1 |
=head1 NAME |
2 |
|
3 |
AnyEvent::Fork - everything you wanted to use fork() for, but couldn't |
4 |
|
5 |
ATTENTION, this is a very early release, and very untested. Consider it a |
6 |
technology preview. |
7 |
|
8 |
=head1 SYNOPSIS |
9 |
|
10 |
use AnyEvent::Fork; |
11 |
|
12 |
################################################################## |
13 |
# create a single new process, tell it to run your worker function |
14 |
|
15 |
AnyEvent::Fork |
16 |
->new |
17 |
->require ("MyModule") |
18 |
->run ("MyModule::worker, sub { |
19 |
my ($master_filehandle) = @_; |
20 |
|
21 |
# now $master_filehandle is connected to the |
22 |
# $slave_filehandle in the new process. |
23 |
}); |
24 |
|
25 |
# MyModule::worker might look like this |
26 |
sub MyModule::worker { |
27 |
my ($slave_filehandle) = @_; |
28 |
|
29 |
# now $slave_filehandle is connected to the $master_filehandle |
30 |
# in the original prorcess. have fun! |
31 |
} |
32 |
|
33 |
################################################################## |
34 |
# create a pool of server processes all accepting on the same socket |
35 |
|
36 |
# create listener socket |
37 |
my $listener = ...; |
38 |
|
39 |
# create a pool template, initialise it and give it the socket |
40 |
my $pool = AnyEvent::Fork |
41 |
->new |
42 |
->require ("Some::Stuff", "My::Server") |
43 |
->send_fh ($listener); |
44 |
|
45 |
# now create 10 identical workers |
46 |
for my $id (1..10) { |
47 |
$pool |
48 |
->fork |
49 |
->send_arg ($id) |
50 |
->run ("My::Server::run"); |
51 |
} |
52 |
|
53 |
# now do other things - maybe use the filehandle provided by run |
54 |
# to wait for the processes to die. or whatever. |
55 |
|
56 |
# My::Server::run might look like this |
57 |
sub My::Server::run { |
58 |
my ($slave, $listener, $id) = @_; |
59 |
|
60 |
close $slave; # we do not use the socket, so close it to save resources |
61 |
|
62 |
# we could go ballistic and use e.g. AnyEvent here, or IO::AIO, |
63 |
# or anything we usually couldn't do in a process forked normally. |
64 |
while (my $socket = $listener->accept) { |
65 |
# do sth. with new socket |
66 |
} |
67 |
} |
68 |
|
69 |
=head1 DESCRIPTION |
70 |
|
71 |
This module allows you to create new processes, without actually forking |
72 |
them from your current process (avoiding the problems of forking), but |
73 |
preserving most of the advantages of fork. |
74 |
|
75 |
It can be used to create new worker processes or new independent |
76 |
subprocesses for short- and long-running jobs, process pools (e.g. for use |
77 |
in pre-forked servers) but also to spawn new external processes (such as |
78 |
CGI scripts from a webserver), which can be faster (and more well behaved) |
79 |
than using fork+exec in big processes. |
80 |
|
81 |
Special care has been taken to make this module useful from other modules, |
82 |
while still supporting specialised environments such as L<App::Staticperl> |
83 |
or L<PAR::Packer>. |
84 |
|
85 |
=head1 PROBLEM STATEMENT |
86 |
|
87 |
There are two ways to implement parallel processing on UNIX like operating |
88 |
systems - fork and process, and fork+exec and process. They have different |
89 |
advantages and disadvantages that I describe below, together with how this |
90 |
module tries to mitigate the disadvantages. |
91 |
|
92 |
=over 4 |
93 |
|
94 |
=item Forking from a big process can be very slow (a 5GB process needs |
95 |
0.05s to fork on my 3.6GHz amd64 GNU/Linux box for example). This overhead |
96 |
is often shared with exec (because you have to fork first), but in some |
97 |
circumstances (e.g. when vfork is used), fork+exec can be much faster. |
98 |
|
99 |
This module can help here by telling a small(er) helper process to fork, |
100 |
or fork+exec instead. |
101 |
|
102 |
=item Forking usually creates a copy-on-write copy of the parent |
103 |
process. Memory (for example, modules or data files that have been |
104 |
will not take additional memory). When exec'ing a new process, modules |
105 |
and data files might need to be loaded again, at extra cpu and memory |
106 |
cost. Likewise when forking, all data structures are copied as well - if |
107 |
the program frees them and replaces them by new data, the child processes |
108 |
will retain the memory even if it isn't used. |
109 |
|
110 |
This module allows the main program to do a controlled fork, and allows |
111 |
modules to exec processes safely at any time. When creating a custom |
112 |
process pool you can take advantage of data sharing via fork without |
113 |
risking to share large dynamic data structures that will blow up child |
114 |
memory usage. |
115 |
|
116 |
=item Exec'ing a new perl process might be difficult and slow. For |
117 |
example, it is not easy to find the correct path to the perl interpreter, |
118 |
and all modules have to be loaded from disk again. Long running processes |
119 |
might run into problems when perl is upgraded for example. |
120 |
|
121 |
This module supports creating pre-initialised perl processes to be used |
122 |
as template, and also tries hard to identify the correct path to the perl |
123 |
interpreter. With a cooperative main program, exec'ing the interpreter |
124 |
might not even be necessary. |
125 |
|
126 |
=item Forking might be impossible when a program is running. For example, |
127 |
POSIX makes it almost impossible to fork from a multithreaded program and |
128 |
do anything useful in the child - strictly speaking, if your perl program |
129 |
uses posix threads (even indirectly via e.g. L<IO::AIO> or L<threads>), |
130 |
you cannot call fork on the perl level anymore, at all. |
131 |
|
132 |
This module can safely fork helper processes at any time, by caling |
133 |
fork+exec in C, in a POSIX-compatible way. |
134 |
|
135 |
=item Parallel processing with fork might be inconvenient or difficult |
136 |
to implement. For example, when a program uses an event loop and creates |
137 |
watchers it becomes very hard to use the event loop from a child |
138 |
program, as the watchers already exist but are only meaningful in the |
139 |
parent. Worse, a module might want to use such a system, not knowing |
140 |
whether another module or the main program also does, leading to problems. |
141 |
|
142 |
This module only lets the main program create pools by forking (because |
143 |
only the main program can know when it is still safe to do so) - all other |
144 |
pools are created by fork+exec, after which such modules can again be |
145 |
loaded. |
146 |
|
147 |
=back |
148 |
|
149 |
=head1 CONCEPTS |
150 |
|
151 |
This module can create new processes either by executing a new perl |
152 |
process, or by forking from an existing "template" process. |
153 |
|
154 |
Each such process comes with its own file handle that can be used to |
155 |
communicate with it (it's actually a socket - one end in the new process, |
156 |
one end in the main process), and among the things you can do in it are |
157 |
load modules, fork new processes, send file handles to it, and execute |
158 |
functions. |
159 |
|
160 |
There are multiple ways to create additional processes to execute some |
161 |
jobs: |
162 |
|
163 |
=over 4 |
164 |
|
165 |
=item fork a new process from the "default" template process, load code, |
166 |
run it |
167 |
|
168 |
This module has a "default" template process which it executes when it is |
169 |
needed the first time. Forking from this process shares the memory used |
170 |
for the perl interpreter with the new process, but loading modules takes |
171 |
time, and the memory is not shared with anything else. |
172 |
|
173 |
This is ideal for when you only need one extra process of a kind, with the |
174 |
option of starting and stipping it on demand. |
175 |
|
176 |
Example: |
177 |
|
178 |
AnyEvent::Fork |
179 |
->new |
180 |
->require ("Some::Module") |
181 |
->run ("Some::Module::run", sub { |
182 |
my ($fork_fh) = @_; |
183 |
}); |
184 |
|
185 |
=item fork a new template process, load code, then fork processes off of |
186 |
it and run the code |
187 |
|
188 |
When you need to have a bunch of processes that all execute the same (or |
189 |
very similar) tasks, then a good way is to create a new template process |
190 |
for them, loading all the modules you need, and then create your worker |
191 |
processes from this new template process. |
192 |
|
193 |
This way, all code (and data structures) that can be shared (e.g. the |
194 |
modules you loaded) is shared between the processes, and each new process |
195 |
consumes relatively little memory of its own. |
196 |
|
197 |
The disadvantage of this approach is that you need to create a template |
198 |
process for the sole purpose of forking new processes from it, but if you |
199 |
only need a fixed number of proceses you can create them, and then destroy |
200 |
the template process. |
201 |
|
202 |
Example: |
203 |
|
204 |
my $template = AnyEvent::Fork->new->require ("Some::Module"); |
205 |
|
206 |
for (1..10) { |
207 |
$template->fork->run ("Some::Module::run", sub { |
208 |
my ($fork_fh) = @_; |
209 |
}); |
210 |
} |
211 |
|
212 |
# at this point, you can keep $template around to fork new processes |
213 |
# later, or you can destroy it, which causes it to vanish. |
214 |
|
215 |
=item execute a new perl interpreter, load some code, run it |
216 |
|
217 |
This is relatively slow, and doesn't allow you to share memory between |
218 |
multiple processes. |
219 |
|
220 |
The only advantage is that you don't have to have a template process |
221 |
hanging around all the time to fork off some new processes, which might be |
222 |
an advantage when there are long time spans where no extra processes are |
223 |
needed. |
224 |
|
225 |
Example: |
226 |
|
227 |
AnyEvent::Fork |
228 |
->new_exec |
229 |
->require ("Some::Module") |
230 |
->run ("Some::Module::run", sub { |
231 |
my ($fork_fh) = @_; |
232 |
}); |
233 |
|
234 |
=back |
235 |
|
236 |
=head1 FUNCTIONS |
237 |
|
238 |
=over 4 |
239 |
|
240 |
=cut |
241 |
|
242 |
package AnyEvent::Fork; |
243 |
|
244 |
use common::sense; |
245 |
|
246 |
use Socket (); |
247 |
|
248 |
use AnyEvent; |
249 |
use AnyEvent::Fork::Util; |
250 |
use AnyEvent::Util (); |
251 |
|
252 |
our $VERSION = $AnyEvent::Fork::Util::VERSION; |
253 |
|
254 |
our $PERL; # the path to the perl interpreter, deduces with various forms of magic |
255 |
|
256 |
=item my $pool = new AnyEvent::Fork key => value... |
257 |
|
258 |
Create a new process pool. The following named parameters are supported: |
259 |
|
260 |
=over 4 |
261 |
|
262 |
=back |
263 |
|
264 |
=cut |
265 |
|
266 |
# the early fork template process |
267 |
our $EARLY; |
268 |
|
269 |
# the empty template process |
270 |
our $TEMPLATE; |
271 |
|
272 |
sub _cmd { |
273 |
my $self = shift; |
274 |
|
275 |
#TODO: maybe append the packet to any existing string command already in the queue |
276 |
|
277 |
# ideally, we would want to use "a (w/a)*" as format string, but perl versions |
278 |
# from at least 5.8.9 to 5.16.3 are all buggy and can't unpack it. |
279 |
push @{ $self->[2] }, pack "N/a*", pack "(w/a*)*", @_; |
280 |
|
281 |
$self->[3] ||= AE::io $self->[1], 1, sub { |
282 |
# send the next "thing" in the queue - either a reference to an fh, |
283 |
# or a plain string. |
284 |
|
285 |
if (ref $self->[2][0]) { |
286 |
# send fh |
287 |
AnyEvent::Fork::Util::fd_send fileno $self->[1], fileno ${ $self->[2][0] } |
288 |
and shift @{ $self->[2] }; |
289 |
|
290 |
} else { |
291 |
# send string |
292 |
my $len = syswrite $self->[1], $self->[2][0] |
293 |
or do { undef $self->[3]; die "AnyEvent::Fork: command write failure: $!" }; |
294 |
|
295 |
substr $self->[2][0], 0, $len, ""; |
296 |
shift @{ $self->[2] } unless length $self->[2][0]; |
297 |
} |
298 |
|
299 |
unless (@{ $self->[2] }) { |
300 |
undef $self->[3]; |
301 |
# invoke run callback |
302 |
$self->[0]->($self->[1]) if $self->[0]; |
303 |
} |
304 |
}; |
305 |
} |
306 |
|
307 |
sub _new { |
308 |
my ($self, $fh) = @_; |
309 |
|
310 |
AnyEvent::Util::fh_nonblocking $fh, 1; |
311 |
|
312 |
$self = bless [ |
313 |
undef, # run callback |
314 |
$fh, |
315 |
[], # write queue - strings or fd's |
316 |
undef, # AE watcher |
317 |
], $self; |
318 |
|
319 |
$self |
320 |
} |
321 |
|
322 |
# fork template from current process, used by AnyEvent::Fork::Early/Template |
323 |
sub _new_fork { |
324 |
my ($fh, $slave) = AnyEvent::Util::portable_socketpair; |
325 |
my $parent = $$; |
326 |
|
327 |
my $pid = fork; |
328 |
|
329 |
if ($pid eq 0) { |
330 |
require AnyEvent::Fork::Serve; |
331 |
$AnyEvent::Fork::Serve::OWNER = $parent; |
332 |
close $fh; |
333 |
$0 = "$_[1] of $parent"; |
334 |
AnyEvent::Fork::Serve::serve ($slave); |
335 |
AnyEvent::Fork::Util::_exit 0; |
336 |
} elsif (!$pid) { |
337 |
die "AnyEvent::Fork::Early/Template: unable to fork template process: $!"; |
338 |
} |
339 |
|
340 |
AnyEvent::Fork->_new ($fh) |
341 |
} |
342 |
|
343 |
=item my $proc = new AnyEvent::Fork |
344 |
|
345 |
Create a new "empty" perl interpreter process and returns its process |
346 |
object for further manipulation. |
347 |
|
348 |
The new process is forked from a template process that is kept around |
349 |
for this purpose. When it doesn't exist yet, it is created by a call to |
350 |
C<new_exec> and kept around for future calls. |
351 |
|
352 |
When the process object is destroyed, it will release the file handle |
353 |
that connects it with the new process. When the new process has not yet |
354 |
called C<run>, then the process will exit. Otherwise, what happens depends |
355 |
entirely on the code that is executed. |
356 |
|
357 |
=cut |
358 |
|
359 |
sub new { |
360 |
my $class = shift; |
361 |
|
362 |
$TEMPLATE ||= $class->new_exec; |
363 |
$TEMPLATE->fork |
364 |
} |
365 |
|
366 |
=item $new_proc = $proc->fork |
367 |
|
368 |
Forks C<$proc>, creating a new process, and returns the process object |
369 |
of the new process. |
370 |
|
371 |
If any of the C<send_> functions have been called before fork, then they |
372 |
will be cloned in the child. For example, in a pre-forked server, you |
373 |
might C<send_fh> the listening socket into the template process, and then |
374 |
keep calling C<fork> and C<run>. |
375 |
|
376 |
=cut |
377 |
|
378 |
sub fork { |
379 |
my ($self) = @_; |
380 |
|
381 |
my ($fh, $slave) = AnyEvent::Util::portable_socketpair; |
382 |
|
383 |
$self->send_fh ($slave); |
384 |
$self->_cmd ("f"); |
385 |
|
386 |
AnyEvent::Fork->_new ($fh) |
387 |
} |
388 |
|
389 |
=item my $proc = new_exec AnyEvent::Fork |
390 |
|
391 |
Create a new "empty" perl interpreter process and returns its process |
392 |
object for further manipulation. |
393 |
|
394 |
Unlike the C<new> method, this method I<always> spawns a new perl process |
395 |
(except in some cases, see L<AnyEvent::Fork::Early> for details). This |
396 |
reduces the amount of memory sharing that is possible, and is also slower. |
397 |
|
398 |
You should use C<new> whenever possible, except when having a template |
399 |
process around is unacceptable. |
400 |
|
401 |
The path to the perl interpreter is divined usign various methods - first |
402 |
C<$^X> is investigated to see if the path ends with something that sounds |
403 |
as if it were the perl interpreter. Failing this, the module falls back to |
404 |
using C<$Config::Config{perlpath}>. |
405 |
|
406 |
=cut |
407 |
|
408 |
sub new_exec { |
409 |
my ($self) = @_; |
410 |
|
411 |
return $EARLY->fork |
412 |
if $EARLY; |
413 |
|
414 |
# first find path of perl |
415 |
my $perl = $; |
416 |
|
417 |
# first we try $^X, but the path must be absolute (always on win32), and end in sth. |
418 |
# that looks like perl. this obviously only works for posix and win32 |
419 |
unless ( |
420 |
(AnyEvent::Fork::Util::WIN32 || $perl =~ m%^/%) |
421 |
&& $perl =~ m%[/\\]perl(?:[0-9]+(\.[0-9]+)+)?(\.exe)?$%i |
422 |
) { |
423 |
# if it doesn't look perlish enough, try Config |
424 |
require Config; |
425 |
$perl = $Config::Config{perlpath}; |
426 |
$perl =~ s/(?:\Q$Config::Config{_exe}\E)?$/$Config::Config{_exe}/; |
427 |
} |
428 |
|
429 |
require Proc::FastSpawn; |
430 |
|
431 |
my ($fh, $slave) = AnyEvent::Util::portable_socketpair; |
432 |
Proc::FastSpawn::fd_inherit (fileno $slave); |
433 |
|
434 |
# new fh's should always be set cloexec (due to $^F), |
435 |
# but hey, not on win32, so we always clear the inherit flag. |
436 |
Proc::FastSpawn::fd_inherit (fileno $fh, 0); |
437 |
|
438 |
# quick. also doesn't work in win32. of course. what did you expect |
439 |
#local $ENV{PERL5LIB} = join ":", grep !ref, @INC; |
440 |
my %env = %ENV; |
441 |
$env{PERL5LIB} = join +(AnyEvent::Fork::Util::WIN32 ? ";" : ":"), grep !ref, @INC; |
442 |
|
443 |
Proc::FastSpawn::spawn ( |
444 |
$perl, |
445 |
["perl", "-MAnyEvent::Fork::Serve", "-e", "AnyEvent::Fork::Serve::me", fileno $slave, $$], |
446 |
[map "$_=$env{$_}", keys %env], |
447 |
) or die "unable to spawn AnyEvent::Fork server: $!"; |
448 |
|
449 |
$self->_new ($fh) |
450 |
} |
451 |
|
452 |
=item $proc = $proc->eval ($perlcode, @args) |
453 |
|
454 |
Evaluates the given C<$perlcode> as ... perl code, while setting C<@_> to |
455 |
the strings specified by C<@args>. |
456 |
|
457 |
This call is meant to do any custom initialisation that might be required |
458 |
(for example, the C<require> method uses it). It's not supposed to be used |
459 |
to completely take over the process, use C<run> for that. |
460 |
|
461 |
The code will usually be executed after this call returns, and there is no |
462 |
way to pass anything back to the calling process. Any evaluation errors |
463 |
will be reported to stderr and cause the process to exit. |
464 |
|
465 |
Returns the process object for easy chaining of method calls. |
466 |
|
467 |
=cut |
468 |
|
469 |
sub eval { |
470 |
my ($self, $code, @args) = @_; |
471 |
|
472 |
$self->_cmd (e => $code, @args); |
473 |
|
474 |
$self |
475 |
} |
476 |
|
477 |
=item $proc = $proc->require ($module, ...) |
478 |
|
479 |
Tries to load the given module(s) into the process |
480 |
|
481 |
Returns the process object for easy chaining of method calls. |
482 |
|
483 |
=cut |
484 |
|
485 |
sub require { |
486 |
my ($self, @modules) = @_; |
487 |
|
488 |
s%::%/%g for @modules; |
489 |
$self->eval ('require "$_.pm" for @_', @modules); |
490 |
|
491 |
$self |
492 |
} |
493 |
|
494 |
=item $proc = $proc->send_fh ($handle, ...) |
495 |
|
496 |
Send one or more file handles (I<not> file descriptors) to the process, |
497 |
to prepare a call to C<run>. |
498 |
|
499 |
The process object keeps a reference to the handles until this is done, |
500 |
so you must not explicitly close the handles. This is most easily |
501 |
accomplished by simply not storing the file handles anywhere after passing |
502 |
them to this method. |
503 |
|
504 |
Returns the process object for easy chaining of method calls. |
505 |
|
506 |
Example: pass an fh to a process, and release it without closing. it will |
507 |
be closed automatically when it is no longer used. |
508 |
|
509 |
$proc->send_fh ($my_fh); |
510 |
undef $my_fh; # free the reference if you want, but DO NOT CLOSE IT |
511 |
|
512 |
=cut |
513 |
|
514 |
sub send_fh { |
515 |
my ($self, @fh) = @_; |
516 |
|
517 |
for my $fh (@fh) { |
518 |
$self->_cmd ("h"); |
519 |
push @{ $self->[2] }, \$fh; |
520 |
} |
521 |
|
522 |
$self |
523 |
} |
524 |
|
525 |
=item $proc = $proc->send_arg ($string, ...) |
526 |
|
527 |
Send one or more argument strings to the process, to prepare a call to |
528 |
C<run>. The strings can be any octet string. |
529 |
|
530 |
Returns the process object for easy chaining of emthod calls. |
531 |
|
532 |
=cut |
533 |
|
534 |
sub send_arg { |
535 |
my ($self, @arg) = @_; |
536 |
|
537 |
$self->_cmd (a => @arg); |
538 |
|
539 |
$self |
540 |
} |
541 |
|
542 |
=item $proc->run ($func, $cb->($fh)) |
543 |
|
544 |
Enter the function specified by the fully qualified name in C<$func> in |
545 |
the process. The function is called with the communication socket as first |
546 |
argument, followed by all file handles and string arguments sent earlier |
547 |
via C<send_fh> and C<send_arg> methods, in the order they were called. |
548 |
|
549 |
If the called function returns, the process exits. |
550 |
|
551 |
Preparing the process can take time - when the process is ready, the |
552 |
callback is invoked with the local communications socket as argument. |
553 |
|
554 |
The process object becomes unusable on return from this function. |
555 |
|
556 |
If the communication socket isn't used, it should be closed on both sides, |
557 |
to save on kernel memory. |
558 |
|
559 |
The socket is non-blocking in the parent, and blocking in the newly |
560 |
created process. The close-on-exec flag is set on both. Even if not used |
561 |
otherwise, the socket can be a good indicator for the existance of the |
562 |
process - if the other process exits, you get a readable event on it, |
563 |
because exiting the process closes the socket (if it didn't create any |
564 |
children using fork). |
565 |
|
566 |
Example: create a template for a process pool, pass a few strings, some |
567 |
file handles, then fork, pass one more string, and run some code. |
568 |
|
569 |
my $pool = AnyEvent::Fork |
570 |
->new |
571 |
->send_arg ("str1", "str2") |
572 |
->send_fh ($fh1, $fh2); |
573 |
|
574 |
for (1..2) { |
575 |
$pool |
576 |
->fork |
577 |
->send_arg ("str3") |
578 |
->run ("Some::function", sub { |
579 |
my ($fh) = @_; |
580 |
|
581 |
# fh is nonblocking, but we trust that the OS can accept these |
582 |
# extra 3 octets anyway. |
583 |
syswrite $fh, "hi #$_\n"; |
584 |
|
585 |
# $fh is being closed here, as we don't store it anywhere |
586 |
}); |
587 |
} |
588 |
|
589 |
# Some::function might look like this - all parameters passed before fork |
590 |
# and after will be passed, in order, after the communications socket. |
591 |
sub Some::function { |
592 |
my ($fh, $str1, $str2, $fh1, $fh2, $str3) = @_; |
593 |
|
594 |
print scalar <$fh>; # prints "hi 1\n" and "hi 2\n" |
595 |
} |
596 |
|
597 |
=cut |
598 |
|
599 |
sub run { |
600 |
my ($self, $func, $cb) = @_; |
601 |
|
602 |
$self->[0] = $cb; |
603 |
$self->_cmd (r => $func); |
604 |
} |
605 |
|
606 |
=back |
607 |
|
608 |
=head1 PORTABILITY NOTES |
609 |
|
610 |
Native win32 perls are somewhat supported (AnyEvent::Fork::Early is a nop, |
611 |
and ::Template is not going to work), and it cost a lot of blood and sweat |
612 |
to make it so, mostly due to the bloody broken perl that nobody seems to |
613 |
care about. The fork emulation is a bad joke - I have yet to see something |
614 |
useful that you cna do with it without running into memory corruption |
615 |
issues or other braindamage. Hrrrr. |
616 |
|
617 |
Cygwin perl is not supported at the moment, as it should implement fd |
618 |
passing, but doesn't, and rolling my own is hard, as cygwin doesn't |
619 |
support enough functionality to do it. |
620 |
|
621 |
=head1 AUTHOR |
622 |
|
623 |
Marc Lehmann <schmorp@schmorp.de> |
624 |
http://home.schmorp.de/ |
625 |
|
626 |
=cut |
627 |
|
628 |
1 |
629 |
|