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 1.9 by root, Thu Apr 4 03:45:12 2013 UTC vs.
Revision 1.30 by root, Sat Apr 6 09:28:45 2013 UTC

1=head1 NAME 1=head1 NAME
2 2
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
5ATTENTION, this is a very early release, and very untested. Consider it a
6technology preview.
7
8=head1 SYNOPSIS 5=head1 SYNOPSIS
9 6
10 use AnyEvent::Fork; 7 use AnyEvent::Fork;
11 8
12 ################################################################## 9 AnyEvent::Fork
10 ->new
11 ->require ("MyModule")
12 ->run ("MyModule::server", my $cv = AE::cv);
13
14 my $fh = $cv->recv;
15
16=head1 DESCRIPTION
17
18This module allows you to create new processes, without actually forking
19them from your current process (avoiding the problems of forking), but
20preserving most of the advantages of fork.
21
22It can be used to create new worker processes or new independent
23subprocesses for short- and long-running jobs, process pools (e.g. for use
24in pre-forked servers) but also to spawn new external processes (such as
25CGI scripts from a web server), which can be faster (and more well behaved)
26than using fork+exec in big processes.
27
28Special care has been taken to make this module useful from other modules,
29while still supporting specialised environments such as L<App::Staticperl>
30or L<PAR::Packer>.
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
45=head1 PROBLEM STATEMENT
46
47There are two traditional ways to implement parallel processing on UNIX
48like operating systems - fork and process, and fork+exec and process. They
49have different advantages and disadvantages that I describe below,
50together with how this module tries to mitigate the disadvantages.
51
52=over 4
53
54=item Forking from a big process can be very slow.
55
56A 5GB process needs 0.05s to fork on my 3.6GHz amd64 GNU/Linux box. This
57overhead is often shared with exec (because you have to fork first), but
58in some circumstances (e.g. when vfork is used), fork+exec can be much
59faster.
60
61This module can help here by telling a small(er) helper process to fork,
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.
64
65=item Forking usually creates a copy-on-write copy of the parent
66process.
67
68For example, modules or data files that are loaded will not use additional
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
72them and replaces them by new data, the child processes will retain the
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.
78
79This module allows the main program to do a controlled fork, and allows
80modules to exec processes safely at any time. When creating a custom
81process pool you can take advantage of data sharing via fork without
82risking to share large dynamic data structures that will blow up child
83memory usage.
84
85In other words, this module puts you into control over what is being
86shared and what isn't, at all times.
87
88=item Exec'ing a new perl process might be difficult.
89
90For example, it is not easy to find the correct path to the perl
91interpreter - C<$^X> might not be a perl interpreter at all.
92
93This module tries hard to identify the correct path to the perl
94interpreter. With a cooperative main program, exec'ing the interpreter
95might not even be necessary, but even without help from the main program,
96it will still work when used from a module.
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
109=item Forking might be impossible when a program is running.
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
118This module can safely fork helper processes at any time, by calling
119fork+exec in C, in a POSIX-compatible way (via L<Proc::FastSpawn>).
120
121=item Parallel processing with fork might be inconvenient or difficult
122to implement. Modules might not work in both parent and child.
123
124For example, when a program uses an event loop and creates watchers it
125becomes very hard to use the event loop from a child program, as the
126watchers already exist but are only meaningful in the parent. Worse, a
127module might want to use such a module, not knowing whether another module
128or the main program also does, leading to problems.
129
130Apart from event loops, graphical toolkits also commonly fall into the
131"unsafe module" category, or just about anything that communicates with
132the external world, such as network libraries and file I/O modules, which
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.
140
141=back
142
143=head1 EXAMPLES
144
13 # create a single new process, tell it to run your worker function 145=head2 Create a single new process, tell it to run your worker function.
14 146
15 AnyEvent::Fork 147 AnyEvent::Fork
16 ->new 148 ->new
17 ->require ("MyModule") 149 ->require ("MyModule")
18 ->run ("MyModule::worker, sub { 150 ->run ("MyModule::worker, sub {
… …
20 152
21 # now $master_filehandle is connected to the 153 # now $master_filehandle is connected to the
22 # $slave_filehandle in the new process. 154 # $slave_filehandle in the new process.
23 }); 155 });
24 156
25 # MyModule::worker might look like this 157MyModule::worker might look like this:
158
26 sub MyModule::worker { 159 sub MyModule::worker {
27 my ($slave_filehandle) = @_; 160 my ($slave_filehandle) = @_;
28 161
29 # now $slave_filehandle is connected to the $master_filehandle 162 # now $slave_filehandle is connected to the $master_filehandle
30 # in the original prorcess. have fun! 163 # in the original prorcess. have fun!
31 } 164 }
32 165
33 ##################################################################
34 # create a pool of server processes all accepting on the same socket 166=head2 Create a pool of server processes all accepting on the same socket.
35 167
36 # create listener socket 168 # create listener socket
37 my $listener = ...; 169 my $listener = ...;
38 170
39 # create a pool template, initialise it and give it the socket 171 # create a pool template, initialise it and give it the socket
… …
51 } 183 }
52 184
53 # now do other things - maybe use the filehandle provided by run 185 # now do other things - maybe use the filehandle provided by run
54 # to wait for the processes to die. or whatever. 186 # to wait for the processes to die. or whatever.
55 187
56 # My::Server::run might look like this 188My::Server::run might look like this:
189
57 sub My::Server::run { 190 sub My::Server::run {
58 my ($slave, $listener, $id) = @_; 191 my ($slave, $listener, $id) = @_;
59 192
60 close $slave; # we do not use the socket, so close it to save resources 193 close $slave; # we do not use the socket, so close it to save resources
61 194
… …
64 while (my $socket = $listener->accept) { 197 while (my $socket = $listener->accept) {
65 # do sth. with new socket 198 # do sth. with new socket
66 } 199 }
67 } 200 }
68 201
69=head1 DESCRIPTION 202=head2 use AnyEvent::Fork as a faster fork+exec
70 203
71This module allows you to create new processes, without actually forking 204This runs /bin/echo hi, with stdout redirected to /tmp/log and stderr to
72them from your current process (avoiding the problems of forking), but 205the communications socket. It is usually faster than fork+exec, but still
73preserving most of the advantages of fork. 206let's you prepare the environment.
74 207
75It can be used to create new worker processes or new independent 208 open my $output, ">/tmp/log" or die "$!";
76subprocesses for short- and long-running jobs, process pools (e.g. for use
77in pre-forked servers) but also to spawn new external processes (such as
78CGI scripts from a webserver), which can be faster (and more well behaved)
79than using fork+exec in big processes.
80 209
81Special care has been taken to make this module useful from other modules, 210 AnyEvent::Fork
82while still supporting specialised environments such as L<App::Staticperl> 211 ->new
83or L<PAR::Packer>. 212 ->eval ('
213 sub run {
214 my ($fh, $output, @cmd) = @_;
84 215
85=head1 PROBLEM STATEMENT 216 # perl will clear close-on-exec on STDOUT/STDERR
217 open STDOUT, ">&", $output or die;
218 open STDERR, ">&", $fh or die;
86 219
87There are two ways to implement parallel processing on UNIX like operating 220 exec @cmd;
88systems - fork and process, and fork+exec and process. They have different 221 }
89advantages and disadvantages that I describe below, together with how this 222 ')
90module tries to mitigate the disadvantages. 223 ->send_fh ($output)
224 ->send_arg ("/bin/echo", "hi")
225 ->run ("run", my $cv = AE::cv);
91 226
92=over 4 227 my $stderr = $cv->recv;
93
94=item Forking from a big process can be very slow (a 5GB process needs
950.05s to fork on my 3.6GHz amd64 GNU/Linux box for example). This overhead
96is often shared with exec (because you have to fork first), but in some
97circumstances (e.g. when vfork is used), fork+exec can be much faster.
98
99This module can help here by telling a small(er) helper process to fork,
100or fork+exec instead.
101
102=item Forking usually creates a copy-on-write copy of the parent
103process. Memory (for example, modules or data files that have been
104will not take additional memory). When exec'ing a new process, modules
105and data files might need to be loaded again, at extra cpu and memory
106cost. Likewise when forking, all data structures are copied as well - if
107the program frees them and replaces them by new data, the child processes
108will retain the memory even if it isn't used.
109
110This module allows the main program to do a controlled fork, and allows
111modules to exec processes safely at any time. When creating a custom
112process pool you can take advantage of data sharing via fork without
113risking to share large dynamic data structures that will blow up child
114memory usage.
115
116=item Exec'ing a new perl process might be difficult and slow. For
117example, it is not easy to find the correct path to the perl interpreter,
118and all modules have to be loaded from disk again. Long running processes
119might run into problems when perl is upgraded for example.
120
121This module supports creating pre-initialised perl processes to be used
122as template, and also tries hard to identify the correct path to the perl
123interpreter. With a cooperative main program, exec'ing the interpreter
124might not even be necessary.
125
126=item Forking might be impossible when a program is running. For example,
127POSIX makes it almost impossible to fork from a multithreaded program and
128do anything useful in the child - strictly speaking, if your perl program
129uses posix threads (even indirectly via e.g. L<IO::AIO> or L<threads>),
130you cannot call fork on the perl level anymore, at all.
131
132This module can safely fork helper processes at any time, by caling
133fork+exec in C, in a POSIX-compatible way.
134
135=item Parallel processing with fork might be inconvenient or difficult
136to implement. For example, when a program uses an event loop and creates
137watchers it becomes very hard to use the event loop from a child
138program, as the watchers already exist but are only meaningful in the
139parent. Worse, a module might want to use such a system, not knowing
140whether another module or the main program also does, leading to problems.
141
142This module only lets the main program create pools by forking (because
143only the main program can know when it is still safe to do so) - all other
144pools are created by fork+exec, after which such modules can again be
145loaded.
146
147=back
148 228
149=head1 CONCEPTS 229=head1 CONCEPTS
150 230
151This module can create new processes either by executing a new perl 231This module can create new processes either by executing a new perl
152process, or by forking from an existing "template" process. 232process, or by forking from an existing "template" process.
… …
169needed the first time. Forking from this process shares the memory used 249needed the first time. Forking from this process shares the memory used
170for the perl interpreter with the new process, but loading modules takes 250for the perl interpreter with the new process, but loading modules takes
171time, and the memory is not shared with anything else. 251time, and the memory is not shared with anything else.
172 252
173This is ideal for when you only need one extra process of a kind, with the 253This is ideal for when you only need one extra process of a kind, with the
174option of starting and stipping it on demand. 254option of starting and stopping it on demand.
175 255
176Example: 256Example:
177 257
178 AnyEvent::Fork 258 AnyEvent::Fork
179 ->new 259 ->new
… …
194modules you loaded) is shared between the processes, and each new process 274modules you loaded) is shared between the processes, and each new process
195consumes relatively little memory of its own. 275consumes relatively little memory of its own.
196 276
197The disadvantage of this approach is that you need to create a template 277The disadvantage of this approach is that you need to create a template
198process for the sole purpose of forking new processes from it, but if you 278process for the sole purpose of forking new processes from it, but if you
199only need a fixed number of proceses you can create them, and then destroy 279only need a fixed number of processes you can create them, and then destroy
200the template process. 280the template process.
201 281
202Example: 282Example:
203 283
204 my $template = AnyEvent::Fork->new->require ("Some::Module"); 284 my $template = AnyEvent::Fork->new->require ("Some::Module");
… …
231 my ($fork_fh) = @_; 311 my ($fork_fh) = @_;
232 }); 312 });
233 313
234=back 314=back
235 315
236=head1 FUNCTIONS 316=head1 THE C<AnyEvent::Fork> CLASS
317
318This module exports nothing, and only implements a single class -
319C<AnyEvent::Fork>.
320
321There are two class constructors that both create new processes - C<new>
322and C<new_exec>. The C<fork> method creates a new process by forking an
323existing one and could be considered a third constructor.
324
325Most of the remaining methods deal with preparing the new process, by
326loading code, evaluating code and sending data to the new process. They
327usually return the process object, so you can chain method calls.
328
329If a process object is destroyed before calling its C<run> method, then
330the process simply exits. After C<run> is called, all responsibility is
331passed to the specified function.
332
333As long as there is any outstanding work to be done, process objects
334resist being destroyed, so there is no reason to store them unless you
335need them later - configure and forget works just fine.
237 336
238=over 4 337=over 4
239 338
240=cut 339=cut
241 340
242package AnyEvent::Fork; 341package AnyEvent::Fork;
243 342
244use common::sense; 343use common::sense;
245 344
246use Socket (); 345use Errno ();
247 346
248use AnyEvent; 347use AnyEvent;
249use AnyEvent::Fork::Util;
250use AnyEvent::Util (); 348use AnyEvent::Util ();
251 349
350use IO::FDPass;
351
352our $VERSION = 0.5;
353
252our $PERL; # the path to the perl interpreter, deduces with various forms of magic 354our $PERL; # the path to the perl interpreter, deduces with various forms of magic
253
254=item my $pool = new AnyEvent::Fork key => value...
255
256Create a new process pool. The following named parameters are supported:
257 355
258=over 4 356=over 4
259 357
260=back 358=back
261 359
… …
268our $TEMPLATE; 366our $TEMPLATE;
269 367
270sub _cmd { 368sub _cmd {
271 my $self = shift; 369 my $self = shift;
272 370
273 #TODO: maybe append the packet to any existing string command already in the queue
274
275 # ideally, we would want to use "a (w/a)*" as format string, but perl versions 371 # ideally, we would want to use "a (w/a)*" as format string, but perl
276 # from at least 5.8.9 to 5.16.3 are all buggy and can't unpack it. 372 # versions from at least 5.8.9 to 5.16.3 are all buggy and can't unpack
277 push @{ $self->[2] }, pack "N/a", pack "(w/a)*", @_; 373 # it.
374 push @{ $self->[2] }, pack "a L/a*", $_[0], $_[1];
278 375
279 $self->[3] ||= AE::io $self->[1], 1, sub { 376 $self->[3] ||= AE::io $self->[1], 1, sub {
377 do {
280 # send the next "thing" in the queue - either a reference to an fh, 378 # send the next "thing" in the queue - either a reference to an fh,
281 # or a plain string. 379 # or a plain string.
282 380
283 if (ref $self->[2][0]) { 381 if (ref $self->[2][0]) {
284 # send fh 382 # send fh
285 AnyEvent::Fork::Util::fd_send fileno $self->[1], fileno ${ $self->[2][0] } 383 unless (IO::FDPass::send fileno $self->[1], fileno ${ $self->[2][0] }) {
384 return if $! == Errno::EAGAIN || $! == Errno::EWOULDBLOCK;
385 undef $self->[3];
386 die "AnyEvent::Fork: file descriptor send failure: $!";
387 }
388
286 and shift @{ $self->[2] }; 389 shift @{ $self->[2] };
287 390
288 } else { 391 } else {
289 # send string 392 # send string
290 my $len = syswrite $self->[1], $self->[2][0] 393 my $len = syswrite $self->[1], $self->[2][0];
394
395 unless ($len) {
396 return if $! == Errno::EAGAIN || $! == Errno::EWOULDBLOCK;
397 undef $self->[3];
291 or do { undef $self->[3]; die "AnyEvent::Fork: command write failure: $!" }; 398 die "AnyEvent::Fork: command write failure: $!";
399 }
292 400
293 substr $self->[2][0], 0, $len, ""; 401 substr $self->[2][0], 0, $len, "";
294 shift @{ $self->[2] } unless length $self->[2][0]; 402 shift @{ $self->[2] } unless length $self->[2][0];
295 } 403 }
404 } while @{ $self->[2] };
296 405
297 unless (@{ $self->[2] }) { 406 # everything written
298 undef $self->[3]; 407 undef $self->[3];
408
299 # invoke run callback 409 # invoke run callback, if any
300 $self->[0]->($self->[1]) if $self->[0]; 410 $self->[4]->($self->[1]) if $self->[4];
301 }
302 }; 411 };
412
413 () # make sure we don't leak the watcher
303} 414}
304 415
305sub _new { 416sub _new {
306 my ($self, $fh) = @_; 417 my ($self, $fh, $pid) = @_;
307 418
308 AnyEvent::Util::fh_nonblocking $fh, 1; 419 AnyEvent::Util::fh_nonblocking $fh, 1;
309 420
310 $self = bless [ 421 $self = bless [
311 undef, # run callback 422 $pid,
312 $fh, 423 $fh,
313 [], # write queue - strings or fd's 424 [], # write queue - strings or fd's
314 undef, # AE watcher 425 undef, # AE watcher
315 ], $self; 426 ], $self;
316 427
… …
327 if ($pid eq 0) { 438 if ($pid eq 0) {
328 require AnyEvent::Fork::Serve; 439 require AnyEvent::Fork::Serve;
329 $AnyEvent::Fork::Serve::OWNER = $parent; 440 $AnyEvent::Fork::Serve::OWNER = $parent;
330 close $fh; 441 close $fh;
331 $0 = "$_[1] of $parent"; 442 $0 = "$_[1] of $parent";
443 $SIG{CHLD} = 'IGNORE';
332 AnyEvent::Fork::Serve::serve ($slave); 444 AnyEvent::Fork::Serve::serve ($slave);
333 AnyEvent::Fork::Util::_exit 0; 445 exit 0;
334 } elsif (!$pid) { 446 } elsif (!$pid) {
335 die "AnyEvent::Fork::Early/Template: unable to fork template process: $!"; 447 die "AnyEvent::Fork::Early/Template: unable to fork template process: $!";
336 } 448 }
337 449
338 AnyEvent::Fork->_new ($fh) 450 AnyEvent::Fork->_new ($fh, $pid)
339} 451}
340 452
341=item my $proc = new AnyEvent::Fork 453=item my $proc = new AnyEvent::Fork
342 454
343Create a new "empty" perl interpreter process and returns its process 455Create a new "empty" perl interpreter process and returns its process
344object for further manipulation. 456object for further manipulation.
345 457
346The new process is forked from a template process that is kept around 458The new process is forked from a template process that is kept around
347for this purpose. When it doesn't exist yet, it is created by a call to 459for this purpose. When it doesn't exist yet, it is created by a call to
348C<new_exec> and kept around for future calls. 460C<new_exec> first and then stays around for future calls.
349
350When the process object is destroyed, it will release the file handle
351that connects it with the new process. When the new process has not yet
352called C<run>, then the process will exit. Otherwise, what happens depends
353entirely on the code that is executed.
354 461
355=cut 462=cut
356 463
357sub new { 464sub new {
358 my $class = shift; 465 my $class = shift;
… …
394reduces the amount of memory sharing that is possible, and is also slower. 501reduces the amount of memory sharing that is possible, and is also slower.
395 502
396You should use C<new> whenever possible, except when having a template 503You should use C<new> whenever possible, except when having a template
397process around is unacceptable. 504process around is unacceptable.
398 505
399The path to the perl interpreter is divined usign various methods - first 506The path to the perl interpreter is divined using various methods - first
400C<$^X> is investigated to see if the path ends with something that sounds 507C<$^X> is investigated to see if the path ends with something that sounds
401as if it were the perl interpreter. Failing this, the module falls back to 508as if it were the perl interpreter. Failing this, the module falls back to
402using C<$Config::Config{perlpath}>. 509using C<$Config::Config{perlpath}>.
403 510
404=cut 511=cut
… …
413 my $perl = $; 520 my $perl = $;
414 521
415 # first we try $^X, but the path must be absolute (always on win32), and end in sth. 522 # first we try $^X, but the path must be absolute (always on win32), and end in sth.
416 # that looks like perl. this obviously only works for posix and win32 523 # that looks like perl. this obviously only works for posix and win32
417 unless ( 524 unless (
418 (AnyEvent::Fork::Util::WIN32 || $perl =~ m%^/%) 525 ($^O eq "MSWin32" || $perl =~ m%^/%)
419 && $perl =~ m%[/\\]perl(?:[0-9]+(\.[0-9]+)+)?(\.exe)?$%i 526 && $perl =~ m%[/\\]perl(?:[0-9]+(\.[0-9]+)+)?(\.exe)?$%i
420 ) { 527 ) {
421 # if it doesn't look perlish enough, try Config 528 # if it doesn't look perlish enough, try Config
422 require Config; 529 require Config;
423 $perl = $Config::Config{perlpath}; 530 $perl = $Config::Config{perlpath};
… …
427 require Proc::FastSpawn; 534 require Proc::FastSpawn;
428 535
429 my ($fh, $slave) = AnyEvent::Util::portable_socketpair; 536 my ($fh, $slave) = AnyEvent::Util::portable_socketpair;
430 Proc::FastSpawn::fd_inherit (fileno $slave); 537 Proc::FastSpawn::fd_inherit (fileno $slave);
431 538
539 # new fh's should always be set cloexec (due to $^F),
540 # but hey, not on win32, so we always clear the inherit flag.
541 Proc::FastSpawn::fd_inherit (fileno $fh, 0);
542
432 # quick. also doesn't work in win32. of course. what did you expect 543 # quick. also doesn't work in win32. of course. what did you expect
433 #local $ENV{PERL5LIB} = join ":", grep !ref, @INC; 544 #local $ENV{PERL5LIB} = join ":", grep !ref, @INC;
434 my %env = %ENV; 545 my %env = %ENV;
435 $env{PERL5LIB} = join +(AnyEvent::Fork::Util::WIN32 ? ";" : ":"), grep !ref, @INC; 546 $env{PERL5LIB} = join +($^O eq "MSWin32" ? ";" : ":"), grep !ref, @INC;
436 547
437 Proc::FastSpawn::spawn ( 548 my $pid = Proc::FastSpawn::spawn (
438 $perl, 549 $perl,
439 ["perl", "-MAnyEvent::Fork::Serve", "-e", "AnyEvent::Fork::Serve::me", fileno $slave, $$], 550 ["perl", "-MAnyEvent::Fork::Serve", "-e", "AnyEvent::Fork::Serve::me", fileno $slave, $$],
440 [map "$_=$env{$_}", keys %env], 551 [map "$_=$env{$_}", keys %env],
441 ) or die "unable to spawn AnyEvent::Fork server: $!"; 552 ) or die "unable to spawn AnyEvent::Fork server: $!";
442 553
443 $self->_new ($fh) 554 $self->_new ($fh, $pid)
555}
556
557=item $pid = $proc->pid
558
559Returns the process id of the process I<iff it is a direct child of the
560process> running AnyEvent::Fork, and C<undef> otherwise.
561
562Normally, only processes created via C<< AnyEvent::Fork->new_exec >> and
563L<AnyEvent::Fork::Template> are direct children, and you are responsible
564to clean up their zombies when they die.
565
566All other processes are not direct children, and will be cleaned up by
567AnyEvent::Fork itself.
568
569=cut
570
571sub pid {
572 $_[0][0]
444} 573}
445 574
446=item $proc = $proc->eval ($perlcode, @args) 575=item $proc = $proc->eval ($perlcode, @args)
447 576
448Evaluates the given C<$perlcode> as ... perl code, while setting C<@_> to 577Evaluates the given C<$perlcode> as ... perl code, while setting C<@_> to
449the strings specified by C<@args>. 578the strings specified by C<@args>, in the "main" package.
450 579
451This call is meant to do any custom initialisation that might be required 580This call is meant to do any custom initialisation that might be required
452(for example, the C<require> method uses it). It's not supposed to be used 581(for example, the C<require> method uses it). It's not supposed to be used
453to completely take over the process, use C<run> for that. 582to completely take over the process, use C<run> for that.
454 583
455The code will usually be executed after this call returns, and there is no 584The code will usually be executed after this call returns, and there is no
456way to pass anything back to the calling process. Any evaluation errors 585way to pass anything back to the calling process. Any evaluation errors
457will be reported to stderr and cause the process to exit. 586will be reported to stderr and cause the process to exit.
458 587
588If you want to execute some code to take over the process (see the
589"fork+exec" example in the SYNOPSIS), you should compile a function via
590C<eval> first, and then call it via C<run>. This also gives you access to
591any arguments passed via the C<send_xxx> methods, such as file handles.
592
459Returns the process object for easy chaining of method calls. 593Returns the process object for easy chaining of method calls.
460 594
461=cut 595=cut
462 596
463sub eval { 597sub eval {
464 my ($self, $code, @args) = @_; 598 my ($self, $code, @args) = @_;
465 599
466 $self->_cmd (e => $code, @args); 600 $self->_cmd (e => pack "(w/a*)*", $code, @args);
467 601
468 $self 602 $self
469} 603}
470 604
471=item $proc = $proc->require ($module, ...) 605=item $proc = $proc->require ($module, ...)
… …
495accomplished by simply not storing the file handles anywhere after passing 629accomplished by simply not storing the file handles anywhere after passing
496them to this method. 630them to this method.
497 631
498Returns the process object for easy chaining of method calls. 632Returns the process object for easy chaining of method calls.
499 633
500Example: pass an fh to a process, and release it without closing. it will 634Example: pass a file handle to a process, and release it without
501be closed automatically when it is no longer used. 635closing. It will be closed automatically when it is no longer used.
502 636
503 $proc->send_fh ($my_fh); 637 $proc->send_fh ($my_fh);
504 undef $my_fh; # free the reference if you want, but DO NOT CLOSE IT 638 undef $my_fh; # free the reference if you want, but DO NOT CLOSE IT
505 639
506=cut 640=cut
… …
519=item $proc = $proc->send_arg ($string, ...) 653=item $proc = $proc->send_arg ($string, ...)
520 654
521Send one or more argument strings to the process, to prepare a call to 655Send one or more argument strings to the process, to prepare a call to
522C<run>. The strings can be any octet string. 656C<run>. The strings can be any octet string.
523 657
658The protocol is optimised to pass a moderate number of relatively short
659strings - while you can pass up to 4GB of data in one go, this is more
660meant to pass some ID information or other startup info, not big chunks of
661data.
662
524Returns the process object for easy chaining of emthod calls. 663Returns the process object for easy chaining of method calls.
525 664
526=cut 665=cut
527 666
528sub send_arg { 667sub send_arg {
529 my ($self, @arg) = @_; 668 my ($self, @arg) = @_;
530 669
531 $self->_cmd (a => @arg); 670 $self->_cmd (a => pack "(w/a*)*", @arg);
532 671
533 $self 672 $self
534} 673}
535 674
536=item $proc->run ($func, $cb->($fh)) 675=item $proc->run ($func, $cb->($fh))
537 676
538Enter the function specified by the fully qualified name in C<$func> in 677Enter the function specified by the function name in C<$func> in the
539the process. The function is called with the communication socket as first 678process. The function is called with the communication socket as first
540argument, followed by all file handles and string arguments sent earlier 679argument, followed by all file handles and string arguments sent earlier
541via C<send_fh> and C<send_arg> methods, in the order they were called. 680via C<send_fh> and C<send_arg> methods, in the order they were called.
542 681
543If the called function returns, the process exits. 682The function name should be fully qualified, but if it isn't, it will be
683looked up in the main package.
544 684
545Preparing the process can take time - when the process is ready, the 685If the called function returns, doesn't exist, or any error occurs, the
686process exits.
687
688Preparing the process is done in the background - when all commands have
546callback is invoked with the local communications socket as argument. 689been sent, the callback is invoked with the local communications socket
690as argument. At this point you can start using the socket in any way you
691like.
547 692
548The process object becomes unusable on return from this function. 693The process object becomes unusable on return from this function - any
694further method calls result in undefined behaviour.
549 695
550If the communication socket isn't used, it should be closed on both sides, 696If the communication socket isn't used, it should be closed on both sides,
551to save on kernel memory. 697to save on kernel memory.
552 698
553The socket is non-blocking in the parent, and blocking in the newly 699The socket is non-blocking in the parent, and blocking in the newly
554created process. The close-on-exec flag is set on both. Even if not used 700created process. The close-on-exec flag is set in both.
701
555otherwise, the socket can be a good indicator for the existance of the 702Even if not used otherwise, the socket can be a good indicator for the
556process - if the other process exits, you get a readable event on it, 703existence of the process - if the other process exits, you get a readable
557because exiting the process closes the socket (if it didn't create any 704event on it, because exiting the process closes the socket (if it didn't
558children using fork). 705create any children using fork).
559 706
560Example: create a template for a process pool, pass a few strings, some 707Example: create a template for a process pool, pass a few strings, some
561file handles, then fork, pass one more string, and run some code. 708file handles, then fork, pass one more string, and run some code.
562 709
563 my $pool = AnyEvent::Fork 710 my $pool = AnyEvent::Fork
… …
571 ->send_arg ("str3") 718 ->send_arg ("str3")
572 ->run ("Some::function", sub { 719 ->run ("Some::function", sub {
573 my ($fh) = @_; 720 my ($fh) = @_;
574 721
575 # fh is nonblocking, but we trust that the OS can accept these 722 # fh is nonblocking, but we trust that the OS can accept these
576 # extra 3 octets anyway. 723 # few octets anyway.
577 syswrite $fh, "hi #$_\n"; 724 syswrite $fh, "hi #$_\n";
578 725
579 # $fh is being closed here, as we don't store it anywhere 726 # $fh is being closed here, as we don't store it anywhere
580 }); 727 });
581 } 728 }
… …
583 # Some::function might look like this - all parameters passed before fork 730 # Some::function might look like this - all parameters passed before fork
584 # and after will be passed, in order, after the communications socket. 731 # and after will be passed, in order, after the communications socket.
585 sub Some::function { 732 sub Some::function {
586 my ($fh, $str1, $str2, $fh1, $fh2, $str3) = @_; 733 my ($fh, $str1, $str2, $fh1, $fh2, $str3) = @_;
587 734
588 print scalar <$fh>; # prints "hi 1\n" and "hi 2\n" 735 print scalar <$fh>; # prints "hi #1\n" and "hi #2\n" in any order
589 } 736 }
590 737
591=cut 738=cut
592 739
593sub run { 740sub run {
594 my ($self, $func, $cb) = @_; 741 my ($self, $func, $cb) = @_;
595 742
596 $self->[0] = $cb; 743 $self->[4] = $cb;
597 $self->_cmd (r => $func); 744 $self->_cmd (r => $func);
598} 745}
599 746
600=back 747=back
601 748
749=head1 PERFORMANCE
750
751Now for some unscientific benchmark numbers (all done on an amd64
752GNU/Linux box). These are intended to give you an idea of the relative
753performance you can expect, they are not meant to be absolute performance
754numbers.
755
756OK, so, I ran a simple benchmark that creates a socket pair, forks, calls
757exit in the child and waits for the socket to close in the parent. I did
758load AnyEvent, EV and AnyEvent::Fork, for a total process size of 5100kB.
759
760 2079 new processes per second, using manual socketpair + fork
761
762Then I did the same thing, but instead of calling fork, I called
763AnyEvent::Fork->new->run ("CORE::exit") and then again waited for the
764socket form the child to close on exit. This does the same thing as manual
765socket pair + fork, except that what is forked is the template process
766(2440kB), and the socket needs to be passed to the server at the other end
767of the socket first.
768
769 2307 new processes per second, using AnyEvent::Fork->new
770
771And finally, using C<new_exec> instead C<new>, using vforks+execs to exec
772a new perl interpreter and compile the small server each time, I get:
773
774 479 vfork+execs per second, using AnyEvent::Fork->new_exec
775
776So how can C<< AnyEvent->new >> be faster than a standard fork, even
777though it uses the same operations, but adds a lot of overhead?
778
779The difference is simply the process size: forking the 6MB process takes
780so much longer than forking the 2.5MB template process that the overhead
781introduced is canceled out.
782
783If the benchmark process grows, the normal fork becomes even slower:
784
785 1340 new processes, manual fork in a 20MB process
786 731 new processes, manual fork in a 200MB process
787 235 new processes, manual fork in a 2000MB process
788
789What that means (to me) is that I can use this module without having a
790very bad conscience because of the extra overhead required to start new
791processes.
792
793=head1 TYPICAL PROBLEMS
794
795This section lists typical problems that remain. I hope by recognising
796them, most can be avoided.
797
798=over 4
799
800=item "leaked" file descriptors for exec'ed processes
801
802POSIX systems inherit file descriptors by default when exec'ing a new
803process. While perl itself laudably sets the close-on-exec flags on new
804file handles, most C libraries don't care, and even if all cared, it's
805often not possible to set the flag in a race-free manner.
806
807That means some file descriptors can leak through. And since it isn't
808possible to know which file descriptors are "good" and "necessary" (or
809even to know which file descriptors are open), there is no good way to
810close the ones that might harm.
811
812As an example of what "harm" can be done consider a web server that
813accepts connections and afterwards some module uses AnyEvent::Fork for the
814first time, causing it to fork and exec a new process, which might inherit
815the network socket. When the server closes the socket, it is still open
816in the child (which doesn't even know that) and the client might conclude
817that the connection is still fine.
818
819For the main program, there are multiple remedies available -
820L<AnyEvent::Fork::Early> is one, creating a process early and not using
821C<new_exec> is another, as in both cases, the first process can be exec'ed
822well before many random file descriptors are open.
823
824In general, the solution for these kind of problems is to fix the
825libraries or the code that leaks those file descriptors.
826
827Fortunately, most of these leaked descriptors do no harm, other than
828sitting on some resources.
829
830=item "leaked" file descriptors for fork'ed processes
831
832Normally, L<AnyEvent::Fork> does start new processes by exec'ing them,
833which closes file descriptors not marked for being inherited.
834
835However, L<AnyEvent::Fork::Early> and L<AnyEvent::Fork::Template> offer
836a way to create these processes by forking, and this leaks more file
837descriptors than exec'ing them, as there is no way to mark descriptors as
838"close on fork".
839
840An example would be modules like L<EV>, L<IO::AIO> or L<Gtk2>. Both create
841pipes for internal uses, and L<Gtk2> might open a connection to the X
842server. L<EV> and L<IO::AIO> can deal with fork, but Gtk2 might have
843trouble with a fork.
844
845The solution is to either not load these modules before use'ing
846L<AnyEvent::Fork::Early> or L<AnyEvent::Fork::Template>, or to delay
847initialising them, for example, by calling C<init Gtk2> manually.
848
849=item exit runs destructors
850
851This only applies to users of Lc<AnyEvent::Fork:Early> and
852L<AnyEvent::Fork::Template>.
853
854When a process created by AnyEvent::Fork exits, it might do so by calling
855exit, or simply letting perl reach the end of the program. At which point
856Perl runs all destructors.
857
858Not all destructors are fork-safe - for example, an object that represents
859the connection to an X display might tell the X server to free resources,
860which is inconvenient when the "real" object in the parent still needs to
861use them.
862
863This is obviously not a problem for L<AnyEvent::Fork::Early>, as you used
864it as the very first thing, right?
865
866It is a problem for L<AnyEvent::Fork::Template> though - and the solution
867is to not create objects with nontrivial destructors that might have an
868effect outside of Perl.
869
870=back
871
602=head1 PORTABILITY NOTES 872=head1 PORTABILITY NOTES
603 873
604Win32 is a loser - code has been written for this platform, pain has been 874Native win32 perls are somewhat supported (AnyEvent::Fork::Early is a nop,
605felt, but in the end, this platform is just too broken - maybe a later 875and ::Template is not going to work), and it cost a lot of blood and sweat
606version can do it. 876to make it so, mostly due to the bloody broken perl that nobody seems to
877care about. The fork emulation is a bad joke - I have yet to see something
878useful that you can do with it without running into memory corruption
879issues or other braindamage. Hrrrr.
880
881Cygwin perl is not supported at the moment, as it should implement fd
882passing, but doesn't, and rolling my own is hard, as cygwin doesn't
883support enough functionality to do it.
884
885=head1 SEE ALSO
886
887L<AnyEvent::Fork::Early> (to avoid executing a perl interpreter),
888L<AnyEvent::Fork::Template> (to create a process by forking the main
889program at a convenient time).
607 890
608=head1 AUTHOR 891=head1 AUTHOR
609 892
610 Marc Lehmann <schmorp@schmorp.de> 893 Marc Lehmann <schmorp@schmorp.de>
611 http://home.schmorp.de/ 894 http://home.schmorp.de/

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines