ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/AnyEvent-Fork/Fork.pm
(Generate patch)

Comparing AnyEvent-Fork/Fork.pm (file contents):
Revision 1.6 by root, Wed Apr 3 08:47:44 2013 UTC vs.
Revision 1.38 by root, Sat Apr 6 20:07:30 2013 UTC

3AnyEvent::Fork - everything you wanted to use fork() for, but couldn't 3AnyEvent::Fork - everything you wanted to use fork() for, but couldn't
4 4
5=head1 SYNOPSIS 5=head1 SYNOPSIS
6 6
7 use AnyEvent::Fork; 7 use AnyEvent::Fork;
8
9 AnyEvent::Fork
10 ->new
11 ->require ("MyModule")
12 ->run ("MyModule::server", my $cv = AE::cv);
13
14 my $fh = $cv->recv;
8 15
9=head1 DESCRIPTION 16=head1 DESCRIPTION
10 17
11This module allows you to create new processes, without actually forking 18This module allows you to create new processes, without actually forking
12them from your current process (avoiding the problems of forking), but 19them from your current process (avoiding the problems of forking), but
13preserving most of the advantages of fork. 20preserving most of the advantages of fork.
14 21
15It can be used to create new worker processes or new independent 22It can be used to create new worker processes or new independent
16subprocesses for short- and long-running jobs, process pools (e.g. for use 23subprocesses for short- and long-running jobs, process pools (e.g. for use
17in pre-forked servers) but also to spawn new external processes (such as 24in pre-forked servers) but also to spawn new external processes (such as
18CGI scripts from a webserver), which can be faster (and more well behaved) 25CGI scripts from a web server), which can be faster (and more well behaved)
19than using fork+exec in big processes. 26than using fork+exec in big processes.
20 27
21Special care has been taken to make this module useful from other modules, 28Special care has been taken to make this module useful from other modules,
22while still supporting specialised environments such as L<App::Staticperl> 29while still supporting specialised environments such as L<App::Staticperl>
23or L<PAR::Packer>. 30or L<PAR::Packer>.
24 31
32=head1 WHAT THIS MODULE IS NOT
33
34This module only creates processes and lets you pass file handles and
35strings to it, and run perl code. It does not implement any kind of RPC -
36there is no back channel from the process back to you, and there is no RPC
37or message passing going on.
38
39If you need some form of RPC, you can either implement it yourself
40in whatever way you like, use some message-passing module such
41as L<AnyEvent::MP>, some pipe such as L<AnyEvent::ZeroMQ>, use
42L<AnyEvent::Handle> on both sides to send e.g. JSON or Storable messages,
43and so on.
44
25=head1 PROBLEM STATEMENT 45=head1 PROBLEM STATEMENT
26 46
27There are two ways to implement parallel processing on UNIX like operating 47There are two traditional ways to implement parallel processing on UNIX
28systems - fork and process, and fork+exec and process. They have different 48like operating systems - fork and process, and fork+exec and process. They
29advantages and disadvantages that I describe below, together with how this 49have different advantages and disadvantages that I describe below,
30module tries to mitigate the disadvantages. 50together with how this module tries to mitigate the disadvantages.
31 51
32=over 4 52=over 4
33 53
34=item Forking from a big process can be very slow (a 5GB process needs 54=item Forking from a big process can be very slow.
350.05s to fork on my 3.6GHz amd64 GNU/Linux box for example). This overhead 55
56A 5GB process needs 0.05s to fork on my 3.6GHz amd64 GNU/Linux box. This
36is often shared with exec (because you have to fork first), but in some 57overhead is often shared with exec (because you have to fork first), but
37circumstances (e.g. when vfork is used), fork+exec can be much faster. 58in some circumstances (e.g. when vfork is used), fork+exec can be much
59faster.
38 60
39This module can help here by telling a small(er) helper process to fork, 61This module can help here by telling a small(er) helper process to fork,
40or fork+exec instead. 62which is faster then forking the main process, and also uses vfork where
63possible. This gives the speed of vfork, with the flexibility of fork.
41 64
42=item Forking usually creates a copy-on-write copy of the parent 65=item Forking usually creates a copy-on-write copy of the parent
43process. Memory (for example, modules or data files that have been 66process.
44will not take additional memory). When exec'ing a new process, modules 67
45and data files might need to be loaded again, at extra cpu and memory 68For example, modules or data files that are loaded will not use additional
46cost. Likewise when forking, all data structures are copied as well - if 69memory after a fork. When exec'ing a new process, modules and data files
70might need to be loaded again, at extra CPU and memory cost. But when
71forking, literally all data structures are copied - if the program frees
47the program frees them and replaces them by new data, the child processes 72them and replaces them by new data, the child processes will retain the
48will retain the memory even if it isn't used. 73old version even if it isn't used, which can suddenly and unexpectedly
74increase memory usage when freeing memory.
75
76The trade-off is between more sharing with fork (which can be good or
77bad), and no sharing with exec.
49 78
50This module allows the main program to do a controlled fork, and allows 79This module allows the main program to do a controlled fork, and allows
51modules to exec processes safely at any time. When creating a custom 80modules to exec processes safely at any time. When creating a custom
52process pool you can take advantage of data sharing via fork without 81process pool you can take advantage of data sharing via fork without
53risking to share large dynamic data structures that will blow up child 82risking to share large dynamic data structures that will blow up child
54memory usage. 83memory usage.
55 84
85In other words, this module puts you into control over what is being
86shared and what isn't, at all times.
87
56=item Exec'ing a new perl process might be difficult and slow. For 88=item Exec'ing a new perl process might be difficult.
89
57example, it is not easy to find the correct path to the perl interpreter, 90For example, it is not easy to find the correct path to the perl
58and all modules have to be loaded from disk again. Long running processes 91interpreter - C<$^X> might not be a perl interpreter at all.
59might run into problems when perl is upgraded for example.
60 92
61This module supports creating pre-initialised perl processes to be used
62as template, and also tries hard to identify the correct path to the perl 93This module tries hard to identify the correct path to the perl
63interpreter. With a cooperative main program, exec'ing the interpreter 94interpreter. With a cooperative main program, exec'ing the interpreter
64might not even be necessary. 95might not even be necessary, but even without help from the main program,
96it will still work when used from a module.
65 97
98=item Exec'ing a new perl process might be slow, as all necessary modules
99have to be loaded from disk again, with no guarantees of success.
100
101Long running processes might run into problems when perl is upgraded
102and modules are no longer loadable because they refer to a different
103perl version, or parts of a distribution are newer than the ones already
104loaded.
105
106This module supports creating pre-initialised perl processes to be used as
107a template for new processes.
108
66=item Forking might be impossible when a program is running. For example, 109=item Forking might be impossible when a program is running.
67POSIX makes it almost impossible to fork from a multithreaded program and
68do anything useful in the child - strictly speaking, if your perl program
69uses posix threads (even indirectly via e.g. L<IO::AIO> or L<threads>),
70you cannot call fork on the perl level anymore, at all.
71 110
111For example, POSIX makes it almost impossible to fork from a
112multi-threaded program while doing anything useful in the child - in
113fact, if your perl program uses POSIX threads (even indirectly via
114e.g. L<IO::AIO> or L<threads>), you cannot call fork on the perl level
115anymore without risking corruption issues on a number of operating
116systems.
117
72This module can safely fork helper processes at any time, by caling 118This module can safely fork helper processes at any time, by calling
73fork+exec in C, in a POSIX-compatible way. 119fork+exec in C, in a POSIX-compatible way (via L<Proc::FastSpawn>).
74 120
75=item Parallel processing with fork might be inconvenient or difficult 121=item Parallel processing with fork might be inconvenient or difficult
122to implement. Modules might not work in both parent and child.
123
76to implement. For example, when a program uses an event loop and creates 124For example, when a program uses an event loop and creates watchers it
77watchers it becomes very hard to use the event loop from a child 125becomes very hard to use the event loop from a child program, as the
78program, as the watchers already exist but are only meaningful in the 126watchers already exist but are only meaningful in the parent. Worse, a
79parent. Worse, a module might want to use such a system, not knowing 127module might want to use such a module, not knowing whether another module
80whether another module or the main program also does, leading to problems. 128or the main program also does, leading to problems.
81 129
82This module only lets the main program create pools by forking (because 130Apart from event loops, graphical toolkits also commonly fall into the
83only the main program can know when it is still safe to do so) - all other 131"unsafe module" category, or just about anything that communicates with
84pools are created by fork+exec, after which such modules can again be 132the external world, such as network libraries and file I/O modules, which
85loaded. 133usually don't like being copied and then allowed to continue in two
134processes.
135
136With this module only the main program is allowed to create new processes
137by forking (because only the main program can know when it is still safe
138to do so) - all other processes are created via fork+exec, which makes it
139possible to use modules such as event loops or window interfaces safely.
86 140
87=back 141=back
142
143=head1 EXAMPLES
144
145=head2 Create a single new process, tell it to run your worker function.
146
147 AnyEvent::Fork
148 ->new
149 ->require ("MyModule")
150 ->run ("MyModule::worker, sub {
151 my ($master_filehandle) = @_;
152
153 # now $master_filehandle is connected to the
154 # $slave_filehandle in the new process.
155 });
156
157C<MyModule> might look like this:
158
159 package MyModule;
160
161 sub worker {
162 my ($slave_filehandle) = @_;
163
164 # now $slave_filehandle is connected to the $master_filehandle
165 # in the original prorcess. have fun!
166 }
167
168=head2 Create a pool of server processes all accepting on the same socket.
169
170 # create listener socket
171 my $listener = ...;
172
173 # create a pool template, initialise it and give it the socket
174 my $pool = AnyEvent::Fork
175 ->new
176 ->require ("Some::Stuff", "My::Server")
177 ->send_fh ($listener);
178
179 # now create 10 identical workers
180 for my $id (1..10) {
181 $pool
182 ->fork
183 ->send_arg ($id)
184 ->run ("My::Server::run");
185 }
186
187 # now do other things - maybe use the filehandle provided by run
188 # to wait for the processes to die. or whatever.
189
190C<My::Server> might look like this:
191
192 package My::Server;
193
194 sub run {
195 my ($slave, $listener, $id) = @_;
196
197 close $slave; # we do not use the socket, so close it to save resources
198
199 # we could go ballistic and use e.g. AnyEvent here, or IO::AIO,
200 # or anything we usually couldn't do in a process forked normally.
201 while (my $socket = $listener->accept) {
202 # do sth. with new socket
203 }
204 }
205
206=head2 use AnyEvent::Fork as a faster fork+exec
207
208This runs C</bin/echo hi>, with stdandard output redirected to /tmp/log
209and standard error redirected to the communications socket. It is usually
210faster than fork+exec, but still lets you prepare the environment.
211
212 open my $output, ">/tmp/log" or die "$!";
213
214 AnyEvent::Fork
215 ->new
216 ->eval ('
217 sub run {
218 my ($fh, $output, @cmd) = @_;
219
220 # perl will clear close-on-exec on STDOUT/STDERR
221 open STDOUT, ">&", $output or die;
222 open STDERR, ">&", $fh or die;
223
224 exec @cmd;
225 }
226 ')
227 ->send_fh ($output)
228 ->send_arg ("/bin/echo", "hi")
229 ->run ("run", my $cv = AE::cv);
230
231 my $stderr = $cv->recv;
88 232
89=head1 CONCEPTS 233=head1 CONCEPTS
90 234
91This module can create new processes either by executing a new perl 235This module can create new processes either by executing a new perl
92process, or by forking from an existing "template" process. 236process, or by forking from an existing "template" process.
109needed the first time. Forking from this process shares the memory used 253needed the first time. Forking from this process shares the memory used
110for the perl interpreter with the new process, but loading modules takes 254for the perl interpreter with the new process, but loading modules takes
111time, and the memory is not shared with anything else. 255time, and the memory is not shared with anything else.
112 256
113This is ideal for when you only need one extra process of a kind, with the 257This is ideal for when you only need one extra process of a kind, with the
114option of starting and stipping it on demand. 258option of starting and stopping it on demand.
259
260Example:
261
262 AnyEvent::Fork
263 ->new
264 ->require ("Some::Module")
265 ->run ("Some::Module::run", sub {
266 my ($fork_fh) = @_;
267 });
115 268
116=item fork a new template process, load code, then fork processes off of 269=item fork a new template process, load code, then fork processes off of
117it and run the code 270it and run the code
118 271
119When you need to have a bunch of processes that all execute the same (or 272When you need to have a bunch of processes that all execute the same (or
125modules you loaded) is shared between the processes, and each new process 278modules you loaded) is shared between the processes, and each new process
126consumes relatively little memory of its own. 279consumes relatively little memory of its own.
127 280
128The disadvantage of this approach is that you need to create a template 281The disadvantage of this approach is that you need to create a template
129process for the sole purpose of forking new processes from it, but if you 282process for the sole purpose of forking new processes from it, but if you
130only need a fixed number of proceses you can create them, and then destroy 283only need a fixed number of processes you can create them, and then destroy
131the template process. 284the template process.
285
286Example:
287
288 my $template = AnyEvent::Fork->new->require ("Some::Module");
289
290 for (1..10) {
291 $template->fork->run ("Some::Module::run", sub {
292 my ($fork_fh) = @_;
293 });
294 }
295
296 # at this point, you can keep $template around to fork new processes
297 # later, or you can destroy it, which causes it to vanish.
132 298
133=item execute a new perl interpreter, load some code, run it 299=item execute a new perl interpreter, load some code, run it
134 300
135This is relatively slow, and doesn't allow you to share memory between 301This is relatively slow, and doesn't allow you to share memory between
136multiple processes. 302multiple processes.
138The only advantage is that you don't have to have a template process 304The only advantage is that you don't have to have a template process
139hanging around all the time to fork off some new processes, which might be 305hanging around all the time to fork off some new processes, which might be
140an advantage when there are long time spans where no extra processes are 306an advantage when there are long time spans where no extra processes are
141needed. 307needed.
142 308
309Example:
310
311 AnyEvent::Fork
312 ->new_exec
313 ->require ("Some::Module")
314 ->run ("Some::Module::run", sub {
315 my ($fork_fh) = @_;
316 });
317
143=back 318=back
144 319
145=head1 FUNCTIONS 320=head1 THE C<AnyEvent::Fork> CLASS
321
322This module exports nothing, and only implements a single class -
323C<AnyEvent::Fork>.
324
325There are two class constructors that both create new processes - C<new>
326and C<new_exec>. The C<fork> method creates a new process by forking an
327existing one and could be considered a third constructor.
328
329Most of the remaining methods deal with preparing the new process, by
330loading code, evaluating code and sending data to the new process. They
331usually return the process object, so you can chain method calls.
332
333If a process object is destroyed before calling its C<run> method, then
334the process simply exits. After C<run> is called, all responsibility is
335passed to the specified function.
336
337As long as there is any outstanding work to be done, process objects
338resist being destroyed, so there is no reason to store them unless you
339need them later - configure and forget works just fine.
146 340
147=over 4 341=over 4
148 342
149=cut 343=cut
150 344
151package AnyEvent::Fork; 345package AnyEvent::Fork;
152 346
153use common::sense; 347use common::sense;
154 348
155use Socket (); 349use Errno ();
156 350
157use AnyEvent; 351use AnyEvent;
158use AnyEvent::Fork::Util;
159use AnyEvent::Util (); 352use AnyEvent::Util ();
160 353
354use IO::FDPass;
355
356our $VERSION = 0.5;
357
161our $PERL; # the path to the perl interpreter, deduces with various forms of magic 358our $PERL; # the path to the perl interpreter, deduces with various forms of magic
162
163=item my $pool = new AnyEvent::Fork key => value...
164
165Create a new process pool. The following named parameters are supported:
166 359
167=over 4 360=over 4
168 361
169=back 362=back
170 363
177our $TEMPLATE; 370our $TEMPLATE;
178 371
179sub _cmd { 372sub _cmd {
180 my $self = shift; 373 my $self = shift;
181 374
182 # ideally, we would want to use "a (w/a)*" as format string, but perl versions 375 # ideally, we would want to use "a (w/a)*" as format string, but perl
183 # from at least 5.8.9 to 5.16.3 are all buggy and can't unpack it. 376 # versions from at least 5.8.9 to 5.16.3 are all buggy and can't unpack
184 push @{ $self->[2] }, pack "N/a", pack "(w/a)*", @_; 377 # it.
378 push @{ $self->[2] }, pack "a L/a*", $_[0], $_[1];
185 379
186 $self->[3] ||= AE::io $self->[1], 1, sub { 380 $self->[3] ||= AE::io $self->[1], 1, sub {
381 do {
382 # send the next "thing" in the queue - either a reference to an fh,
383 # or a plain string.
384
187 if (ref $self->[2][0]) { 385 if (ref $self->[2][0]) {
386 # send fh
188 AnyEvent::Fork::Util::fd_send fileno $self->[1], fileno ${ $self->[2][0] } 387 unless (IO::FDPass::send fileno $self->[1], fileno ${ $self->[2][0] }) {
388 return if $! == Errno::EAGAIN || $! == Errno::EWOULDBLOCK;
389 undef $self->[3];
390 die "AnyEvent::Fork: file descriptor send failure: $!";
391 }
392
189 and shift @{ $self->[2] }; 393 shift @{ $self->[2] };
190 394
191 } else { 395 } else {
396 # send string
192 my $len = syswrite $self->[1], $self->[2][0] 397 my $len = syswrite $self->[1], $self->[2][0];
398
399 unless ($len) {
400 return if $! == Errno::EAGAIN || $! == Errno::EWOULDBLOCK;
401 undef $self->[3];
193 or do { undef $self->[3]; die "AnyEvent::Fork: command write failure: $!" }; 402 die "AnyEvent::Fork: command write failure: $!";
403 }
194 404
195 substr $self->[2][0], 0, $len, ""; 405 substr $self->[2][0], 0, $len, "";
196 shift @{ $self->[2] } unless length $self->[2][0]; 406 shift @{ $self->[2] } unless length $self->[2][0];
197 } 407 }
408 } while @{ $self->[2] };
198 409
199 unless (@{ $self->[2] }) { 410 # everything written
200 undef $self->[3]; 411 undef $self->[3];
412
413 # invoke run callback, if any
201 $self->[0]->($self->[1]) if $self->[0]; 414 $self->[4]->($self->[1]) if $self->[4];
202 }
203 }; 415 };
416
417 () # make sure we don't leak the watcher
204} 418}
205 419
206sub _new { 420sub _new {
207 my ($self, $fh) = @_; 421 my ($self, $fh, $pid) = @_;
208 422
209 AnyEvent::Util::fh_nonblocking $fh, 1; 423 AnyEvent::Util::fh_nonblocking $fh, 1;
210 424
211 $self = bless [ 425 $self = bless [
212 undef, # run callback 426 $pid,
213 $fh, 427 $fh,
214 [], # write queue - strings or fd's 428 [], # write queue - strings or fd's
215 undef, # AE watcher 429 undef, # AE watcher
216 ], $self; 430 ], $self;
217 431
218# my ($a, $b) = AnyEvent::Util::portable_socketpair;
219
220# queue_cmd $template, "Iabc";
221# push @{ $template->[2] }, \$b;
222
223# use Coro::AnyEvent; Coro::AnyEvent::sleep 1;
224# undef $b;
225# die "x" . <$a>;
226
227 $self 432 $self
228} 433}
229 434
230# fork template from current process, used by AnyEvent::Fork::Early/Template 435# fork template from current process, used by AnyEvent::Fork::Early/Template
231sub _new_fork { 436sub _new_fork {
232 my ($fh, $slave) = AnyEvent::Util::portable_socketpair; 437 my ($fh, $slave) = AnyEvent::Util::portable_socketpair;
438 my $parent = $$;
439
233 my $pid = fork; 440 my $pid = fork;
234 441
235 if ($pid eq 0) { 442 if ($pid eq 0) {
236 require AnyEvent::Fork::Serve; 443 require AnyEvent::Fork::Serve;
444 $AnyEvent::Fork::Serve::OWNER = $parent;
237 close $fh; 445 close $fh;
446 $0 = "$_[1] of $parent";
447 $SIG{CHLD} = 'IGNORE';
238 AnyEvent::Fork::Serve::serve ($slave); 448 AnyEvent::Fork::Serve::serve ($slave);
239 AnyEvent::Fork::Util::_exit 0; 449 exit 0;
240 } elsif (!$pid) { 450 } elsif (!$pid) {
241 die "AnyEvent::Fork::Early/Template: unable to fork template process: $!"; 451 die "AnyEvent::Fork::Early/Template: unable to fork template process: $!";
242 } 452 }
243 453
244 AnyEvent::Fork->_new ($fh) 454 AnyEvent::Fork->_new ($fh, $pid)
245} 455}
246 456
247=item my $proc = new AnyEvent::Fork 457=item my $proc = new AnyEvent::Fork
248 458
249Create a new "empty" perl interpreter process and returns its process 459Create a new "empty" perl interpreter process and returns its process
250object for further manipulation. 460object for further manipulation.
251 461
252The new process is forked from a template process that is kept around 462The new process is forked from a template process that is kept around
253for this purpose. When it doesn't exist yet, it is created by a call to 463for this purpose. When it doesn't exist yet, it is created by a call to
254C<new_exec> and kept around for future calls. 464C<new_exec> first and then stays around for future calls.
255 465
256=cut 466=cut
257 467
258sub new { 468sub new {
259 my $class = shift; 469 my $class = shift;
295reduces the amount of memory sharing that is possible, and is also slower. 505reduces the amount of memory sharing that is possible, and is also slower.
296 506
297You should use C<new> whenever possible, except when having a template 507You should use C<new> whenever possible, except when having a template
298process around is unacceptable. 508process around is unacceptable.
299 509
300The path to the perl interpreter is divined usign various methods - first 510The path to the perl interpreter is divined using various methods - first
301C<$^X> is investigated to see if the path ends with something that sounds 511C<$^X> is investigated to see if the path ends with something that sounds
302as if it were the perl interpreter. Failing this, the module falls back to 512as if it were the perl interpreter. Failing this, the module falls back to
303using C<$Config::Config{perlpath}>. 513using C<$Config::Config{perlpath}>.
304 514
305=cut 515=cut
314 my $perl = $; 524 my $perl = $;
315 525
316 # first we try $^X, but the path must be absolute (always on win32), and end in sth. 526 # first we try $^X, but the path must be absolute (always on win32), and end in sth.
317 # that looks like perl. this obviously only works for posix and win32 527 # that looks like perl. this obviously only works for posix and win32
318 unless ( 528 unless (
319 (AnyEvent::Fork::Util::WIN32 || $perl =~ m%^/%) 529 ($^O eq "MSWin32" || $perl =~ m%^/%)
320 && $perl =~ m%[/\\]perl(?:[0-9]+(\.[0-9]+)+)?(\.exe)?$%i 530 && $perl =~ m%[/\\]perl(?:[0-9]+(\.[0-9]+)+)?(\.exe)?$%i
321 ) { 531 ) {
322 # if it doesn't look perlish enough, try Config 532 # if it doesn't look perlish enough, try Config
323 require Config; 533 require Config;
324 $perl = $Config::Config{perlpath}; 534 $perl = $Config::Config{perlpath};
328 require Proc::FastSpawn; 538 require Proc::FastSpawn;
329 539
330 my ($fh, $slave) = AnyEvent::Util::portable_socketpair; 540 my ($fh, $slave) = AnyEvent::Util::portable_socketpair;
331 Proc::FastSpawn::fd_inherit (fileno $slave); 541 Proc::FastSpawn::fd_inherit (fileno $slave);
332 542
543 # new fh's should always be set cloexec (due to $^F),
544 # but hey, not on win32, so we always clear the inherit flag.
545 Proc::FastSpawn::fd_inherit (fileno $fh, 0);
546
333 # quick. also doesn't work in win32. of course. what did you expect 547 # quick. also doesn't work in win32. of course. what did you expect
334 #local $ENV{PERL5LIB} = join ":", grep !ref, @INC; 548 #local $ENV{PERL5LIB} = join ":", grep !ref, @INC;
335 my %env = %ENV; 549 my %env = %ENV;
336 $env{PERL5LIB} = join ":", grep !ref, @INC; 550 $env{PERL5LIB} = join +($^O eq "MSWin32" ? ";" : ":"), grep !ref, @INC;
337 551
338 Proc::FastSpawn::spawn ( 552 my $pid = Proc::FastSpawn::spawn (
339 $perl, 553 $perl,
340 ["perl", "-MAnyEvent::Fork::Serve", "-e", "AnyEvent::Fork::Serve::me", fileno $slave], 554 ["perl", "-MAnyEvent::Fork::Serve", "-e", "AnyEvent::Fork::Serve::me", fileno $slave, $$],
341 [map "$_=$env{$_}", keys %env], 555 [map "$_=$env{$_}", keys %env],
342 ) or die "unable to spawn AnyEvent::Fork server: $!"; 556 ) or die "unable to spawn AnyEvent::Fork server: $!";
343 557
344 $self->_new ($fh) 558 $self->_new ($fh, $pid)
559}
560
561=item $pid = $proc->pid
562
563Returns the process id of the process I<iff it is a direct child of the
564process running AnyEvent::Fork>, and C<undef> otherwise.
565
566Normally, only processes created via C<< AnyEvent::Fork->new_exec >> and
567L<AnyEvent::Fork::Template> are direct children, and you are responsible
568to clean up their zombies when they die.
569
570All other processes are not direct children, and will be cleaned up by
571AnyEvent::Fork itself.
572
573=cut
574
575sub pid {
576 $_[0][0]
577}
578
579=item $proc = $proc->eval ($perlcode, @args)
580
581Evaluates the given C<$perlcode> as ... perl code, while setting C<@_> to
582the strings specified by C<@args>, in the "main" package.
583
584This call is meant to do any custom initialisation that might be required
585(for example, the C<require> method uses it). It's not supposed to be used
586to completely take over the process, use C<run> for that.
587
588The code will usually be executed after this call returns, and there is no
589way to pass anything back to the calling process. Any evaluation errors
590will be reported to stderr and cause the process to exit.
591
592If you want to execute some code (that isn't in a module) to take over the
593process, you should compile a function via C<eval> first, and then call
594it via C<run>. This also gives you access to any arguments passed via the
595C<send_xxx> methods, such as file handles. See the L<use AnyEvent::Fork as
596a faster fork+exec> example to see it in action.
597
598Returns the process object for easy chaining of method calls.
599
600=cut
601
602sub eval {
603 my ($self, $code, @args) = @_;
604
605 $self->_cmd (e => pack "(w/a*)*", $code, @args);
606
607 $self
345} 608}
346 609
347=item $proc = $proc->require ($module, ...) 610=item $proc = $proc->require ($module, ...)
348 611
349Tries to load the given modules into the process 612Tries to load the given module(s) into the process
350 613
351Returns the process object for easy chaining of method calls. 614Returns the process object for easy chaining of method calls.
615
616=cut
617
618sub require {
619 my ($self, @modules) = @_;
620
621 s%::%/%g for @modules;
622 $self->eval ('require "$_.pm" for @_', @modules);
623
624 $self
625}
352 626
353=item $proc = $proc->send_fh ($handle, ...) 627=item $proc = $proc->send_fh ($handle, ...)
354 628
355Send one or more file handles (I<not> file descriptors) to the process, 629Send one or more file handles (I<not> file descriptors) to the process,
356to prepare a call to C<run>. 630to prepare a call to C<run>.
357 631
358The process object keeps a reference to the handles until this is done, 632The process object keeps a reference to the handles until they have
359so you must not explicitly close the handles. This is most easily 633been passed over to the process, so you must not explicitly close the
360accomplished by simply not storing the file handles anywhere after passing 634handles. This is most easily accomplished by simply not storing the file
361them to this method. 635handles anywhere after passing them to this method - when AnyEvent::Fork
636is finished using them, perl will automatically close them.
362 637
363Returns the process object for easy chaining of method calls. 638Returns the process object for easy chaining of method calls.
639
640Example: pass a file handle to a process, and release it without
641closing. It will be closed automatically when it is no longer used.
642
643 $proc->send_fh ($my_fh);
644 undef $my_fh; # free the reference if you want, but DO NOT CLOSE IT
364 645
365=cut 646=cut
366 647
367sub send_fh { 648sub send_fh {
368 my ($self, @fh) = @_; 649 my ($self, @fh) = @_;
376} 657}
377 658
378=item $proc = $proc->send_arg ($string, ...) 659=item $proc = $proc->send_arg ($string, ...)
379 660
380Send one or more argument strings to the process, to prepare a call to 661Send one or more argument strings to the process, to prepare a call to
381C<run>. The strings can be any octet string. 662C<run>. The strings can be any octet strings.
382 663
664The protocol is optimised to pass a moderate number of relatively short
665strings - while you can pass up to 4GB of data in one go, this is more
666meant to pass some ID information or other startup info, not big chunks of
667data.
