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.
23 Do as little as you possibly can in your handler; notice how all we do is
24 set a global variable and then raise an exception. That's because on most
25 systems, libraries are not re-entrant; particularly, memory allocation and
26 I/O routines are not. That means that doing nearly I<anything> in your
27 handler could in theory trigger a memory fault and subsequent core dump.
32 die "Somebody sent me a SIG$signame";
34 $SIG{INT} = 'catch_zap'; # could fail in modules
35 $SIG{INT} = \&catch_zap; # best strategy
37 The names of the signals are the ones listed out by C<kill -l> on your
38 system, or you can retrieve them from the Config module. Set up an
39 @signame list indexed by number to get the name and a %signo table
40 indexed by name to get the number:
43 defined $Config{sig_name} || die "No sigs?";
44 foreach $name (split(' ', $Config{sig_name})) {
50 So to check whether signal 17 and SIGALRM were the same, do just this:
52 print "signal #17 = $signame[17]\n";
54 print "SIGALRM is $signo{ALRM}\n";
57 You may also choose to assign the strings C<'IGNORE'> or C<'DEFAULT'> as
58 the handler, in which case Perl will try to discard the signal or do the
61 On most UNIX platforms, the C<CHLD> (sometimes also known as C<CLD>) signal
62 has special behavior with respect to a value of C<'IGNORE'>.
63 Setting C<$SIG{CHLD}> to C<'IGNORE'> on such a platform has the effect of
64 not creating zombie processes when the parent process fails to C<wait()>
65 on its child processes (i.e. child processes are automatically reaped).
66 Calling C<wait()> with C<$SIG{CHLD}> set to C<'IGNORE'> usually returns
67 C<-1> on such platforms.
69 Some signals can be neither trapped nor ignored, such as
70 the KILL and STOP (but not the TSTP) signals. One strategy for
71 temporarily ignoring signals is to use a local() statement, which will be
72 automatically restored once your block is exited. (Remember that local()
73 values are "inherited" by functions called from within that block.)
76 local $SIG{INT} = 'IGNORE';
80 # interrupts still ignored, for now...
83 Sending a signal to a negative process ID means that you send the signal
84 to the entire Unix process-group. This code sends a hang-up signal to all
85 processes in the current process group (and sets $SIG{HUP} to IGNORE so
86 it doesn't kill itself):
89 local $SIG{HUP} = 'IGNORE';
91 # snazzy writing of: kill('HUP', -$$)
94 Another interesting signal to send is signal number zero. This doesn't
95 actually affect another process, but instead checks whether it's alive
96 or has changed its UID.
98 unless (kill 0 => $kid_pid) {
99 warn "something wicked happened to $kid_pid";
102 You might also want to employ anonymous functions for simple signal
105 $SIG{INT} = sub { die "\nOutta here!\n" };
107 But that will be problematic for the more complicated handlers that need
108 to reinstall themselves. Because Perl's signal mechanism is currently
109 based on the signal(3) function from the C library, you may sometimes be so
110 misfortunate as to run on systems where that function is "broken", that
111 is, it behaves in the old unreliable SysV way rather than the newer, more
112 reasonable BSD and POSIX fashion. So you'll see defensive people writing
113 signal handlers like this:
117 # loathe sysV: it makes us not only reinstate
118 # the handler, but place it after the wait
119 $SIG{CHLD} = \&REAPER;
121 $SIG{CHLD} = \&REAPER;
122 # now do something that forks...
124 or even the more elaborate:
126 use POSIX ":sys_wait_h";
129 while ($child = waitpid(-1,WNOHANG)) {
130 $Kid_Status{$child} = $?;
132 $SIG{CHLD} = \&REAPER; # still loathe sysV
134 $SIG{CHLD} = \&REAPER;
135 # do something that forks...
137 Signal handling is also used for timeouts in Unix, While safely
138 protected within an C<eval{}> block, you set a signal handler to trap
139 alarm signals and then schedule to have one delivered to you in some
140 number of seconds. Then try your blocking operation, clearing the alarm
141 when it's done but not before you've exited your C<eval{}> block. If it
142 goes off, you'll use die() to jump out of the block, much as you might
143 using longjmp() or throw() in other languages.
148 local $SIG{ALRM} = sub { die "alarm clock restart" };
150 flock(FH, 2); # blocking write lock
153 if ($@ and $@ !~ /alarm clock restart/) { die }
155 For more complex signal handling, you might see the standard POSIX
156 module. Lamentably, this is almost entirely undocumented, but
157 the F<t/lib/posix.t> file from the Perl source distribution has some
162 A named pipe (often referred to as a FIFO) is an old Unix IPC
163 mechanism for processes communicating on the same machine. It works
164 just like a regular, connected anonymous pipes, except that the
165 processes rendezvous using a filename and don't have to be related.
167 To create a named pipe, use the Unix command mknod(1) or on some
168 systems, mkfifo(1). These may not be in your normal path.
170 # system return val is backwards, so && not ||
172 $ENV{PATH} .= ":/etc:/usr/etc";
173 if ( system('mknod', $path, 'p')
174 && system('mkfifo', $path) )
176 die "mk{nod,fifo} $path failed";
180 A fifo is convenient when you want to connect a process to an unrelated
181 one. When you open a fifo, the program will block until there's something
184 For example, let's say you'd like to have your F<.signature> file be a
185 named pipe that has a Perl program on the other end. Now every time any
186 program (like a mailer, news reader, finger program, etc.) tries to read
187 from that file, the reading program will block and your program will
188 supply the new signature. We'll use the pipe-checking file test B<-p>
189 to find out whether anyone (or anything) has accidentally removed our fifo.
192 $FIFO = '.signature';
193 $ENV{PATH} .= ":/etc:/usr/games";
198 system('mknod', $FIFO, 'p')
199 && die "can't mknod $FIFO: $!";
202 # next line blocks until there's a reader
203 open (FIFO, "> $FIFO") || die "can't write $FIFO: $!";
204 print FIFO "John Smith (smith\@host.org)\n", `fortune -s`;
206 sleep 2; # to avoid dup signals
211 By installing Perl code to deal with signals, you're exposing yourself
212 to danger from two things. First, few system library functions are
213 re-entrant. If the signal interrupts while Perl is executing one function
214 (like malloc(3) or printf(3)), and your signal handler then calls the
215 same function again, you could get unpredictable behavior--often, a
216 core dump. Second, Perl isn't itself re-entrant at the lowest levels.
217 If the signal interrupts Perl while Perl is changing its own internal
218 data structures, similarly unpredictable behaviour may result.
220 There are two things you can do, knowing this: be paranoid or be
221 pragmatic. The paranoid approach is to do as little as possible in your
222 signal handler. Set an existing integer variable that already has a
223 value, and return. This doesn't help you if you're in a slow system call,
224 which will just restart. That means you have to C<die> to longjump(3) out
225 of the handler. Even this is a little cavalier for the true paranoiac,
226 who avoids C<die> in a handler because the system I<is> out to get you.
227 The pragmatic approach is to say ``I know the risks, but prefer the
228 convenience'', and to do anything you want in your signal handler,
229 prepared to clean up core dumps now and again.
231 To forbid signal handlers altogether would bars you from
232 many interesting programs, including virtually everything in this manpage,
233 since you could no longer even write SIGCHLD handlers. Their dodginess
234 is expected to be addresses in the 5.005 release.
237 =head1 Using open() for IPC
239 Perl's basic open() statement can also be used for unidirectional interprocess
240 communication by either appending or prepending a pipe symbol to the second
241 argument to open(). Here's how to start something up in a child process you
244 open(SPOOLER, "| cat -v | lpr -h 2>/dev/null")
245 || die "can't fork: $!";
246 local $SIG{PIPE} = sub { die "spooler pipe broke" };
247 print SPOOLER "stuff\n";
248 close SPOOLER || die "bad spool: $! $?";
250 And here's how to start up a child process you intend to read from:
252 open(STATUS, "netstat -an 2>&1 |")
253 || die "can't fork: $!";
255 next if /^(tcp|udp)/;
258 close STATUS || die "bad netstat: $! $?";
260 If one can be sure that a particular program is a Perl script that is
261 expecting filenames in @ARGV, the clever programmer can write something
264 % program f1 "cmd1|" - f2 "cmd2|" f3 < tmpfile
266 and irrespective of which shell it's called from, the Perl program will
267 read from the file F<f1>, the process F<cmd1>, standard input (F<tmpfile>
268 in this case), the F<f2> file, the F<cmd2> command, and finally the F<f3>
269 file. Pretty nifty, eh?
271 You might notice that you could use backticks for much the
272 same effect as opening a pipe for reading:
274 print grep { !/^(tcp|udp)/ } `netstat -an 2>&1`;
275 die "bad netstat" if $?;
277 While this is true on the surface, it's much more efficient to process the
278 file one line or record at a time because then you don't have to read the
279 whole thing into memory at once. It also gives you finer control of the
280 whole process, letting you to kill off the child process early if you'd
283 Be careful to check both the open() and the close() return values. If
284 you're I<writing> to a pipe, you should also trap SIGPIPE. Otherwise,
285 think of what happens when you start up a pipe to a command that doesn't
286 exist: the open() will in all likelihood succeed (it only reflects the
287 fork()'s success), but then your output will fail--spectacularly. Perl
288 can't know whether the command worked because your command is actually
289 running in a separate process whose exec() might have failed. Therefore,
290 while readers of bogus commands return just a quick end of file, writers
291 to bogus command will trigger a signal they'd better be prepared to
294 open(FH, "|bogus") or die "can't fork: $!";
295 print FH "bang\n" or die "can't write: $!";
296 close FH or die "can't close: $!";
298 That won't blow up until the close, and it will blow up with a SIGPIPE.
299 To catch it, you could use this:
301 $SIG{PIPE} = 'IGNORE';
302 open(FH, "|bogus") or die "can't fork: $!";
303 print FH "bang\n" or die "can't write: $!";
304 close FH or die "can't close: status=$?";
308 Both the main process and any child processes it forks share the same
309 STDIN, STDOUT, and STDERR filehandles. If both processes try to access
310 them at once, strange things can happen. You'll certainly want to any
311 stdio flush output buffers before forking. You may also want to close
312 or reopen the filehandles for the child. You can get around this by
313 opening your pipe with open(), but on some systems this means that the
314 child process cannot outlive the parent.
316 =head2 Background Processes
318 You can run a command in the background with:
322 The command's STDOUT and STDERR (and possibly STDIN, depending on your
323 shell) will be the same as the parent's. You won't need to catch
324 SIGCHLD because of the double-fork taking place (see below for more
327 =head2 Complete Dissociation of Child from Parent
329 In some cases (starting server processes, for instance) you'll want to
330 completely dissociate the child process from the parent. This is
331 often called daemonization. A well behaved daemon will also chdir()
332 to the root directory (so it doesn't prevent unmounting the filesystem
333 containing the directory from which it was launched) and redirect its
334 standard file descriptors from and to F</dev/null> (so that random
335 output doesn't wind up on the user's terminal).
340 chdir '/' or die "Can't chdir to /: $!";
341 open STDIN, '/dev/null' or die "Can't read /dev/null: $!";
342 open STDOUT, '>/dev/null'
343 or die "Can't write to /dev/null: $!";
344 defined(my $pid = fork) or die "Can't fork: $!";
346 setsid or die "Can't start a new session: $!";
347 open STDERR, '>&STDOUT' or die "Can't dup stdout: $!";
350 The fork() has to come before the setsid() to ensure that you aren't a
351 process group leader (the setsid() will fail if you are). If your
352 system doesn't have the setsid() function, open F</dev/tty> and use the
353 C<TIOCNOTTY> ioctl() on it instead. See L<tty(4)> for details.
355 Non-Unix users should check their Your_OS::Process module for other
358 =head2 Safe Pipe Opens
360 Another interesting approach to IPC is making your single program go
361 multiprocess and communicate between (or even amongst) yourselves. The
362 open() function will accept a file argument of either C<"-|"> or C<"|-">
363 to do a very interesting thing: it forks a child connected to the
364 filehandle you've opened. The child is running the same program as the
365 parent. This is useful for safely opening a file when running under an
366 assumed UID or GID, for example. If you open a pipe I<to> minus, you can
367 write to the filehandle you opened and your kid will find it in his
368 STDIN. If you open a pipe I<from> minus, you can read from the filehandle
369 you opened whatever your kid writes to his STDOUT.
375 $pid = open(KID_TO_WRITE, "|-");
376 unless (defined $pid) {
377 warn "cannot fork: $!";
378 die "bailing out" if $sleep_count++ > 6;
381 } until defined $pid;
384 print KID_TO_WRITE @some_data;
385 close(KID_TO_WRITE) || warn "kid exited $?";
387 ($EUID, $EGID) = ($UID, $GID); # suid progs only
388 open (FILE, "> /safe/file")
389 || die "can't open /safe/file: $!";
391 print FILE; # child's STDIN is parent's KID
393 exit; # don't forget this
396 Another common use for this construct is when you need to execute
397 something without the shell's interference. With system(), it's
398 straightforward, but you can't use a pipe open or backticks safely.
399 That's because there's no way to stop the shell from getting its hands on
400 your arguments. Instead, use lower-level control to call exec() directly.
402 Here's a safe backtick or pipe open for read:
404 # add error processing as above
405 $pid = open(KID_TO_READ, "-|");
408 while (<KID_TO_READ>) {
409 # do something interesting
411 close(KID_TO_READ) || warn "kid exited $?";
414 ($EUID, $EGID) = ($UID, $GID); # suid only
415 exec($program, @options, @args)
416 || die "can't exec program: $!";
421 And here's a safe pipe open for writing:
423 # add error processing as above
424 $pid = open(KID_TO_WRITE, "|-");
425 $SIG{ALRM} = sub { die "whoops, $program pipe broke" };
431 close(KID_TO_WRITE) || warn "kid exited $?";
434 ($EUID, $EGID) = ($UID, $GID);
435 exec($program, @options, @args)
436 || die "can't exec program: $!";
440 Note that these operations are full Unix forks, which means they may not be
441 correctly implemented on alien systems. Additionally, these are not true
442 multithreading. If you'd like to learn more about threading, see the
443 F<modules> file mentioned below in the SEE ALSO section.
445 =head2 Bidirectional Communication with Another Process
447 While this works reasonably well for unidirectional communication, what
448 about bidirectional communication? The obvious thing you'd like to do
449 doesn't actually work:
451 open(PROG_FOR_READING_AND_WRITING, "| some program |")
453 and if you forget to use the B<-w> flag, then you'll miss out
454 entirely on the diagnostic message:
456 Can't do bidirectional pipe at -e line 1.
458 If you really want to, you can use the standard open2() library function
459 to catch both ends. There's also an open3() for tridirectional I/O so you
460 can also catch your child's STDERR, but doing so would then require an
461 awkward select() loop and wouldn't allow you to use normal Perl input
464 If you look at its source, you'll see that open2() uses low-level
465 primitives like Unix pipe() and exec() calls to create all the connections.
466 While it might have been slightly more efficient by using socketpair(), it
467 would have then been even less portable than it already is. The open2()
468 and open3() functions are unlikely to work anywhere except on a Unix
469 system or some other one purporting to be POSIX compliant.
471 Here's an example of using open2():
475 $pid = open2(*Reader, *Writer, "cat -u -n" );
476 Writer->autoflush(); # default here, actually
477 print Writer "stuff\n";
480 The problem with this is that Unix buffering is really going to
481 ruin your day. Even though your C<Writer> filehandle is auto-flushed,
482 and the process on the other end will get your data in a timely manner,
483 you can't usually do anything to force it to give it back to you
484 in a similarly quick fashion. In this case, we could, because we
485 gave I<cat> a B<-u> flag to make it unbuffered. But very few Unix
486 commands are designed to operate over pipes, so this seldom works
487 unless you yourself wrote the program on the other end of the
490 A solution to this is the nonstandard F<Comm.pl> library. It uses
491 pseudo-ttys to make your program behave more reasonably:
494 $ph = open_proc('cat -n');
496 print $ph "a line\n";
497 print "got back ", scalar <$ph>;
500 This way you don't have to have control over the source code of the
501 program you're using. The F<Comm> library also has expect()
502 and interact() functions. Find the library (and we hope its
503 successor F<IPC::Chat>) at your nearest CPAN archive as detailed
504 in the SEE ALSO section below.
506 The newer Expect.pm module from CPAN also addresses this kind of thing.
507 This module requires two other modules from CPAN: IO::Pty and IO::Stty.
508 It sets up a pseudo-terminal to interact with programs that insist on
509 using talking to the terminal device driver. If your system is
510 amongst those supported, this may be your best bet.
512 =head2 Bidirectional Communication with Yourself
514 If you want, you may make low-level pipe() and fork()
515 to stitch this together by hand. This example only
516 talks to itself, but you could reopen the appropriate
517 handles to STDIN and STDOUT and call other processes.
520 # pipe1 - bidirectional communication using two pipe pairs
521 # designed for the socketpair-challenged
522 use IO::Handle; # thousands of lines just for autoflush :-(
523 pipe(PARENT_RDR, CHILD_WTR); # XXX: failure?
524 pipe(CHILD_RDR, PARENT_WTR); # XXX: failure?
525 CHILD_WTR->autoflush(1);
526 PARENT_WTR->autoflush(1);
529 close PARENT_RDR; close PARENT_WTR;
530 print CHILD_WTR "Parent Pid $$ is sending this\n";
531 chomp($line = <CHILD_RDR>);
532 print "Parent Pid $$ just read this: `$line'\n";
533 close CHILD_RDR; close CHILD_WTR;
536 die "cannot fork: $!" unless defined $pid;
537 close CHILD_RDR; close CHILD_WTR;
538 chomp($line = <PARENT_RDR>);
539 print "Child Pid $$ just read this: `$line'\n";
540 print PARENT_WTR "Child Pid $$ is sending this\n";
541 close PARENT_RDR; close PARENT_WTR;
545 But you don't actually have to make two pipe calls. If you
546 have the socketpair() system call, it will do this all for you.
549 # pipe2 - bidirectional communication using socketpair
550 # "the best ones always go both ways"
553 use IO::Handle; # thousands of lines just for autoflush :-(
554 # We say AF_UNIX because although *_LOCAL is the
555 # POSIX 1003.1g form of the constant, many machines
556 # still don't have it.
557 socketpair(CHILD, PARENT, AF_UNIX, SOCK_STREAM, PF_UNSPEC)
558 or die "socketpair: $!";
561 PARENT->autoflush(1);
565 print CHILD "Parent Pid $$ is sending this\n";
566 chomp($line = <CHILD>);
567 print "Parent Pid $$ just read this: `$line'\n";
571 die "cannot fork: $!" unless defined $pid;
573 chomp($line = <PARENT>);
574 print "Child Pid $$ just read this: `$line'\n";
575 print PARENT "Child Pid $$ is sending this\n";
580 =head1 Sockets: Client/Server Communication
582 While not limited to Unix-derived operating systems (e.g., WinSock on PCs
583 provides socket support, as do some VMS libraries), you may not have
584 sockets on your system, in which case this section probably isn't going to do
585 you much good. With sockets, you can do both virtual circuits (i.e., TCP
586 streams) and datagrams (i.e., UDP packets). You may be able to do even more
587 depending on your system.
589 The Perl function calls for dealing with sockets have the same names as
590 the corresponding system calls in C, but their arguments tend to differ
591 for two reasons: first, Perl filehandles work differently than C file
592 descriptors. Second, Perl already knows the length of its strings, so you
593 don't need to pass that information.
595 One of the major problems with old socket code in Perl was that it used
596 hard-coded values for some of the constants, which severely hurt
597 portability. If you ever see code that does anything like explicitly
598 setting C<$AF_INET = 2>, you know you're in for big trouble: An
599 immeasurably superior approach is to use the C<Socket> module, which more
600 reliably grants access to various constants and functions you'll need.
602 If you're not writing a server/client for an existing protocol like
603 NNTP or SMTP, you should give some thought to how your server will
604 know when the client has finished talking, and vice-versa. Most
605 protocols are based on one-line messages and responses (so one party
606 knows the other has finished when a "\n" is received) or multi-line
607 messages and responses that end with a period on an empty line
608 ("\n.\n" terminates a message/response).
610 =head2 Internet Line Terminators
612 The Internet line terminator is "\015\012". Under ASCII variants of
613 Unix, that could usually be written as "\r\n", but under other systems,
614 "\r\n" might at times be "\015\015\012", "\012\012\015", or something
615 completely different. The standards specify writing "\015\012" to be
616 conformant (be strict in what you provide), but they also recommend
617 accepting a lone "\012" on input (but be lenient in what you require).
618 We haven't always been very good about that in the code in this manpage,
619 but unless you're on a Mac, you'll probably be ok.
621 =head2 Internet TCP Clients and Servers
623 Use Internet-domain sockets when you want to do client-server
624 communication that might extend to machines outside of your own system.
626 Here's a sample TCP client using Internet-domain sockets:
631 my ($remote,$port, $iaddr, $paddr, $proto, $line);
633 $remote = shift || 'localhost';
634 $port = shift || 2345; # random port
635 if ($port =~ /\D/) { $port = getservbyname($port, 'tcp') }
636 die "No port" unless $port;
637 $iaddr = inet_aton($remote) || die "no host: $remote";
638 $paddr = sockaddr_in($port, $iaddr);
640 $proto = getprotobyname('tcp');
641 socket(SOCK, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
642 connect(SOCK, $paddr) || die "connect: $!";
643 while (defined($line = <SOCK>)) {
647 close (SOCK) || die "close: $!";
650 And here's a corresponding server to go along with it. We'll
651 leave the address as INADDR_ANY so that the kernel can choose
652 the appropriate interface on multihomed hosts. If you want sit
653 on a particular interface (like the external side of a gateway
654 or firewall machine), you should fill this in with your real address
659 BEGIN { $ENV{PATH} = '/usr/ucb:/bin' }
664 sub logmsg { print "$0 $$: @_ at ", scalar localtime, "\n" }
666 my $port = shift || 2345;
667 my $proto = getprotobyname('tcp');
668 $port = $1 if $port =~ /(\d+)/; # untaint port number
670 socket(Server, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
671 setsockopt(Server, SOL_SOCKET, SO_REUSEADDR,
672 pack("l", 1)) || die "setsockopt: $!";
673 bind(Server, sockaddr_in($port, INADDR_ANY)) || die "bind: $!";
674 listen(Server,SOMAXCONN) || die "listen: $!";
676 logmsg "server started on port $port";
680 $SIG{CHLD} = \&REAPER;
682 for ( ; $paddr = accept(Client,Server); close Client) {
683 my($port,$iaddr) = sockaddr_in($paddr);
684 my $name = gethostbyaddr($iaddr,AF_INET);
686 logmsg "connection from $name [",
687 inet_ntoa($iaddr), "]
690 print Client "Hello there, $name, it's now ",
691 scalar localtime, $EOL;
694 And here's a multithreaded version. It's multithreaded in that
695 like most typical servers, it spawns (forks) a slave server to
696 handle the client request so that the master server can quickly
697 go back to service a new client.
701 BEGIN { $ENV{PATH} = '/usr/ucb:/bin' }
706 sub spawn; # forward declaration
707 sub logmsg { print "$0 $$: @_ at ", scalar localtime, "\n" }
709 my $port = shift || 2345;
710 my $proto = getprotobyname('tcp');
711 $port = $1 if $port =~ /(\d+)/; # untaint port number
713 socket(Server, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
714 setsockopt(Server, SOL_SOCKET, SO_REUSEADDR,
715 pack("l", 1)) || die "setsockopt: $!";
716 bind(Server, sockaddr_in($port, INADDR_ANY)) || die "bind: $!";
717 listen(Server,SOMAXCONN) || die "listen: $!";
719 logmsg "server started on port $port";
726 $SIG{CHLD} = \&REAPER; # loathe sysV
727 logmsg "reaped $waitedpid" . ($? ? " with exit $?" : '');
730 $SIG{CHLD} = \&REAPER;
732 for ( $waitedpid = 0;
733 ($paddr = accept(Client,Server)) || $waitedpid;
734 $waitedpid = 0, close Client)
736 next if $waitedpid and not $paddr;
737 my($port,$iaddr) = sockaddr_in($paddr);
738 my $name = gethostbyaddr($iaddr,AF_INET);
740 logmsg "connection from $name [",
741 inet_ntoa($iaddr), "]
745 print "Hello there, $name, it's now ", scalar localtime, $EOL;
746 exec '/usr/games/fortune' # XXX: `wrong' line terminators
747 or confess "can't exec fortune: $!";
755 unless (@_ == 0 && $coderef && ref($coderef) eq 'CODE') {
756 confess "usage: spawn CODEREF";
760 if (!defined($pid = fork)) {
761 logmsg "cannot fork: $!";
765 return; # I'm the parent
767 # else I'm the child -- go spawn
769 open(STDIN, "<&Client") || die "can't dup client to stdin";
770 open(STDOUT, ">&Client") || die "can't dup client to stdout";
771 ## open(STDERR, ">&STDOUT") || die "can't dup stdout to stderr";
775 This server takes the trouble to clone off a child version via fork() for
776 each incoming request. That way it can handle many requests at once,
777 which you might not always want. Even if you don't fork(), the listen()
778 will allow that many pending connections. Forking servers have to be
779 particularly careful about cleaning up their dead children (called
780 "zombies" in Unix parlance), because otherwise you'll quickly fill up your
783 We suggest that you use the B<-T> flag to use taint checking (see L<perlsec>)
784 even if we aren't running setuid or setgid. This is always a good idea
785 for servers and other programs run on behalf of someone else (like CGI
786 scripts), because it lessens the chances that people from the outside will
787 be able to compromise your system.
789 Let's look at another TCP client. This one connects to the TCP "time"
790 service on a number of different machines and shows how far their clocks
791 differ from the system on which it's being run:
797 my $SECS_of_70_YEARS = 2208988800;
798 sub ctime { scalar localtime(shift) }
800 my $iaddr = gethostbyname('localhost');
801 my $proto = getprotobyname('tcp');
802 my $port = getservbyname('time', 'tcp');
803 my $paddr = sockaddr_in(0, $iaddr);
807 printf "%-24s %8s %s\n", "localhost", 0, ctime(time());
809 foreach $host (@ARGV) {
810 printf "%-24s ", $host;
811 my $hisiaddr = inet_aton($host) || die "unknown host";
812 my $hispaddr = sockaddr_in($port, $hisiaddr);
813 socket(SOCKET, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
814 connect(SOCKET, $hispaddr) || die "bind: $!";
816 read(SOCKET, $rtime, 4);
818 my $histime = unpack("N", $rtime) - $SECS_of_70_YEARS ;
819 printf "%8d %s\n", $histime - time, ctime($histime);
822 =head2 Unix-Domain TCP Clients and Servers
824 That's fine for Internet-domain clients and servers, but what about local
825 communications? While you can use the same setup, sometimes you don't
826 want to. Unix-domain sockets are local to the current host, and are often
827 used internally to implement pipes. Unlike Internet domain sockets, Unix
828 domain sockets can show up in the file system with an ls(1) listing.
831 srw-rw-rw- 1 root 0 Oct 31 07:23 /dev/log
833 You can test for these with Perl's B<-S> file test:
835 unless ( -S '/dev/log' ) {
836 die "something's wicked with the print system";
839 Here's a sample Unix-domain client:
844 my ($rendezvous, $line);
846 $rendezvous = shift || '/tmp/catsock';
847 socket(SOCK, PF_UNIX, SOCK_STREAM, 0) || die "socket: $!";
848 connect(SOCK, sockaddr_un($rendezvous)) || die "connect: $!";
849 while (defined($line = <SOCK>)) {
854 And here's a corresponding server. You don't have to worry about silly
855 network terminators here because Unix domain sockets are guaranteed
856 to be on the localhost, and thus everything works right.
863 BEGIN { $ENV{PATH} = '/usr/ucb:/bin' }
864 sub logmsg { print "$0 $$: @_ at ", scalar localtime, "\n" }
866 my $NAME = '/tmp/catsock';
867 my $uaddr = sockaddr_un($NAME);
868 my $proto = getprotobyname('tcp');
870 socket(Server,PF_UNIX,SOCK_STREAM,0) || die "socket: $!";
872 bind (Server, $uaddr) || die "bind: $!";
873 listen(Server,SOMAXCONN) || die "listen: $!";
875 logmsg "server started on $NAME";
881 $SIG{CHLD} = \&REAPER; # loathe sysV
882 logmsg "reaped $waitedpid" . ($? ? " with exit $?" : '');
885 $SIG{CHLD} = \&REAPER;
888 for ( $waitedpid = 0;
889 accept(Client,Server) || $waitedpid;
890 $waitedpid = 0, close Client)
893 logmsg "connection on $NAME";
895 print "Hello there, it's now ", scalar localtime, "\n";
896 exec '/usr/games/fortune' or die "can't exec fortune: $!";
900 As you see, it's remarkably similar to the Internet domain TCP server, so
901 much so, in fact, that we've omitted several duplicate functions--spawn(),
902 logmsg(), ctime(), and REAPER()--which are exactly the same as in the
905 So why would you ever want to use a Unix domain socket instead of a
906 simpler named pipe? Because a named pipe doesn't give you sessions. You
907 can't tell one process's data from another's. With socket programming,
908 you get a separate session for each client: that's why accept() takes two
911 For example, let's say that you have a long running database server daemon
912 that you want folks from the World Wide Web to be able to access, but only
913 if they go through a CGI interface. You'd have a small, simple CGI
914 program that does whatever checks and logging you feel like, and then acts
915 as a Unix-domain client and connects to your private server.
917 =head1 TCP Clients with IO::Socket
919 For those preferring a higher-level interface to socket programming, the
920 IO::Socket module provides an object-oriented approach. IO::Socket is
921 included as part of the standard Perl distribution as of the 5.004
922 release. If you're running an earlier version of Perl, just fetch
923 IO::Socket from CPAN, where you'll also find find modules providing easy
924 interfaces to the following systems: DNS, FTP, Ident (RFC 931), NIS and
925 NISPlus, NNTP, Ping, POP3, SMTP, SNMP, SSLeay, Telnet, and Time--just
928 =head2 A Simple Client
930 Here's a client that creates a TCP connection to the "daytime"
931 service at port 13 of the host name "localhost" and prints out everything
932 that the server there cares to provide.
936 $remote = IO::Socket::INET->new(
938 PeerAddr => "localhost",
939 PeerPort => "daytime(13)",
941 or die "cannot connect to daytime port at localhost";
942 while ( <$remote> ) { print }
944 When you run this program, you should get something back that
947 Wed May 14 08:40:46 MDT 1997
949 Here are what those parameters to the C<new> constructor mean:
955 This is which protocol to use. In this case, the socket handle returned
956 will be connected to a TCP socket, because we want a stream-oriented
957 connection, that is, one that acts pretty much like a plain old file.
958 Not all sockets are this of this type. For example, the UDP protocol
959 can be used to make a datagram socket, used for message-passing.
963 This is the name or Internet address of the remote host the server is
964 running on. We could have specified a longer name like C<"www.perl.com">,
965 or an address like C<"204.148.40.9">. For demonstration purposes, we've
966 used the special hostname C<"localhost">, which should always mean the
967 current machine you're running on. The corresponding Internet address
968 for localhost is C<"127.1">, if you'd rather use that.
972 This is the service name or port number we'd like to connect to.
973 We could have gotten away with using just C<"daytime"> on systems with a
974 well-configured system services file,[FOOTNOTE: The system services file
975 is in I</etc/services> under Unix] but just in case, we've specified the
976 port number (13) in parentheses. Using just the number would also have
977 worked, but constant numbers make careful programmers nervous.
981 Notice how the return value from the C<new> constructor is used as
982 a filehandle in the C<while> loop? That's what's called an indirect
983 filehandle, a scalar variable containing a filehandle. You can use
984 it the same way you would a normal filehandle. For example, you
985 can read one line from it this way:
989 all remaining lines from is this way:
993 and send a line of data to it this way:
995 print $handle "some data\n";
997 =head2 A Webget Client
999 Here's a simple client that takes a remote host to fetch a document
1000 from, and then a list of documents to get from that host. This is a
1001 more interesting client than the previous one because it first sends
1002 something to the server before fetching the server's response.
1006 unless (@ARGV > 1) { die "usage: $0 host document ..." }
1007 $host = shift(@ARGV);
1010 foreach $document ( @ARGV ) {
1011 $remote = IO::Socket::INET->new( Proto => "tcp",
1013 PeerPort => "http(80)",
1015 unless ($remote) { die "cannot connect to http daemon on $host" }
1016 $remote->autoflush(1);
1017 print $remote "GET $document HTTP/1.0" . $BLANK;
1018 while ( <$remote> ) { print }
1022 The web server handing the "http" service, which is assumed to be at
1023 its standard port, number 80. If your the web server you're trying to
1024 connect to is at a different port (like 1080 or 8080), you should specify
1025 as the named-parameter pair, C<PeerPort =E<gt> 8080>. The C<autoflush>
1026 method is used on the socket because otherwise the system would buffer
1027 up the output we sent it. (If you're on a Mac, you'll also need to
1028 change every C<"\n"> in your code that sends data over the network to
1029 be a C<"\015\012"> instead.)
1031 Connecting to the server is only the first part of the process: once you
1032 have the connection, you have to use the server's language. Each server
1033 on the network has its own little command language that it expects as
1034 input. The string that we send to the server starting with "GET" is in
1035 HTTP syntax. In this case, we simply request each specified document.
1036 Yes, we really are making a new connection for each document, even though
1037 it's the same host. That's the way you always used to have to speak HTTP.
1038 Recent versions of web browsers may request that the remote server leave
1039 the connection open a little while, but the server doesn't have to honor
1042 Here's an example of running that program, which we'll call I<webget>:
1044 % webget www.perl.com /guanaco.html
1045 HTTP/1.1 404 File Not Found
1046 Date: Thu, 08 May 1997 18:02:32 GMT
1047 Server: Apache/1.2b6
1049 Content-type: text/html
1051 <HEAD><TITLE>404 File Not Found</TITLE></HEAD>
1052 <BODY><H1>File Not Found</H1>
1053 The requested URL /guanaco.html was not found on this server.<P>
1056 Ok, so that's not very interesting, because it didn't find that
1057 particular document. But a long response wouldn't have fit on this page.
1059 For a more fully-featured version of this program, you should look to
1060 the I<lwp-request> program included with the LWP modules from CPAN.
1062 =head2 Interactive Client with IO::Socket
1064 Well, that's all fine if you want to send one command and get one answer,
1065 but what about setting up something fully interactive, somewhat like
1066 the way I<telnet> works? That way you can type a line, get the answer,
1067 type a line, get the answer, etc.
1069 This client is more complicated than the two we've done so far, but if
1070 you're on a system that supports the powerful C<fork> call, the solution
1071 isn't that rough. Once you've made the connection to whatever service
1072 you'd like to chat with, call C<fork> to clone your process. Each of
1073 these two identical process has a very simple job to do: the parent
1074 copies everything from the socket to standard output, while the child
1075 simultaneously copies everything from standard input to the socket.
1076 To accomplish the same thing using just one process would be I<much>
1077 harder, because it's easier to code two processes to do one thing than it
1078 is to code one process to do two things. (This keep-it-simple principle
1079 a cornerstones of the Unix philosophy, and good software engineering as
1080 well, which is probably why it's spread to other systems.)
1087 my ($host, $port, $kidpid, $handle, $line);
1089 unless (@ARGV == 2) { die "usage: $0 host port" }
1090 ($host, $port) = @ARGV;
1092 # create a tcp connection to the specified host and port
1093 $handle = IO::Socket::INET->new(Proto => "tcp",
1096 or die "can't connect to port $port on $host: $!";
1098 $handle->autoflush(1); # so output gets there right away
1099 print STDERR "[Connected to $host:$port]\n";
1101 # split the program into two processes, identical twins
1102 die "can't fork: $!" unless defined($kidpid = fork());
1104 # the if{} block runs only in the parent process
1106 # copy the socket to standard output
1107 while (defined ($line = <$handle>)) {
1110 kill("TERM", $kidpid); # send SIGTERM to child
1112 # the else{} block runs only in the child process
1114 # copy standard input to the socket
1115 while (defined ($line = <STDIN>)) {
1116 print $handle $line;
1120 The C<kill> function in the parent's C<if> block is there to send a
1121 signal to our child process (current running in the C<else> block)
1122 as soon as the remote server has closed its end of the connection.
1124 If the remote server sends data a byte at time, and you need that
1125 data immediately without waiting for a newline (which might not happen),
1126 you may wish to replace the C<while> loop in the parent with the
1130 while (sysread($handle, $byte, 1) == 1) {
1134 Making a system call for each byte you want to read is not very efficient
1135 (to put it mildly) but is the simplest to explain and works reasonably
1138 =head1 TCP Servers with IO::Socket
1140 As always, setting up a server is little bit more involved than running a client.
1141 The model is that the server creates a special kind of socket that
1142 does nothing but listen on a particular port for incoming connections.
1143 It does this by calling the C<IO::Socket::INET-E<gt>new()> method with
1144 slightly different arguments than the client did.
1150 This is which protocol to use. Like our clients, we'll
1151 still specify C<"tcp"> here.
1156 port in the C<LocalPort> argument, which we didn't do for the client.
1157 This is service name or port number for which you want to be the
1158 server. (Under Unix, ports under 1024 are restricted to the
1159 superuser.) In our sample, we'll use port 9000, but you can use
1160 any port that's not currently in use on your system. If you try
1161 to use one already in used, you'll get an "Address already in use"
1162 message. Under Unix, the C<netstat -a> command will show
1163 which services current have servers.
1167 The C<Listen> parameter is set to the maximum number of
1168 pending connections we can accept until we turn away incoming clients.
1169 Think of it as a call-waiting queue for your telephone.
1170 The low-level Socket module has a special symbol for the system maximum, which
1175 The C<Reuse> parameter is needed so that we restart our server
1176 manually without waiting a few minutes to allow system buffers to
1181 Once the generic server socket has been created using the parameters
1182 listed above, the server then waits for a new client to connect
1183 to it. The server blocks in the C<accept> method, which eventually an
1184 bidirectional connection to the remote client. (Make sure to autoflush
1185 this handle to circumvent buffering.)
1187 To add to user-friendliness, our server prompts the user for commands.
1188 Most servers don't do this. Because of the prompt without a newline,
1189 you'll have to use the C<sysread> variant of the interactive client above.
1191 This server accepts one of five different commands, sending output
1192 back to the client. Note that unlike most network servers, this one
1193 only handles one incoming client at a time. Multithreaded servers are
1194 covered in Chapter 6 of the Camel.
1196 Here's the code. We'll
1200 use Net::hostent; # for OO version of gethostbyaddr
1202 $PORT = 9000; # pick something not in use
1204 $server = IO::Socket::INET->new( Proto => 'tcp',
1206 Listen => SOMAXCONN,
1209 die "can't setup server" unless $server;
1210 print "[Server $0 accepting clients]\n";
1212 while ($client = $server->accept()) {
1213 $client->autoflush(1);
1214 print $client "Welcome to $0; type help for command list.\n";
1215 $hostinfo = gethostbyaddr($client->peeraddr);
1216 printf "[Connect from %s]\n", $hostinfo->name || $client->peerhost;
1217 print $client "Command? ";
1218 while ( <$client>) {
1219 next unless /\S/; # blank line
1220 if (/quit|exit/i) { last; }
1221 elsif (/date|time/i) { printf $client "%s\n", scalar localtime; }
1222 elsif (/who/i ) { print $client `who 2>&1`; }
1223 elsif (/cookie/i ) { print $client `/usr/games/fortune 2>&1`; }
1224 elsif (/motd/i ) { print $client `cat /etc/motd 2>&1`; }
1226 print $client "Commands: quit date who cookie motd\n";
1229 print $client "Command? ";
1234 =head1 UDP: Message Passing
1236 Another kind of client-server setup is one that uses not connections, but
1237 messages. UDP communications involve much lower overhead but also provide
1238 less reliability, as there are no promises that messages will arrive at
1239 all, let alone in order and unmangled. Still, UDP offers some advantages
1240 over TCP, including being able to "broadcast" or "multicast" to a whole
1241 bunch of destination hosts at once (usually on your local subnet). If you
1242 find yourself overly concerned about reliability and start building checks
1243 into your message system, then you probably should use just TCP to start
1246 Here's a UDP program similar to the sample Internet TCP client given
1247 earlier. However, instead of checking one host at a time, the UDP version
1248 will check many of them asynchronously by simulating a multicast and then
1249 using select() to do a timed-out wait for I/O. To do something similar
1250 with TCP, you'd have to use a different socket handle for each host.
1257 my ( $count, $hisiaddr, $hispaddr, $histime,
1258 $host, $iaddr, $paddr, $port, $proto,
1259 $rin, $rout, $rtime, $SECS_of_70_YEARS);
1261 $SECS_of_70_YEARS = 2208988800;
1263 $iaddr = gethostbyname(hostname());
1264 $proto = getprotobyname('udp');
1265 $port = getservbyname('time', 'udp');
1266 $paddr = sockaddr_in(0, $iaddr); # 0 means let kernel pick
1268 socket(SOCKET, PF_INET, SOCK_DGRAM, $proto) || die "socket: $!";
1269 bind(SOCKET, $paddr) || die "bind: $!";
1272 printf "%-12s %8s %s\n", "localhost", 0, scalar localtime time;
1276 $hisiaddr = inet_aton($host) || die "unknown host";
1277 $hispaddr = sockaddr_in($port, $hisiaddr);
1278 defined(send(SOCKET, 0, 0, $hispaddr)) || die "send $host: $!";
1282 vec($rin, fileno(SOCKET), 1) = 1;
1284 # timeout after 10.0 seconds
1285 while ($count && select($rout = $rin, undef, undef, 10.0)) {
1287 ($hispaddr = recv(SOCKET, $rtime, 4, 0)) || die "recv: $!";
1288 ($port, $hisiaddr) = sockaddr_in($hispaddr);
1289 $host = gethostbyaddr($hisiaddr, AF_INET);
1290 $histime = unpack("N", $rtime) - $SECS_of_70_YEARS ;
1291 printf "%-12s ", $host;
1292 printf "%8d %s\n", $histime - time, scalar localtime($histime);
1298 While System V IPC isn't so widely used as sockets, it still has some
1299 interesting uses. You can't, however, effectively use SysV IPC or
1300 Berkeley mmap() to have shared memory so as to share a variable amongst
1301 several processes. That's because Perl would reallocate your string when
1302 you weren't wanting it to.
1304 Here's a small example showing shared memory usage.
1306 use IPC::SysV qw(IPC_PRIVATE IPC_RMID S_IRWXU S_IRWXG S_IRWXO);
1309 $key = shmget(IPC_PRIVATE, $size, S_IRWXU|S_IRWXG|S_IRWXO) || die "$!";
1310 print "shm key $key\n";
1312 $message = "Message #1";
1313 shmwrite($key, $message, 0, 60) || die "$!";
1314 print "wrote: '$message'\n";
1315 shmread($key, $buff, 0, 60) || die "$!";
1316 print "read : '$buff'\n";
1318 # the buffer of shmread is zero-character end-padded.
1319 substr($buff, index($buff, "\0")) = '';
1320 print "un" unless $buff eq $message;
1323 print "deleting shm $key\n";
1324 shmctl($key, IPC_RMID, 0) || die "$!";
1326 Here's an example of a semaphore:
1328 use IPC::SysV qw(IPC_CREAT);
1331 $key = semget($IPC_KEY, 10, 0666 | IPC_CREAT ) || die "$!";
1332 print "shm key $key\n";
1334 Put this code in a separate file to be run in more than one process.
1335 Call the file F<take>:
1337 # create a semaphore
1340 $key = semget($IPC_KEY, 0 , 0 );
1341 die if !defined($key);
1347 # wait for semaphore to be zero
1349 $opstring1 = pack("sss", $semnum, $semop, $semflag);
1351 # Increment the semaphore count
1353 $opstring2 = pack("sss", $semnum, $semop, $semflag);
1354 $opstring = $opstring1 . $opstring2;
1356 semop($key,$opstring) || die "$!";
1358 Put this code in a separate file to be run in more than one process.
1359 Call this file F<give>:
1361 # 'give' the semaphore
1362 # run this in the original process and you will see
1363 # that the second process continues
1366 $key = semget($IPC_KEY, 0, 0);
1367 die if !defined($key);
1372 # Decrement the semaphore count
1374 $opstring = pack("sss", $semnum, $semop, $semflag);
1376 semop($key,$opstring) || die "$!";
1378 The SysV IPC code above was written long ago, and it's definitely
1379 clunky looking. For a more modern look, see the IPC::SysV module
1380 which is included with Perl starting from Perl 5.005.
1384 Most of these routines quietly but politely return C<undef> when they
1385 fail instead of causing your program to die right then and there due to
1386 an uncaught exception. (Actually, some of the new I<Socket> conversion
1387 functions croak() on bad arguments.) It is therefore essential to
1388 check return values from these functions. Always begin your socket
1389 programs this way for optimal success, and don't forget to add B<-T>
1390 taint checking flag to the #! line for servers:
1399 All these routines create system-specific portability problems. As noted
1400 elsewhere, Perl is at the mercy of your C libraries for much of its system
1401 behaviour. It's probably safest to assume broken SysV semantics for
1402 signals and to stick with simple TCP and UDP socket operations; e.g., don't
1403 try to pass open file descriptors over a local UDP datagram socket if you
1404 want your code to stand a chance of being portable.
1406 As mentioned in the signals section, because few vendors provide C
1407 libraries that are safely re-entrant, the prudent programmer will do
1408 little else within a handler beyond setting a numeric variable that
1409 already exists; or, if locked into a slow (restarting) system call,
1410 using die() to raise an exception and longjmp(3) out. In fact, even
1411 these may in some cases cause a core dump. It's probably best to avoid
1412 signals except where they are absolutely inevitable. This
1413 will be addressed in a future release of Perl.
1417 Tom Christiansen, with occasional vestiges of Larry Wall's original
1418 version and suggestions from the Perl Porters.
1422 There's a lot more to networking than this, but this should get you
1425 For intrepid programmers, the indispensable textbook is I<Unix Network
1426 Programming> by W. Richard Stevens (published by Addison-Wesley). Note
1427 that most books on networking address networking from the perspective of
1428 a C programmer; translation to Perl is left as an exercise for the reader.
1430 The IO::Socket(3) manpage describes the object library, and the Socket(3)
1431 manpage describes the low-level interface to sockets. Besides the obvious
1432 functions in L<perlfunc>, you should also check out the F<modules> file
1433 at your nearest CPAN site. (See L<perlmodlib> or best yet, the F<Perl
1434 FAQ> for a description of what CPAN is and where to get it.)
1436 Section 5 of the F<modules> file is devoted to "Networking, Device Control
1437 (modems), and Interprocess Communication", and contains numerous unbundled
1438 modules numerous networking modules, Chat and Expect operations, CGI
1439 programming, DCE, FTP, IPC, NNTP, Proxy, Ptty, RPC, SNMP, SMTP, Telnet,
1440 Threads, and ToolTalk--just to name a few.