Commit | Line | Data |
---|---|---|
a0d0e21e LW |
1 | =head1 NAME |
2 | ||
cb1a09d0 | 3 | perlipc - Perl interprocess communication (signals, fifos, pipes, safe subprocceses, sockets, and semaphores) |
a0d0e21e LW |
4 | |
5 | =head1 DESCRIPTION | |
6 | ||
4633a7c4 LW |
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. | |
10 | ||
11 | =head1 Signals | |
12 | ||
13 | Perl uses a simple signal handling model: the %SIG hash contains names or | |
14 | references of user-installed signal handlers. These handlers will be called | |
15 | with an argument which is the name of the signal that triggered it. A | |
16 | signal may be generated intentionally from a particular keyboard sequence like | |
17 | control-C or control-Z, sent to you from an another process, or | |
18 | triggered automatically by the kernel when special events transpire, like | |
19 | a child process exiting, your process running out of stack space, or | |
20 | hitting file size limit. | |
21 | ||
22 | For example, to trap an interrupt signal, set up a handler like this. | |
23 | Notice how all we do is set with a global variable and then raise an | |
24 | exception. That's because on most systems libraries are not | |
25 | re-entrant, so calling any print() functions (or even anything that needs to | |
26 | malloc(3) more memory) could in theory trigger a memory fault | |
27 | and subsequent core dump. | |
28 | ||
29 | sub catch_zap { | |
30 | my $signame = shift; | |
31 | $shucks++; | |
32 | die "Somebody sent me a SIG$signame"; | |
33 | } | |
34 | $SIG{INT} = 'catch_zap'; # could fail in modules | |
35 | $SIG{INT} = \&catch_zap; # best strategy | |
36 | ||
37 | The names of the signals are the ones listed out by C<kill -l> on your | |
38 | system, or you can retrieve them from the Config module. Set up an | |
39 | @signame list indexed by number to get the name and a %signo table | |
40 | indexed by name to get the number: | |
41 | ||
42 | use Config; | |
43 | defined $Config{sig_name} || die "No sigs?"; | |
44 | foreach $name (split(' ', $Config{sig_name})) { | |
45 | $signo{$name} = $i; | |
46 | $signame[$i] = $name; | |
47 | $i++; | |
48 | } | |
49 | ||
50 | So to check whether signal 17 and SIGALRM were the same, just do this: | |
51 | ||
52 | print "signal #17 = $signame[17]\n"; | |
53 | if ($signo{ALRM}) { | |
54 | print "SIGALRM is $signo{ALRM}\n"; | |
55 | } | |
56 | ||
57 | You may also choose to assign the strings C<'IGNORE'> or C<'DEFAULT'> as | |
58 | the handler, in which case Perl will try to discard the signal or do the | |
59 | default thing. Some signals can be neither trapped nor ignored, such as | |
60 | the KILL and STOP (but not the TSTP) signals. One strategy for | |
61 | temporarily ignoring signals is to use a local() statement, which will be | |
62 | automatically restored once your block is exited. (Remember that local() | |
63 | values are "inherited" by functions called from within that block.) | |
64 | ||
65 | sub precious { | |
66 | local $SIG{INT} = 'IGNORE'; | |
67 | &more_functions; | |
68 | } | |
69 | sub more_functions { | |
70 | # interrupts still ignored, for now... | |
71 | } | |
72 | ||
73 | Sending a signal to a negative process ID means that you send the signal | |
74 | to the entire Unix process-group. This code send a hang-up signal to all | |
75 | processes in the current process group I<except for> the current process | |
76 | itself: | |
77 | ||
78 | { | |
79 | local $SIG{HUP} = 'IGNORE'; | |
80 | kill HUP => -$$; | |
81 | # snazzy writing of: kill('HUP', -$$) | |
82 | } | |
a0d0e21e | 83 | |
4633a7c4 LW |
84 | Another interesting signal to send is signal number zero. This doesn't |
85 | actually affect another process, but instead checks whether it's alive | |
86 | or has changed its UID. | |
a0d0e21e | 87 | |
4633a7c4 LW |
88 | unless (kill 0 => $kid_pid) { |
89 | warn "something wicked happened to $kid_pid"; | |
90 | } | |
a0d0e21e | 91 | |
4633a7c4 LW |
92 | You might also want to employ anonymous functions for simple signal |
93 | handlers: | |
a0d0e21e | 94 | |
4633a7c4 | 95 | $SIG{INT} = sub { die "\nOutta here!\n" }; |
a0d0e21e | 96 | |
4633a7c4 LW |
97 | But that will be problematic for the more complicated handlers that need |
98 | to re-install themselves. Because Perl's signal mechanism is currently | |
99 | based on the signal(3) function from the C library, you may somtimes be so | |
100 | misfortunate as to run on systems where that function is "broken", that | |
101 | is, it behaves in the old unreliable SysV way rather than the newer, more | |
102 | reasonable BSD and POSIX fashion. So you'll see defensive people writing | |
103 | signal handlers like this: | |
a0d0e21e | 104 | |
4633a7c4 LW |
105 | sub REAPER { |
106 | $SIG{CHLD} = \&REAPER; # loathe sysV | |
107 | $waitedpid = wait; | |
108 | } | |
109 | $SIG{CHLD} = \&REAPER; | |
110 | # now do something that forks... | |
111 | ||
112 | or even the more elaborate: | |
113 | ||
114 | use POSIX "wait_h"; | |
115 | sub REAPER { | |
116 | my $child; | |
117 | $SIG{CHLD} = \&REAPER; # loathe sysV | |
118 | while ($child = waitpid(-1,WNOHANG)) { | |
119 | $Kid_Status{$child} = $?; | |
120 | } | |
121 | } | |
122 | $SIG{CHLD} = \&REAPER; | |
123 | # do something that forks... | |
124 | ||
125 | Signal handling is also used for timeouts in Unix, While safely | |
126 | protected within an C<eval{}> block, you set a signal handler to trap | |
127 | alarm signals and then schedule to have one delivered to you in some | |
128 | number of seconds. Then try your blocking operation, clearing the alarm | |
129 | when it's done but not before you've exited your C<eval{}> block. If it | |
130 | goes off, you'll use die() to jump out of the block, much as you might | |
131 | using longjmp() or throw() in other languages. | |
132 | ||
133 | Here's an example: | |
134 | ||
135 | eval { | |
136 | local $SIG{ALRM} = sub { die "alarm clock restart" }; | |
137 | alarm 10; | |
138 | flock(FH, 2); # blocking write lock | |
139 | alarm 0; | |
140 | }; | |
141 | if ($@ and $@ !~ /alarm clock restart/) { die } | |
142 | ||
143 | For more complex signal handling, you might see the standard POSIX | |
144 | module. Lamentably, this is almost entirely undocumented, but | |
145 | the F<t/lib/posix.t> file from the Perl source distribution has some | |
146 | examples in it. | |
147 | ||
148 | =head1 Named Pipes | |
149 | ||
150 | A named pipe (often referred to as a FIFO) is an old Unix IPC | |
151 | mechanism for processes communicating on the same machine. It works | |
152 | just like a regular, connected anonymous pipes, except that the | |
153 | processes rendezvous using a filename and don't have to be related. | |
154 | ||
155 | To create a named pipe, use the Unix command mknod(1) or on some | |
156 | systems, mkfifo(1). These may not be in your normal path. | |
157 | ||
158 | # system return val is backwards, so && not || | |
159 | # | |
160 | $ENV{PATH} .= ":/etc:/usr/etc"; | |
161 | if ( system('mknod', $path, 'p') | |
162 | && system('mkfifo', $path) ) | |
163 | { | |
164 | die "mk{nod,fifo} $path failed; | |
165 | } | |
166 | ||
167 | ||
168 | A fifo is convenient when you want to connect a process to an unrelated | |
169 | one. When you open a fifo, the program will block until there's something | |
170 | on the other end. | |
171 | ||
172 | For example, let's say you'd like to have your F<.signature> file be a | |
173 | named pipe that has a Perl program on the other end. Now every time any | |
174 | program (like a mailer, newsreader, finger program, etc.) tries to read | |
175 | from that file, the reading program will block and your program will | |
176 | supply the the new signature. We'll use the pipe-checking file test B<-p> | |
177 | to find out whether anyone (or anything) has accidentally removed our fifo. | |
178 | ||
179 | chdir; # go home | |
180 | $FIFO = '.signature'; | |
181 | $ENV{PATH} .= ":/etc:/usr/games"; | |
182 | ||
183 | while (1) { | |
184 | unless (-p $FIFO) { | |
185 | unlink $FIFO; | |
186 | system('mknod', $FIFO, 'p') | |
187 | && die "can't mknod $FIFO: $!"; | |
188 | } | |
189 | ||
190 | # next line blocks until there's a reader | |
191 | open (FIFO, "> $FIFO") || die "can't write $FIFO: $!"; | |
192 | print FIFO "John Smith (smith\@host.org)\n", `fortune -s`; | |
193 | close FIFO; | |
194 | sleep 2; # to avoid dup sigs | |
195 | } | |
a0d0e21e | 196 | |
a0d0e21e | 197 | |
4633a7c4 LW |
198 | =head1 Using open() for IPC |
199 | ||
200 | Perl's basic open() statement can also be used for unidirectional interprocess | |
201 | communication by either appending or prepending a pipe symbol to the second | |
202 | argument to open(). Here's how to start something up a child process you | |
203 | intend to write to: | |
204 | ||
205 | open(SPOOLER, "| cat -v | lpr -h 2>/dev/null") | |
206 | || die "can't fork: $!"; | |
207 | local $SIG{PIPE} = sub { die "spooler pipe broke" }; | |
208 | print SPOOLER "stuff\n"; | |
209 | close SPOOLER || die "bad spool: $! $?"; | |
210 | ||
211 | And here's how to start up a child process you intend to read from: | |
212 | ||
213 | open(STATUS, "netstat -an 2>&1 |") | |
214 | || die "can't fork: $!"; | |
215 | while (<STATUS>) { | |
216 | next if /^(tcp|udp)/; | |
217 | print; | |
218 | } | |
219 | close SPOOLER || die "bad netstat: $! $?"; | |
220 | ||
221 | If one can be sure that a particular program is a Perl script that is | |
222 | expecting filenames in @ARGV, the clever programmer can write something | |
223 | like this: | |
224 | ||
225 | $ program f1 "cmd1|" - f2 "cmd2|" f3 < tmpfile | |
226 | ||
227 | and irrespective of which shell it's called from, the Perl program will | |
228 | read from the file F<f1>, the process F<cmd1>, standard input (F<tmpfile> | |
229 | in this case), the F<f2> file, the F<cmd2> command, and finally the F<f3> | |
230 | file. Pretty nifty, eh? | |
231 | ||
232 | You might notice that you could use backticks for much the | |
233 | same effect as opening a pipe for reading: | |
234 | ||
235 | print grep { !/^(tcp|udp)/ } `netstat -an 2>&1`; | |
236 | die "bad netstat" if $?; | |
237 | ||
238 | While this is true on the surface, it's much more efficient to process the | |
239 | file one line or record at a time because then you don't have to read the | |
240 | whole thing into memory at once. It also gives you finer control of the | |
241 | whole process, letting you to kill off the child process early if you'd | |
242 | like. | |
243 | ||
244 | Be careful to check both the open() and the close() return values. If | |
245 | you're I<writing> to a pipe, you should also trap SIGPIPE. Otherwise, | |
246 | think of what happens when you start up a pipe to a command that doesn't | |
247 | exist: the open() will in all likelihood succeed (it only reflects the | |
248 | fork()'s success), but then your output will fail--spectacularly. Perl | |
249 | can't know whether the command worked because your command is actually | |
250 | running in a separate process whose exec() might have failed. Therefore, | |
251 | while readers of bogus commands just return a quick end of file, writers | |
252 | to bogus command will trigger a signal they'd better be prepared to | |
253 | handle. Consider: | |
254 | ||
255 | open(FH, "|bogus"); | |
256 | print FH "bang\n"; | |
257 | close FH; | |
258 | ||
259 | =head2 Safe Pipe Opens | |
260 | ||
261 | Another interesting approach to IPC is making your single program go | |
262 | multiprocess and communicate between (or even amongst) yourselves. The | |
263 | open() function will accept a file argument of either C<"-|"> or C<"|-"> | |
264 | to do a very interesting thing: it forks a child connected to the | |
265 | filehandle you've opened. The child is running the same program as the | |
266 | parent. This is useful for safely opening a file when running under an | |
267 | assumed UID or GID, for example. If you open a pipe I<to> minus, you can | |
268 | write to the filehandle you opened and your kid will find it in his | |
269 | STDIN. If you open a pipe I<from> minus, you can read from the filehandle | |
270 | you opened whatever your kid writes to his STDOUT. | |
271 | ||
272 | use English; | |
273 | my $sleep_count = 0; | |
274 | ||
275 | do { | |
276 | $pid = open(KID, "-|"); | |
277 | unless (defined $pid) { | |
278 | warn "cannot fork: $!"; | |
279 | die "bailing out" if $sleep_count++ > 6; | |
280 | sleep 10; | |
281 | } | |
282 | } until defined $pid; | |
283 | ||
284 | if ($pid) { # parent | |
285 | print KID @some_data; | |
286 | close(KID) || warn "kid exited $?"; | |
287 | } else { # child | |
288 | ($EUID, $EGID) = ($UID, $GID); # suid progs only | |
289 | open (FILE, "> /safe/file") | |
290 | || die "can't open /safe/file: $!"; | |
291 | while (<STDIN>) { | |
292 | print FILE; # child's STDIN is parent's KID | |
293 | } | |
294 | exit; # don't forget this | |
295 | } | |
296 | ||
297 | Another common use for this construct is when you need to execute | |
298 | something without the shell's interference. With system(), it's | |
299 | straigh-forward, but you can't use a pipe open or backticks safely. | |
300 | That's because there's no way to stop the shell from getting its hands on | |
301 | your arguments. Instead, use lower-level control to call exec() directly. | |
302 | ||
303 | Here's a safe backtick or pipe open for read: | |
304 | ||
305 | # add error processing as above | |
306 | $pid = open(KID, "-|"); | |
307 | ||
308 | if ($pid) { # parent | |
309 | while (<KID>) { | |
310 | # do something interesting | |
311 | } | |
312 | close(KID) || warn "kid exited $?"; | |
313 | ||
314 | } else { # child | |
315 | ($EUID, $EGID) = ($UID, $GID); # suid only | |
316 | exec($program, @options, @args) | |
317 | || die "can't exec program: $!"; | |
318 | # NOTREACHED | |
319 | } | |
320 | ||
321 | ||
322 | And here's a safe pipe open for writing: | |
323 | ||
324 | # add error processing as above | |
325 | $pid = open(KID, "|-"); | |
326 | $SIG{ALRM} = sub { die "whoops, $program pipe broke" }; | |
327 | ||
328 | if ($pid) { # parent | |
329 | for (@data) { | |
330 | print KID; | |
331 | } | |
332 | close(KID) || warn "kid exited $?"; | |
333 | ||
334 | } else { # child | |
335 | ($EUID, $EGID) = ($UID, $GID); | |
336 | exec($program, @options, @args) | |
337 | || die "can't exec program: $!"; | |
338 | # NOTREACHED | |
339 | } | |
340 | ||
341 | Note that these operations are full Unix forks, which means they may not be | |
342 | correctly implemented on alien systems. Additionally, these are not true | |
343 | multithreading. If you'd like to learn more about threading, see the | |
344 | F<modules> file mentioned below in the L<SEE ALSO> section. | |
345 | ||
346 | =head2 Bidirectional Communication | |
347 | ||
348 | While this works reasonably well for unidirectional communication, what | |
349 | about bidirectional communication? The obvious thing you'd like to do | |
350 | doesn't actually work: | |
351 | ||
352 | open(KID, "| some program |") | |
353 | ||
354 | and if you forgot to use the B<-w> flag, then you'll miss out | |
355 | entirely on the diagnostic message: | |
356 | ||
357 | Can't do bidirectional pipe at -e line 1. | |
358 | ||
359 | If you really want to, you can use the standard open2() library function | |
360 | to catch both ends. There's also an open3() for tridirectional I/O so you | |
361 | can also catch your child's STDERR, but doing so would then require an | |
362 | awkward select() loop and wouldn't allow you to use normal Perl input | |
363 | operations. | |
364 | ||
365 | If you look at its source, you'll see that open2() uses low-level | |
366 | primitives like Unix pipe() and exec() to create all the connections. | |
367 | While it might have been slightly more efficient by using socketpair(), it | |
368 | would have then been even less portable than it already is. The open2() | |
369 | and open3() functions are unlikely to work anywhere except on a Unix | |
370 | system or some other one purporting to be POSIX compliant. | |
371 | ||
372 | Here's an example of using open2(): | |
373 | ||
374 | use FileHandle; | |
375 | use IPC::Open2; | |
376 | $pid = open2( \*Reader, \*Writer, "cat -u -n" ); | |
377 | Writer->autoflush(); # default here, actually | |
378 | print Writer "stuff\n"; | |
379 | $got = <Reader>; | |
380 | ||
381 | The problem with this is that Unix buffering is going to really | |
382 | ruin your day. Even though your C<Writer> filehandle is autoflushed, | |
383 | and the process on the other end will get your data in a timely manner, | |
384 | you can't usually do anything to force it to actually give it back to you | |
385 | in a similarly quick fashion. In this case, we could, because we | |
386 | gave I<cat> a B<-u> flag to make it unbuffered. But very few Unix | |
387 | commands are designed to operate over pipes, so this seldom works | |
388 | unless you yourself wrote the program on the other end of the | |
389 | double-ended pipe. | |
390 | ||
391 | A solution to this is the non-standard F<Comm.pl> library. It uses | |
392 | pseudo-ttys to make your program behave more reasonably: | |
393 | ||
394 | require 'Comm.pl'; | |
395 | $ph = open_proc('cat -n'); | |
396 | for (1..10) { | |
397 | print $ph "a line\n"; | |
398 | print "got back ", scalar <$ph>; | |
399 | } | |
a0d0e21e | 400 | |
4633a7c4 LW |
401 | This way you don't have to have control over the source code of the |
402 | program you're using. The F<Comm> library also has expect() | |
403 | and interact() functions. Find the library (and hopefully its | |
404 | successor F<IPC::Chat>) at your nearest CPAN archive as detailed | |
405 | in the L<SEE ALSO> section below. | |
a0d0e21e | 406 | |
4633a7c4 | 407 | =head1 Sockets: Client/Server Communication |
a0d0e21e | 408 | |
4633a7c4 LW |
409 | While not limited to Unix-derived operating systems (e.g. WinSock on PCs |
410 | provides socket support, as do some VMS libraries), you may not have | |
411 | sockets on your system, in which this section probably isn't going to do | |
412 | you much good. With sockets, you can do both virtual circuits (i.e. TCP | |
413 | streams) and datagrams (i.e. UDP packets). You may be able to do even more | |
414 | depending on your system. | |
415 | ||
416 | The Perl function calls for dealing with sockets have the same names as | |
417 | the corresponding system calls in C, but their arguments tend to differ | |
418 | for two reasons: first, Perl filehandles work differently than C file | |
419 | descriptors. Second, Perl already knows the length of its strings, so you | |
420 | don't need to pass that information. | |
a0d0e21e | 421 | |
4633a7c4 LW |
422 | One of the major problems with old socket code in Perl was that it used |
423 | hard-coded values for some of the constants, which severely hurt | |
424 | portability. If you ever see code that does anything like explicitly | |
425 | setting C<$AF_INET = 2>, you know you're in for big trouble: An | |
426 | immeasurably superior approach is to use the C<Socket> module, which more | |
427 | reliably grants access to various constants and functions you'll need. | |
a0d0e21e | 428 | |
4633a7c4 | 429 | =head2 Internet TCP Clients and Servers |
a0d0e21e | 430 | |
4633a7c4 LW |
431 | Use Internet-domain sockets when you want to do client-server |
432 | communication that might extend to machines outside of your own system. | |
433 | ||
434 | Here's a sample TCP client using Internet-domain sockets: | |
435 | ||
436 | #!/usr/bin/perl -w | |
437 | require 5.002; | |
438 | use strict; | |
439 | use Socket; | |
440 | my ($remote,$port, $iaddr, $paddr, $proto, $line); | |
441 | ||
442 | $remote = shift || 'localhost'; | |
443 | $port = shift || 2345; # random port | |
444 | if ($port =~ /\D/) { $port = getservbyname($port, 'tcp') } | |
445 | die "No port" unless $port; | |
446 | $iaddr = inet_aton($remote) || die "no host: $remote"; | |
447 | $paddr = sockaddr_in($port, $iaddr); | |
448 | ||
449 | $proto = getprotobyname('tcp'); | |
450 | socket(SOCK, PF_INET, SOCK_STREAM, $proto) || die "socket: $!"; | |
451 | connect(SOCK, $paddr) || die "connect: $!"; | |
452 | while ($line = <SOCK>) { | |
453 | print $line; | |
454 | } | |
455 | ||
456 | close (SOCK) || die "close: $!"; | |
457 | exit; | |
458 | ||
459 | And here's a corresponding server to go along with it. We'll | |
460 | leave the address as INADDR_ANY so that the kernel can choose | |
461 | the appropriate interface on multihomed hosts: | |
462 | ||
463 | #!/usr/bin/perl -Tw | |
464 | require 5.002; | |
465 | use strict; | |
466 | BEGIN { $ENV{PATH} = '/usr/ucb:/bin' } | |
a0d0e21e | 467 | use Socket; |
4633a7c4 | 468 | use Carp; |
a0d0e21e | 469 | |
4633a7c4 LW |
470 | sub spawn; # forward declaration |
471 | sub logmsg { print "$0 $$: @_ at ", scalar localtime, "\n" } | |
a0d0e21e | 472 | |
4633a7c4 LW |
473 | my $port = shift || 2345; |
474 | my $proto = getprotobyname('tcp'); | |
475 | socket(SERVER, PF_INET, SOCK_STREAM, $proto) || die "socket: $!"; | |
476 | setsockopt(SERVER, SOL_SOCKET, SO_REUSEADDR, 1) || die "setsockopt: $!"; | |
477 | bind(SERVER, sockaddr_in($port, INADDR_ANY)) || die "bind: $!"; | |
478 | listen(SERVER,5) || die "listen: $!"; | |
a0d0e21e | 479 | |
4633a7c4 | 480 | logmsg "server started on port $port"; |
a0d0e21e | 481 | |
4633a7c4 LW |
482 | my $waitedpid = 0; |
483 | my $paddr; | |
a0d0e21e | 484 | |
4633a7c4 LW |
485 | sub REAPER { |
486 | $SIG{CHLD} = \&REAPER; # loathe sysV | |
487 | $waitedpid = wait; | |
488 | logmsg "reaped $waitedpid" . ($? ? " with exit $?" : ''); | |
489 | } | |
490 | ||
491 | $SIG{CHLD} = \&REAPER; | |
492 | ||
493 | for ( $waitedpid = 0; | |
494 | ($paddr = accept(CLIENT,SERVER)) || $waitedpid; | |
495 | $waitedpid = 0, close CLIENT) | |
496 | { | |
497 | next if $waitedpid; | |
498 | my($port,$iaddr) = sockaddr_in($paddr); | |
499 | my $name = gethostbyaddr($iaddr,AF_INET); | |
500 | ||
501 | logmsg "connection from $name [", | |
502 | inet_ntoa($iaddr), "] | |
503 | at port $port"; | |
a0d0e21e | 504 | |
4633a7c4 LW |
505 | spawn sub { |
506 | print "Hello there, $name, it's now ", scalar localtime, "\n"; | |
507 | exec '/usr/games/fortune' | |
508 | or confess "can't exec fortune: $!"; | |
509 | }; | |
a0d0e21e | 510 | |
4633a7c4 | 511 | } |
a0d0e21e | 512 | |
4633a7c4 LW |
513 | sub spawn { |
514 | my $coderef = shift; | |
a0d0e21e | 515 | |
4633a7c4 LW |
516 | unless (@_ == 0 && $coderef && ref($coderef) eq 'CODE') { |
517 | confess "usage: spawn CODEREF"; | |
a0d0e21e | 518 | } |
4633a7c4 LW |
519 | |
520 | my $pid; | |
521 | if (!defined($pid = fork)) { | |
522 | logmsg "cannot fork: $!"; | |
523 | return; | |
524 | } elsif ($pid) { | |
525 | logmsg "begat $pid"; | |
526 | return; # i'm the parent | |
527 | } | |
528 | # else i'm the child -- go spawn | |
529 | ||
530 | open(STDIN, "<&CLIENT") || die "can't dup client to stdin"; | |
531 | open(STDOUT, ">&CLIENT") || die "can't dup client to stdout"; | |
532 | ## open(STDERR, ">&STDOUT") || die "can't dup stdout to stderr"; | |
533 | exit &$coderef(); | |
534 | } | |
535 | ||
536 | This server takes the trouble to clone off a child version via fork() for | |
537 | each incoming request. That way it can handle many requests at once, | |
538 | which you might not always want. Even if you don't fork(), the listen() | |
539 | will allow that many pending connections. Forking servers have to be | |
540 | particularly careful about cleaning up their dead children (called | |
541 | "zombies" in Unix parlance), because otherwise you'll quickly fill up your | |
542 | process table. | |
543 | ||
544 | We suggest that you use the B<-T> flag to use taint checking (see L<perlsec>) | |
545 | even if we aren't running setuid or setgid. This is always a good idea | |
546 | for servers and other programs run on behalf of someone else (like CGI | |
547 | scripts), because it lessens the chances that people from the outside will | |
548 | be able to compromise your system. | |
549 | ||
550 | Let's look at another TCP client. This one connects to the TCP "time" | |
551 | service on a number of different machines and shows how far their clocks | |
552 | differ from the system on which it's being run: | |
553 | ||
554 | #!/usr/bin/perl -w | |
555 | require 5.002; | |
556 | use strict; | |
557 | use Socket; | |
558 | ||
559 | my $SECS_of_70_YEARS = 2208988800; | |
560 | sub ctime { scalar localtime(shift) } | |
561 | ||
562 | my $iaddr = gethostbyname('localhost'); | |
563 | my $proto = getprotobyname('tcp'); | |
564 | my $port = getservbyname('time', 'tcp'); | |
565 | my $paddr = sockaddr_in(0, $iaddr); | |
566 | my($host); | |
567 | ||
568 | $| = 1; | |
569 | printf "%-24s %8s %s\n", "localhost", 0, ctime(time()); | |
570 | ||
571 | foreach $host (@ARGV) { | |
572 | printf "%-24s ", $host; | |
573 | my $hisiaddr = inet_aton($host) || die "unknown host"; | |
574 | my $hispaddr = sockaddr_in($port, $hisiaddr); | |
575 | socket(SOCKET, PF_INET, SOCK_STREAM, $proto) || die "socket: $!"; | |
576 | connect(SOCKET, $hispaddr) || die "bind: $!"; | |
577 | my $rtime = ' '; | |
578 | read(SOCKET, $rtime, 4); | |
579 | close(SOCKET); | |
580 | my $histime = unpack("N", $rtime) - $SECS_of_70_YEARS ; | |
581 | printf "%8d %s\n", $histime - time, ctime($histime); | |
a0d0e21e LW |
582 | } |
583 | ||
4633a7c4 LW |
584 | =head2 Unix-Domain TCP Clients and Servers |
585 | ||
586 | That's fine for Internet-domain clients and servers, but what local | |
587 | communications? While you can use the same setup, sometimes you don't | |
588 | want to. Unix-domain sockets are local to the current host, and are often | |
589 | used internally to implement pipes. Unlike Internet domain sockets, UNIX | |
590 | domain sockets can show up in the file system with an ls(1) listing. | |
591 | ||
592 | $ ls -l /dev/log | |
593 | srw-rw-rw- 1 root 0 Oct 31 07:23 /dev/log | |
a0d0e21e | 594 | |
4633a7c4 LW |
595 | You can test for these with Perl's B<-S> file test: |
596 | ||
597 | unless ( -S '/dev/log' ) { | |
598 | die "something's wicked with the print system"; | |
599 | } | |
600 | ||
601 | Here's a sample Unix-domain client: | |
602 | ||
603 | #!/usr/bin/perl -w | |
604 | require 5.002; | |
605 | use Socket; | |
606 | use strict; | |
607 | my ($rendezvous, $line); | |
608 | ||
609 | $rendezvous = shift || '/tmp/catsock'; | |
610 | socket(SOCK, PF_UNIX, SOCK_STREAM, 0) || die "socket: $!"; | |
611 | connect(SOCK, sockaddr_un($remote)) || die "connect: $!"; | |
612 | while ($line = <SOCK>) { | |
613 | print $line; | |
614 | } | |
615 | exit; | |
616 | ||
617 | And here's a corresponding server. | |
618 | ||
619 | #!/usr/bin/perl -Tw | |
620 | require 5.002; | |
621 | use strict; | |
622 | use Socket; | |
623 | use Carp; | |
624 | ||
625 | BEGIN { $ENV{PATH} = '/usr/ucb:/bin' } | |
626 | ||
627 | my $NAME = '/tmp/catsock'; | |
628 | my $uaddr = sockaddr_un($NAME); | |
629 | my $proto = getprotobyname('tcp'); | |
630 | ||
631 | socket(SERVER,PF_UNIX,SOCK_STREAM,0) || die "socket: $!"; | |
632 | unlink($NAME); | |
633 | bind (SERVER, $uaddr) || die "bind: $!"; | |
634 | listen(SERVER,5) || die "listen: $!"; | |
635 | ||
636 | logmsg "server started on $NAME"; | |
637 | ||
638 | $SIG{CHLD} = \&REAPER; | |
639 | ||
640 | for ( $waitedpid = 0; | |
641 | accept(CLIENT,SERVER) || $waitedpid; | |
642 | $waitedpid = 0, close CLIENT) | |
643 | { | |
644 | next if $waitedpid; | |
645 | logmsg "connection on $NAME"; | |
646 | spawn sub { | |
647 | print "Hello there, it's now ", scalar localtime, "\n"; | |
648 | exec '/usr/games/fortune' or die "can't exec fortune: $!"; | |
649 | }; | |
650 | } | |
651 | ||
652 | As you see, it's remarkably similar to the Internet domain TCP server, so | |
653 | much so, in fact, that we've omitted several duplicate functions--spawn(), | |
654 | logmsg(), ctime(), and REAPER()--which are exactly the same as in the | |
655 | other server. | |
656 | ||
657 | So why would you ever want to use a Unix domain socket instead of a | |
658 | simpler named pipe? Because a named pipe doesn't give you sessions. You | |
659 | can't tell one process's data from another's. With socket programming, | |
660 | you get a separate session for each client: that's why accept() takes two | |
661 | arguments. | |
662 | ||
663 | For example, let's say that you have a long running database server daemon | |
664 | that you want folks from the World Wide Web to be able to access, but only | |
665 | if they go through a CGI interface. You'd have a small, simple CGI | |
666 | program that does whatever checks and logging you feel like, and then acts | |
667 | as a Unix-domain client and connects to your private server. | |
668 | ||
669 | =head2 UDP: Message Passing | |
670 | ||
671 | Another kind of client-server setup is one that uses not connections, but | |
672 | messages. UDP communications involve much lower overhead but also provide | |
673 | less reliability, as there are no promises that messages will arrive at | |
674 | all, let alone in order and unmangled. Still, UDP offers some advantages | |
675 | over TCP, including being able to "broadcast" or "multicast" to a whole | |
676 | bunch of destination hosts at once (usually on your local subnet). If you | |
677 | find yourself overly concerned about reliability and start building checks | |
678 | into your message system, then you probably should just use TCP to start | |
679 | with. | |
680 | ||
681 | Here's a UDP program similar to the sample Internet TCP client given | |
682 | above. However, instead of checking one host at a time, the UDP version | |
683 | will check many of them asynchronously by simulating a multicast and then | |
684 | using select() to do a timed-out wait for I/O. To do something similar | |
685 | with TCP, you'd have to use a different socket handle for each host. | |
686 | ||
687 | #!/usr/bin/perl -w | |
688 | use strict; | |
689 | require 5.002; | |
690 | use Socket; | |
691 | use Sys::Hostname; | |
692 | ||
693 | my ( $count, $hisiaddr, $hispaddr, $histime, | |
694 | $host, $iaddr, $paddr, $port, $proto, | |
695 | $rin, $rout, $rtime, $SECS_of_70_YEARS); | |
696 | ||
697 | $SECS_of_70_YEARS = 2208988800; | |
698 | ||
699 | $iaddr = gethostbyname(hostname()); | |
700 | $proto = getprotobyname('udp'); | |
701 | $port = getservbyname('time', 'udp'); | |
702 | $paddr = sockaddr_in(0, $iaddr); # 0 means let kernel pick | |
703 | ||
704 | socket(SOCKET, PF_INET, SOCK_DGRAM, $proto) || die "socket: $!"; | |
705 | bind(SOCKET, $paddr) || die "bind: $!"; | |
706 | ||
707 | $| = 1; | |
708 | printf "%-12s %8s %s\n", "localhost", 0, scalar localtime time; | |
709 | $count = 0; | |
710 | for $host (@ARGV) { | |
711 | $count++; | |
712 | $hisiaddr = inet_aton($host) || die "unknown host"; | |
713 | $hispaddr = sockaddr_in($port, $hisiaddr); | |
714 | defined(send(SOCKET, 0, 0, $hispaddr)) || die "send $host: $!"; | |
715 | } | |
716 | ||
717 | $rin = ''; | |
718 | vec($rin, fileno(SOCKET), 1) = 1; | |
719 | ||
720 | # timeout after 10.0 seconds | |
721 | while ($count && select($rout = $rin, undef, undef, 10.0)) { | |
722 | $rtime = ''; | |
723 | ($hispaddr = recv(SOCKET, $rtime, 4, 0)) || die "recv: $!"; | |
724 | ($port, $hisiaddr) = sockaddr_in($hispaddr); | |
725 | $host = gethostbyaddr($hisiaddr, AF_INET); | |
726 | $histime = unpack("N", $rtime) - $SECS_of_70_YEARS ; | |
727 | printf "%-12s ", $host; | |
728 | printf "%8d %s\n", $histime - time, scalar localtime($histime); | |
729 | $count--; | |
730 | } | |
731 | ||
732 | =head1 SysV IPC | |
733 | ||
734 | While System V IPC isn't so widely used as sockets, it still has some | |
735 | interesting uses. You can't, however, effectively use SysV IPC or | |
736 | Berkeley mmap() to have shared memory so as to share a variable amongst | |
737 | several processes. That's because Perl would reallocate your string when | |
738 | you weren't wanting it to. | |
739 | ||
740 | ||
741 | Here's a small example showing shared memory usage. | |
a0d0e21e LW |
742 | |
743 | $IPC_PRIVATE = 0; | |
744 | $IPC_RMID = 0; | |
745 | $size = 2000; | |
746 | $key = shmget($IPC_PRIVATE, $size , 0777 ); | |
4633a7c4 | 747 | die unless defined $key; |
a0d0e21e LW |
748 | |
749 | $message = "Message #1"; | |
750 | shmwrite($key, $message, 0, 60 ) || die "$!"; | |
751 | shmread($key,$buff,0,60) || die "$!"; | |
752 | ||
753 | print $buff,"\n"; | |
754 | ||
755 | print "deleting $key\n"; | |
756 | shmctl($key ,$IPC_RMID, 0) || die "$!"; | |
757 | ||
758 | Here's an example of a semaphore: | |
759 | ||
760 | $IPC_KEY = 1234; | |
761 | $IPC_RMID = 0; | |
762 | $IPC_CREATE = 0001000; | |
763 | $key = semget($IPC_KEY, $nsems , 0666 | $IPC_CREATE ); | |
764 | die if !defined($key); | |
765 | print "$key\n"; | |
766 | ||
767 | Put this code in a separate file to be run in more that one process | |
768 | Call the file F<take>: | |
769 | ||
770 | # create a semaphore | |
771 | ||
772 | $IPC_KEY = 1234; | |
773 | $key = semget($IPC_KEY, 0 , 0 ); | |
774 | die if !defined($key); | |
775 | ||
776 | $semnum = 0; | |
777 | $semflag = 0; | |
778 | ||
779 | # 'take' semaphore | |
780 | # wait for semaphore to be zero | |
781 | $semop = 0; | |
782 | $opstring1 = pack("sss", $semnum, $semop, $semflag); | |
783 | ||
784 | # Increment the semaphore count | |
785 | $semop = 1; | |
786 | $opstring2 = pack("sss", $semnum, $semop, $semflag); | |
787 | $opstring = $opstring1 . $opstring2; | |
788 | ||
789 | semop($key,$opstring) || die "$!"; | |
790 | ||
791 | Put this code in a separate file to be run in more that one process | |
792 | Call this file F<give>: | |
793 | ||
4633a7c4 | 794 | # 'give' the semaphore |
a0d0e21e LW |
795 | # run this in the original process and you will see |
796 | # that the second process continues | |
797 | ||
798 | $IPC_KEY = 1234; | |
799 | $key = semget($IPC_KEY, 0, 0); | |
800 | die if !defined($key); | |
801 | ||
802 | $semnum = 0; | |
803 | $semflag = 0; | |
804 | ||
805 | # Decrement the semaphore count | |
806 | $semop = -1; | |
807 | $opstring = pack("sss", $semnum, $semop, $semflag); | |
808 | ||
809 | semop($key,$opstring) || die "$!"; | |
810 | ||
4633a7c4 LW |
811 | =head1 WARNING |
812 | ||
813 | The SysV IPC code above was written long ago, and it's definitely clunky | |
814 | looking. It should at the very least be made to C<use strict> and | |
815 | C<require "sys/ipc.ph">. Better yet, perhaps someone should create an | |
816 | C<IPC::SysV> module the way we have the C<Socket> module for normal | |
817 | client-server communications. | |
818 | ||
819 | (... time passes) | |
820 | ||
821 | Voila! Check out the IPC::SysV modules written by Jack Shirazi. You can | |
822 | find them at a CPAN store near you. | |
823 | ||
824 | =head1 NOTES | |
825 | ||
826 | If you are running under version 5.000 (dubious) or 5.001, you can still | |
827 | use most of the examples in this document. You may have to remove the | |
828 | C<use strict> and some of the my() statements for 5.000, and for both | |
829 | you'll have to load in version 1.2 of the F<Socket.pm> module, which | |
830 | was/is/shall-be included in I<perl5.001o>. | |
831 | ||
832 | Most of these routines quietly but politely return C<undef> when they fail | |
833 | instead of causing your program to die right then and there due to an | |
834 | uncaught exception. (Actually, some of the new I<Socket> conversion | |
835 | functions croak() on bad arguments.) It is therefore essential | |
836 | that you should check the return values fo these functions. Always begin | |
837 | your socket programs this way for optimal success, and don't forget to add | |
838 | B<-T> taint checking flag to the pound-bang line for servers: | |
839 | ||
840 | #!/usr/bin/perl -w | |
841 | require 5.002; | |
842 | use strict; | |
843 | use sigtrap; | |
844 | use Socket; | |
845 | ||
846 | =head1 BUGS | |
847 | ||
848 | All these routines create system-specific portability problems. As noted | |
849 | elsewhere, Perl is at the mercy of your C libraries for much of its system | |
850 | behaviour. It's probably safest to assume broken SysV semantics for | |
851 | signals and to stick with simple TCP and UDP socket operations; e.g. don't | |
852 | try to pass open filedescriptors over a local UDP datagram socket if you | |
853 | want your code to stand a chance of being portable. | |
854 | ||
855 | Because few vendors provide C libraries that are safely | |
856 | re-entrant, the prudent programmer will do little else within | |
857 | a handler beyond die() to raise an exception and longjmp(3) out. | |
858 | ||
859 | =head1 AUTHOR | |
860 | ||
861 | Tom Christiansen, with occasional vestiges of Larry Wall's original | |
862 | version. | |
863 | ||
864 | =head1 SEE ALSO | |
865 | ||
866 | Besides the obvious functions in L<perlfunc>, you should also check out | |
867 | the F<modules> file at your nearest CPAN site. (See L<perlmod> or best | |
868 | yet, the F<Perl FAQ> for a description of what CPAN is and where to get it.) | |
869 | Section 5 of the F<modules> file is devoted to "Networking, Device Control | |
870 | (modems) and Interprocess Communication", and contains numerous unbundled | |
871 | modules numerous networking modules, Chat and Expect operations, CGI | |
872 | programming, DCE, FTP, IPC, NNTP, Proxy, Ptty, RPC, SNMP, SMTP, Telnet, | |
873 | Threads, and ToolTalk--just to name a few. |