668
383Returns the process object for easy chaining of emthod calls. 669Returns the process object for easy chaining of method calls.
384 670
385=cut 671=cut
386 672
387sub send_arg { 673sub send_arg {
388 my ($self, @arg) = @_; 674 my ($self, @arg) = @_;
389 675
390 $self->_cmd (a => @arg); 676 $self->_cmd (a => pack "(w/a*)*", @arg);
391 677
392 $self 678 $self
393} 679}
394 680
395=item $proc->run ($func, $cb->($fh)) 681=item $proc->run ($func, $cb->($fh))
396 682
397Enter the function specified by the fully qualified name in C<$func> in 683Enter the function specified by the function name in C<$func> in the
398the process. The function is called with the communication socket as first 684process. The function is called with the communication socket as first
399argument, followed by all file handles and string arguments sent earlier 685argument, followed by all file handles and string arguments sent earlier
400via C<send_fh> and C<send_arg> methods, in the order they were called. 686via C<send_fh> and C<send_arg> methods, in the order they were called.
401 687
402If the called function returns, the process exits.
403
404Preparing the process can take time - when the process is ready, the
405callback is invoked with the local communications socket as argument.
406
407The process object becomes unusable on return from this function. 688The process object becomes unusable on return from this function - any
689further method calls result in undefined behaviour.
690
691The function name should be fully qualified, but if it isn't, it will be
692looked up in the C<main> package.
693
694If the called function returns, doesn't exist, or any error occurs, the
695process exits.
696
697Preparing the process is done in the background - when all commands have
698been sent, the callback is invoked with the local communications socket
699as argument. At this point you can start using the socket in any way you
700like.
408 701
409If the communication socket isn't used, it should be closed on both sides, 702If the communication socket isn't used, it should be closed on both sides,
410to save on kernel memory. 703to save on kernel memory.
411 704
412The socket is non-blocking in the parent, and blocking in the newly 705The socket is non-blocking in the parent, and blocking in the newly
413created process. The close-on-exec flag is set on both. Even if not used 706created process. The close-on-exec flag is set in both.
707
414otherwise, the socket can be a good indicator for the existance of the 708Even if not used otherwise, the socket can be a good indicator for the
415process - if the othe rprocess exits, you get a readable event on it, 709existence of the process - if the other process exits, you get a readable
416because exiting the process closes the socket (if it didn't create any 710event on it, because exiting the process closes the socket (if it didn't
417children using fork). 711create any children using fork).
712
713Example: create a template for a process pool, pass a few strings, some
714file handles, then fork, pass one more string, and run some code.
715
716 my $pool = AnyEvent::Fork
717 ->new
718 ->send_arg ("str1", "str2")
719 ->send_fh ($fh1, $fh2);
720
721 for (1..2) {
722 $pool
723 ->fork
724 ->send_arg ("str3")
725 ->run ("Some::function", sub {
726 my ($fh) = @_;
727
728 # fh is nonblocking, but we trust that the OS can accept these
729 # few octets anyway.
730 syswrite $fh, "hi #$_\n";
731
732 # $fh is being closed here, as we don't store it anywhere
733 });
734 }
735
736 # Some::function might look like this - all parameters passed before fork
737 # and after will be passed, in order, after the communications socket.
738 sub Some::function {
739 my ($fh, $str1, $str2, $fh1, $fh2, $str3) = @_;
740
741 print scalar <$fh>; # prints "hi #1\n" and "hi #2\n" in any order
742 }
418 743
419=cut 744=cut
420 745
421sub run { 746sub run {
422 my ($self, $func, $cb) = @_; 747 my ($self, $func, $cb) = @_;
423 748
424 $self->[0] = $cb; 749 $self->[4] = $cb;
425 $self->_cmd ("r", $func); 750 $self->_cmd (r => $func);
426} 751}
427 752
428=back 753=back
754
755=head1 PERFORMANCE
756
757Now for some unscientific benchmark numbers (all done on an amd64
758GNU/Linux box). These are intended to give you an idea of the relative
759performance you can expect, they are not meant to be absolute performance
760numbers.
761
762OK, so, I ran a simple benchmark that creates a socket pair, forks, calls
763exit in the child and waits for the socket to close in the parent. I did
764load AnyEvent, EV and AnyEvent::Fork, for a total process size of 5100kB.
765
766 2079 new processes per second, using manual socketpair + fork
767
768Then I did the same thing, but instead of calling fork, I called
769AnyEvent::Fork->new->run ("CORE::exit") and then again waited for the
770socket form the child to close on exit. This does the same thing as manual
771socket pair + fork, except that what is forked is the template process
772(2440kB), and the socket needs to be passed to the server at the other end
773of the socket first.
774
775 2307 new processes per second, using AnyEvent::Fork->new
776
777And finally, using C<new_exec> instead C<new>, using vforks+execs to exec
778a new perl interpreter and compile the small server each time, I get:
779
780 479 vfork+execs per second, using AnyEvent::Fork->new_exec
781
782So how can C<< AnyEvent->new >> be faster than a standard fork, even
783though it uses the same operations, but adds a lot of overhead?
784
785The difference is simply the process size: forking the 5MB process takes
786so much longer than forking the 2.5MB template process that the extra
787overhead introduced is canceled out.
788
789If the benchmark process grows, the normal fork becomes even slower:
790
791 1340 new processes, manual fork of a 20MB process
792 731 new processes, manual fork of a 200MB process
793 235 new processes, manual fork of a 2000MB process
794
795What that means (to me) is that I can use this module without having a bad
796conscience because of the extra overhead required to start new processes.
797
798=head1 TYPICAL PROBLEMS
799
800This section lists typical problems that remain. I hope by recognising
801them, most can be avoided.
802
803=over 4
804
805=item leaked file descriptors for exec'ed processes
806
807POSIX systems inherit file descriptors by default when exec'ing a new
808process. While perl itself laudably sets the close-on-exec flags on new
809file handles, most C libraries don't care, and even if all cared, it's
810often not possible to set the flag in a race-free manner.
811
812That means some file descriptors can leak through. And since it isn't
813possible to know which file descriptors are "good" and "necessary" (or
814even to know which file descriptors are open), there is no good way to
815close the ones that might harm.
816
817As an example of what "harm" can be done consider a web server that
818accepts connections and afterwards some module uses AnyEvent::Fork for the
819first time, causing it to fork and exec a new process, which might inherit
820the network socket. When the server closes the socket, it is still open
821in the child (which doesn't even know that) and the client might conclude
822that the connection is still fine.
823
824For the main program, there are multiple remedies available -
825L<AnyEvent::Fork::Early> is one, creating a process early and not using
826C<new_exec> is another, as in both cases, the first process can be exec'ed
827well before many random file descriptors are open.
828
829In general, the solution for these kind of problems is to fix the
830libraries or the code that leaks those file descriptors.
831
832Fortunately, most of these leaked descriptors do no harm, other than
833sitting on some resources.
834
835=item leaked file descriptors for fork'ed processes
836
837Normally, L<AnyEvent::Fork> does start new processes by exec'ing them,
838which closes file descriptors not marked for being inherited.
839
840However, L<AnyEvent::Fork::Early> and L<AnyEvent::Fork::Template> offer
841a way to create these processes by forking, and this leaks more file
842descriptors than exec'ing them, as there is no way to mark descriptors as
843"close on fork".
844
845An example would be modules like L<EV>, L<IO::AIO> or L<Gtk2>. Both create
846pipes for internal uses, and L<Gtk2> might open a connection to the X
847server. L<EV> and L<IO::AIO> can deal with fork, but Gtk2 might have
848trouble with a fork.
849
850The solution is to either not load these modules before use'ing
851L<AnyEvent::Fork::Early> or L<AnyEvent::Fork::Template>, or to delay
852initialising them, for example, by calling C<init Gtk2> manually.
853
854=item exiting calls object destructors
855
856This only applies to users of L<AnyEvent::Fork:Early> and
857L<AnyEvent::Fork::Template>, or when initialiasing code creates objects
858that reference external resources.
859
860When a process created by AnyEvent::Fork exits, it might do so by calling
861exit, or simply letting perl reach the end of the program. At which point
862Perl runs all destructors.
863
864Not all destructors are fork-safe - for example, an object that represents
865the connection to an X display might tell the X server to free resources,
866which is inconvenient when the "real" object in the parent still needs to
867use them.
868
869This is obviously not a problem for L<AnyEvent::Fork::Early>, as you used
870it as the very first thing, right?
871
872It is a problem for L<AnyEvent::Fork::Template> though - and the solution
873is to not create objects with nontrivial destructors that might have an
874effect outside of Perl.
875
876=back
877
878=head1 PORTABILITY NOTES
879
880Native win32 perls are somewhat supported (AnyEvent::Fork::Early is a nop,
881and ::Template is not going to work), and it cost a lot of blood and sweat
882to make it so, mostly due to the bloody broken perl that nobody seems to
883care about. The fork emulation is a bad joke - I have yet to see something
884useful that you can do with it without running into memory corruption
885issues or other braindamage. Hrrrr.
886
887Cygwin perl is not supported at the moment due to some hilarious
888shortcomings of its API - see L<IO::FDPoll> for more details.
889
890=head1 SEE ALSO
891
892L<AnyEvent::Fork::Early> (to avoid executing a perl interpreter),
893L<AnyEvent::Fork::Template> (to create a process by forking the main
894program at a convenient time).
429 895
430=head1 AUTHOR 896=head1 AUTHOR
431 897
432 Marc Lehmann <schmorp@schmorp.de> 898 Marc Lehmann <schmorp@schmorp.de>
433 http://home.schmorp.de/ 899 http://home.schmorp.de/

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines