This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
perlipc: strict safety, consistency, cleanup
[perl5.git] / pod / perlipc.pod
CommitLineData
a0d0e21e
LW
1=head1 NAME
2
184e9718 3perlipc - Perl interprocess communication (signals, fifos, pipes, safe subprocesses, sockets, and semaphores)
a0d0e21e
LW
4
5=head1 DESCRIPTION
6
4633a7c4
LW
7The basic IPC facilities of Perl are built out of the good old Unix
8signals, named pipes, pipe opens, the Berkeley socket routines, and SysV
9IPC calls. Each is used in slightly different situations.
10
11=head1 Signals
12
490f90af
JH
13Perl uses a simple signal handling model: the %SIG hash contains names
14or references of user-installed signal handlers. These handlers will
15be called with an argument which is the name of the signal that
16triggered it. A signal may be generated intentionally from a
17particular keyboard sequence like control-C or control-Z, sent to you
18from another process, or triggered automatically by the kernel when
cf21866a
TC
19special events transpire, like a child process exiting, your own process
20running out of stack space, or hitting a process file-size limit.
4633a7c4 21
a11adca0 22For example, to trap an interrupt signal, set up a handler like this:
4633a7c4 23
73af1a12
DG
24 our $shucks;
25
4633a7c4 26 sub catch_zap {
322c2516
SF
27 my $signame = shift;
28 $shucks++;
29 die "Somebody sent me a SIG$signame";
54310121 30 }
82f82fdb 31 $SIG{INT} = __PACKAGE__ . "::catch_zap";
4633a7c4
LW
32 $SIG{INT} = \&catch_zap; # best strategy
33
e6aa8b84 34Prior to Perl 5.8.0 it was necessary to do as little as you possibly
490f90af
JH
35could in your handler; notice how all we do is set a global variable
36and then raise an exception. That's because on most systems,
37libraries are not re-entrant; particularly, memory allocation and I/O
38routines are not. That meant that doing nearly I<anything> in your
39handler could in theory trigger a memory fault and subsequent core
ec488bcf 40dump - see L</Deferred Signals (Safe Signals)> below.
a11adca0 41
4633a7c4 42The names of the signals are the ones listed out by C<kill -l> on your
de7ba517 43system, or you can retrieve them using the CPAN module L<IPC::Signal>.
4633a7c4 44
cf21866a 45You may also choose to assign the strings C<"IGNORE"> or C<"DEFAULT"> as
4633a7c4 46the handler, in which case Perl will try to discard the signal or do the
f648820c
GS
47default thing.
48
19799a22 49On most Unix platforms, the C<CHLD> (sometimes also known as C<CLD>) signal
cf21866a
TC
50has special behavior with respect to a value of C<"IGNORE">.
51Setting C<$SIG{CHLD}> to C<"IGNORE"> on such a platform has the effect of
f648820c 52not creating zombie processes when the parent process fails to C<wait()>
cf21866a
TC
53on its child processes (i.e., child processes are automatically reaped).
54Calling C<wait()> with C<$SIG{CHLD}> set to C<"IGNORE"> usually returns
f648820c
GS
55C<-1> on such platforms.
56
cf21866a 57Some signals can be neither trapped nor ignored, such as the KILL and STOP
de7ba517
LT
58(but not the TSTP) signals. Note that ignoring signals makes them disappear.
59If you only want them blocked temporarily without them getting lost you'll
60have to use POSIX' sigprocmask.
4633a7c4
LW
61
62Sending a signal to a negative process ID means that you send the signal
cf21866a 63to the entire Unix process group. This code sends a hang-up signal to all
82f82fdb 64processes in the current process group, and also sets $SIG{HUP} to C<"IGNORE">
cf21866a 65so it doesn't kill itself:
4633a7c4 66
cf21866a 67 # block scope for local
4633a7c4 68 {
cf21866a 69 local $SIG{HUP} = "IGNORE";
6fc8325d
Z
70 kill HUP => -getpgrp();
71 # snazzy writing of: kill("HUP", -getpgrp())
4633a7c4 72 }
a0d0e21e 73
4633a7c4 74Another interesting signal to send is signal number zero. This doesn't
1e9c1022 75actually affect a child process, but instead checks whether it's alive
de7ba517 76or has changed its UIDs.
a0d0e21e 77
4633a7c4 78 unless (kill 0 => $kid_pid) {
322c2516 79 warn "something wicked happened to $kid_pid";
54310121 80 }
a0d0e21e 81
de7ba517
LT
82Signal number zero may fail because you lack permission to send the
83signal when directed at a process whose real or saved UID is not
84identical to the real or effective UID of the sending process, even
85though the process is alive. You may be able to determine the cause of
86failure using C<$!> or C<%!>.
1e9c1022 87
cf21866a 88 unless (kill(0 => $pid) || $!{EPERM}) {
322c2516 89 warn "$pid looks dead";
1e9c1022
JL
90 }
91
4633a7c4
LW
92You might also want to employ anonymous functions for simple signal
93handlers:
a0d0e21e 94
4633a7c4 95 $SIG{INT} = sub { die "\nOutta here!\n" };
a0d0e21e 96
de7ba517
LT
97SIGCHLD handlers require some special care. If a second child dies
98while in the signal handler caused by the first death, we won't get
99another signal. So must loop here else we will leave the unreaped child
100as a zombie. And the next time two children die we get another zombie.
101And so on.
4633a7c4 102
6a3992aa 103 use POSIX ":sys_wait_h";
de7ba517
LT
104 $SIG{CHLD} = sub {
105 while ((my $child = waitpid(-1, WNOHANG)) > 0) {
322c2516
SF
106 $Kid_Status{$child} = $?;
107 }
de7ba517 108 };
4633a7c4
LW
109 # do something that forks...
110
cf21866a
TC
111Be careful: qx(), system(), and some modules for calling external commands
112do a fork(), then wait() for the result. Thus, your signal handler
de7ba517
LT
113will be called. Because wait() was already called by system() or qx(),
114the wait() in the signal handler will see no more zombies and will
115therefore block.
0a18a49b 116
cf21866a 117The best way to prevent this issue is to use waitpid(), as in the following
0a18a49b
MH
118example:
119
120 use POSIX ":sys_wait_h"; # for nonblocking read
121
122 my %children;
123
124 $SIG{CHLD} = sub {
125 # don't change $! and $? outside handler
cf21866a 126 local ($!, $?);
fd440202
TC
127 while ( (my $pid = waitpid(-1, WNOHANG)) > 0 ) {
128 delete $children{$pid};
129 cleanup_child($pid, $?);
130 }
0a18a49b
MH
131 };
132
133 while (1) {
134 my $pid = fork();
cf21866a 135 die "cannot fork" unless defined $pid;
0a18a49b
MH
136 if ($pid == 0) {
137 # ...
138 exit 0;
139 } else {
cf21866a 140 $children{$pid}=1;
0a18a49b
MH
141 # ...
142 system($command);
143 # ...
144 }
145 }
146
147Signal handling is also used for timeouts in Unix. While safely
4633a7c4
LW
148protected within an C<eval{}> block, you set a signal handler to trap
149alarm signals and then schedule to have one delivered to you in some
150number of seconds. Then try your blocking operation, clearing the alarm
151when it's done but not before you've exited your C<eval{}> block. If it
de7ba517 152goes off, you'll use die() to jump out of the block.
4633a7c4
LW
153
154Here's an example:
155
cf21866a 156 my $ALARM_EXCEPTION = "alarm clock restart";
54310121 157 eval {
cf21866a 158 local $SIG{ALRM} = sub { die $ALARM_EXCEPTION };
54310121 159 alarm 10;
08ab6dc8 160 flock($fh, 2) # blocking write lock
cf21866a 161 || die "cannot flock: $!";
54310121 162 alarm 0;
4633a7c4 163 };
cf21866a 164 if ($@ && $@ !~ quotemeta($ALARM_EXCEPTION)) { die }
4633a7c4 165
8a4f6ac2
GS
166If the operation being timed out is system() or qx(), this technique
167is liable to generate zombies. If this matters to you, you'll
168need to do your own fork() and exec(), and kill the errant child process.
169
4633a7c4 170For more complex signal handling, you might see the standard POSIX
cb0ee57a
LM
171module. Lamentably, this is almost entirely undocumented, but the
172F<ext/POSIX/t/sigaction.t> file from the Perl source distribution has
173some examples in it.
4633a7c4 174
28494392
SB
175=head2 Handling the SIGHUP Signal in Daemons
176
177A process that usually starts when the system boots and shuts down
178when the system is shut down is called a daemon (Disk And Execution
179MONitor). If a daemon process has a configuration file which is
180modified after the process has been started, there should be a way to
cf21866a
TC
181tell that process to reread its configuration file without stopping
182the process. Many daemons provide this mechanism using a C<SIGHUP>
183signal handler. When you want to tell the daemon to reread the file,
184simply send it the C<SIGHUP> signal.
28494392 185
28494392
SB
186The following example implements a simple daemon, which restarts
187itself every time the C<SIGHUP> signal is received. The actual code is
cf21866a
TC
188located in the subroutine C<code()>, which just prints some debugging
189info to show that it works; it should be replaced with the real code.
28494392 190
8bc5de20
SF
191 #!/usr/bin/perl
192
193 use strict;
194 use warnings;
d6fd60d6 195
28494392
SB
196 use POSIX ();
197 use FindBin ();
198 use File::Basename ();
8bc5de20 199 use File::Spec::Functions qw(catfile);
d6fd60d6 200
cf21866a 201 $| = 1;
d6fd60d6 202
28494392
SB
203 # make the daemon cross-platform, so exec always calls the script
204 # itself with the right path, no matter how the script was invoked.
205 my $script = File::Basename::basename($0);
cf21866a 206 my $SELF = catfile($FindBin::Bin, $script);
d6fd60d6 207
28494392 208 # POSIX unmasks the sigprocmask properly
de7ba517 209 $SIG{HUP} = sub {
28494392 210 print "got SIGHUP\n";
cf21866a 211 exec($SELF, @ARGV) || die "$0: couldn't restart: $!";
de7ba517 212 };
d6fd60d6 213
28494392 214 code();
d6fd60d6 215
28494392
SB
216 sub code {
217 print "PID: $$\n";
218 print "ARGV: @ARGV\n";
cf21866a 219 my $count = 0;
8bc5de20 220 while (1) {
28494392 221 sleep 2;
8bc5de20 222 print ++$count, "\n";
28494392
SB
223 }
224 }
28494392
SB
225
226
ffc145e8 227=head2 Deferred Signals (Safe Signals)
5a964f20 228
e6aa8b84 229Before Perl 5.8.0, installing Perl code to deal with signals exposed you to
cf21866a
TC
230danger from two things. First, few system library functions are
231re-entrant. If the signal interrupts while Perl is executing one function
232(like malloc(3) or printf(3)), and your signal handler then calls the same
233function again, you could get unpredictable behavior--often, a core dump.
234Second, Perl isn't itself re-entrant at the lowest levels. If the signal
235interrupts Perl while Perl is changing its own internal data structures,
236similarly unpredictable behavior may result.
5a964f20 237
a11adca0
NIS
238There were two things you could do, knowing this: be paranoid or be
239pragmatic. The paranoid approach was to do as little as possible in your
5a964f20
TC
240signal handler. Set an existing integer variable that already has a
241value, and return. This doesn't help you if you're in a slow system call,
7b34eba2 242which will just restart. That means you have to C<die> to longjmp(3) out
5a964f20
TC
243of the handler. Even this is a little cavalier for the true paranoiac,
244who avoids C<die> in a handler because the system I<is> out to get you.
b432a672
AL
245The pragmatic approach was to say "I know the risks, but prefer the
246convenience", and to do anything you wanted in your signal handler,
a11adca0
NIS
247and be prepared to clean up core dumps now and again.
248
e6aa8b84 249Perl 5.8.0 and later avoid these problems by "deferring" signals. That is,
cf21866a
TC
250when the signal is delivered to the process by the system (to the C code
251that implements Perl) a flag is set, and the handler returns immediately.
252Then at strategic "safe" points in the Perl interpreter (e.g. when it is
253about to execute a new opcode) the flags are checked and the Perl level
254handler from %SIG is executed. The "deferred" scheme allows much more
255flexibility in the coding of signal handlers as we know the Perl
de7ba517
LT
256interpreter is in a safe state, and that we are not in a system library
257function when the handler is called. However the implementation does
cf21866a 258differ from previous Perls in the following ways:
5a964f20 259
a11adca0 260=over 4
5a964f20 261
e188fdae
CB
262=item Long-running opcodes
263
cf21866a 264As the Perl interpreter looks at signal flags only when it is about
e188fdae
CB
265to execute a new opcode, a signal that arrives during a long-running
266opcode (e.g. a regular expression operation on a very large string) will
267not be seen until the current opcode completes.
268
82f82fdb 269If a signal of any given type fires multiple times during an opcode
e188fdae 270(such as from a fine-grained timer), the handler for that signal will
cf21866a 271be called only once, after the opcode completes; all other
e188fdae
CB
272instances will be discarded. Furthermore, if your system's signal queue
273gets flooded to the point that there are signals that have been raised
274but not yet caught (and thus not deferred) at the time an opcode
275completes, those signals may well be caught and deferred during
276subsequent opcodes, with sometimes surprising results. For example, you
277may see alarms delivered even after calling C<alarm(0)> as the latter
278stops the raising of alarms but does not cancel the delivery of alarms
279raised but not yet caught. Do not depend on the behaviors described in
280this paragraph as they are side effects of the current implementation and
281may change in future versions of Perl.
a11adca0 282
a11adca0
NIS
283=item Interrupting IO
284
cf21866a
TC
285When a signal is delivered (e.g., SIGINT from a control-C) the operating
286system breaks into IO operations like I<read>(2), which is used to
287implement Perl's readline() function, the C<< <> >> operator. On older
288Perls the handler was called immediately (and as C<read> is not "unsafe",
289this worked well). With the "deferred" scheme the handler is I<not> called
290immediately, and if Perl is using the system's C<stdio> library that
291library may restart the C<read> without returning to Perl to give it a
292chance to call the %SIG handler. If this happens on your system the
293solution is to use the C<:perlio> layer to do IO--at least on those handles
294that you want to be able to break into with signals. (The C<:perlio> layer
295checks the signal flags and calls %SIG handlers before resuming IO
296operation.)
297
e6aa8b84 298The default in Perl 5.8.0 and later is to automatically use
490f90af 299the C<:perlio> layer.
a11adca0 300
abf9167d
DM
301Note that it is not advisable to access a file handle within a signal
302handler where that signal has interrupted an I/O operation on that same
303handle. While perl will at least try hard not to crash, there are no
304guarantees of data integrity; for example, some data might get dropped or
305written twice.
306
cf21866a
TC
307Some networking library functions like gethostbyname() are known to have
308their own implementations of timeouts which may conflict with your
309timeouts. If you have problems with such functions, try using the POSIX
310sigaction() function, which bypasses Perl safe signals. Be warned that
311this does subject you to possible memory corruption, as described above.
312
313Instead of setting C<$SIG{ALRM}>:
91d81acc 314
e399c6ae
SB
315 local $SIG{ALRM} = sub { die "alarm" };
316
317try something like the following:
318
e46aa1dd
KW
319 use POSIX qw(SIGALRM);
320 POSIX::sigaction(SIGALRM,
321 POSIX::SigAction->new(sub { die "alarm" }))
de7ba517 322 || die "Error setting SIGALRM handler: $!\n";
91d81acc 323
a1966b02 324Another way to disable the safe signal behavior locally is to use
cf21866a
TC
325the C<Perl::Unsafe::Signals> module from CPAN, which affects
326all signals.
a1966b02 327
9ce5b4ad
SG
328=item Restartable system calls
329
330On systems that supported it, older versions of Perl used the
331SA_RESTART flag when installing %SIG handlers. This meant that
332restartable system calls would continue rather than returning when
333a signal arrived. In order to deliver deferred signals promptly,
82f82fdb 334Perl 5.8.0 and later do I<not> use SA_RESTART. Consequently,
9ce5b4ad
SG
335restartable system calls can fail (with $! set to C<EINTR>) in places
336where they previously would have succeeded.
337
cf21866a 338The default C<:perlio> layer retries C<read>, C<write>
82f82fdb 339and C<close> as described above; interrupted C<wait> and
9ce5b4ad
SG
340C<waitpid> calls will always be retried.
341
a11adca0
NIS
342=item Signals as "faults"
343
cf21866a 344Certain signals like SEGV, ILL, and BUS are generated by virtual memory
c69ca1d4 345addressing errors and similar "faults". These are normally fatal: there is
de7ba517 346little a Perl-level handler can do with them. So Perl delivers them
e188fdae 347immediately rather than attempting to defer them.
a11adca0
NIS
348
349=item Signals triggered by operating system state
350
490f90af 351On some operating systems certain signal handlers are supposed to "do
cf21866a 352something" before returning. One example can be CHLD or CLD, which
490f90af
JH
353indicates a child process has completed. On some operating systems the
354signal handler is expected to C<wait> for the completed child
355process. On such systems the deferred signal scheme will not work for
cf21866a
TC
356those signals: it does not do the C<wait>. Again the failure will
357look like a loop as the operating system will reissue the signal because
358there are completed child processes that have not yet been C<wait>ed for.
a11adca0 359
818c4caa 360=back
a0d0e21e 361
cf21866a 362If you want the old signal behavior back despite possible
4ffa73a3 363memory corruption, set the environment variable C<PERL_SIGNALS> to
cf21866a 364C<"unsafe">. This feature first appeared in Perl 5.8.1.
4ffa73a3 365
9eed50dc
DM
366=head1 Named Pipes
367
368A named pipe (often referred to as a FIFO) is an old Unix IPC
369mechanism for processes communicating on the same machine. It works
370just like regular anonymous pipes, except that the
371processes rendezvous using a filename and need not be related.
372
373To create a named pipe, use the C<POSIX::mkfifo()> function.
374
375 use POSIX qw(mkfifo);
376 mkfifo($path, 0700) || die "mkfifo $path failed: $!";
377
378You can also use the Unix command mknod(1), or on some
379systems, mkfifo(1). These may not be in your normal path, though.
380
381 # system return val is backwards, so && not ||
382 #
383 $ENV{PATH} .= ":/etc:/usr/etc";
384 if ( system("mknod", $path, "p")
385 && system("mkfifo", $path) )
386 {
387 die "mk{nod,fifo} $path failed";
388 }
389
390
391A fifo is convenient when you want to connect a process to an unrelated
392one. When you open a fifo, the program will block until there's something
393on the other end.
394
395For example, let's say you'd like to have your F<.signature> file be a
396named pipe that has a Perl program on the other end. Now every time any
397program (like a mailer, news reader, finger program, etc.) tries to read
398from that file, the reading program will read the new signature from your
399program. We'll use the pipe-checking file-test operator, B<-p>, to find
400out whether anyone (or anything) has accidentally removed our fifo.
401
402 chdir(); # go home
403 my $FIFO = ".signature";
404
405 while (1) {
406 unless (-p $FIFO) {
407 unlink $FIFO; # discard any failure, will catch later
408 require POSIX; # delayed loading of heavy module
409 POSIX::mkfifo($FIFO, 0700)
08ab6dc8 410 || die "can't mkfifo $FIFO: $!";
9eed50dc
DM
411 }
412
413 # next line blocks till there's a reader
08ab6dc8
DB
414 open (my $fh, ">", $FIFO) || die "can't open $FIFO: $!";
415 print $fh "John Smith (smith\@host.org)\n", `fortune -s`;
416 close($fh) || die "can't close $FIFO: $!";
9eed50dc
DM
417 sleep 2; # to avoid dup signals
418 }
419
4633a7c4
LW
420=head1 Using open() for IPC
421
490f90af 422Perl's basic open() statement can also be used for unidirectional
08ab6dc8
DB
423interprocess communication by specifying the open mode as C<|-> or C<-|>.
424Here's how to start
490f90af 425something up in a child process you intend to write to:
4633a7c4 426
08ab6dc8 427 open(my $spooler, "|-", "cat -v | lpr -h 2>/dev/null")
cf21866a 428 || die "can't fork: $!";
4633a7c4 429 local $SIG{PIPE} = sub { die "spooler pipe broke" };
08ab6dc8
DB
430 print $spooler "stuff\n";
431 close $spooler || die "bad spool: $! $?";
4633a7c4
LW
432
433And here's how to start up a child process you intend to read from:
434
08ab6dc8 435 open(my $status, "-|", "netstat -an 2>&1")
cf21866a 436 || die "can't fork: $!";
08ab6dc8 437 while (<$status>) {
322c2516
SF
438 next if /^(tcp|udp)/;
439 print;
54310121 440 }
08ab6dc8 441 close $status || die "bad netstat: $! $?";
4633a7c4 442
08ab6dc8
DB
443Be aware that these operations are full Unix forks, which means they may
444not be correctly implemented on all alien systems. See L<perlport/open>
445for portability details.
446
447In the two-argument form of open(), a pipe open can be achieved by
448either appending or prepending a pipe symbol to the second argument:
449
450 open(my $spooler, "| cat -v | lpr -h 2>/dev/null")
451 || die "can't fork: $!";
452 open(my $status, "netstat -an 2>&1 |")
453 || die "can't fork: $!";
454
455This can be used even on systems that do not support forking, but this
456possibly allows code intended to read files to unexpectedly execute
457programs. If one can be sure that a particular program is a Perl script
458expecting filenames in @ARGV using the two-argument form of open() or the
459C<< <> >> operator, the clever programmer can write something like this:
4633a7c4 460
5a964f20 461 % program f1 "cmd1|" - f2 "cmd2|" f3 < tmpfile
4633a7c4 462
cf21866a 463and no matter which sort of shell it's called from, the Perl program will
4633a7c4
LW
464read from the file F<f1>, the process F<cmd1>, standard input (F<tmpfile>
465in this case), the F<f2> file, the F<cmd2> command, and finally the F<f3>
466file. Pretty nifty, eh?
467
54310121 468You might notice that you could use backticks for much the
4633a7c4
LW
469same effect as opening a pipe for reading:
470
471 print grep { !/^(tcp|udp)/ } `netstat -an 2>&1`;
cf21866a 472 die "bad netstatus ($?)" if $?;
4633a7c4
LW
473
474While this is true on the surface, it's much more efficient to process the
475file one line or record at a time because then you don't have to read the
19799a22 476whole thing into memory at once. It also gives you finer control of the
cf21866a 477whole process, letting you kill off the child process early if you'd like.
4633a7c4 478
cf21866a 479Be careful to check the return values from both open() and close(). If
4633a7c4
LW
480you're I<writing> to a pipe, you should also trap SIGPIPE. Otherwise,
481think of what happens when you start up a pipe to a command that doesn't
482exist: the open() will in all likelihood succeed (it only reflects the
483fork()'s success), but then your output will fail--spectacularly. Perl
cf21866a 484can't know whether the command worked, because your command is actually
4633a7c4 485running in a separate process whose exec() might have failed. Therefore,
cf21866a
TC
486while readers of bogus commands return just a quick EOF, writers
487to bogus commands will get hit with a signal, which they'd best be prepared
488to handle. Consider:
4633a7c4 489
08ab6dc8
DB
490 open(my $fh, "|-", "bogus") || die "can't fork: $!";
491 print $fh "bang\n"; # neither necessary nor sufficient
492 # to check print retval!
493 close($fh) || die "can't close: $!";
5a964f20 494
cf21866a
TC
495The reason for not checking the return value from print() is because of
496pipe buffering; physical writes are delayed. That won't blow up until the
497close, and it will blow up with a SIGPIPE. To catch it, you could use
498this:
5a964f20 499
cf21866a 500 $SIG{PIPE} = "IGNORE";
08ab6dc8
DB
501 open(my $fh, "|-", "bogus") || die "can't fork: $!";
502 print $fh "bang\n";
503 close($fh) || die "can't close: status=$?";
4633a7c4 504
68dc0745
PP
505=head2 Filehandles
506
5a964f20
TC
507Both the main process and any child processes it forks share the same
508STDIN, STDOUT, and STDERR filehandles. If both processes try to access
45bc9206 509them at once, strange things can happen. You may also want to close
5a964f20
TC
510or reopen the filehandles for the child. You can get around this by
511opening your pipe with open(), but on some systems this means that the
512child process cannot outlive the parent.
68dc0745
PP
513
514=head2 Background Processes
515
516You can run a command in the background with:
517
7b05b7e3 518 system("cmd &");
68dc0745
PP
519
520The command's STDOUT and STDERR (and possibly STDIN, depending on your
521shell) will be the same as the parent's. You won't need to catch
cf21866a 522SIGCHLD because of the double-fork taking place; see below for details.
68dc0745
PP
523
524=head2 Complete Dissociation of Child from Parent
525
526In some cases (starting server processes, for instance) you'll want to
893af57a 527completely dissociate the child process from the parent. This is
cf21866a
TC
528often called daemonization. A well-behaved daemon will also chdir()
529to the root directory so it doesn't prevent unmounting the filesystem
530containing the directory from which it was launched, and redirect its
531standard file descriptors from and to F</dev/null> so that random
532output doesn't wind up on the user's terminal.
893af57a 533
e46aa1dd 534 use POSIX "setsid";
893af57a 535
e46aa1dd 536 sub daemonize {
08ab6dc8
DB
537 chdir("/") || die "can't chdir to /: $!";
538 open(STDIN, "<", "/dev/null") || die "can't read /dev/null: $!";
539 open(STDOUT, ">", "/dev/null") || die "can't write /dev/null: $!";
540 defined(my $pid = fork()) || die "can't fork: $!";
541 exit if $pid; # non-zero now means I am the parent
542 (setsid() != -1) || die "Can't start a new session: $!";
543 open(STDERR, ">&", STDOUT) || die "can't dup stdout: $!";
e46aa1dd 544 }
5a964f20 545
cf21866a
TC
546The fork() has to come before the setsid() to ensure you aren't a
547process group leader; the setsid() will fail if you are. If your
893af57a 548system doesn't have the setsid() function, open F</dev/tty> and use the
f979aebc 549C<TIOCNOTTY> ioctl() on it instead. See tty(4) for details.
5a964f20 550
82f82fdb 551Non-Unix users should check their C<< I<Your_OS>::Process >> module for
cf21866a 552other possible solutions.
68dc0745 553
4633a7c4
LW
554=head2 Safe Pipe Opens
555
556Another interesting approach to IPC is making your single program go
cf21866a 557multiprocess and communicate between--or even amongst--yourselves. The
08ab6dc8 558two-argument form of the
4633a7c4
LW
559open() function will accept a file argument of either C<"-|"> or C<"|-">
560to do a very interesting thing: it forks a child connected to the
561filehandle you've opened. The child is running the same program as the
562parent. This is useful for safely opening a file when running under an
563assumed UID or GID, for example. If you open a pipe I<to> minus, you can
cf21866a 564write to the filehandle you opened and your kid will find it in I<his>
4633a7c4 565STDIN. If you open a pipe I<from> minus, you can read from the filehandle
cf21866a 566you opened whatever your kid writes to I<his> STDOUT.
4633a7c4 567
cf21866a
TC
568 my $PRECIOUS = "/path/to/some/safe/file";
569 my $sleep_count;
570 my $pid;
08ab6dc8 571 my $kid_to_write;
4633a7c4 572
54310121 573 do {
08ab6dc8 574 $pid = open($kid_to_write, "|-");
322c2516
SF
575 unless (defined $pid) {
576 warn "cannot fork: $!";
577 die "bailing out" if $sleep_count++ > 6;
578 sleep 10;
579 }
4633a7c4
LW
580 } until defined $pid;
581
82f82fdb 582 if ($pid) { # I am the parent
08ab6dc8
DB
583 print $kid_to_write @some_data;
584 close($kid_to_write) || warn "kid exited $?";
cf21866a
TC
585 } else { # I am the child
586 # drop permissions in setuid and/or setgid programs:
c43b1046 587 ($>, $)) = ($<, $();
08ab6dc8 588 open (my $outfile, ">", $PRECIOUS)
cf21866a 589 || die "can't open $PRECIOUS: $!";
322c2516 590 while (<STDIN>) {
08ab6dc8 591 print $outfile; # child STDIN is parent $kid_to_write
322c2516 592 }
08ab6dc8 593 close($outfile) || die "can't close $PRECIOUS: $!";
cf21866a 594 exit(0); # don't forget this!!
54310121 595 }
4633a7c4
LW
596
597Another common use for this construct is when you need to execute
598something without the shell's interference. With system(), it's
54310121 599straightforward, but you can't use a pipe open or backticks safely.
4633a7c4
LW
600That's because there's no way to stop the shell from getting its hands on
601your arguments. Instead, use lower-level control to call exec() directly.
602
54310121 603Here's a safe backtick or pipe open for read:
4633a7c4 604
08ab6dc8
DB
605 my $pid = open(my $kid_to_read, "-|");
606 defined($pid) || die "can't fork: $!";
4633a7c4 607
cf21866a 608 if ($pid) { # parent
08ab6dc8 609 while (<$kid_to_read>) {
cf21866a 610 # do something interesting
322c2516 611 }
08ab6dc8 612 close($kid_to_read) || warn "kid exited $?";
4633a7c4 613
cf21866a 614 } else { # child
c43b1046 615 ($>, $)) = ($<, $(); # suid only
322c2516 616 exec($program, @options, @args)
08ab6dc8 617 || die "can't exec program: $!";
322c2516 618 # NOTREACHED
54310121 619 }
4633a7c4 620
4633a7c4
LW
621And here's a safe pipe open for writing:
622
08ab6dc8
DB
623 my $pid = open(my $kid_to_write, "|-");
624 defined($pid) || die "can't fork: $!";
cf21866a 625
76c0e0db 626 $SIG{PIPE} = sub { die "whoops, $program pipe broke" };
4633a7c4 627
cf21866a 628 if ($pid) { # parent
08ab6dc8
DB
629 print $kid_to_write @data;
630 close($kid_to_write) || warn "kid exited $?";
4633a7c4 631
cf21866a 632 } else { # child
c43b1046 633 ($>, $)) = ($<, $();
322c2516 634 exec($program, @options, @args)
08ab6dc8 635 || die "can't exec program: $!";
322c2516 636 # NOTREACHED
54310121 637 }
4633a7c4 638
c40e8e9b 639It is very easy to dead-lock a process using this form of open(), or
82f82fdb 640indeed with any use of pipe() with multiple subprocesses. The
cf21866a 641example above is "safe" because it is simple and calls exec(). See
c40e8e9b
SV
642L</"Avoiding Pipe Deadlocks"> for general safety principles, but there
643are extra gotchas with Safe Pipe Opens.
644
08ab6dc8 645In particular, if you opened the pipe using C<open $fh, "|-">, then you
c40e8e9b
SV
646cannot simply use close() in the parent process to close an unwanted
647writer. Consider this code:
648
08ab6dc8 649 my $pid = open(my $writer, "|-"); # fork open a kid
cf21866a 650 defined($pid) || die "first fork failed: $!";
c40e8e9b
SV
651 if ($pid) {
652 if (my $sub_pid = fork()) {
cf21866a 653 defined($sub_pid) || die "second fork failed: $!";
08ab6dc8 654 close($writer) || die "couldn't close writer: $!";
cf21866a 655 # now do something else...
c40e8e9b
SV
656 }
657 else {
08ab6dc8 658 # first write to $writer
cf21866a
TC
659 # ...
660 # then when finished
08ab6dc8 661 close($writer) || die "couldn't close writer: $!";
cf21866a 662 exit(0);
c40e8e9b
SV
663 }
664 }
665 else {
cf21866a
TC
666 # first do something with STDIN, then
667 exit(0);
c40e8e9b
SV
668 }
669
08ab6dc8
DB
670In the example above, the true parent does not want to write to the $writer
671filehandle, so it closes it. However, because $writer was opened using
672C<open $fh, "|-">, it has a special behavior: closing it calls
cf21866a 673waitpid() (see L<perlfunc/waitpid>), which waits for the subprocess
c40e8e9b 674to exit. If the child process ends up waiting for something happening
cf21866a 675in the section marked "do something else", you have deadlock.
c40e8e9b 676
cf21866a 677This can also be a problem with intermediate subprocesses in more
c40e8e9b 678complicated code, which will call waitpid() on all open filehandles
cf21866a 679during global destruction--in no predictable order.
c40e8e9b
SV
680
681To solve this, you must manually use pipe(), fork(), and the form of
cf21866a 682open() which sets one file descriptor to another, as shown below:
c40e8e9b 683
08ab6dc8 684 pipe(my $reader, my $writer) || die "pipe failed: $!";
c43b1046 685 my $pid = fork();
08ab6dc8 686 defined($pid) || die "first fork failed: $!";
c40e8e9b 687 if ($pid) {
08ab6dc8 688 close $reader;
c40e8e9b 689 if (my $sub_pid = fork()) {
08ab6dc8
DB
690 defined($sub_pid) || die "first fork failed: $!";
691 close($writer) || die "can't close writer: $!";
c40e8e9b
SV
692 }
693 else {
08ab6dc8 694 # write to $writer...
cf21866a
TC
695 # ...
696 # then when finished
08ab6dc8 697 close($writer) || die "can't close writer: $!";
cf21866a 698 exit(0);
c40e8e9b 699 }
08ab6dc8 700 # write to $writer...
c40e8e9b
SV
701 }
702 else {
08ab6dc8
DB
703 open(STDIN, "<&", $reader) || die "can't reopen STDIN: $!";
704 close($writer) || die "can't close writer: $!";
c40e8e9b 705 # do something...
cf21866a 706 exit(0);
c40e8e9b
SV
707 }
708
cf21866a
TC
709Since Perl 5.8.0, you can also use the list form of C<open> for pipes.
710This is preferred when you wish to avoid having the shell interpret
711metacharacters that may be in your command string.
307eac13 712
cf21866a 713So for example, instead of using:
307eac13 714
08ab6dc8 715 open(my $ps_pipe, "-|", "ps aux") || die "can't open ps pipe: $!";
307eac13 716
cf21866a 717One would use either of these:
4633a7c4 718
08ab6dc8
DB
719 open(my $ps_pipe, "-|", "ps", "aux")
720 || die "can't open ps pipe: $!";
c40e8e9b 721
c43b1046 722 my @ps_args = qw[ ps aux ];
08ab6dc8
DB
723 open(my $ps_pipe, "-|", @ps_args)
724 || die "can't open @ps_args|: $!";
c40e8e9b 725
08ab6dc8 726Because there are more than three arguments to open(), it forks the ps(1)
cf21866a 727command I<without> spawning a shell, and reads its standard output via the
08ab6dc8 728C<$ps_pipe> filehandle. The corresponding syntax to I<write> to command
82f82fdb 729pipes is to use C<"|-"> in place of C<"-|">.
c40e8e9b 730
cf21866a
TC
731This was admittedly a rather silly example, because you're using string
732literals whose content is perfectly safe. There is therefore no cause to
faa783ac 733resort to the harder-to-read, multi-argument form of pipe open(). However,
cf21866a
TC
734whenever you cannot be assured that the program arguments are free of shell
735metacharacters, the fancier form of open() should be used. For example:
c40e8e9b 736
c43b1046 737 my @grep_args = ("egrep", "-i", $some_pattern, @many_files);
08ab6dc8 738 open(my $grep_pipe, "-|", @grep_args)
cf21866a
TC
739 || die "can't open @grep_args|: $!";
740
741Here the multi-argument form of pipe open() is preferred because the
742pattern and indeed even the filenames themselves might hold metacharacters.
743
cf21866a
TC
744=head2 Avoiding Pipe Deadlocks
745
746Whenever you have more than one subprocess, you must be careful that each
747closes whichever half of any pipes created for interprocess communication
748it is not using. This is because any child process reading from the pipe
749and expecting an EOF will never receive it, and therefore never exit. A
750single process closing a pipe is not enough to close it; the last process
751with the pipe open must close it for it to read EOF.
752
753Certain built-in Unix features help prevent this most of the time. For
754instance, filehandles have a "close on exec" flag, which is set I<en masse>
755under control of the C<$^F> variable. This is so any filehandles you
756didn't explicitly route to the STDIN, STDOUT or STDERR of a child
757I<program> will be automatically closed.
758
759Always explicitly and immediately call close() on the writable end of any
760pipe, unless that process is actually writing to it. Even if you don't
761explicitly call close(), Perl will still close() all filehandles during
762global destruction. As previously discussed, if those filehandles have
763been opened with Safe Pipe Open, this will result in calling waitpid(),
764which may again deadlock.
c40e8e9b 765
7b05b7e3 766=head2 Bidirectional Communication with Another Process
4633a7c4
LW
767
768While this works reasonably well for unidirectional communication, what
cf21866a 769about bidirectional communication? The most obvious approach doesn't work:
4633a7c4 770
cf21866a 771 # THIS DOES NOT WORK!!
08ab6dc8 772 open(my $prog_for_reading_and_writing, "| some program |")
4633a7c4 773
cf21866a
TC
774If you forget to C<use warnings>, you'll miss out entirely on the
775helpful diagnostic message:
4633a7c4
LW
776
777 Can't do bidirectional pipe at -e line 1.
778
cf21866a 779If you really want to, you can use the standard open2() from the
c43b1046
DB
780L<IPC::Open2> module to catch both ends. There's also an open3() in
781L<IPC::Open3> for tridirectional I/O so you can also catch your child's
cf21866a
TC
782STDERR, but doing so would then require an awkward select() loop and
783wouldn't allow you to use normal Perl input operations.
4633a7c4
LW
784
785If you look at its source, you'll see that open2() uses low-level
cf21866a
TC
786primitives like the pipe() and exec() syscalls to create all the
787connections. Although it might have been more efficient by using
788socketpair(), this would have been even less portable than it already
789is. The open2() and open3() functions are unlikely to work anywhere
790except on a Unix system, or at least one purporting POSIX compliance.
791
792=for TODO
793Hold on, is this even true? First it says that socketpair() is avoided
82f82fdb 794for portability, but then it says it probably won't work except on
cf21866a 795Unixy systems anyway. Which one of those is true?
4633a7c4
LW
796
797Here's an example of using open2():
798
4633a7c4 799 use IPC::Open2;
c43b1046 800 my $pid = open2(my $reader, my $writer, "cat -un");
08ab6dc8 801 print $writer "stuff\n";
c43b1046
DB
802 my $got = <$reader>;
803 waitpid $pid, 0;
4633a7c4 804
cf21866a 805The problem with this is that buffering is really going to ruin your
08ab6dc8 806day. Even though your C<$writer> filehandle is auto-flushed so the process
cf21866a
TC
807on the other end gets your data in a timely manner, you can't usually do
808anything to force that process to give its data to you in a similarly quick
809fashion. In this special case, we could actually so, because we gave
810I<cat> a B<-u> flag to make it unbuffered. But very few commands are
811designed to operate over pipes, so this seldom works unless you yourself
812wrote the program on the other end of the double-ended pipe.
813
814A solution to this is to use a library which uses pseudottys to make your
815program behave more reasonably. This way you don't have to have control
816over the source code of the program you're using. The C<Expect> module
817from CPAN also addresses this kind of thing. This module requires two
818other modules from CPAN, C<IO::Pty> and C<IO::Stty>. It sets up a pseudo
819terminal to interact with programs that insist on talking to the terminal
820device driver. If your system is supported, this may be your best bet.
c8db1d39 821
5a964f20
TC
822=head2 Bidirectional Communication with Yourself
823
cf21866a
TC
824If you want, you may make low-level pipe() and fork() syscalls to stitch
825this together by hand. This example only talks to itself, but you could
826reopen the appropriate handles to STDIN and STDOUT and call other processes.
827(The following example lacks proper error checking.)
5a964f20 828
c43b1046 829 #!/usr/bin/perl
e46aa1dd
KW
830 # pipe1 - bidirectional communication using two pipe pairs
831 # designed for the socketpair-challenged
c43b1046
DB
832 use strict;
833 use warnings;
834 use IO::Handle; # enable autoflush method before Perl 5.14
08ab6dc8
DB
835 pipe(my $parent_rdr, my $child_wtr); # XXX: check failure?
836 pipe(my $child_rdr, my $parent_wtr); # XXX: check failure?
837 $child_wtr->autoflush(1);
838 $parent_wtr->autoflush(1);
e46aa1dd
KW
839
840 if ($pid = fork()) {
08ab6dc8
DB
841 close $parent_rdr;
842 close $parent_wtr;
843 print $child_wtr "Parent Pid $$ is sending this\n";
c43b1046 844 chomp(my $line = <$child_rdr>);
e46aa1dd 845 print "Parent Pid $$ just read this: '$line'\n";
08ab6dc8 846 close $child_rdr; close $child_wtr;
e46aa1dd
KW
847 waitpid($pid, 0);
848 } else {
849 die "cannot fork: $!" unless defined $pid;
08ab6dc8
DB
850 close $child_rdr;
851 close $child_wtr;
c43b1046 852 chomp(my $line = <$parent_rdr>);
e46aa1dd 853 print "Child Pid $$ just read this: '$line'\n";
08ab6dc8
DB
854 print $parent_wtr "Child Pid $$ is sending this\n";
855 close $parent_rdr;
856 close $parent_wtr;
e46aa1dd
KW
857 exit(0);
858 }
5a964f20 859
a11adca0 860But you don't actually have to make two pipe calls. If you
5a964f20
TC
861have the socketpair() system call, it will do this all for you.
862
c43b1046 863 #!/usr/bin/perl
e46aa1dd
KW
864 # pipe2 - bidirectional communication using socketpair
865 # "the best ones always go both ways"
866
c43b1046
DB
867 use strict;
868 use warnings;
e46aa1dd 869 use Socket;
c43b1046 870 use IO::Handle; # enable autoflush method before Perl 5.14
e46aa1dd
KW
871
872 # We say AF_UNIX because although *_LOCAL is the
873 # POSIX 1003.1g form of the constant, many machines
874 # still don't have it.
08ab6dc8 875 socketpair(my $child, my $parent, AF_UNIX, SOCK_STREAM, PF_UNSPEC)
e46aa1dd
KW
876 || die "socketpair: $!";
877
08ab6dc8
DB
878 $child->autoflush(1);
879 $parent->autoflush(1);
e46aa1dd
KW
880
881 if ($pid = fork()) {
08ab6dc8
DB
882 close $parent;
883 print $child "Parent Pid $$ is sending this\n";
c43b1046 884 chomp(my $line = <$child>);
e46aa1dd 885 print "Parent Pid $$ just read this: '$line'\n";
08ab6dc8 886 close $child;
e46aa1dd
KW
887 waitpid($pid, 0);
888 } else {
889 die "cannot fork: $!" unless defined $pid;
08ab6dc8 890 close $child;
c43b1046 891 chomp(my $line = <$parent>);
e46aa1dd 892 print "Child Pid $$ just read this: '$line'\n";
08ab6dc8
DB
893 print $parent "Child Pid $$ is sending this\n";
894 close $parent;
e46aa1dd
KW
895 exit(0);
896 }
5a964f20 897
4633a7c4 898=head1 Sockets: Client/Server Communication
a0d0e21e 899
cf21866a
TC
900While not entirely limited to Unix-derived operating systems (e.g., WinSock
901on PCs provides socket support, as do some VMS libraries), you might not have
902sockets on your system, in which case this section probably isn't going to
903do you much good. With sockets, you can do both virtual circuits like TCP
904streams and datagrams like UDP packets. You may be able to do even more
4633a7c4
LW
905depending on your system.
906
cf21866a 907The Perl functions for dealing with sockets have the same names as
4633a7c4 908the corresponding system calls in C, but their arguments tend to differ
cf21866a 909for two reasons. First, Perl filehandles work differently than C file
4633a7c4
LW
910descriptors. Second, Perl already knows the length of its strings, so you
911don't need to pass that information.
a0d0e21e 912
cf21866a
TC
913One of the major problems with ancient, antemillennial socket code in Perl
914was that it used hard-coded values for some of the constants, which
915severely hurt portability. If you ever see code that does anything like
82f82fdb 916explicitly setting C<$AF_INET = 2>, you know you're in for big trouble.
c43b1046 917An immeasurably superior approach is to use the L<Socket> module, which more
cf21866a 918reliably grants access to the various constants and functions you'll need.
a0d0e21e 919
68dc0745
PP
920If you're not writing a server/client for an existing protocol like
921NNTP or SMTP, you should give some thought to how your server will
922know when the client has finished talking, and vice-versa. Most
923protocols are based on one-line messages and responses (so one party
4a6725af 924knows the other has finished when a "\n" is received) or multi-line
68dc0745
PP
925messages and responses that end with a period on an empty line
926("\n.\n" terminates a message/response).
927
5a964f20
TC
928=head2 Internet Line Terminators
929
930The Internet line terminator is "\015\012". Under ASCII variants of
931Unix, that could usually be written as "\r\n", but under other systems,
932"\r\n" might at times be "\015\015\012", "\012\012\015", or something
933completely different. The standards specify writing "\015\012" to be
934conformant (be strict in what you provide), but they also recommend
cf21866a 935accepting a lone "\012" on input (be lenient in what you require).
5a964f20 936We haven't always been very good about that in the code in this manpage,
82f82fdb 937but unless you're on a Mac from way back in its pre-Unix dark ages, you'll
cf21866a 938probably be ok.
5a964f20 939
4633a7c4 940=head2 Internet TCP Clients and Servers
a0d0e21e 941
4633a7c4
LW
942Use Internet-domain sockets when you want to do client-server
943communication that might extend to machines outside of your own system.
944
945Here's a sample TCP client using Internet-domain sockets:
946
c43b1046 947 #!/usr/bin/perl
4633a7c4 948 use strict;
c43b1046 949 use warnings;
4633a7c4 950 use Socket;
4633a7c4 951
c43b1046
DB
952 my $remote = shift || "localhost";
953 my $port = shift || 2345; # random port
cf21866a 954 if ($port =~ /\D/) { $port = getservbyname($port, "tcp") }
4633a7c4 955 die "No port" unless $port;
c43b1046
DB
956 my $iaddr = inet_aton($remote) || die "no host: $remote";
957 my $paddr = sockaddr_in($port, $iaddr);
4633a7c4 958
c43b1046 959 my $proto = getprotobyname("tcp");
08ab6dc8
DB
960 socket(my $sock, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
961 connect($sock, $paddr) || die "connect: $!";
c43b1046 962 while (my $line = <$sock>) {
322c2516 963 print $line;
54310121 964 }
4633a7c4 965
08ab6dc8 966 close ($sock) || die "close: $!";
cf21866a 967 exit(0);
4633a7c4
LW
968
969And here's a corresponding server to go along with it. We'll
cf21866a 970leave the address as C<INADDR_ANY> so that the kernel can choose
54310121 971the appropriate interface on multihomed hosts. If you want sit
c07a80fd 972on a particular interface (like the external side of a gateway
cf21866a 973or firewall machine), fill this in with your real address instead.
c07a80fd 974
c43b1046 975 #!/usr/bin/perl -T
e46aa1dd 976 use strict;
c43b1046 977 use warnings;
e46aa1dd
KW
978 BEGIN { $ENV{PATH} = "/usr/bin:/bin" }
979 use Socket;
980 use Carp;
981 my $EOL = "\015\012";
c07a80fd 982
e46aa1dd 983 sub logmsg { print "$0 $$: @_ at ", scalar localtime(), "\n" }
c07a80fd 984
e46aa1dd
KW
985 my $port = shift || 2345;
986 die "invalid port" unless $port =~ /^ \d+ $/x;
51ee6500 987
e46aa1dd 988 my $proto = getprotobyname("tcp");
6a3992aa 989
08ab6dc8
DB
990 socket(my $server, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
991 setsockopt($server, SOL_SOCKET, SO_REUSEADDR, pack("l", 1))
992 || die "setsockopt: $!";
993 bind($server, sockaddr_in($port, INADDR_ANY)) || die "bind: $!";
994 listen($server, SOMAXCONN) || die "listen: $!";
c07a80fd 995
e46aa1dd 996 logmsg "server started on port $port";
c07a80fd 997
c43b1046 998 for (my $paddr; $paddr = accept(my $client, $server); close $client) {
e46aa1dd
KW
999 my($port, $iaddr) = sockaddr_in($paddr);
1000 my $name = gethostbyaddr($iaddr, AF_INET);
c07a80fd 1001
e46aa1dd
KW
1002 logmsg "connection from $name [",
1003 inet_ntoa($iaddr), "]
1004 at port $port";
c07a80fd 1005
08ab6dc8 1006 print $client "Hello there, $name, it's now ",
e46aa1dd
KW
1007 scalar localtime(), $EOL;
1008 }
c07a80fd 1009
5e220227 1010And here's a multitasking version. It's multitasked in that
cf21866a 1011like most typical servers, it spawns (fork()s) a slave server to
c07a80fd
PP
1012handle the client request so that the master server can quickly
1013go back to service a new client.
4633a7c4 1014
c43b1046 1015 #!/usr/bin/perl -T
e46aa1dd 1016 use strict;
c43b1046 1017 use warnings;
e46aa1dd
KW
1018 BEGIN { $ENV{PATH} = "/usr/bin:/bin" }
1019 use Socket;
1020 use Carp;
1021 my $EOL = "\015\012";
a0d0e21e 1022
e46aa1dd
KW
1023 sub spawn; # forward declaration
1024 sub logmsg { print "$0 $$: @_ at ", scalar localtime(), "\n" }
a0d0e21e 1025
e46aa1dd
KW
1026 my $port = shift || 2345;
1027 die "invalid port" unless $port =~ /^ \d+ $/x;
51ee6500 1028
e46aa1dd 1029 my $proto = getprotobyname("tcp");
54310121 1030
08ab6dc8
DB
1031 socket(my $server, PF_INET, SOCK_STREAM, $proto) || die "socket: $!";
1032 setsockopt($server, SOL_SOCKET, SO_REUSEADDR, pack("l", 1))
1033 || die "setsockopt: $!";
1034 bind($server, sockaddr_in($port, INADDR_ANY)) || die "bind: $!";
1035 listen($server, SOMAXCONN) || die "listen: $!";
a0d0e21e 1036
e46aa1dd 1037 logmsg "server started on port $port";
a0d0e21e 1038
e46aa1dd 1039 my $waitedpid = 0;
c5ae6365 1040
e46aa1dd
KW
1041 use POSIX ":sys_wait_h";
1042 use Errno;
c5ae6365 1043
e46aa1dd
KW
1044 sub REAPER {
1045 local $!; # don't let waitpid() overwrite current error
1046 while ((my $pid = waitpid(-1, WNOHANG)) > 0 && WIFEXITED($?)) {
1047 logmsg "reaped $waitedpid" . ($? ? " with exit $?" : "");
1048 }
1049 $SIG{CHLD} = \&REAPER; # loathe SysV
1050 }
c5ae6365 1051
e46aa1dd
KW
1052 $SIG{CHLD} = \&REAPER;
1053
1054 while (1) {
c43b1046 1055 my $paddr = accept(my $client, $server) || do {
e46aa1dd
KW
1056 # try again if accept() returned because got a signal
1057 next if $!{EINTR};
1058 die "accept: $!";
1059 };
1060 my ($port, $iaddr) = sockaddr_in($paddr);
1061 my $name = gethostbyaddr($iaddr, AF_INET);
1062
1063 logmsg "connection from $name [",
1064 inet_ntoa($iaddr),
1065 "] at port $port";
1066
08ab6dc8 1067 spawn $client, sub {
e46aa1dd
KW
1068 $| = 1;
1069 print "Hello there, $name, it's now ",
1070 scalar localtime(),
1071 $EOL;
1072 exec "/usr/games/fortune" # XXX: "wrong" line terminators
1073 or confess "can't exec fortune: $!";
1074 };
08ab6dc8 1075 close $client;
e46aa1dd 1076 }
a0d0e21e 1077
e46aa1dd 1078 sub spawn {
08ab6dc8 1079 my $client = shift;
e46aa1dd 1080 my $coderef = shift;
c5ae6365 1081
e46aa1dd 1082 unless (@_ == 0 && $coderef && ref($coderef) eq "CODE") {
08ab6dc8 1083 confess "usage: spawn CLIENT CODEREF";
e46aa1dd 1084 }
c5ae6365 1085
e46aa1dd
KW
1086 my $pid;
1087 unless (defined($pid = fork())) {
1088 logmsg "cannot fork: $!";
1089 return;
1090 }
1091 elsif ($pid) {
1092 logmsg "begat $pid";
1093 return; # I'm the parent
1094 }
1095 # else I'm the child -- go spawn
c5ae6365 1096
08ab6dc8
DB
1097 open(STDIN, "<&", $client) || die "can't dup client to stdin";
1098 open(STDOUT, ">&", $client) || die "can't dup client to stdout";
1099 ## open(STDERR, ">&", STDOUT) || die "can't dup stdout to stderr";
e46aa1dd
KW
1100 exit($coderef->());
1101 }
4633a7c4 1102
c5ae6365
AW
1103This server takes the trouble to clone off a child version via fork()
1104for each incoming request. That way it can handle many requests at
1105once, which you might not always want. Even if you don't fork(), the
1106listen() will allow that many pending connections. Forking servers
1107have to be particularly careful about cleaning up their dead children
1108(called "zombies" in Unix parlance), because otherwise you'll quickly
1109fill up your process table. The REAPER subroutine is used here to
1110call waitpid() for any child processes that have finished, thereby
1111ensuring that they terminate cleanly and don't join the ranks of the
1112living dead.
1113
1114Within the while loop we call accept() and check to see if it returns
cf21866a
TC
1115a false value. This would normally indicate a system error needs
1116to be reported. However, the introduction of safe signals (see
e6aa8b84 1117L</Deferred Signals (Safe Signals)> above) in Perl 5.8.0 means that
cf21866a
TC
1118accept() might also be interrupted when the process receives a signal.
1119This typically happens when one of the forked subprocesses exits and
82f82fdb 1120notifies the parent process with a CHLD signal.
c5ae6365 1121
cf21866a
TC
1122If accept() is interrupted by a signal, $! will be set to EINTR.
1123If this happens, we can safely continue to the next iteration of
c5ae6365 1124the loop and another call to accept(). It is important that your
82f82fdb 1125signal handling code not modify the value of $!, or else this test
cf21866a
TC
1126will likely fail. In the REAPER subroutine we create a local version
1127of $! before calling waitpid(). When waitpid() sets $! to ECHILD as
82f82fdb 1128it inevitably does when it has no more children waiting, it
cf21866a 1129updates the local copy and leaves the original unchanged.
4633a7c4 1130
cf21866a 1131You should use the B<-T> flag to enable taint checking (see L<perlsec>)
4633a7c4 1132even if we aren't running setuid or setgid. This is always a good idea
cf21866a 1133for servers or any program run on behalf of someone else (like CGI
4633a7c4
LW
1134scripts), because it lessens the chances that people from the outside will
1135be able to compromise your system.
1136
1137Let's look at another TCP client. This one connects to the TCP "time"
1138service on a number of different machines and shows how far their clocks
1139differ from the system on which it's being run:
1140
c43b1046 1141 #!/usr/bin/perl
4633a7c4 1142 use strict;
c43b1046 1143 use warnings;
4633a7c4
LW
1144 use Socket;
1145
cf21866a
TC
1146 my $SECS_OF_70_YEARS = 2208988800;
1147 sub ctime { scalar localtime(shift() || time()) }
4633a7c4 1148
cf21866a
TC
1149 my $iaddr = gethostbyname("localhost");
1150 my $proto = getprotobyname("tcp");
1151 my $port = getservbyname("time", "tcp");
4633a7c4 1152 my $paddr = sockaddr_in(0, $iaddr);
4633a7c4
LW
1153
1154 $| = 1;
cf21866a 1155 printf "%-24s %8s %s\n", "localhost", 0, ctime();
4633a7c4 1156
c43b1046 1157 foreach my $host (@ARGV) {
322c2516
SF
1158 printf "%-24s ", $host;
1159 my $hisiaddr = inet_aton($host) || die "unknown host";
1160 my $hispaddr = sockaddr_in($port, $hisiaddr);
08ab6dc8 1161 socket(my $socket, PF_INET, SOCK_STREAM, $proto)
cf21866a 1162 || die "socket: $!";
08ab6dc8 1163 connect($socket, $hispaddr) || die "connect: $!";
cf21866a 1164 my $rtime = pack("C4", ());
08ab6dc8
DB
1165 read($socket, $rtime, 4);
1166 close($socket);
cf21866a
TC
1167 my $histime = unpack("N", $rtime) - $SECS_OF_70_YEARS;
1168 printf "%8d %s\n", $histime - time(), ctime($histime);
a0d0e21e
LW
1169 }
1170
4633a7c4
LW
1171=head2 Unix-Domain TCP Clients and Servers
1172
a2eb9003 1173That's fine for Internet-domain clients and servers, but what about local
4633a7c4
LW
1174communications? While you can use the same setup, sometimes you don't
1175want to. Unix-domain sockets are local to the current host, and are often
54310121 1176used internally to implement pipes. Unlike Internet domain sockets, Unix
4633a7c4
LW
1177domain sockets can show up in the file system with an ls(1) listing.
1178
5a964f20 1179 % ls -l /dev/log
4633a7c4 1180 srw-rw-rw- 1 root 0 Oct 31 07:23 /dev/log
a0d0e21e 1181
4633a7c4
LW
1182You can test for these with Perl's B<-S> file test:
1183
cf21866a 1184 unless (-S "/dev/log") {
322c2516 1185 die "something's wicked with the log system";
54310121 1186 }
4633a7c4
LW
1187
1188Here's a sample Unix-domain client:
1189
c43b1046 1190 #!/usr/bin/perl
4633a7c4
LW
1191 use Socket;
1192 use strict;
c43b1046 1193 use warnings;
4633a7c4 1194
c43b1046 1195 my $rendezvous = shift || "catsock";
08ab6dc8
DB
1196 socket(my $sock, PF_UNIX, SOCK_STREAM, 0) || die "socket: $!";
1197 connect($sock, sockaddr_un($rendezvous)) || die "connect: $!";
c43b1046 1198 while (defined(my $line = <$sock>)) {
322c2516 1199 print $line;
54310121 1200 }
cf21866a 1201 exit(0);
4633a7c4 1202
5a964f20
TC
1203And here's a corresponding server. You don't have to worry about silly
1204network terminators here because Unix domain sockets are guaranteed
1205to be on the localhost, and thus everything works right.
4633a7c4 1206
c43b1046 1207 #!/usr/bin/perl -T
4633a7c4 1208 use strict;
c43b1046 1209 use warnings;
4633a7c4
LW
1210 use Socket;
1211 use Carp;
1212
cf21866a 1213 BEGIN { $ENV{PATH} = "/usr/bin:/bin" }
5865a7df 1214 sub spawn; # forward declaration
cf21866a 1215 sub logmsg { print "$0 $$: @_ at ", scalar localtime(), "\n" }
4633a7c4 1216
cf21866a 1217 my $NAME = "catsock";
4633a7c4 1218 my $uaddr = sockaddr_un($NAME);
cf21866a 1219 my $proto = getprotobyname("tcp");
4633a7c4 1220
08ab6dc8 1221 socket(my $server, PF_UNIX, SOCK_STREAM, 0) || die "socket: $!";
4633a7c4 1222 unlink($NAME);
08ab6dc8
DB
1223 bind ($server, $uaddr) || die "bind: $!";
1224 listen($server, SOMAXCONN) || die "listen: $!";
4633a7c4
LW
1225
1226 logmsg "server started on $NAME";
1227
5a964f20
TC
1228 my $waitedpid;
1229
816229cf 1230 use POSIX ":sys_wait_h";
5a964f20 1231 sub REAPER {
322c2516 1232 my $child;
cf21866a
TC
1233 while (($waitedpid = waitpid(-1, WNOHANG)) > 0) {
1234 logmsg "reaped $waitedpid" . ($? ? " with exit $?" : "");
322c2516
SF
1235 }
1236 $SIG{CHLD} = \&REAPER; # loathe SysV
5a964f20
TC
1237 }
1238
4633a7c4
LW
1239 $SIG{CHLD} = \&REAPER;
1240
5a964f20 1241
54310121 1242 for ( $waitedpid = 0;
08ab6dc8
DB
1243 accept(my $client, $server) || $waitedpid;
1244 $waitedpid = 0, close $client)
4633a7c4 1245 {
322c2516
SF
1246 next if $waitedpid;
1247 logmsg "connection on $NAME";
08ab6dc8 1248 spawn $client, sub {
cf21866a
TC
1249 print "Hello there, it's now ", scalar localtime(), "\n";
1250 exec("/usr/games/fortune") || die "can't exec fortune: $!";
322c2516 1251 };
54310121 1252 }
4633a7c4 1253
5865a7df 1254 sub spawn {
08ab6dc8 1255 my $client = shift();
cf21866a 1256 my $coderef = shift();
322c2516 1257
cf21866a 1258 unless (@_ == 0 && $coderef && ref($coderef) eq "CODE") {
08ab6dc8 1259 confess "usage: spawn CLIENT CODEREF";
322c2516
SF
1260 }
1261
1262 my $pid;
cf21866a 1263 unless (defined($pid = fork())) {
322c2516
SF
1264 logmsg "cannot fork: $!";
1265 return;
82f82fdb 1266 }
cf21866a 1267 elsif ($pid) {
322c2516
SF
1268 logmsg "begat $pid";
1269 return; # I'm the parent
82f82fdb 1270 }
cf21866a
TC
1271 else {
1272 # I'm the child -- go spawn
322c2516 1273 }
322c2516 1274
08ab6dc8
DB
1275 open(STDIN, "<&", $client)
1276 || die "can't dup client to stdin";
1277 open(STDOUT, ">&", $client)
1278 || die "can't dup client to stdout";
1279 ## open(STDERR, ">&", STDOUT)
1280 ## || die "can't dup stdout to stderr";
cf21866a 1281 exit($coderef->());
5865a7df
NC
1282 }
1283
4633a7c4
LW
1284As you see, it's remarkably similar to the Internet domain TCP server, so
1285much so, in fact, that we've omitted several duplicate functions--spawn(),
cf21866a 1286logmsg(), ctime(), and REAPER()--which are the same as in the other server.
4633a7c4
LW
1287
1288So why would you ever want to use a Unix domain socket instead of a
1289simpler named pipe? Because a named pipe doesn't give you sessions. You
1290can't tell one process's data from another's. With socket programming,
cf21866a 1291you get a separate session for each client; that's why accept() takes two
4633a7c4
LW
1292arguments.
1293
cf21866a
TC
1294For example, let's say that you have a long-running database server daemon
1295that you want folks to be able to access from the Web, but only
4633a7c4
LW
1296if they go through a CGI interface. You'd have a small, simple CGI
1297program that does whatever checks and logging you feel like, and then acts
1298as a Unix-domain client and connects to your private server.
1299
7b05b7e3
TC
1300=head1 TCP Clients with IO::Socket
1301
1302For those preferring a higher-level interface to socket programming, the
e6aa8b84
BF
1303IO::Socket module provides an object-oriented approach. If for some reason
1304you lack this module, you can just fetch IO::Socket from CPAN, where you'll also
cf21866a
TC
1305find modules providing easy interfaces to the following systems: DNS, FTP,
1306Ident (RFC 931), NIS and NISPlus, NNTP, Ping, POP3, SMTP, SNMP, SSLeay,
1307Telnet, and Time--to name just a few.
7b05b7e3
TC
1308
1309=head2 A Simple Client
1310
1311Here's a client that creates a TCP connection to the "daytime"
1312service at port 13 of the host name "localhost" and prints out everything
1313that the server there cares to provide.
1314
c43b1046
DB
1315 #!/usr/bin/perl
1316 use strict;
1317 use warnings;
7b05b7e3 1318 use IO::Socket;
c43b1046 1319 my $remote = IO::Socket::INET->new(
322c2516
SF
1320 Proto => "tcp",
1321 PeerAddr => "localhost",
1322 PeerPort => "daytime(13)",
1323 )
e46aa1dd 1324 || die "can't connect to daytime service on localhost";
cf21866a 1325 while (<$remote>) { print }
7b05b7e3
TC
1326
1327When you run this program, you should get something back that
1328looks like this:
1329
1330 Wed May 14 08:40:46 MDT 1997
1331
cf21866a 1332Here are what those parameters to the new() constructor mean:
7b05b7e3 1333
13a2d996 1334=over 4
7b05b7e3
TC
1335
1336=item C<Proto>
1337
1338This is which protocol to use. In this case, the socket handle returned
1339will be connected to a TCP socket, because we want a stream-oriented
1340connection, that is, one that acts pretty much like a plain old file.
1341Not all sockets are this of this type. For example, the UDP protocol
1342can be used to make a datagram socket, used for message-passing.
1343
1344=item C<PeerAddr>
1345
1346This is the name or Internet address of the remote host the server is
1347running on. We could have specified a longer name like C<"www.perl.com">,
cf21866a 1348or an address like C<"207.171.7.72">. For demonstration purposes, we've
7b05b7e3
TC
1349used the special hostname C<"localhost">, which should always mean the
1350current machine you're running on. The corresponding Internet address
cf21866a 1351for localhost is C<"127.0.0.1">, if you'd rather use that.
7b05b7e3
TC
1352
1353=item C<PeerPort>
1354
1355This is the service name or port number we'd like to connect to.
1356We could have gotten away with using just C<"daytime"> on systems with a
1357well-configured system services file,[FOOTNOTE: The system services file
cf21866a
TC
1358is found in I</etc/services> under Unixy systems.] but here we've specified the
1359port number (13) in parentheses. Using just the number would have also
1360worked, but numeric literals make careful programmers nervous.
7b05b7e3
TC
1361
1362=back
1363
7b05b7e3
TC
1364=head2 A Webget Client
1365
1366Here's a simple client that takes a remote host to fetch a document
cf21866a 1367from, and then a list of files to get from that host. This is a
7b05b7e3
TC
1368more interesting client than the previous one because it first sends
1369something to the server before fetching the server's response.
1370
c43b1046
DB
1371 #!/usr/bin/perl
1372 use strict;
1373 use warnings;
7b05b7e3 1374 use IO::Socket;
cf21866a 1375 unless (@ARGV > 1) { die "usage: $0 host url ..." }
c43b1046
DB
1376 my $host = shift(@ARGV);
1377 my $EOL = "\015\012";
1378 my $BLANK = $EOL x 2;
cf21866a 1379 for my $document (@ARGV) {
c43b1046
DB
1380 my $remote = IO::Socket::INET->new( Proto => "tcp",
1381 PeerAddr => $host,
1382 PeerPort => "http(80)",
cf21866a 1383 ) || die "cannot connect to httpd on $host";
322c2516
SF
1384 $remote->autoflush(1);
1385 print $remote "GET $document HTTP/1.0" . $BLANK;
1386 while ( <$remote> ) { print }
1387 close $remote;
7b05b7e3
TC
1388 }
1389
cf21866a
TC
1390The web server handling the HTTP service is assumed to be at
1391its standard port, number 80. If the server you're trying to
1392connect to is at a different port, like 1080 or 8080, you should specify it
c47ff5f1 1393as the named-parameter pair, C<< PeerPort => 8080 >>. The C<autoflush>
7b05b7e3 1394method is used on the socket because otherwise the system would buffer
cf21866a
TC
1395up the output we sent it. (If you're on a prehistoric Mac, you'll also
1396need to change every C<"\n"> in your code that sends data over the network
1397to be a C<"\015\012"> instead.)
7b05b7e3
TC
1398
1399Connecting to the server is only the first part of the process: once you
1400have the connection, you have to use the server's language. Each server
1401on the network has its own little command language that it expects as
1402input. The string that we send to the server starting with "GET" is in
1403HTTP syntax. In this case, we simply request each specified document.
1404Yes, we really are making a new connection for each document, even though
1405it's the same host. That's the way you always used to have to speak HTTP.
1406Recent versions of web browsers may request that the remote server leave
1407the connection open a little while, but the server doesn't have to honor
1408such a request.
1409
1410Here's an example of running that program, which we'll call I<webget>:
1411
5a964f20 1412 % webget www.perl.com /guanaco.html
7b05b7e3
TC
1413 HTTP/1.1 404 File Not Found
1414 Date: Thu, 08 May 1997 18:02:32 GMT
1415 Server: Apache/1.2b6
1416 Connection: close
1417 Content-type: text/html
1418
1419 <HEAD><TITLE>404 File Not Found</TITLE></HEAD>
1420 <BODY><H1>File Not Found</H1>
1421 The requested URL /guanaco.html was not found on this server.<P>
1422 </BODY>
1423
1424Ok, so that's not very interesting, because it didn't find that
1425particular document. But a long response wouldn't have fit on this page.
1426
cf21866a 1427For a more featureful version of this program, you should look to
7b05b7e3
TC
1428the I<lwp-request> program included with the LWP modules from CPAN.
1429
1430=head2 Interactive Client with IO::Socket
1431
1432Well, that's all fine if you want to send one command and get one answer,
1433but what about setting up something fully interactive, somewhat like
1434the way I<telnet> works? That way you can type a line, get the answer,
1435type a line, get the answer, etc.
1436
1437This client is more complicated than the two we've done so far, but if
1438you're on a system that supports the powerful C<fork> call, the solution
1439isn't that rough. Once you've made the connection to whatever service
1440you'd like to chat with, call C<fork> to clone your process. Each of
1441these two identical process has a very simple job to do: the parent
1442copies everything from the socket to standard output, while the child
1443simultaneously copies everything from standard input to the socket.
1444To accomplish the same thing using just one process would be I<much>
1445harder, because it's easier to code two processes to do one thing than it
1446is to code one process to do two things. (This keep-it-simple principle
5a964f20
TC
1447a cornerstones of the Unix philosophy, and good software engineering as
1448well, which is probably why it's spread to other systems.)
7b05b7e3
TC
1449
1450Here's the code:
1451
c43b1046 1452 #!/usr/bin/perl
7b05b7e3 1453 use strict;
c43b1046 1454 use warnings;
7b05b7e3 1455 use IO::Socket;
7b05b7e3
TC
1456
1457 unless (@ARGV == 2) { die "usage: $0 host port" }
c43b1046 1458 my ($host, $port) = @ARGV;
7b05b7e3
TC
1459
1460 # create a tcp connection to the specified host and port
c43b1046
DB
1461 my $handle = IO::Socket::INET->new(Proto => "tcp",
1462 PeerAddr => $host,
1463 PeerPort => $port)
cf21866a 1464 || die "can't connect to port $port on $host: $!";
7b05b7e3 1465
cf21866a 1466 $handle->autoflush(1); # so output gets there right away
7b05b7e3
TC
1467 print STDERR "[Connected to $host:$port]\n";
1468
1469 # split the program into two processes, identical twins
c43b1046 1470 die "can't fork: $!" unless defined(my $kidpid = fork());
7b05b7e3
TC
1471
1472 # the if{} block runs only in the parent process
1473 if ($kidpid) {
322c2516 1474 # copy the socket to standard output
c43b1046 1475 while (defined (my $line = <$handle>)) {
322c2516
SF
1476 print STDOUT $line;
1477 }
cf21866a 1478 kill("TERM", $kidpid); # send SIGTERM to child
7b05b7e3
TC
1479 }
1480 # the else{} block runs only in the child process
1481 else {
322c2516 1482 # copy standard input to the socket
c43b1046 1483 while (defined (my $line = <STDIN>)) {
322c2516
SF
1484 print $handle $line;
1485 }
cf21866a 1486 exit(0); # just in case
7b05b7e3
TC
1487 }
1488
1489The C<kill> function in the parent's C<if> block is there to send a
cf21866a 1490signal to our child process, currently running in the C<else> block,
7b05b7e3
TC
1491as soon as the remote server has closed its end of the connection.
1492
7b05b7e3
TC
1493If the remote server sends data a byte at time, and you need that
1494data immediately without waiting for a newline (which might not happen),
1495you may wish to replace the C<while> loop in the parent with the
1496following:
1497
1498 my $byte;
1499 while (sysread($handle, $byte, 1) == 1) {
322c2516 1500 print STDOUT $byte;
7b05b7e3
TC
1501 }
1502
1503Making a system call for each byte you want to read is not very efficient
1504(to put it mildly) but is the simplest to explain and works reasonably
1505well.
1506
1507=head1 TCP Servers with IO::Socket
1508
5a964f20 1509As always, setting up a server is little bit more involved than running a client.
7b05b7e3
TC
1510The model is that the server creates a special kind of socket that
1511does nothing but listen on a particular port for incoming connections.
c47ff5f1 1512It does this by calling the C<< IO::Socket::INET->new() >> method with
7b05b7e3
TC
1513slightly different arguments than the client did.
1514
13a2d996 1515=over 4
7b05b7e3
TC
1516
1517=item Proto
1518
1519This is which protocol to use. Like our clients, we'll
1520still specify C<"tcp"> here.
1521
1522=item LocalPort
1523
1524We specify a local
1525port in the C<LocalPort> argument, which we didn't do for the client.
1526This is service name or port number for which you want to be the
1527server. (Under Unix, ports under 1024 are restricted to the
1528superuser.) In our sample, we'll use port 9000, but you can use
1529any port that's not currently in use on your system. If you try
1530to use one already in used, you'll get an "Address already in use"
19799a22 1531message. Under Unix, the C<netstat -a> command will show
7b05b7e3
TC
1532which services current have servers.
1533
1534=item Listen
1535
1536The C<Listen> parameter is set to the maximum number of
1537pending connections we can accept until we turn away incoming clients.
1538Think of it as a call-waiting queue for your telephone.
1539The low-level Socket module has a special symbol for the system maximum, which
1540is SOMAXCONN.
1541
1542=item Reuse
1543
1544The C<Reuse> parameter is needed so that we restart our server
1545manually without waiting a few minutes to allow system buffers to
1546clear out.
1547
1548=back
1549
1550Once the generic server socket has been created using the parameters
1551listed above, the server then waits for a new client to connect
d1be9408
JF
1552to it. The server blocks in the C<accept> method, which eventually accepts a
1553bidirectional connection from the remote client. (Make sure to autoflush
7b05b7e3
TC
1554this handle to circumvent buffering.)
1555
1556To add to user-friendliness, our server prompts the user for commands.
1557Most servers don't do this. Because of the prompt without a newline,
1558you'll have to use the C<sysread> variant of the interactive client above.
1559
cf21866a
TC
1560This server accepts one of five different commands, sending output back to
1561the client. Unlike most network servers, this one handles only one
5e220227 1562incoming client at a time. Multitasking servers are covered in
faa783ac 1563Chapter 16 of the Camel.
7b05b7e3 1564
c43b1046 1565Here's the code.
7b05b7e3 1566
c43b1046
DB
1567 #!/usr/bin/perl
1568 use strict;
1569 use warnings;
7b05b7e3 1570 use IO::Socket;
cf21866a 1571 use Net::hostent; # for OOish version of gethostbyaddr
7b05b7e3 1572
c43b1046 1573 my $PORT = 9000; # pick something not in use
7b05b7e3 1574
c43b1046
DB
1575 my $server = IO::Socket::INET->new( Proto => "tcp",
1576 LocalPort => $PORT,
1577 Listen => SOMAXCONN,
1578 Reuse => 1);
7b05b7e3
TC
1579
1580 die "can't setup server" unless $server;
1581 print "[Server $0 accepting clients]\n";
1582
c43b1046 1583 while (my $client = $server->accept()) {
7b05b7e3
TC
1584 $client->autoflush(1);
1585 print $client "Welcome to $0; type help for command list.\n";
c43b1046 1586 my $hostinfo = gethostbyaddr($client->peeraddr);
e46aa1dd
KW
1587 printf "[Connect from %s]\n",
1588 $hostinfo ? $hostinfo->name : $client->peerhost;
7b05b7e3
TC
1589 print $client "Command? ";
1590 while ( <$client>) {
e46aa1dd
KW
1591 next unless /\S/; # blank line
1592 if (/quit|exit/i) { last }
1593 elsif (/date|time/i) { printf $client "%s\n", scalar localtime() }
1594 elsif (/who/i ) { print $client `who 2>&1` }
1595 elsif (/cookie/i ) { print $client `/usr/games/fortune 2>&1` }
1596 elsif (/motd/i ) { print $client `cat /etc/motd 2>&1` }
7b05b7e3
TC
1597 else {
1598 print $client "Commands: quit date who cookie motd\n";
1599 }
1600 } continue {
1601 print $client "Command? ";
1602 }
1603 close $client;
1604 }
1605
1606=head1 UDP: Message Passing
4633a7c4
LW
1607
1608Another kind of client-server setup is one that uses not connections, but
1609messages. UDP communications involve much lower overhead but also provide
1610less reliability, as there are no promises that messages will arrive at
1611all, let alone in order and unmangled. Still, UDP offers some advantages
1612over TCP, including being able to "broadcast" or "multicast" to a whole
1613bunch of destination hosts at once (usually on your local subnet). If you
1614find yourself overly concerned about reliability and start building checks
6a3992aa 1615into your message system, then you probably should use just TCP to start
4633a7c4
LW
1616with.
1617
cf21866a
TC
1618UDP datagrams are I<not> a bytestream and should not be treated as such.
1619This makes using I/O mechanisms with internal buffering like stdio (i.e.
1620print() and friends) especially cumbersome. Use syswrite(), or better
1621send(), like in the example below.
90034919 1622
4633a7c4 1623Here's a UDP program similar to the sample Internet TCP client given
7b05b7e3 1624earlier. However, instead of checking one host at a time, the UDP version
4633a7c4
LW
1625will check many of them asynchronously by simulating a multicast and then
1626using select() to do a timed-out wait for I/O. To do something similar
1627with TCP, you'd have to use a different socket handle for each host.
1628
c43b1046 1629 #!/usr/bin/perl
e46aa1dd 1630 use strict;
c43b1046 1631 use warnings;
e46aa1dd
KW
1632 use Socket;
1633 use Sys::Hostname;
1634
c43b1046 1635 my $SECS_OF_70_YEARS = 2_208_988_800;
e46aa1dd 1636
c43b1046
DB
1637 my $iaddr = gethostbyname(hostname());
1638 my $proto = getprotobyname("udp");
1639 my $port = getservbyname("time", "udp");
1640 my $paddr = sockaddr_in(0, $iaddr); # 0 means let kernel pick
e46aa1dd 1641
08ab6dc8
DB
1642 socket(my $socket, PF_INET, SOCK_DGRAM, $proto) || die "socket: $!";
1643 bind($socket, $paddr) || die "bind: $!";
e46aa1dd
KW
1644
1645 $| = 1;
1646 printf "%-12s %8s %s\n", "localhost", 0, scalar localtime();
c43b1046
DB
1647 my $count = 0;
1648 for my $host (@ARGV) {
e46aa1dd 1649 $count++;
c43b1046
DB
1650 my $hisiaddr = inet_aton($host) || die "unknown host";
1651 my $hispaddr = sockaddr_in($port, $hisiaddr);
08ab6dc8 1652 defined(send($socket, 0, 0, $hispaddr)) || die "send $host: $!";
e46aa1dd 1653 }
4633a7c4 1654
c43b1046 1655 my $rout = my $rin = "";
08ab6dc8 1656 vec($rin, fileno($socket), 1) = 1;
e46aa1dd
KW
1657
1658 # timeout after 10.0 seconds
1659 while ($count && select($rout = $rin, undef, undef, 10.0)) {
c43b1046
DB
1660 my $rtime = "";
1661 my $hispaddr = recv($socket, $rtime, 4, 0) || die "recv: $!";
1662 my ($port, $hisiaddr) = sockaddr_in($hispaddr);
1663 my $host = gethostbyaddr($hisiaddr, AF_INET);
1664 my $histime = unpack("N", $rtime) - $SECS_OF_70_YEARS;
e46aa1dd
KW
1665 printf "%-12s ", $host;
1666 printf "%8d %s\n", $histime - time(), scalar localtime($histime);
1667 $count--;
1668 }
4633a7c4 1669
cf21866a
TC
1670This example does not include any retries and may consequently fail to
1671contact a reachable host. The most prominent reason for this is congestion
1672of the queues on the sending host if the number of hosts to contact is
1673sufficiently large.
90034919 1674
4633a7c4
LW
1675=head1 SysV IPC
1676
1677While System V IPC isn't so widely used as sockets, it still has some
cf21866a
TC
1678interesting uses. However, you cannot use SysV IPC or Berkeley mmap() to
1679have a variable shared amongst several processes. That's because Perl
1680would reallocate your string when you weren't wanting it to. You might
1681look into the C<IPC::Shareable> or C<threads::shared> modules for that.
4633a7c4 1682
54310121 1683Here's a small example showing shared memory usage.
a0d0e21e 1684
7b34eba2 1685 use IPC::SysV qw(IPC_PRIVATE IPC_RMID S_IRUSR S_IWUSR);
0ade1984 1686
c43b1046
DB
1687 my $size = 2000;
1688 my $id = shmget(IPC_PRIVATE, $size, S_IRUSR | S_IWUSR);
cf21866a 1689 defined($id) || die "shmget: $!";
41d6edb2 1690 print "shm key $id\n";
a0d0e21e 1691
c43b1046 1692 my $message = "Message #1";
cf21866a 1693 shmwrite($id, $message, 0, 60) || die "shmwrite: $!";
0ade1984 1694 print "wrote: '$message'\n";
c43b1046 1695 shmread($id, my $buff, 0, 60) || die "shmread: $!";
0ade1984 1696 print "read : '$buff'\n";
a0d0e21e 1697
0ade1984 1698 # the buffer of shmread is zero-character end-padded.
b18b5ffd 1699 substr($buff, index($buff, "\0")) = "";
0ade1984
JH
1700 print "un" unless $buff eq $message;
1701 print "swell\n";
a0d0e21e 1702
41d6edb2 1703 print "deleting shm $id\n";
cf21866a 1704 shmctl($id, IPC_RMID, 0) || die "shmctl: $!";
a0d0e21e
LW
1705
1706Here's an example of a semaphore:
1707
0ade1984
JH
1708 use IPC::SysV qw(IPC_CREAT);
1709
c43b1046
DB
1710 my $IPC_KEY = 1234;
1711 my $id = semget($IPC_KEY, 10, 0666 | IPC_CREAT);
3389bcf7
JL
1712 defined($id) || die "semget: $!";
1713 print "sem id $id\n";
a0d0e21e 1714
a2eb9003 1715Put this code in a separate file to be run in more than one process.
a0d0e21e
LW
1716Call the file F<take>:
1717
1718 # create a semaphore
1719
c43b1046
DB
1720 my $IPC_KEY = 1234;
1721 my $id = semget($IPC_KEY, 0, 0);
3389bcf7 1722 defined($id) || die "semget: $!";
a0d0e21e 1723
c43b1046
DB
1724 my $semnum = 0;
1725 my $semflag = 0;
a0d0e21e 1726
cf21866a 1727 # "take" semaphore
a0d0e21e 1728 # wait for semaphore to be zero
c43b1046
DB
1729 my $semop = 0;
1730 my $opstring1 = pack("s!s!s!", $semnum, $semop, $semflag);
a0d0e21e
LW
1731
1732 # Increment the semaphore count
1733 $semop = 1;
c43b1046
DB
1734 my $opstring2 = pack("s!s!s!", $semnum, $semop, $semflag);
1735 my $opstring = $opstring1 . $opstring2;
a0d0e21e 1736
cf21866a 1737 semop($id, $opstring) || die "semop: $!";
a0d0e21e 1738
a2eb9003 1739Put this code in a separate file to be run in more than one process.
a0d0e21e
LW
1740Call this file F<give>:
1741
cf21866a 1742 # "give" the semaphore
a0d0e21e
LW
1743 # run this in the original process and you will see
1744 # that the second process continues
1745
c43b1046
DB
1746 my $IPC_KEY = 1234;
1747 my $id = semget($IPC_KEY, 0, 0);
cf21866a 1748 die unless defined($id);
a0d0e21e 1749
c43b1046
DB
1750 my $semnum = 0;
1751 my $semflag = 0;
a0d0e21e
LW
1752
1753 # Decrement the semaphore count
c43b1046
DB
1754 my $semop = -1;
1755 my $opstring = pack("s!s!s!", $semnum, $semop, $semflag);
a0d0e21e 1756
cf21866a 1757 semop($id, $opstring) || die "semop: $!";
a0d0e21e 1758
7b05b7e3 1759The SysV IPC code above was written long ago, and it's definitely
e6aa8b84 1760clunky looking. For a more modern look, see the IPC::SysV module.
4633a7c4 1761
41d6edb2
JH
1762A small example demonstrating SysV message queues:
1763
7b34eba2 1764 use IPC::SysV qw(IPC_PRIVATE IPC_RMID IPC_CREAT S_IRUSR S_IWUSR);
41d6edb2 1765
7b34eba2 1766 my $id = msgget(IPC_PRIVATE, IPC_CREAT | S_IRUSR | S_IWUSR);
cf21866a 1767 defined($id) || die "msgget failed: $!";
41d6edb2 1768
cf21866a 1769 my $sent = "message";
e343e2e2 1770 my $type_sent = 1234;
cf21866a
TC
1771
1772 msgsnd($id, pack("l! a*", $type_sent, $sent), 0)
1773 || die "msgsnd failed: $!";
1774
1775 msgrcv($id, my $rcvd_buf, 60, 0, 0)
1776 || die "msgrcv failed: $!";
1777
1778 my($type_rcvd, $rcvd) = unpack("l! a*", $rcvd_buf);
1779
1780 if ($rcvd eq $sent) {
1781 print "okay\n";
41d6edb2 1782 } else {
cf21866a 1783 print "not okay\n";
41d6edb2
JH
1784 }
1785
cf21866a
TC
1786 msgctl($id, IPC_RMID, 0) || die "msgctl failed: $!\n";
1787
4633a7c4
LW
1788=head1 NOTES
1789
5a964f20
TC
1790Most of these routines quietly but politely return C<undef> when they
1791fail instead of causing your program to die right then and there due to
1792an uncaught exception. (Actually, some of the new I<Socket> conversion
cf21866a 1793functions do croak() on bad arguments.) It is therefore essential to
5a964f20 1794check return values from these functions. Always begin your socket
cf21866a
TC
1795programs this way for optimal success, and don't forget to add the B<-T>
1796taint-checking flag to the C<#!> line for servers:
4633a7c4 1797
c43b1046 1798 #!/usr/bin/perl -T
4633a7c4 1799 use strict;
c43b1046 1800 use warnings;
4633a7c4
LW
1801 use sigtrap;
1802 use Socket;
1803
1804=head1 BUGS
1805
cf21866a 1806These routines all create system-specific portability problems. As noted
4633a7c4 1807elsewhere, Perl is at the mercy of your C libraries for much of its system
cf21866a 1808behavior. It's probably safest to assume broken SysV semantics for
6a3992aa 1809signals and to stick with simple TCP and UDP socket operations; e.g., don't
a2eb9003 1810try to pass open file descriptors over a local UDP datagram socket if you
4633a7c4
LW
1811want your code to stand a chance of being portable.
1812
4633a7c4
LW
1813=head1 AUTHOR
1814
1815Tom Christiansen, with occasional vestiges of Larry Wall's original
7b05b7e3 1816version and suggestions from the Perl Porters.
4633a7c4
LW
1817
1818=head1 SEE ALSO
1819
7b05b7e3
TC
1820There's a lot more to networking than this, but this should get you
1821started.
1822
cf21866a
TC
1823For intrepid programmers, the indispensable textbook is I<Unix Network
1824Programming, 2nd Edition, Volume 1> by W. Richard Stevens (published by
1825Prentice-Hall). Most books on networking address the subject from the
1826perspective of a C programmer; translation to Perl is left as an exercise
1827for the reader.
7b05b7e3
TC
1828
1829The IO::Socket(3) manpage describes the object library, and the Socket(3)
1830manpage describes the low-level interface to sockets. Besides the obvious
cf21866a
TC
1831functions in L<perlfunc>, you should also check out the F<modules> file at
1832your nearest CPAN site, especially
82f82fdb 1833L<http://www.cpan.org/modules/00modlist.long.html#ID5_Networking_>.
cf21866a 1834See L<perlmodlib> or best yet, the F<Perl FAQ> for a description
82f82fdb 1835of what CPAN is and where to get it if the previous link doesn't work
cf21866a
TC
1836for you.
1837
1838Section 5 of CPAN's F<modules> file is devoted to "Networking, Device
1839Control (modems), and Interprocess Communication", and contains numerous
1840unbundled modules numerous networking modules, Chat and Expect operations,
1841CGI programming, DCE, FTP, IPC, NNTP, Proxy, Ptty, RPC, SNMP, SMTP, Telnet,
1842Threads, and ToolTalk--to name just a few.