/catsock';
908 socket(SOCK, PF_UNIX, SOCK_STREAM, 0) || die "socket: $!";
909 connect(SOCK, sockaddr_un($rendezvous)) || die "connect: $!";
910 while (defined($line = <SOCK>)) {
915 And here's a corresponding server. You don't have to worry about silly
916 network terminators here because Unix domain sockets are guaranteed
917 to be on the localhost, and thus everything works right.
924 BEGIN { $ENV{PATH} = '/usr/ucb:/bin' }
925 sub spawn; # forward declaration
926 sub logmsg { print "$0 $$: @_ at ", scalar localtime, "\n" }
928 my $NAME = '/tmp/catsock';
929 my $uaddr = sockaddr_un($NAME);
930 my $proto = getprotobyname('tcp');
932 socket(Server,PF_UNIX,SOCK_STREAM,0) || die "socket: $!";
934 bind (Server, $uaddr) || die "bind: $!";
935 listen(Server,SOMAXCONN) || die "listen: $!";
937 logmsg "server started on $NAME";
941 use POSIX ":sys_wait_h";
944 while (($waitedpid = waitpid(-1,WNOHANG)) > 0) {
945 logmsg "reaped $waitedpid" . ($? ? " with exit $?" : '');
947 $SIG{CHLD} = \&REAPER; # loathe sysV
950 $SIG{CHLD} = \&REAPER;
953 for ( $waitedpid = 0;
954 accept(Client,Server) || $waitedpid;
955 $waitedpid = 0, close Client)
958 logmsg "connection on $NAME";
960 print "Hello there, it's now ", scalar localtime, "\n";
961 exec '/usr/games/fortune' or die "can't exec fortune: $!";
968 unless (@_ == 0 && $coderef && ref($coderef) eq 'CODE') {
969 confess "usage: spawn CODEREF";
973 if (!defined($pid = fork)) {
974 logmsg "cannot fork: $!";
978 return; # I'm the parent
980 # else I'm the child -- go spawn
982 open(STDIN, "<&Client") || die "can't dup client to stdin";
983 open(STDOUT, ">&Client") || die "can't dup client to stdout";
984 ## open(STDERR, ">&STDOUT") || die "can't dup stdout to stderr";
988 As you see, it's remarkably similar to the Internet domain TCP server, so
989 much so, in fact, that we've omitted several duplicate functions--spawn(),
990 logmsg(), ctime(), and REAPER()--which are exactly the same as in the
993 So why would you ever want to use a Unix domain socket instead of a
994 simpler named pipe? Because a named pipe doesn't give you sessions. You
995 can't tell one process's data from another's. With socket programming,
996 you get a separate session for each client: that's why accept() takes two
999 For example, let's say that you have a long running database server daemon
1000 that you want folks from the World Wide Web to be able to access, but only
1001 if they go through a CGI interface. You'd have a small, simple CGI
1002 program that does whatever checks and logging you feel like, and then acts
1003 as a Unix-domain client and connects to your private server.
1005 =head1 TCP Clients with IO::Socket
1007 For those preferring a higher-level interface to socket programming, the
1008 IO::Socket module provides an object-oriented approach. IO::Socket is
1009 included as part of the standard Perl distribution as of the 5.004
1010 release. If you're running an earlier version of Perl, just fetch
1011 IO::Socket from CPAN, where you'll also find modules providing easy
1012 interfaces to the following systems: DNS, FTP, Ident (RFC 931), NIS and
1013 NISPlus, NNTP, Ping, POP3, SMTP, SNMP, SSLeay, Telnet, and Time--just
1016 =head2 A Simple Client
1018 Here's a client that creates a TCP connection to the "daytime"
1019 service at port 13 of the host name "localhost" and prints out everything
1020 that the server there cares to provide.
1024 $remote = IO::Socket::INET->new(
1026 PeerAddr => "localhost",
1027 PeerPort => "daytime(13)",
1029 or die "cannot connect to daytime port at localhost";
1030 while ( <$remote> ) { print }
1032 When you run this program, you should get something back that
1035 Wed May 14 08:40:46 MDT 1997
1037 Here are what those parameters to the C<new> constructor mean:
1043 This is which protocol to use. In this case, the socket handle returned
1044 will be connected to a TCP socket, because we want a stream-oriented
1045 connection, that is, one that acts pretty much like a plain old file.
1046 Not all sockets are this of this type. For example, the UDP protocol
1047 can be used to make a datagram socket, used for message-passing.
1051 This is the name or Internet address of the remote host the server is
1052 running on. We could have specified a longer name like C<"www.perl.com">,
1053 or an address like C<"204.148.40.9">. For demonstration purposes, we've
1054 used the special hostname C<"localhost">, which should always mean the
1055 current machine you're running on. The corresponding Internet address
1056 for localhost is C<"127.1">, if you'd rather use that.
1060 This is the service name or port number we'd like to connect to.
1061 We could have gotten away with using just C<"daytime"> on systems with a
1062 well-configured system services file,[FOOTNOTE: The system services file
1063 is in I</etc/services> under Unix] but just in case, we've specified the
1064 port number (13) in parentheses. Using just the number would also have
1065 worked, but constant numbers make careful programmers nervous.
1069 Notice how the return value from the C<new> constructor is used as
1070 a filehandle in the C<while> loop? That's what's called an indirect
1071 filehandle, a scalar variable containing a filehandle. You can use
1072 it the same way you would a normal filehandle. For example, you
1073 can read one line from it this way:
1077 all remaining lines from is this way:
1081 and send a line of data to it this way:
1083 print $handle "some data\n";
1085 =head2 A Webget Client
1087 Here's a simple client that takes a remote host to fetch a document
1088 from, and then a list of documents to get from that host. This is a
1089 more interesting client than the previous one because it first sends
1090 something to the server before fetching the server's response.
1094 unless (@ARGV > 1) { die "usage: $0 host document ..." }
1095 $host = shift(@ARGV);
1098 foreach $document ( @ARGV ) {
1099 $remote = IO::Socket::INET->new( Proto => "tcp",
1101 PeerPort => "http(80)",
1103 unless ($remote) { die "cannot connect to http daemon on $host" }
1104 $remote->autoflush(1);
1105 print $remote "GET $document HTTP/1.0" . $BLANK;
1106 while ( <$remote> ) { print }
1110 The web server handing the "http" service, which is assumed to be at
1111 its standard port, number 80. If the web server you're trying to
1112 connect to is at a different port (like 1080 or 8080), you should specify
1113 as the named-parameter pair, C<< PeerPort => 8080 >>. The C<autoflush>
1114 method is used on the socket because otherwise the system would buffer
1115 up the output we sent it. (If you're on a Mac, you'll also need to
1116 change every C<"\n"> in your code that sends data over the network to
1117 be a C<"\015\012"> instead.)
1119 Connecting to the server is only the first part of the process: once you
1120 have the connection, you have to use the server's language. Each server
1121 on the network has its own little command language that it expects as
1122 input. The string that we send to the server starting with "GET" is in
1123 HTTP syntax. In this case, we simply request each specified document.
1124 Yes, we really are making a new connection for each document, even though
1125 it's the same host. That's the way you always used to have to speak HTTP.
1126 Recent versions of web browsers may request that the remote server leave
1127 the connection open a little while, but the server doesn't have to honor
1130 Here's an example of running that program, which we'll call I<webget>:
1132 % webget www.perl.com /guanaco.html
1133 HTTP/1.1 404 File Not Found
1134 Date: Thu, 08 May 1997 18:02:32 GMT
1135 Server: Apache/1.2b6
1137 Content-type: text/html
1139 <HEAD><TITLE>404 File Not Found</TITLE></HEAD>
1140 <BODY><H1>File Not Found</H1>
1141 The requested URL /guanaco.html was not found on this server.<P>
1144 Ok, so that's not very interesting, because it didn't find that
1145 particular document. But a long response wouldn't have fit on this page.
1147 For a more fully-featured version of this program, you should look to
1148 the I<lwp-request> program included with the LWP modules from CPAN.
1150 =head2 Interactive Client with IO::Socket
1152 Well, that's all fine if you want to send one command and get one answer,
1153 but what about setting up something fully interactive, somewhat like
1154 the way I<telnet> works? That way you can type a line, get the answer,
1155 type a line, get the answer, etc.
1157 This client is more complicated than the two we've done so far, but if
1158 you're on a system that supports the powerful C<fork> call, the solution
1159 isn't that rough. Once you've made the connection to whatever service
1160 you'd like to chat with, call C<fork> to clone your process. Each of
1161 these two identical process has a very simple job to do: the parent
1162 copies everything from the socket to standard output, while the child
1163 simultaneously copies everything from standard input to the socket.
1164 To accomplish the same thing using just one process would be I<much>
1165 harder, because it's easier to code two processes to do one thing than it
1166 is to code one process to do two things. (This keep-it-simple principle
1167 a cornerstones of the Unix philosophy, and good software engineering as
1168 well, which is probably why it's spread to other systems.)
1175 my ($host, $port, $kidpid, $handle, $line);
1177 unless (@ARGV == 2) { die "usage: $0 host port" }
1178 ($host, $port) = @ARGV;
1180 # create a tcp connection to the specified host and port
1181 $handle = IO::Socket::INET->new(Proto => "tcp",
1184 or die "can't connect to port $port on $host: $!";
1186 $handle->autoflush(1); # so output gets there right away
1187 print STDERR "[Connected to $host:$port]\n";
1189 # split the program into two processes, identical twins
1190 die "can't fork: $!" unless defined($kidpid = fork());
1192 # the if{} block runs only in the parent process
1194 # copy the socket to standard output
1195 while (defined ($line = <$handle>)) {
1198 kill("TERM", $kidpid); # send SIGTERM to child
1200 # the else{} block runs only in the child process
1202 # copy standard input to the socket
1203 while (defined ($line = <STDIN>)) {
1204 print $handle $line;
1208 The C<kill> function in the parent's C<if> block is there to send a
1209 signal to our child process (current running in the C<else> block)
1210 as soon as the remote server has closed its end of the connection.
1212 If the remote server sends data a byte at time, and you need that
1213 data immediately without waiting for a newline (which might not happen),
1214 you may wish to replace the C<while> loop in the parent with the
1218 while (sysread($handle, $byte, 1) == 1) {
1222 Making a system call for each byte you want to read is not very efficient
1223 (to put it mildly) but is the simplest to explain and works reasonably
1226 =head1 TCP Servers with IO::Socket
1228 As always, setting up a server is little bit more involved than running a client.
1229 The model is that the server creates a special kind of socket that
1230 does nothing but listen on a particular port for incoming connections.
1231 It does this by calling the C<< IO::Socket::INET->new() >> method with
1232 slightly different arguments than the client did.
1238 This is which protocol to use. Like our clients, we'll
1239 still specify C<"tcp"> here.
1244 port in the C<LocalPort> argument, which we didn't do for the client.
1245 This is service name or port number for which you want to be the
1246 server. (Under Unix, ports under 1024 are restricted to the
1247 superuser.) In our sample, we'll use port 9000, but you can use
1248 any port that's not currently in use on your system. If you try
1249 to use one already in used, you'll get an "Address already in use"
1250 message. Under Unix, the C<netstat -a> command will show
1251 which services current have servers.
1255 The C<Listen> parameter is set to the maximum number of
1256 pending connections we can accept until we turn away incoming clients.
1257 Think of it as a call-waiting queue for your telephone.
1258 The low-level Socket module has a special symbol for the system maximum, which
1263 The C<Reuse> parameter is needed so that we restart our server
1264 manually without waiting a few minutes to allow system buffers to
1269 Once the generic server socket has been created using the parameters
1270 listed above, the server then waits for a new client to connect
1271 to it. The server blocks in the C<accept> method, which eventually accepts a
1272 bidirectional connection from the remote client. (Make sure to autoflush
1273 this handle to circumvent buffering.)
1275 To add to user-friendliness, our server prompts the user for commands.
1276 Most servers don't do this. Because of the prompt without a newline,
1277 you'll have to use the C<sysread> variant of the interactive client above.
1279 This server accepts one of five different commands, sending output
1280 back to the client. Note that unlike most network servers, this one
1281 only handles one incoming client at a time. Multithreaded servers are
1282 covered in Chapter 6 of the Camel.
1284 Here's the code. We'll
1288 use Net::hostent; # for OO version of gethostbyaddr
1290 $PORT = 9000; # pick something not in use
1292 $server = IO::Socket::INET->new( Proto => 'tcp',
1294 Listen => SOMAXCONN,
1297 die "can't setup server" unless $server;
1298 print "[Server $0 accepting clients]\n";
1300 while ($client = $server->accept()) {
1301 $client->autoflush(1);
1302 print $client "Welcome to $0; type help for command list.\n";
1303 $hostinfo = gethostbyaddr($client->peeraddr);
1304 printf "[Connect from %s]\n", $hostinfo->name || $client->peerhost;
1305 print $client "Command? ";
1306 while ( <$client>) {
1307 next unless /\S/; # blank line
1308 if (/quit|exit/i) { last; }
1309 elsif (/date|time/i) { printf $client "%s\n", scalar localtime; }
1310 elsif (/who/i ) { print $client `who 2>&1`; }
1311 elsif (/cookie/i ) { print $client `/usr/games/fortune 2>&1`; }
1312 elsif (/motd/i ) { print $client `cat /etc/motd 2>&1`; }
1314 print $client "Commands: quit date who cookie motd\n";
1317 print $client "Command? ";
1322 =head1 UDP: Message Passing
1324 Another kind of client-server setup is one that uses not connections, but
1325 messages. UDP communications involve much lower overhead but also provide
1326 less reliability, as there are no promises that messages will arrive at
1327 all, let alone in order and unmangled. Still, UDP offers some advantages
1328 over TCP, including being able to "broadcast" or "multicast" to a whole
1329 bunch of destination hosts at once (usually on your local subnet). If you
1330 find yourself overly concerned about reliability and start building checks
1331 into your message system, then you probably should use just TCP to start
1334 Note that UDP datagrams are I<not> a bytestream and should not be treated
1335 as such. This makes using I/O mechanisms with internal buffering
1336 like stdio (i.e. print() and friends) especially cumbersome. Use syswrite(),
1337 or better send(), like in the example below.
1339 Here's a UDP program similar to the sample Internet TCP client given
1340 earlier. However, instead of checking one host at a time, the UDP version
1341 will check many of them asynchronously by simulating a multicast and then
1342 using select() to do a timed-out wait for I/O. To do something similar
1343 with TCP, you'd have to use a different socket handle for each host.
1350 my ( $count, $hisiaddr, $hispaddr, $histime,
1351 $host, $iaddr, $paddr, $port, $proto,
1352 $rin, $rout, $rtime, $SECS_of_70_YEARS);
1354 $SECS_of_70_YEARS = 2208988800;
1356 $iaddr = gethostbyname(hostname());
1357 $proto = getprotobyname('udp');
1358 $port = getservbyname('time', 'udp');
1359 $paddr = sockaddr_in(0, $iaddr); # 0 means let kernel pick
1361 socket(SOCKET, PF_INET, SOCK_DGRAM, $proto) || die "socket: $!";
1362 bind(SOCKET, $paddr) || die "bind: $!";
1365 printf "%-12s %8s %s\n", "localhost", 0, scalar localtime time;
1369 $hisiaddr = inet_aton($host) || die "unknown host";
1370 $hispaddr = sockaddr_in($port, $hisiaddr);
1371 defined(send(SOCKET, 0, 0, $hispaddr)) || die "send $host: $!";
1375 vec($rin, fileno(SOCKET), 1) = 1;
1377 # timeout after 10.0 seconds
1378 while ($count && select($rout = $rin, undef, undef, 10.0)) {
1380 ($hispaddr = recv(SOCKET, $rtime, 4, 0)) || die "recv: $!";
1381 ($port, $hisiaddr) = sockaddr_in($hispaddr);
1382 $host = gethostbyaddr($hisiaddr, AF_INET);
1383 $histime = unpack("N", $rtime) - $SECS_of_70_YEARS ;
1384 printf "%-12s ", $host;
1385 printf "%8d %s\n", $histime - time, scalar localtime($histime);
1389 Note that this example does not include any retries and may consequently
1390 fail to contact a reachable host. The most prominent reason for this
1391 is congestion of the queues on the sending host if the number of
1392 list of hosts to contact is sufficiently large.
1396 While System V IPC isn't so widely used as sockets, it still has some
1397 interesting uses. You can't, however, effectively use SysV IPC or
1398 Berkeley mmap() to have shared memory so as to share a variable amongst
1399 several processes. That's because Perl would reallocate your string when
1400 you weren't wanting it to.
1402 Here's a small example showing shared memory usage.
1404 use IPC::SysV qw(IPC_PRIVATE IPC_RMID S_IRWXU);
1407 $id = shmget(IPC_PRIVATE, $size, S_IRWXU) || die "$!";
1408 print "shm key $id\n";
1410 $message = "Message #1";
1411 shmwrite($id, $message, 0, 60) || die "$!";
1412 print "wrote: '$message'\n";
1413 shmread($id, $buff, 0, 60) || die "$!";
1414 print "read : '$buff'\n";
1416 # the buffer of shmread is zero-character end-padded.
1417 substr($buff, index($buff, "\0")) = '';
1418 print "un" unless $buff eq $message;
1421 print "deleting shm $id\n";
1422 shmctl($id, IPC_RMID, 0) || die "$!";
1424 Here's an example of a semaphore:
1426 use IPC::SysV qw(IPC_CREAT);
1429 $id = semget($IPC_KEY, 10, 0666 | IPC_CREAT ) || die "$!";
1430 print "shm key $id\n";
1432 Put this code in a separate file to be run in more than one process.
1433 Call the file F<take>:
1435 # create a semaphore
1438 $id = semget($IPC_KEY, 0 , 0 );
1439 die if !defined($id);
1445 # wait for semaphore to be zero
1447 $opstring1 = pack("s!s!s!", $semnum, $semop, $semflag);
1449 # Increment the semaphore count
1451 $opstring2 = pack("s!s!s!", $semnum, $semop, $semflag);
1452 $opstring = $opstring1 . $opstring2;
1454 semop($id,$opstring) || die "$!";
1456 Put this code in a separate file to be run in more than one process.
1457 Call this file F<give>:
1459 # 'give' the semaphore
1460 # run this in the original process and you will see
1461 # that the second process continues
1464 $id = semget($IPC_KEY, 0, 0);
1465 die if !defined($id);
1470 # Decrement the semaphore count
1472 $opstring = pack("s!s!s!", $semnum, $semop, $semflag);
1474 semop($id,$opstring) || die "$!";
1476 The SysV IPC code above was written long ago, and it's definitely
1477 clunky looking. For a more modern look, see the IPC::SysV module
1478 which is included with Perl starting from Perl 5.005.
1480 A small example demonstrating SysV message queues:
1482 use IPC::SysV qw(IPC_PRIVATE IPC_RMID IPC_CREAT S_IRWXU);
1484 my $id = msgget(IPC_PRIVATE, IPC_CREAT | S_IRWXU);
1486 my $sent = "message";
1492 if (msgsnd($id, pack("l! a*", $type_sent, $sent), 0)) {
1493 if (msgrcv($id, $rcvd, 60, 0, 0)) {
1494 ($type_rcvd, $rcvd) = unpack("l! a*", $rcvd);
1495 if ($rcvd eq $sent) {
1501 die "# msgrcv failed\n";
1504 die "# msgsnd failed\n";
1506 msgctl($id, IPC_RMID, 0) || die "# msgctl failed: $!\n";
1508 die "# msgget failed\n";
1513 Most of these routines quietly but politely return C<undef> when they
1514 fail instead of causing your program to die right then and there due to
1515 an uncaught exception. (Actually, some of the new I<Socket> conversion
1516 functions croak() on bad arguments.) It is therefore essential to
1517 check return values from these functions. Always begin your socket
1518 programs this way for optimal success, and don't forget to add B<-T>
1519 taint checking flag to the #! line for servers:
1528 All these routines create system-specific portability problems. As noted
1529 elsewhere, Perl is at the mercy of your C libraries for much of its system
1530 behaviour. It's probably safest to assume broken SysV semantics for
1531 signals and to stick with simple TCP and UDP socket operations; e.g., don't
1532 try to pass open file descriptors over a local UDP datagram socket if you
1533 want your code to stand a chance of being portable.
1535 As mentioned in the signals section, because few vendors provide C
1536 libraries that are safely re-entrant, the prudent programmer will do
1537 little else within a handler beyond setting a numeric variable that
1538 already exists; or, if locked into a slow (restarting) system call,
1539 using die() to raise an exception and longjmp(3) out. In fact, even
1540 these may in some cases cause a core dump. It's probably best to avoid
1541 signals except where they are absolutely inevitable. This
1542 will be addressed in a future release of Perl.
1546 Tom Christiansen, with occasional vestiges of Larry Wall's original
1547 version and suggestions from the Perl Porters.
1551 There's a lot more to networking than this, but this should get you
1554 For intrepid programmers, the indispensable textbook is I<Unix Network
1555 Programming> by W. Richard Stevens (published by Addison-Wesley). Note
1556 that most books on networking address networking from the perspective of
1557 a C programmer; translation to Perl is left as an exercise for the reader.
1559 The IO::Socket(3) manpage describes the object library, and the Socket(3)
1560 manpage describes the low-level interface to sockets. Besides the obvious
1561 functions in L<perlfunc>, you should also check out the F<modules> file
1562 at your nearest CPAN site. (See L<perlmodlib> or best yet, the F<Perl
1563 FAQ> for a description of what CPAN is and where to get it.)
1565 Section 5 of the F<modules> file is devoted to "Networking, Device Control
1566 (modems), and Interprocess Communication", and contains numerous unbundled
1567 modules numerous networking modules, Chat and Expect operations, CGI
1568 programming, DCE, FTP, IPC, NNTP, Proxy, Ptty, RPC, SNMP, SMTP, Telnet,
1569 Threads, and ToolTalk--just to name a few.