3 perlipc - Perl interprocess communication (signals, fifos, pipes, safe subprocesses, sockets, and semaphores)
7 The basic IPC facilities of Perl are built out of the good old Unix
8 signals, named pipes, pipe opens, the Berkeley socket routines, and SysV
9 IPC calls. Each is used in slightly different situations.
13 Perl uses a simple signal handling model: the %SIG hash contains names
14 or references of user-installed signal handlers. These handlers will
15 be called with an argument which is the name of the signal that
16 triggered it. A signal may be generated intentionally from a
17 particular keyboard sequence like control-C or control-Z, sent to you
18 from another process, or triggered automatically by the kernel when
19 special events transpire, like a child process exiting, your own process
20 running out of stack space, or hitting a process file-size limit.
22 For example, to trap an interrupt signal, set up a handler like this:
29 die "Somebody sent me a SIG$signame";
31 $SIG{INT} = __PACKAGE__ . "::catch_zap";
32 $SIG{INT} = \&catch_zap; # best strategy
34 Prior to Perl 5.8.0 it was necessary to do as little as you possibly
35 could in your handler; notice how all we do is set a global variable
36 and then raise an exception. That's because on most systems,
37 libraries are not re-entrant; particularly, memory allocation and I/O
38 routines are not. That meant that doing nearly I<anything> in your
39 handler could in theory trigger a memory fault and subsequent core
40 dump - see L</Deferred Signals (Safe Signals)> below.
42 The names of the signals are the ones listed out by C<kill -l> on your
43 system, or you can retrieve them using the CPAN module L<IPC::Signal>.
45 You may also choose to assign the strings C<"IGNORE"> or C<"DEFAULT"> as
46 the handler, in which case Perl will try to discard the signal or do the
49 On most Unix platforms, the C<CHLD> (sometimes also known as C<CLD>) signal
50 has special behavior with respect to a value of C<"IGNORE">.
51 Setting C<$SIG{CHLD}> to C<"IGNORE"> on such a platform has the effect of
52 not creating zombie processes when the parent process fails to C<wait()>
53 on its child processes (i.e., child processes are automatically reaped).
54 Calling C<wait()> with C<$SIG{CHLD}> set to C<"IGNORE"> usually returns
55 C<-1> on such platforms.
57 Some signals can be neither trapped nor ignored, such as the KILL and STOP
58 (but not the TSTP) signals. Note that ignoring signals makes them disappear.
59 If you only want them blocked temporarily without them getting lost you'll
60 have to use POSIX' sigprocmask.
62 Sending a signal to a negative process ID means that you send the signal
63 to the entire Unix process group. This code sends a hang-up signal to all
64 processes in the current process group, and also sets $SIG{HUP} to C<"IGNORE">
65 so it doesn't kill itself:
67 # block scope for local
69 local $SIG{HUP} = "IGNORE";
71 # snazzy writing of: kill("HUP", -$$)
74 Another interesting signal to send is signal number zero. This doesn't
75 actually affect a child process, but instead checks whether it's alive
76 or has changed its UIDs.
78 unless (kill 0 => $kid_pid) {
79 warn "something wicked happened to $kid_pid";
82 Signal number zero may fail because you lack permission to send the
83 signal when directed at a process whose real or saved UID is not
84 identical to the real or effective UID of the sending process, even
85 though the process is alive. You may be able to determine the cause of
86 failure using C<$!> or C<%!>.
88 unless (kill(0 => $pid) || $!{EPERM}) {
89 warn "$pid looks dead";
92 You might also want to employ anonymous functions for simple signal
95 $SIG{INT} = sub { die "\nOutta here!\n" };
97 SIGCHLD handlers require some special care. If a second child dies
98 while in the signal handler caused by the first death, we won't get
99 another signal. So must loop here else we will leave the unreaped child
100 as a zombie. And the next time two children die we get another zombie.
103 use POSIX ":sys_wait_h";
105 while ((my $child = waitpid(-1, WNOHANG)) > 0) {
106 $Kid_Status{$child} = $?;
109 # do something that forks...
111 Be careful: qx(), system(), and some modules for calling external commands
112 do a fork(), then wait() for the result. Thus, your signal handler
113 will be called. Because wait() was already called by system() or qx(),
114 the wait() in the signal handler will see no more zombies and will
117 The best way to prevent this issue is to use waitpid(), as in the following
120 use POSIX ":sys_wait_h"; # for nonblocking read
125 # don't change $! and $? outside handler
127 while ( (my $pid = waitpid(-1, WNOHANG)) > 0 ) {
128 delete $children{$pid};
129 cleanup_child($pid, $?);
135 die "cannot fork" unless defined $pid;
147 Signal handling is also used for timeouts in Unix. While safely
148 protected within an C<eval{}> block, you set a signal handler to trap
149 alarm signals and then schedule to have one delivered to you in some
150 number of seconds. Then try your blocking operation, clearing the alarm
151 when it's done but not before you've exited your C<eval{}> block. If it
152 goes off, you'll use die() to jump out of the block.
156 my $ALARM_EXCEPTION = "alarm clock restart";
158 local $SIG{ALRM} = sub { die $ALARM_EXCEPTION };
160 flock(FH, 2) # blocking write lock
161 || die "cannot flock: $!";
164 if ($@ && $@ !~ quotemeta($ALARM_EXCEPTION)) { die }
166 If the operation being timed out is system() or qx(), this technique
167 is liable to generate zombies. If this matters to you, you'll
168 need to do your own fork() and exec(), and kill the errant child process.
170 For more complex signal handling, you might see the standard POSIX
171 module. Lamentably, this is almost entirely undocumented, but
172 the F<t/lib/posix.t> file from the Perl source distribution has some
175 =head2 Handling the SIGHUP Signal in Daemons
177 A process that usually starts when the system boots and shuts down
178 when the system is shut down is called a daemon (Disk And Execution
179 MONitor). If a daemon process has a configuration file which is
180 modified after the process has been started, there should be a way to
181 tell that process to reread its configuration file without stopping
182 the process. Many daemons provide this mechanism using a C<SIGHUP>
183 signal handler. When you want to tell the daemon to reread the file,
184 simply send it the C<SIGHUP> signal.
186 The following example implements a simple daemon, which restarts
187 itself every time the C<SIGHUP> signal is received. The actual code is
188 located in the subroutine C<code()>, which just prints some debugging
189 info to show that it works; it should be replaced with the real code.
195 use File::Basename ();
196 use File::Spec::Functions;
200 # make the daemon cross-platform, so exec always calls the script
201 # itself with the right path, no matter how the script was invoked.
202 my $script = File::Basename::basename($0);
203 my $SELF = catfile($FindBin::Bin, $script);
205 # POSIX unmasks the sigprocmask properly
207 print "got SIGHUP\n";
208 exec($SELF, @ARGV) || die "$0: couldn't restart: $!";
215 print "ARGV: @ARGV\n";
224 =head2 Deferred Signals (Safe Signals)
226 Before Perl 5.8.0, installing Perl code to deal with signals exposed you to
227 danger from two things. First, few system library functions are
228 re-entrant. If the signal interrupts while Perl is executing one function
229 (like malloc(3) or printf(3)), and your signal handler then calls the same
230 function again, you could get unpredictable behavior--often, a core dump.
231 Second, Perl isn't itself re-entrant at the lowest levels. If the signal
232 interrupts Perl while Perl is changing its own internal data structures,
233 similarly unpredictable behavior may result.
235 There were two things you could do, knowing this: be paranoid or be
236 pragmatic. The paranoid approach was to do as little as possible in your
237 signal handler. Set an existing integer variable that already has a
238 value, and return. This doesn't help you if you're in a slow system call,
239 which will just restart. That means you have to C<die> to longjmp(3) out
240 of the handler. Even this is a little cavalier for the true paranoiac,
241 who avoids C<die> in a handler because the system I<is> out to get you.
242 The pragmatic approach was to say "I know the risks, but prefer the
243 convenience", and to do anything you wanted in your signal handler,
244 and be prepared to clean up core dumps now and again.
246 Perl 5.8.0 and later avoid these problems by "deferring" signals. That is,
247 when the signal is delivered to the process by the system (to the C code
248 that implements Perl) a flag is set, and the handler returns immediately.
249 Then at strategic "safe" points in the Perl interpreter (e.g. when it is
250 about to execute a new opcode) the flags are checked and the Perl level
251 handler from %SIG is executed. The "deferred" scheme allows much more
252 flexibility in the coding of signal handlers as we know the Perl
253 interpreter is in a safe state, and that we are not in a system library
254 function when the handler is called. However the implementation does
255 differ from previous Perls in the following ways:
259 =item Long-running opcodes
261 As the Perl interpreter looks at signal flags only when it is about
262 to execute a new opcode, a signal that arrives during a long-running
263 opcode (e.g. a regular expression operation on a very large string) will
264 not be seen until the current opcode completes.
266 If a signal of any given type fires multiple times during an opcode
267 (such as from a fine-grained timer), the handler for that signal will
268 be called only once, after the opcode completes; all other
269 instances will be discarded. Furthermore, if your system's signal queue
270 gets flooded to the point that there are signals that have been raised
271 but not yet caught (and thus not deferred) at the time an opcode
272 completes, those signals may well be caught and deferred during
273 subsequent opcodes, with sometimes surprising results. For example, you
274 may see alarms delivered even after calling C<alarm(0)> as the latter
275 stops the raising of alarms but does not cancel the delivery of alarms
276 raised but not yet caught. Do not depend on the behaviors described in
277 this paragraph as they are side effects of the current implementation and
278 may change in future versions of Perl.
280 =item Interrupting IO
282 When a signal is delivered (e.g., SIGINT from a control-C) the operating
283 system breaks into IO operations like I<read>(2), which is used to
284 implement Perl's readline() function, the C<< <> >> operator. On older
285 Perls the handler was called immediately (and as C<read> is not "unsafe",
286 this worked well). With the "deferred" scheme the handler is I<not> called
287 immediately, and if Perl is using the system's C<stdio> library that
288 library may restart the C<read> without returning to Perl to give it a
289 chance to call the %SIG handler. If this happens on your system the
290 solution is to use the C<:perlio> layer to do IO--at least on those handles
291 that you want to be able to break into with signals. (The C<:perlio> layer
292 checks the signal flags and calls %SIG handlers before resuming IO
295 The default in Perl 5.8.0 and later is to automatically use
296 the C<:perlio> layer.
298 Note that it is not advisable to access a file handle within a signal
299 handler where that signal has interrupted an I/O operation on that same
300 handle. While perl will at least try hard not to crash, there are no
301 guarantees of data integrity; for example, some data might get dropped or
304 Some networking library functions like gethostbyname() are known to have
305 their own implementations of timeouts which may conflict with your
306 timeouts. If you have problems with such functions, try using the POSIX
307 sigaction() function, which bypasses Perl safe signals. Be warned that
308 this does subject you to possible memory corruption, as described above.
310 Instead of setting C<$SIG{ALRM}>:
312 local $SIG{ALRM} = sub { die "alarm" };
314 try something like the following:
316 use POSIX qw(SIGALRM);
317 POSIX::sigaction(SIGALRM, POSIX::SigAction->new(sub { die "alarm" }))
318 || die "Error setting SIGALRM handler: $!\n";
320 Another way to disable the safe signal behavior locally is to use
321 the C<Perl::Unsafe::Signals> module from CPAN, which affects
324 =item Restartable system calls
326 On systems that supported it, older versions of Perl used the
327 SA_RESTART flag when installing %SIG handlers. This meant that
328 restartable system calls would continue rather than returning when
329 a signal arrived. In order to deliver deferred signals promptly,
330 Perl 5.8.0 and later do I<not> use SA_RESTART. Consequently,
331 restartable system calls can fail (with $! set to C<EINTR>) in places
332 where they previously would have succeeded.
334 The default C<:perlio> layer retries C<read>, C<write>
335 and C<close> as described above; interrupted C<wait> and
336 C<waitpid> calls will always be retried.
338 =item Signals as "faults"
340 Certain signals like SEGV, ILL, and BUS are generated by virtual memory
341 addressing errors and similar "faults". These are normally fatal: there is
342 little a Perl-level handler can do with them. So Perl delivers them
343 immediately rather than attempting to defer them.
345 =item Signals triggered by operating system state
347 On some operating systems certain signal handlers are supposed to "do
348 something" before returning. One example can be CHLD or CLD, which
349 indicates a child process has completed. On some operating systems the
350 signal handler is expected to C<wait> for the completed child
351 process. On such systems the deferred signal scheme will not work for
352 those signals: it does not do the C<wait>. Again the failure will
353 look like a loop as the operating system will reissue the signal because
354 there are completed child processes that have not yet been C<wait>ed for.
358 If you want the old signal behavior back despite possible
359 memory corruption, set the environment variable C<PERL_SIGNALS> to
360 C<"unsafe">. This feature first appeared in Perl 5.8.1.
364 A named pipe (often referred to as a FIFO) is an old Unix IPC
365 mechanism for processes communicating on the same machine. It works
366 just like regular anonymous pipes, except that the
367 processes rendezvous using a filename and need not be related.
369 To create a named pipe, use the C<POSIX::mkfifo()> function.
371 use POSIX qw(mkfifo);
372 mkfifo($path, 0700) || die "mkfifo $path failed: $!";
374 You can also use the Unix command mknod(1), or on some
375 systems, mkfifo(1). These may not be in your normal path, though.
377 # system return val is backwards, so && not ||
379 $ENV{PATH} .= ":/etc:/usr/etc";
380 if ( system("mknod", $path, "p")
381 && system("mkfifo", $path) )
383 die "mk{nod,fifo} $path failed";
387 A fifo is convenient when you want to connect a process to an unrelated
388 one. When you open a fifo, the program will block until there's something
391 For example, let's say you'd like to have your F<.signature> file be a
392 named pipe that has a Perl program on the other end. Now every time any
393 program (like a mailer, news reader, finger program, etc.) tries to read
394 from that file, the reading program will read the new signature from your
395 program. We'll use the pipe-checking file-test operator, B<-p>, to find
396 out whether anyone (or anything) has accidentally removed our fifo.
399 my $FIFO = ".signature";
403 unlink $FIFO; # discard any failure, will catch later
404 require POSIX; # delayed loading of heavy module
405 POSIX::mkfifo($FIFO, 0700)
406 || die "can't mkfifo $FIFO: $!";
409 # next line blocks till there's a reader
410 open (FIFO, "> $FIFO") || die "can't open $FIFO: $!";
411 print FIFO "John Smith (smith\@host.org)\n", `fortune -s`;
412 close(FIFO) || die "can't close $FIFO: $!";
413 sleep 2; # to avoid dup signals
416 =head1 Using open() for IPC
418 Perl's basic open() statement can also be used for unidirectional
419 interprocess communication by either appending or prepending a pipe
420 symbol to the second argument to open(). Here's how to start
421 something up in a child process you intend to write to:
423 open(SPOOLER, "| cat -v | lpr -h 2>/dev/null")
424 || die "can't fork: $!";
425 local $SIG{PIPE} = sub { die "spooler pipe broke" };
426 print SPOOLER "stuff\n";
427 close SPOOLER || die "bad spool: $! $?";
429 And here's how to start up a child process you intend to read from:
431 open(STATUS, "netstat -an 2>&1 |")
432 || die "can't fork: $!";
434 next if /^(tcp|udp)/;
437 close STATUS || die "bad netstat: $! $?";
439 If one can be sure that a particular program is a Perl script expecting
440 filenames in @ARGV, the clever programmer can write something like this:
442 % program f1 "cmd1|" - f2 "cmd2|" f3 < tmpfile
444 and no matter which sort of shell it's called from, the Perl program will
445 read from the file F<f1>, the process F<cmd1>, standard input (F<tmpfile>
446 in this case), the F<f2> file, the F<cmd2> command, and finally the F<f3>
447 file. Pretty nifty, eh?
449 You might notice that you could use backticks for much the
450 same effect as opening a pipe for reading:
452 print grep { !/^(tcp|udp)/ } `netstat -an 2>&1`;
453 die "bad netstatus ($?)" if $?;
455 While this is true on the surface, it's much more efficient to process the
456 file one line or record at a time because then you don't have to read the
457 whole thing into memory at once. It also gives you finer control of the
458 whole process, letting you kill off the child process early if you'd like.
460 Be careful to check the return values from both open() and close(). If
461 you're I<writing> to a pipe, you should also trap SIGPIPE. Otherwise,
462 think of what happens when you start up a pipe to a command that doesn't
463 exist: the open() will in all likelihood succeed (it only reflects the
464 fork()'s success), but then your output will fail--spectacularly. Perl
465 can't know whether the command worked, because your command is actually
466 running in a separate process whose exec() might have failed. Therefore,
467 while readers of bogus commands return just a quick EOF, writers
468 to bogus commands will get hit with a signal, which they'd best be prepared
471 open(FH, "|bogus") || die "can't fork: $!";
472 print FH "bang\n"; # neither necessary nor sufficient
473 # to check print retval!
474 close(FH) || die "can't close: $!";
476 The reason for not checking the return value from print() is because of
477 pipe buffering; physical writes are delayed. That won't blow up until the
478 close, and it will blow up with a SIGPIPE. To catch it, you could use
481 $SIG{PIPE} = "IGNORE";
482 open(FH, "|bogus") || die "can't fork: $!";
484 close(FH) || die "can't close: status=$?";
488 Both the main process and any child processes it forks share the same
489 STDIN, STDOUT, and STDERR filehandles. If both processes try to access
490 them at once, strange things can happen. You may also want to close
491 or reopen the filehandles for the child. You can get around this by
492 opening your pipe with open(), but on some systems this means that the
493 child process cannot outlive the parent.
495 =head2 Background Processes
497 You can run a command in the background with:
501 The command's STDOUT and STDERR (and possibly STDIN, depending on your
502 shell) will be the same as the parent's. You won't need to catch
503 SIGCHLD because of the double-fork taking place; see below for details.
505 =head2 Complete Dissociation of Child from Parent
507 In some cases (starting server processes, for instance) you'll want to
508 completely dissociate the child process from the parent. This is
509 often called daemonization. A well-behaved daemon will also chdir()
510 to the root directory so it doesn't prevent unmounting the filesystem
511 containing the directory from which it was launched, and redirect its
512 standard file descriptors from and to F</dev/null> so that random
513 output doesn't wind up on the user's terminal.
518 chdir("/") || die "can't chdir to /: $!";
519 open(STDIN, "< /dev/null") || die "can't read /dev/null: $!";
520 open(STDOUT, "> /dev/null") || die "can't write to /dev/null: $!";
521 defined(my $pid = fork()) || die "can't fork: $!";
522 exit if $pid; # non-zero now means I am the parent
523 (setsid() != -1) || die "Can't start a new session: $!";
524 open(STDERR, ">&STDOUT") || die "can't dup stdout: $!";
527 The fork() has to come before the setsid() to ensure you aren't a
528 process group leader; the setsid() will fail if you are. If your
529 system doesn't have the setsid() function, open F</dev/tty> and use the
530 C<TIOCNOTTY> ioctl() on it instead. See tty(4) for details.
532 Non-Unix users should check their C<< I<Your_OS>::Process >> module for
533 other possible solutions.
535 =head2 Safe Pipe Opens
537 Another interesting approach to IPC is making your single program go
538 multiprocess and communicate between--or even amongst--yourselves. The
539 open() function will accept a file argument of either C<"-|"> or C<"|-">
540 to do a very interesting thing: it forks a child connected to the
541 filehandle you've opened. The child is running the same program as the
542 parent. This is useful for safely opening a file when running under an
543 assumed UID or GID, for example. If you open a pipe I<to> minus, you can
544 write to the filehandle you opened and your kid will find it in I<his>
545 STDIN. If you open a pipe I<from> minus, you can read from the filehandle
546 you opened whatever your kid writes to I<his> STDOUT.
549 my $PRECIOUS = "/path/to/some/safe/file";
554 $pid = open(KID_TO_WRITE, "|-");
555 unless (defined $pid) {
556 warn "cannot fork: $!";
557 die "bailing out" if $sleep_count++ > 6;
560 } until defined $pid;
562 if ($pid) { # I am the parent
563 print KID_TO_WRITE @some_data;
564 close(KID_TO_WRITE) || warn "kid exited $?";
565 } else { # I am the child
566 # drop permissions in setuid and/or setgid programs:
567 ($EUID, $EGID) = ($UID, $GID);
568 open (OUTFILE, "> $PRECIOUS")
569 || die "can't open $PRECIOUS: $!";
571 print OUTFILE; # child's STDIN is parent's KID_TO_WRITE
573 close(OUTFILE) || die "can't close $PRECIOUS: $!";
574 exit(0); # don't forget this!!
577 Another common use for this construct is when you need to execute
578 something without the shell's interference. With system(), it's
579 straightforward, but you can't use a pipe open or backticks safely.
580 That's because there's no way to stop the shell from getting its hands on
581 your arguments. Instead, use lower-level control to call exec() directly.
583 Here's a safe backtick or pipe open for read:
585 my $pid = open(KID_TO_READ, "-|");
586 defined($pid) || die "can't fork: $!";
589 while (<KID_TO_READ>) {
590 # do something interesting
592 close(KID_TO_READ) || warn "kid exited $?";
595 ($EUID, $EGID) = ($UID, $GID); # suid only
596 exec($program, @options, @args)
597 || die "can't exec program: $!";
601 And here's a safe pipe open for writing:
603 my $pid = open(KID_TO_WRITE, "|-");
604 defined($pid) || die "can't fork: $!";
606 $SIG{PIPE} = sub { die "whoops, $program pipe broke" };
609 print KID_TO_WRITE @data;
610 close(KID_TO_WRITE) || warn "kid exited $?";
613 ($EUID, $EGID) = ($UID, $GID);
614 exec($program, @options, @args)
615 || die "can't exec program: $!";
619 It is very easy to dead-lock a process using this form of open(), or
620 indeed with any use of pipe() with multiple subprocesses. The
621 example above is "safe" because it is simple and calls exec(). See
622 L</"Avoiding Pipe Deadlocks"> for general safety principles, but there
623 are extra gotchas with Safe Pipe Opens.
625 In particular, if you opened the pipe using C<open FH, "|-">, then you
626 cannot simply use close() in the parent process to close an unwanted
627 writer. Consider this code:
629 my $pid = open(WRITER, "|-"); # fork open a kid
630 defined($pid) || die "first fork failed: $!";
632 if (my $sub_pid = fork()) {
633 defined($sub_pid) || die "second fork failed: $!";
634 close(WRITER) || die "couldn't close WRITER: $!";
635 # now do something else...
638 # first write to WRITER
641 close(WRITER) || die "couldn't close WRITER: $!";
646 # first do something with STDIN, then
650 In the example above, the true parent does not want to write to the WRITER
651 filehandle, so it closes it. However, because WRITER was opened using
652 C<open FH, "|-">, it has a special behavior: closing it calls
653 waitpid() (see L<perlfunc/waitpid>), which waits for the subprocess
654 to exit. If the child process ends up waiting for something happening
655 in the section marked "do something else", you have deadlock.
657 This can also be a problem with intermediate subprocesses in more
658 complicated code, which will call waitpid() on all open filehandles
659 during global destruction--in no predictable order.
661 To solve this, you must manually use pipe(), fork(), and the form of
662 open() which sets one file descriptor to another, as shown below:
664 pipe(READER, WRITER) || die "pipe failed: $!";
666 defined($pid) || die "first fork failed: $!";
669 if (my $sub_pid = fork()) {
670 defined($sub_pid) || die "first fork failed: $!";
671 close(WRITER) || die "can't close WRITER: $!";
677 close(WRITER) || die "can't close WRITER: $!";
683 open(STDIN, "<&READER") || die "can't reopen STDIN: $!";
684 close(WRITER) || die "can't close WRITER: $!";
689 Since Perl 5.8.0, you can also use the list form of C<open> for pipes.
690 This is preferred when you wish to avoid having the shell interpret
691 metacharacters that may be in your command string.
693 So for example, instead of using:
695 open(PS_PIPE, "ps aux|") || die "can't open ps pipe: $!";
697 One would use either of these:
699 open(PS_PIPE, "-|", "ps", "aux")
700 || die "can't open ps pipe: $!";
702 @ps_args = qw[ ps aux ];
703 open(PS_PIPE, "-|", @ps_args)
704 || die "can't open @ps_args|: $!";
706 Because there are more than three arguments to open(), forks the ps(1)
707 command I<without> spawning a shell, and reads its standard output via the
708 C<PS_PIPE> filehandle. The corresponding syntax to I<write> to command
709 pipes is to use C<"|-"> in place of C<"-|">.
711 This was admittedly a rather silly example, because you're using string
712 literals whose content is perfectly safe. There is therefore no cause to
713 resort to the harder-to-read, multi-argument form of pipe open(). However,
714 whenever you cannot be assured that the program arguments are free of shell
715 metacharacters, the fancier form of open() should be used. For example:
717 @grep_args = ("egrep", "-i", $some_pattern, @many_files);
718 open(GREP_PIPE, "-|", @grep_args)
719 || die "can't open @grep_args|: $!";
721 Here the multi-argument form of pipe open() is preferred because the
722 pattern and indeed even the filenames themselves might hold metacharacters.
724 Be aware that these operations are full Unix forks, which means they may
725 not be correctly implemented on all alien systems.
727 =head2 Avoiding Pipe Deadlocks
729 Whenever you have more than one subprocess, you must be careful that each
730 closes whichever half of any pipes created for interprocess communication
731 it is not using. This is because any child process reading from the pipe
732 and expecting an EOF will never receive it, and therefore never exit. A
733 single process closing a pipe is not enough to close it; the last process
734 with the pipe open must close it for it to read EOF.
736 Certain built-in Unix features help prevent this most of the time. For
737 instance, filehandles have a "close on exec" flag, which is set I<en masse>
738 under control of the C<$^F> variable. This is so any filehandles you
739 didn't explicitly route to the STDIN, STDOUT or STDERR of a child
740 I<program> will be automatically closed.
742 Always explicitly and immediately call close() on the writable end of any
743 pipe, unless that process is actually writing to it. Even if you don't
744 explicitly call close(), Perl will still close() all filehandles during
745 global destruction. As previously discussed, if those filehandles have
746 been opened with Safe Pipe Open, this will result in calling waitpid(),
747 which may again deadlock.
749 =head2 Bidirectional Communication with Another Process
751 While this works reasonably well for unidirectional communication, what
752 about bidirectional communication? The most obvious approach doesn't work:
754 # THIS DOES NOT WORK!!
755 open(PROG_FOR_READING_AND_WRITING, "| some program |")
757 If you forget to C<use warnings>, you'll miss out entirely on the
758 helpful diagnostic message:
760 Can't do bidirectional pipe at -e line 1.
762 If you really want to, you can use the standard open2() from the
763 C<IPC::Open2> module to catch both ends. There's also an open3() in
764 C<IPC::Open3> for tridirectional I/O so you can also catch your child's
765 STDERR, but doing so would then require an awkward select() loop and
766 wouldn't allow you to use normal Perl input operations.
768 If you look at its source, you'll see that open2() uses low-level
769 primitives like the pipe() and exec() syscalls to create all the
770 connections. Although it might have been more efficient by using
771 socketpair(), this would have been even less portable than it already
772 is. The open2() and open3() functions are unlikely to work anywhere
773 except on a Unix system, or at least one purporting POSIX compliance.
776 Hold on, is this even true? First it says that socketpair() is avoided
777 for portability, but then it says it probably won't work except on
778 Unixy systems anyway. Which one of those is true?
780 Here's an example of using open2():
784 $pid = open2(*Reader, *Writer, "cat -un");
785 print Writer "stuff\n";
788 The problem with this is that buffering is really going to ruin your
789 day. Even though your C<Writer> filehandle is auto-flushed so the process
790 on the other end gets your data in a timely manner, you can't usually do
791 anything to force that process to give its data to you in a similarly quick
792 fashion. In this special case, we could actually so, because we gave
793 I<cat> a B<-u> flag to make it unbuffered. But very few commands are
794 designed to operate over pipes, so this seldom works unless you yourself
795 wrote the program on the other end of the double-ended pipe.
797 A solution to this is to use a library which uses pseudottys to make your
798 program behave more reasonably. This way you don't have to have control
799 over the source code of the program you're using. The C<Expect> module
800 from CPAN also addresses this kind of thing. This module requires two
801 other modules from CPAN, C<IO::Pty> and C<IO::Stty>. It sets up a pseudo
802 terminal to interact with programs that insist on talking to the terminal
803 device driver. If your system is supported, this may be your best bet.
805 =head2 Bidirectional Communication with Yourself
807 If you want, you may make low-level pipe() and fork() syscalls to stitch
808 this together by hand. This example only talks to itself, but you could
809 reopen the appropriate handles to STDIN and STDOUT and call other processes.
810 (The following example lacks proper error checking.)
813 # pipe1 - bidirectional communication using two pipe pairs
814 # designed for the socketpair-challenged
815 use IO::Handle; # thousands of lines just for autoflush :-(
816 pipe(PARENT_RDR, CHILD_WTR); # XXX: check failure?
817 pipe(CHILD_RDR, PARENT_WTR); # XXX: check failure?
818 CHILD_WTR->autoflush(1);
819 PARENT_WTR->autoflush(1);
824 print CHILD_WTR "Parent Pid $$ is sending this\n";
825 chomp($line = <CHILD_RDR>);
826 print "Parent Pid $$ just read this: '$line'\n";
827 close CHILD_RDR; close CHILD_WTR;
830 die "cannot fork: $!" unless defined $pid;
833 chomp($line = <PARENT_RDR>);
834 print "Child Pid $$ just read this: '$line'\n";
835 print PARENT_WTR "Child Pid $$ is sending this\n";
841 But you don't actually have to make two pipe calls. If you
842 have the socketpair() system call, it will do this all for you.
845 # pipe2 - bidirectional communication using socketpair
846 # "the best ones always go both ways"
849 use IO::Handle; # thousands of lines just for autoflush :-(
851 # We say AF_UNIX because although *_LOCAL is the
852 # POSIX 1003.1g form of the constant, many machines
853 # still don't have it.
854 socketpair(CHILD, PARENT, AF_UNIX, SOCK_STREAM, PF_UNSPEC)
855 || die "socketpair: $!";
858 PARENT->autoflush(1);
862 print CHILD "Parent Pid $$ is sending this\n";
863 chomp($line = <CHILD>);
864 print "Parent Pid $$ just read this: '$line'\n";
868 die "cannot fork: $!" unless defined $pid;
870 chomp($line = <PARENT>);
871 print "Child Pid $$ just read this: '$line'\n";
872 print PARENT "Child Pid $$ is sending this\n";
877 =head1 Sockets: Client/Server Communication
879 While not entirely limited to Unix-derived operating systems (e.g., WinSock
880 on PCs provides socket support, as do some VMS libraries), you might not have
881 sockets on your system, in which case this section probably isn't going to
882 do you much good. With sockets, you can do both virtual circuits like TCP
883 streams and datagrams like UDP packets. You may be able to do even more
884 depending on your system.
886 The Perl functions for dealing with sockets have the same names as
887 the corresponding system calls in C, but their arguments tend to differ
888 for two reasons. First, Perl filehandles work differently than C file
889 descriptors. Second, Perl already knows the length of its strings, so you
890 don't need to pass that information.
892 One of the major problems with ancient, antemillennial socket code in Perl
893 was that it used hard-coded values for some of the constants, which
894 severely hurt portability. If you ever see code that does anything like
895 explicitly setting C<$AF_INET = 2>, you know you're in for big trouble.
896 An immeasurably superior approach is to use the C<Socket> module, which more
897 reliably grants access to the various constants and functions you'll need.
899 If you're not writing a server/client for an existing protocol like
900 NNTP or SMTP, you should give some thought to how your server will
901 know when the client has finished talking, and vice-versa. Most
902 protocols are based on one-line messages and responses (so one party
903 knows the other has finished when a "\n" is received) or multi-line
904 messages and responses that end with a period on an empty line
905 ("\n.\n" terminates a message/response).
907 =head2 Internet Line Terminators
909 The Internet line terminator is "\015\012". Under ASCII variants of
910 Unix, that could usually be written as "\r\n", but under other systems,
911 "\r\n" might at times be "\015\015\012", "\012\012\015", or something
912 completely different. The standards specify writing "\015\012" to be
913 conformant (be strict in what you provide), but they also recommend
914 accepting a lone "\012" on input (be lenient in what you require).
915 We haven't always been very good about that in the code in this manpage,
916 but unless you're on a Mac from way back in its pre-Unix dark ages, you'll
919 =head2 Internet TCP Clients and Servers
921 Use Internet-domain sockets when you want to do client-server
922 communication that might extend to machines outside of your own system.
924 Here's a sample TCP client using Internet-domain sockets:
929 my ($remote, $port, $iaddr, $paddr, $proto, $line);
931 $remote = shift || "localhost";
932 $port = shift || 2345; # random port
933 if ($port =~ /\D/) { $port = getservbyname($port, "tcp") }
934 die "No port" unless $port;
935 $iaddr = inet_aton($remote) || die "no host: $remote";
936 $paddr = sockaddr_in($port, $iaddr);
938 $proto = getprotobyname("tcp");
939 socket(SOCK, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
940 connect(SOCK, $paddr) || die "connect: $!";
941 while ($line = <SOCK>) {
945 close (SOCK) || die "close: $!";
948 And here's a corresponding server to go along with it. We'll
949 leave the address as C<INADDR_ANY> so that the kernel can choose
950 the appropriate interface on multihomed hosts. If you want sit
951 on a particular interface (like the external side of a gateway
952 or firewall machine), fill this in with your real address instead.
956 BEGIN { $ENV{PATH} = "/usr/bin:/bin" }
959 my $EOL = "\015\012";
961 sub logmsg { print "$0 $$: @_ at ", scalar localtime(), "\n" }
963 my $port = shift || 2345;
964 die "invalid port" unless if $port =~ /^ \d+ $/x;
966 my $proto = getprotobyname("tcp");
968 socket(Server, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
969 setsockopt(Server, SOL_SOCKET, SO_REUSEADDR, pack("l", 1))
970 || die "setsockopt: $!";
971 bind(Server, sockaddr_in($port, INADDR_ANY)) || die "bind: $!";
972 listen(Server, SOMAXCONN) || die "listen: $!";
974 logmsg "server started on port $port";
978 $SIG{CHLD} = \&REAPER;
980 for ( ; $paddr = accept(Client, Server); close Client) {
981 my($port, $iaddr) = sockaddr_in($paddr);
982 my $name = gethostbyaddr($iaddr, AF_INET);
984 logmsg "connection from $name [",
985 inet_ntoa($iaddr), "]
988 print Client "Hello there, $name, it's now ",
989 scalar localtime(), $EOL;
992 And here's a multitasking version. It's multitasked in that
993 like most typical servers, it spawns (fork()s) a slave server to
994 handle the client request so that the master server can quickly
995 go back to service a new client.
999 BEGIN { $ENV{PATH} = "/usr/bin:/bin" }
1002 my $EOL = "\015\012";
1004 sub spawn; # forward declaration
1005 sub logmsg { print "$0 $$: @_ at ", scalar localtime(), "\n" }
1007 my $port = shift || 2345;
1008 die "invalid port" unless $port =~ /^ \d+ $/x;
1010 my $proto = getprotobyname("tcp");
1012 socket(Server, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
1013 setsockopt(Server, SOL_SOCKET, SO_REUSEADDR, pack("l", 1))
1014 || die "setsockopt: $!";
1015 bind(Server, sockaddr_in($port, INADDR_ANY)) || die "bind: $!";
1016 listen(Server, SOMAXCONN) || die "listen: $!";
1018 logmsg "server started on port $port";
1023 use POSIX ":sys_wait_h";
1027 local $!; # don't let waitpid() overwrite current error
1028 while ((my $pid = waitpid(-1, WNOHANG)) > 0 && WIFEXITED($?)) {
1029 logmsg "reaped $waitedpid" . ($? ? " with exit $?" : "");
1031 $SIG{CHLD} = \&REAPER; # loathe SysV
1034 $SIG{CHLD} = \&REAPER;
1037 $paddr = accept(Client, Server) || do {
1038 # try again if accept() returned because got a signal
1042 my ($port, $iaddr) = sockaddr_in($paddr);
1043 my $name = gethostbyaddr($iaddr, AF_INET);
1045 logmsg "connection from $name [",
1051 print "Hello there, $name, it's now ", scalar localtime(), $EOL;
1052 exec "/usr/games/fortune" # XXX: "wrong" line terminators
1053 or confess "can't exec fortune: $!";
1059 my $coderef = shift;
1061 unless (@_ == 0 && $coderef && ref($coderef) eq "CODE") {
1062 confess "usage: spawn CODEREF";
1066 unless (defined($pid = fork())) {
1067 logmsg "cannot fork: $!";
1071 logmsg "begat $pid";
1072 return; # I'm the parent
1074 # else I'm the child -- go spawn
1076 open(STDIN, "<&Client") || die "can't dup client to stdin";
1077 open(STDOUT, ">&Client") || die "can't dup client to stdout";
1078 ## open(STDERR, ">&STDOUT") || die "can't dup stdout to stderr";
1082 This server takes the trouble to clone off a child version via fork()
1083 for each incoming request. That way it can handle many requests at
1084 once, which you might not always want. Even if you don't fork(), the
1085 listen() will allow that many pending connections. Forking servers
1086 have to be particularly careful about cleaning up their dead children
1087 (called "zombies" in Unix parlance), because otherwise you'll quickly
1088 fill up your process table. The REAPER subroutine is used here to
1089 call waitpid() for any child processes that have finished, thereby
1090 ensuring that they terminate cleanly and don't join the ranks of the
1093 Within the while loop we call accept() and check to see if it returns
1094 a false value. This would normally indicate a system error needs
1095 to be reported. However, the introduction of safe signals (see
1096 L</Deferred Signals (Safe Signals)> above) in Perl 5.8.0 means that
1097 accept() might also be interrupted when the process receives a signal.
1098 This typically happens when one of the forked subprocesses exits and
1099 notifies the parent process with a CHLD signal.
1101 If accept() is interrupted by a signal, $! will be set to EINTR.
1102 If this happens, we can safely continue to the next iteration of
1103 the loop and another call to accept(). It is important that your
1104 signal handling code not modify the value of $!, or else this test
1105 will likely fail. In the REAPER subroutine we create a local version
1106 of $! before calling waitpid(). When waitpid() sets $! to ECHILD as
1107 it inevitably does when it has no more children waiting, it
1108 updates the local copy and leaves the original unchanged.
1110 You should use the B<-T> flag to enable taint checking (see L<perlsec>)
1111 even if we aren't running setuid or setgid. This is always a good idea
1112 for servers or any program run on behalf of someone else (like CGI
1113 scripts), because it lessens the chances that people from the outside will
1114 be able to compromise your system.
1116 Let's look at another TCP client. This one connects to the TCP "time"
1117 service on a number of different machines and shows how far their clocks
1118 differ from the system on which it's being run:
1124 my $SECS_OF_70_YEARS = 2208988800;
1125 sub ctime { scalar localtime(shift() || time()) }
1127 my $iaddr = gethostbyname("localhost");
1128 my $proto = getprotobyname("tcp");
1129 my $port = getservbyname("time", "tcp");
1130 my $paddr = sockaddr_in(0, $iaddr);
1134 printf "%-24s %8s %s\n", "localhost", 0, ctime();
1136 foreach $host (@ARGV) {
1137 printf "%-24s ", $host;
1138 my $hisiaddr = inet_aton($host) || die "unknown host";
1139 my $hispaddr = sockaddr_in($port, $hisiaddr);
1140 socket(SOCKET, PF_INET, SOCK_STREAM, $proto)
1141 || die "socket: $!";
1142 connect(SOCKET, $hispaddr) || die "connect: $!";
1143 my $rtime = pack("C4", ());
1144 read(SOCKET, $rtime, 4);
1146 my $histime = unpack("N", $rtime) - $SECS_OF_70_YEARS;
1147 printf "%8d %s\n", $histime - time(), ctime($histime);
1150 =head2 Unix-Domain TCP Clients and Servers
1152 That's fine for Internet-domain clients and servers, but what about local
1153 communications? While you can use the same setup, sometimes you don't
1154 want to. Unix-domain sockets are local to the current host, and are often
1155 used internally to implement pipes. Unlike Internet domain sockets, Unix
1156 domain sockets can show up in the file system with an ls(1) listing.
1159 srw-rw-rw- 1 root 0 Oct 31 07:23 /dev/log
1161 You can test for these with Perl's B<-S> file test:
1163 unless (-S "/dev/log") {
1164 die "something's wicked with the log system";
1167 Here's a sample Unix-domain client:
1172 my ($rendezvous, $line);
1174 $rendezvous = shift || "catsock";
1175 socket(SOCK, PF_UNIX, SOCK_STREAM, 0) || die "socket: $!";
1176 connect(SOCK, sockaddr_un($rendezvous)) || die "connect: $!";
1177 while (defined($line = <SOCK>)) {
1182 And here's a corresponding server. You don't have to worry about silly
1183 network terminators here because Unix domain sockets are guaranteed
1184 to be on the localhost, and thus everything works right.
1191 BEGIN { $ENV{PATH} = "/usr/bin:/bin" }
1192 sub spawn; # forward declaration
1193 sub logmsg { print "$0 $$: @_ at ", scalar localtime(), "\n" }
1195 my $NAME = "catsock";
1196 my $uaddr = sockaddr_un($NAME);
1197 my $proto = getprotobyname("tcp");
1199 socket(Server, PF_UNIX, SOCK_STREAM, 0) || die "socket: $!";
1201 bind (Server, $uaddr) || die "bind: $!";
1202 listen(Server, SOMAXCONN) || die "listen: $!";
1204 logmsg "server started on $NAME";
1208 use POSIX ":sys_wait_h";
1211 while (($waitedpid = waitpid(-1, WNOHANG)) > 0) {
1212 logmsg "reaped $waitedpid" . ($? ? " with exit $?" : "");
1214 $SIG{CHLD} = \&REAPER; # loathe SysV
1217 $SIG{CHLD} = \&REAPER;
1220 for ( $waitedpid = 0;
1221 accept(Client, Server) || $waitedpid;
1222 $waitedpid = 0, close Client)
1225 logmsg "connection on $NAME";
1227 print "Hello there, it's now ", scalar localtime(), "\n";
1228 exec("/usr/games/fortune") || die "can't exec fortune: $!";
1233 my $coderef = shift();
1235 unless (@_ == 0 && $coderef && ref($coderef) eq "CODE") {
1236 confess "usage: spawn CODEREF";
1240 unless (defined($pid = fork())) {
1241 logmsg "cannot fork: $!";
1245 logmsg "begat $pid";
1246 return; # I'm the parent
1249 # I'm the child -- go spawn
1252 open(STDIN, "<&Client") || die "can't dup client to stdin";
1253 open(STDOUT, ">&Client") || die "can't dup client to stdout";
1254 ## open(STDERR, ">&STDOUT") || die "can't dup stdout to stderr";
1258 As you see, it's remarkably similar to the Internet domain TCP server, so
1259 much so, in fact, that we've omitted several duplicate functions--spawn(),
1260 logmsg(), ctime(), and REAPER()--which are the same as in the other server.
1262 So why would you ever want to use a Unix domain socket instead of a
1263 simpler named pipe? Because a named pipe doesn't give you sessions. You
1264 can't tell one process's data from another's. With socket programming,
1265 you get a separate session for each client; that's why accept() takes two
1268 For example, let's say that you have a long-running database server daemon
1269 that you want folks to be able to access from the Web, but only
1270 if they go through a CGI interface. You'd have a small, simple CGI
1271 program that does whatever checks and logging you feel like, and then acts
1272 as a Unix-domain client and connects to your private server.
1274 =head1 TCP Clients with IO::Socket
1276 For those preferring a higher-level interface to socket programming, the
1277 IO::Socket module provides an object-oriented approach. If for some reason
1278 you lack this module, you can just fetch IO::Socket from CPAN, where you'll also
1279 find modules providing easy interfaces to the following systems: DNS, FTP,
1280 Ident (RFC 931), NIS and NISPlus, NNTP, Ping, POP3, SMTP, SNMP, SSLeay,
1281 Telnet, and Time--to name just a few.
1283 =head2 A Simple Client
1285 Here's a client that creates a TCP connection to the "daytime"
1286 service at port 13 of the host name "localhost" and prints out everything
1287 that the server there cares to provide.
1291 $remote = IO::Socket::INET->new(
1293 PeerAddr => "localhost",
1294 PeerPort => "daytime(13)",
1296 || die "can't connect to daytime service on localhost";
1297 while (<$remote>) { print }
1299 When you run this program, you should get something back that
1302 Wed May 14 08:40:46 MDT 1997
1304 Here are what those parameters to the new() constructor mean:
1310 This is which protocol to use. In this case, the socket handle returned
1311 will be connected to a TCP socket, because we want a stream-oriented
1312 connection, that is, one that acts pretty much like a plain old file.
1313 Not all sockets are this of this type. For example, the UDP protocol
1314 can be used to make a datagram socket, used for message-passing.
1318 This is the name or Internet address of the remote host the server is
1319 running on. We could have specified a longer name like C<"www.perl.com">,
1320 or an address like C<"207.171.7.72">. For demonstration purposes, we've
1321 used the special hostname C<"localhost">, which should always mean the
1322 current machine you're running on. The corresponding Internet address
1323 for localhost is C<"127.0.0.1">, if you'd rather use that.
1327 This is the service name or port number we'd like to connect to.
1328 We could have gotten away with using just C<"daytime"> on systems with a
1329 well-configured system services file,[FOOTNOTE: The system services file
1330 is found in I</etc/services> under Unixy systems.] but here we've specified the
1331 port number (13) in parentheses. Using just the number would have also
1332 worked, but numeric literals make careful programmers nervous.
1336 Notice how the return value from the C<new> constructor is used as
1337 a filehandle in the C<while> loop? That's what's called an I<indirect
1338 filehandle>, a scalar variable containing a filehandle. You can use
1339 it the same way you would a normal filehandle. For example, you
1340 can read one line from it this way:
1344 all remaining lines from is this way:
1348 and send a line of data to it this way:
1350 print $handle "some data\n";
1352 =head2 A Webget Client
1354 Here's a simple client that takes a remote host to fetch a document
1355 from, and then a list of files to get from that host. This is a
1356 more interesting client than the previous one because it first sends
1357 something to the server before fetching the server's response.
1361 unless (@ARGV > 1) { die "usage: $0 host url ..." }
1362 $host = shift(@ARGV);
1365 for my $document (@ARGV) {
1366 $remote = IO::Socket::INET->new( Proto => "tcp",
1368 PeerPort => "http(80)",
1369 ) || die "cannot connect to httpd on $host";
1370 $remote->autoflush(1);
1371 print $remote "GET $document HTTP/1.0" . $BLANK;
1372 while ( <$remote> ) { print }
1376 The web server handling the HTTP service is assumed to be at
1377 its standard port, number 80. If the server you're trying to
1378 connect to is at a different port, like 1080 or 8080, you should specify it
1379 as the named-parameter pair, C<< PeerPort => 8080 >>. The C<autoflush>
1380 method is used on the socket because otherwise the system would buffer
1381 up the output we sent it. (If you're on a prehistoric Mac, you'll also
1382 need to change every C<"\n"> in your code that sends data over the network
1383 to be a C<"\015\012"> instead.)
1385 Connecting to the server is only the first part of the process: once you
1386 have the connection, you have to use the server's language. Each server
1387 on the network has its own little command language that it expects as
1388 input. The string that we send to the server starting with "GET" is in
1389 HTTP syntax. In this case, we simply request each specified document.
1390 Yes, we really are making a new connection for each document, even though
1391 it's the same host. That's the way you always used to have to speak HTTP.
1392 Recent versions of web browsers may request that the remote server leave
1393 the connection open a little while, but the server doesn't have to honor
1396 Here's an example of running that program, which we'll call I<webget>:
1398 % webget www.perl.com /guanaco.html
1399 HTTP/1.1 404 File Not Found
1400 Date: Thu, 08 May 1997 18:02:32 GMT
1401 Server: Apache/1.2b6
1403 Content-type: text/html
1405 <HEAD><TITLE>404 File Not Found</TITLE></HEAD>
1406 <BODY><H1>File Not Found</H1>
1407 The requested URL /guanaco.html was not found on this server.<P>
1410 Ok, so that's not very interesting, because it didn't find that
1411 particular document. But a long response wouldn't have fit on this page.
1413 For a more featureful version of this program, you should look to
1414 the I<lwp-request> program included with the LWP modules from CPAN.
1416 =head2 Interactive Client with IO::Socket
1418 Well, that's all fine if you want to send one command and get one answer,
1419 but what about setting up something fully interactive, somewhat like
1420 the way I<telnet> works? That way you can type a line, get the answer,
1421 type a line, get the answer, etc.
1423 This client is more complicated than the two we've done so far, but if
1424 you're on a system that supports the powerful C<fork> call, the solution
1425 isn't that rough. Once you've made the connection to whatever service
1426 you'd like to chat with, call C<fork> to clone your process. Each of
1427 these two identical process has a very simple job to do: the parent
1428 copies everything from the socket to standard output, while the child
1429 simultaneously copies everything from standard input to the socket.
1430 To accomplish the same thing using just one process would be I<much>
1431 harder, because it's easier to code two processes to do one thing than it
1432 is to code one process to do two things. (This keep-it-simple principle
1433 a cornerstones of the Unix philosophy, and good software engineering as
1434 well, which is probably why it's spread to other systems.)
1441 my ($host, $port, $kidpid, $handle, $line);
1443 unless (@ARGV == 2) { die "usage: $0 host port" }
1444 ($host, $port) = @ARGV;
1446 # create a tcp connection to the specified host and port
1447 $handle = IO::Socket::INET->new(Proto => "tcp",
1450 || die "can't connect to port $port on $host: $!";
1452 $handle->autoflush(1); # so output gets there right away
1453 print STDERR "[Connected to $host:$port]\n";
1455 # split the program into two processes, identical twins
1456 die "can't fork: $!" unless defined($kidpid = fork());
1458 # the if{} block runs only in the parent process
1460 # copy the socket to standard output
1461 while (defined ($line = <$handle>)) {
1464 kill("TERM", $kidpid); # send SIGTERM to child
1466 # the else{} block runs only in the child process
1468 # copy standard input to the socket
1469 while (defined ($line = <STDIN>)) {
1470 print $handle $line;
1472 exit(0); # just in case
1475 The C<kill> function in the parent's C<if> block is there to send a
1476 signal to our child process, currently running in the C<else> block,
1477 as soon as the remote server has closed its end of the connection.
1479 If the remote server sends data a byte at time, and you need that
1480 data immediately without waiting for a newline (which might not happen),
1481 you may wish to replace the C<while> loop in the parent with the
1485 while (sysread($handle, $byte, 1) == 1) {
1489 Making a system call for each byte you want to read is not very efficient
1490 (to put it mildly) but is the simplest to explain and works reasonably
1493 =head1 TCP Servers with IO::Socket
1495 As always, setting up a server is little bit more involved than running a client.
1496 The model is that the server creates a special kind of socket that
1497 does nothing but listen on a particular port for incoming connections.
1498 It does this by calling the C<< IO::Socket::INET->new() >> method with
1499 slightly different arguments than the client did.
1505 This is which protocol to use. Like our clients, we'll
1506 still specify C<"tcp"> here.
1511 port in the C<LocalPort> argument, which we didn't do for the client.
1512 This is service name or port number for which you want to be the
1513 server. (Under Unix, ports under 1024 are restricted to the
1514 superuser.) In our sample, we'll use port 9000, but you can use
1515 any port that's not currently in use on your system. If you try
1516 to use one already in used, you'll get an "Address already in use"
1517 message. Under Unix, the C<netstat -a> command will show
1518 which services current have servers.
1522 The C<Listen> parameter is set to the maximum number of
1523 pending connections we can accept until we turn away incoming clients.
1524 Think of it as a call-waiting queue for your telephone.
1525 The low-level Socket module has a special symbol for the system maximum, which
1530 The C<Reuse> parameter is needed so that we restart our server
1531 manually without waiting a few minutes to allow system buffers to
1536 Once the generic server socket has been created using the parameters
1537 listed above, the server then waits for a new client to connect
1538 to it. The server blocks in the C<accept> method, which eventually accepts a
1539 bidirectional connection from the remote client. (Make sure to autoflush
1540 this handle to circumvent buffering.)
1542 To add to user-friendliness, our server prompts the user for commands.
1543 Most servers don't do this. Because of the prompt without a newline,
1544 you'll have to use the C<sysread> variant of the interactive client above.
1546 This server accepts one of five different commands, sending output back to
1547 the client. Unlike most network servers, this one handles only one
1548 incoming client at a time. Multitasking servers are covered in
1549 Chapter 16 of the Camel.
1551 Here's the code. We'll
1555 use Net::hostent; # for OOish version of gethostbyaddr
1557 $PORT = 9000; # pick something not in use
1559 $server = IO::Socket::INET->new( Proto => "tcp",
1561 Listen => SOMAXCONN,
1564 die "can't setup server" unless $server;
1565 print "[Server $0 accepting clients]\n";
1567 while ($client = $server->accept()) {
1568 $client->autoflush(1);
1569 print $client "Welcome to $0; type help for command list.\n";
1570 $hostinfo = gethostbyaddr($client->peeraddr);
1571 printf "[Connect from %s]\n", $hostinfo ? $hostinfo->name : $client->peerhost;
1572 print $client "Command? ";
1573 while ( <$client>) {
1574 next unless /\S/; # blank line
1575 if (/quit|exit/i) { last }
1576 elsif (/date|time/i) { printf $client "%s\n", scalar localtime() }
1577 elsif (/who/i ) { print $client `who 2>&1` }
1578 elsif (/cookie/i ) { print $client `/usr/games/fortune 2>&1` }
1579 elsif (/motd/i ) { print $client `cat /etc/motd 2>&1` }
1581 print $client "Commands: quit date who cookie motd\n";
1584 print $client "Command? ";
1589 =head1 UDP: Message Passing
1591 Another kind of client-server setup is one that uses not connections, but
1592 messages. UDP communications involve much lower overhead but also provide
1593 less reliability, as there are no promises that messages will arrive at
1594 all, let alone in order and unmangled. Still, UDP offers some advantages
1595 over TCP, including being able to "broadcast" or "multicast" to a whole
1596 bunch of destination hosts at once (usually on your local subnet). If you
1597 find yourself overly concerned about reliability and start building checks
1598 into your message system, then you probably should use just TCP to start
1601 UDP datagrams are I<not> a bytestream and should not be treated as such.
1602 This makes using I/O mechanisms with internal buffering like stdio (i.e.
1603 print() and friends) especially cumbersome. Use syswrite(), or better
1604 send(), like in the example below.
1606 Here's a UDP program similar to the sample Internet TCP client given
1607 earlier. However, instead of checking one host at a time, the UDP version
1608 will check many of them asynchronously by simulating a multicast and then
1609 using select() to do a timed-out wait for I/O. To do something similar
1610 with TCP, you'd have to use a different socket handle for each host.
1617 my ( $count, $hisiaddr, $hispaddr, $histime,
1618 $host, $iaddr, $paddr, $port, $proto,
1619 $rin, $rout, $rtime, $SECS_OF_70_YEARS);
1621 $SECS_OF_70_YEARS = 2_208_988_800;
1623 $iaddr = gethostbyname(hostname());
1624 $proto = getprotobyname("udp");
1625 $port = getservbyname("time", "udp");
1626 $paddr = sockaddr_in(0, $iaddr); # 0 means let kernel pick
1628 socket(SOCKET, PF_INET, SOCK_DGRAM, $proto) || die "socket: $!";
1629 bind(SOCKET, $paddr) || die "bind: $!";
1632 printf "%-12s %8s %s\n", "localhost", 0, scalar localtime();
1636 $hisiaddr = inet_aton($host) || die "unknown host";
1637 $hispaddr = sockaddr_in($port, $hisiaddr);
1638 defined(send(SOCKET, 0, 0, $hispaddr)) || die "send $host: $!";
1642 vec($rin, fileno(SOCKET), 1) = 1;
1644 # timeout after 10.0 seconds
1645 while ($count && select($rout = $rin, undef, undef, 10.0)) {
1647 $hispaddr = recv(SOCKET, $rtime, 4, 0) || die "recv: $!";
1648 ($port, $hisiaddr) = sockaddr_in($hispaddr);
1649 $host = gethostbyaddr($hisiaddr, AF_INET);
1650 $histime = unpack("N", $rtime) - $SECS_OF_70_YEARS;
1651 printf "%-12s ", $host;
1652 printf "%8d %s\n", $histime - time(), scalar localtime($histime);
1656 This example does not include any retries and may consequently fail to
1657 contact a reachable host. The most prominent reason for this is congestion
1658 of the queues on the sending host if the number of hosts to contact is
1663 While System V IPC isn't so widely used as sockets, it still has some
1664 interesting uses. However, you cannot use SysV IPC or Berkeley mmap() to
1665 have a variable shared amongst several processes. That's because Perl
1666 would reallocate your string when you weren't wanting it to. You might
1667 look into the C<IPC::Shareable> or C<threads::shared> modules for that.
1669 Here's a small example showing shared memory usage.
1671 use IPC::SysV qw(IPC_PRIVATE IPC_RMID S_IRUSR S_IWUSR);
1674 $id = shmget(IPC_PRIVATE, $size, S_IRUSR | S_IWUSR);
1675 defined($id) || die "shmget: $!";
1676 print "shm key $id\n";
1678 $message = "Message #1";
1679 shmwrite($id, $message, 0, 60) || die "shmwrite: $!";
1680 print "wrote: '$message'\n";
1681 shmread($id, $buff, 0, 60) || die "shmread: $!";
1682 print "read : '$buff'\n";
1684 # the buffer of shmread is zero-character end-padded.
1685 substr($buff, index($buff, "\0")) = "";
1686 print "un" unless $buff eq $message;
1689 print "deleting shm $id\n";
1690 shmctl($id, IPC_RMID, 0) || die "shmctl: $!";
1692 Here's an example of a semaphore:
1694 use IPC::SysV qw(IPC_CREAT);
1697 $id = semget($IPC_KEY, 10, 0666 | IPC_CREAT);
1698 defined($id) || die "semget: $!";
1699 print "sem id $id\n";
1701 Put this code in a separate file to be run in more than one process.
1702 Call the file F<take>:
1704 # create a semaphore
1707 $id = semget($IPC_KEY, 0, 0);
1708 defined($id) || die "semget: $!";
1714 # wait for semaphore to be zero
1716 $opstring1 = pack("s!s!s!", $semnum, $semop, $semflag);
1718 # Increment the semaphore count
1720 $opstring2 = pack("s!s!s!", $semnum, $semop, $semflag);
1721 $opstring = $opstring1 . $opstring2;
1723 semop($id, $opstring) || die "semop: $!";
1725 Put this code in a separate file to be run in more than one process.
1726 Call this file F<give>:
1728 # "give" the semaphore
1729 # run this in the original process and you will see
1730 # that the second process continues
1733 $id = semget($IPC_KEY, 0, 0);
1734 die unless defined($id);
1739 # Decrement the semaphore count
1741 $opstring = pack("s!s!s!", $semnum, $semop, $semflag);
1743 semop($id, $opstring) || die "semop: $!";
1745 The SysV IPC code above was written long ago, and it's definitely
1746 clunky looking. For a more modern look, see the IPC::SysV module.
1748 A small example demonstrating SysV message queues:
1750 use IPC::SysV qw(IPC_PRIVATE IPC_RMID IPC_CREAT S_IRUSR S_IWUSR);
1752 my $id = msgget(IPC_PRIVATE, IPC_CREAT | S_IRUSR | S_IWUSR);
1753 defined($id) || die "msgget failed: $!";
1755 my $sent = "message";
1756 my $type_sent = 1234;
1758 msgsnd($id, pack("l! a*", $type_sent, $sent), 0)
1759 || die "msgsnd failed: $!";
1761 msgrcv($id, my $rcvd_buf, 60, 0, 0)
1762 || die "msgrcv failed: $!";
1764 my($type_rcvd, $rcvd) = unpack("l! a*", $rcvd_buf);
1766 if ($rcvd eq $sent) {
1772 msgctl($id, IPC_RMID, 0) || die "msgctl failed: $!\n";
1776 Most of these routines quietly but politely return C<undef> when they
1777 fail instead of causing your program to die right then and there due to
1778 an uncaught exception. (Actually, some of the new I<Socket> conversion
1779 functions do croak() on bad arguments.) It is therefore essential to
1780 check return values from these functions. Always begin your socket
1781 programs this way for optimal success, and don't forget to add the B<-T>
1782 taint-checking flag to the C<#!> line for servers:
1791 These routines all create system-specific portability problems. As noted
1792 elsewhere, Perl is at the mercy of your C libraries for much of its system
1793 behavior. It's probably safest to assume broken SysV semantics for
1794 signals and to stick with simple TCP and UDP socket operations; e.g., don't
1795 try to pass open file descriptors over a local UDP datagram socket if you
1796 want your code to stand a chance of being portable.
1800 Tom Christiansen, with occasional vestiges of Larry Wall's original
1801 version and suggestions from the Perl Porters.
1805 There's a lot more to networking than this, but this should get you
1808 For intrepid programmers, the indispensable textbook is I<Unix Network
1809 Programming, 2nd Edition, Volume 1> by W. Richard Stevens (published by
1810 Prentice-Hall). Most books on networking address the subject from the
1811 perspective of a C programmer; translation to Perl is left as an exercise
1814 The IO::Socket(3) manpage describes the object library, and the Socket(3)
1815 manpage describes the low-level interface to sockets. Besides the obvious
1816 functions in L<perlfunc>, you should also check out the F<modules> file at
1817 your nearest CPAN site, especially
1818 L<http://www.cpan.org/modules/00modlist.long.html#ID5_Networking_>.
1819 See L<perlmodlib> or best yet, the F<Perl FAQ> for a description
1820 of what CPAN is and where to get it if the previous link doesn't work
1823 Section 5 of CPAN's F<modules> file is devoted to "Networking, Device
1824 Control (modems), and Interprocess Communication", and contains numerous
1825 unbundled modules numerous networking modules, Chat and Expect operations,
1826 CGI programming, DCE, FTP, IPC, NNTP, Proxy, Ptty, RPC, SNMP, SMTP, Telnet,
1827 Threads, and ToolTalk--to name just a few.