[ViewVC] Diff of: cvs/AnyEvent-Fork/Fork.pm

Comparing AnyEvent-Fork/Fork.pm (file contents):
Revision 1.9 by root, Thu Apr 4 03:45:12 2013 UTC vs.
Revision 1.39 by root, Sat Apr 6 22:39:37 2013 UTC

…		…
1	=head1 NAME	1	=head1 NAME
2		2
3	AnyEvent::Fork - everything you wanted to use fork() for, but couldn't	3	AnyEvent::Fork - everything you wanted to use fork() for, but couldn't
4		4
5	ATTENTION, this is a very early release, and very untested. Consider it a
6	technology 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
		18	This module allows you to create new processes, without actually forking
		19	them from your current process (avoiding the problems of forking), but
		20	preserving most of the advantages of fork.
		21
		22	It can be used to create new worker processes or new independent
		23	subprocesses for short- and long-running jobs, process pools (e.g. for use
		24	in pre-forked servers) but also to spawn new external processes (such as
		25	CGI scripts from a web server), which can be faster (and more well behaved)
		26	than using fork+exec in big processes.
		27
		28	Special care has been taken to make this module useful from other modules,
		29	while still supporting specialised environments such as L<App::Staticperl>
		30	or L<PAR::Packer>.
		31
		32	=head2 WHAT THIS MODULE IS NOT
		33
		34	This module only creates processes and lets you pass file handles and
		35	strings to it, and run perl code. It does not implement any kind of RPC -
		36	there is no back channel from the process back to you, and there is no RPC
		37	or message passing going on.
		38
		39	If you need some form of RPC, you can either implement it yourself
		40	in whatever way you like, use some message-passing module such
		41	as L<AnyEvent::MP>, some pipe such as L<AnyEvent::ZeroMQ>, use
		42	L<AnyEvent::Handle> on both sides to send e.g. JSON or Storable messages,
		43	and so on.
		44
		45	=head2 COMPARISON TO OTHER MODULES
		46
		47	There is an abundance of modules on CPAN that do "something fork", such as
		48	L<Parallel::ForkManager>, L<AnyEvent::ForkManager>, L<AnyEvent::Worker>
		49	or L<AnyEvent::Subprocess>. There are modules that implement their own
		50	process management, such as L<AnyEvent::DBI>.
		51
		52	The problems that all these modules try to solve are real, however, none
		53	of them (from what I have seen) tackle the very real problems of unwanted
		54	memory sharing, efficiency, not being able to use event processing or
		55	similar modules in the processes they create.
		56
		57	This module doesn't try to replace any of them - instead it tries to solve
		58	the problem of creating processes with a minimum of fuss and overhead (and
		59	also luxury). Ideally, most of these would use AnyEvent::Fork internally,
		60	except they were written before AnyEvent:Fork was available, so obviously
		61	had to roll their own.
		62
		63	=head2 PROBLEM STATEMENT
		64
		65	There are two traditional ways to implement parallel processing on UNIX
		66	like operating systems - fork and process, and fork+exec and process. They
		67	have different advantages and disadvantages that I describe below,
		68	together with how this module tries to mitigate the disadvantages.
		69
		70	=over 4
		71
		72	=item Forking from a big process can be very slow.
		73
		74	A 5GB process needs 0.05s to fork on my 3.6GHz amd64 GNU/Linux box. This
		75	overhead is often shared with exec (because you have to fork first), but
		76	in some circumstances (e.g. when vfork is used), fork+exec can be much
		77	faster.
		78
		79	This module can help here by telling a small(er) helper process to fork,
		80	which is faster then forking the main process, and also uses vfork where
		81	possible. This gives the speed of vfork, with the flexibility of fork.
		82
		83	=item Forking usually creates a copy-on-write copy of the parent
		84	process.
		85
		86	For example, modules or data files that are loaded will not use additional
		87	memory after a fork. When exec'ing a new process, modules and data files
		88	might need to be loaded again, at extra CPU and memory cost. But when
		89	forking, literally all data structures are copied - if the program frees
		90	them and replaces them by new data, the child processes will retain the
		91	old version even if it isn't used, which can suddenly and unexpectedly
		92	increase memory usage when freeing memory.
		93
		94	The trade-off is between more sharing with fork (which can be good or
		95	bad), and no sharing with exec.
		96
		97	This module allows the main program to do a controlled fork, and allows
		98	modules to exec processes safely at any time. When creating a custom
		99	process pool you can take advantage of data sharing via fork without
		100	risking to share large dynamic data structures that will blow up child
		101	memory usage.
		102
		103	In other words, this module puts you into control over what is being
		104	shared and what isn't, at all times.
		105
		106	=item Exec'ing a new perl process might be difficult.
		107
		108	For example, it is not easy to find the correct path to the perl
		109	interpreter - C<$^X> might not be a perl interpreter at all.
		110
		111	This module tries hard to identify the correct path to the perl
		112	interpreter. With a cooperative main program, exec'ing the interpreter
		113	might not even be necessary, but even without help from the main program,
		114	it will still work when used from a module.
		115
		116	=item Exec'ing a new perl process might be slow, as all necessary modules
		117	have to be loaded from disk again, with no guarantees of success.
		118
		119	Long running processes might run into problems when perl is upgraded
		120	and modules are no longer loadable because they refer to a different
		121	perl version, or parts of a distribution are newer than the ones already
		122	loaded.
		123
		124	This module supports creating pre-initialised perl processes to be used as
		125	a template for new processes.
		126
		127	=item Forking might be impossible when a program is running.
		128
		129	For example, POSIX makes it almost impossible to fork from a
		130	multi-threaded program while doing anything useful in the child - in
		131	fact, if your perl program uses POSIX threads (even indirectly via
		132	e.g. L<IO::AIO> or L<threads>), you cannot call fork on the perl level
		133	anymore without risking corruption issues on a number of operating
		134	systems.
		135
		136	This module can safely fork helper processes at any time, by calling
		137	fork+exec in C, in a POSIX-compatible way (via L<Proc::FastSpawn>).
		138
		139	=item Parallel processing with fork might be inconvenient or difficult
		140	to implement. Modules might not work in both parent and child.
		141
		142	For example, when a program uses an event loop and creates watchers it
		143	becomes very hard to use the event loop from a child program, as the
		144	watchers already exist but are only meaningful in the parent. Worse, a
		145	module might want to use such a module, not knowing whether another module
		146	or the main program also does, leading to problems.
		147
		148	Apart from event loops, graphical toolkits also commonly fall into the
		149	"unsafe module" category, or just about anything that communicates with
		150	the external world, such as network libraries and file I/O modules, which
		151	usually don't like being copied and then allowed to continue in two
		152	processes.
		153
		154	With this module only the main program is allowed to create new processes
		155	by forking (because only the main program can know when it is still safe
		156	to do so) - all other processes are created via fork+exec, which makes it
		157	possible to use modules such as event loops or window interfaces safely.
		158
		159	=back
		160
		161	=head1 EXAMPLES
		162
13	# create a single new process, tell it to run your worker function	163	=head2 Create a single new process, tell it to run your worker function.
14		164
15	AnyEvent::Fork	165	AnyEvent::Fork
16	->new	166	->new
17	->require ("MyModule")	167	->require ("MyModule")
18	->run ("MyModule::worker, sub {	168	->run ("MyModule::worker, sub {
20		170
21	# now $master_filehandle is connected to the	171	# now $master_filehandle is connected to the
22	# $slave_filehandle in the new process.	172	# $slave_filehandle in the new process.
23	});	173	});
24		174
25	# MyModule::worker might look like this	175	C<MyModule> might look like this:
		176
		177	package MyModule;
		178
26	sub MyModule::worker {	179	sub worker {
27	my ($slave_filehandle) = @_;	180	my ($slave_filehandle) = @_;
28		181
29	# now $slave_filehandle is connected to the $master_filehandle	182	# now $slave_filehandle is connected to the $master_filehandle
30	# in the original prorcess. have fun!	183	# in the original prorcess. have fun!
31	}	184	}
32		185
33	##################################################################
34	# create a pool of server processes all accepting on the same socket	186	=head2 Create a pool of server processes all accepting on the same socket.
35		187
36	# create listener socket	188	# create listener socket
37	my $listener = ...;	189	my $listener = ...;
38		190
39	# create a pool template, initialise it and give it the socket	191	# create a pool template, initialise it and give it the socket
…		…
51	}	203	}
52		204
53	# now do other things - maybe use the filehandle provided by run	205	# now do other things - maybe use the filehandle provided by run
54	# to wait for the processes to die. or whatever.	206	# to wait for the processes to die. or whatever.
55		207
56	# My::Server::run might look like this	208	C<My::Server> might look like this:
57	sub My::Server::run {	209
		210	package My::Server;
		211
		212	sub run {
58	my ($slave, $listener, $id) = @_;	213	my ($slave, $listener, $id) = @_;
59		214
60	close $slave; # we do not use the socket, so close it to save resources	215	close $slave; # we do not use the socket, so close it to save resources
61		216
62	# we could go ballistic and use e.g. AnyEvent here, or IO::AIO,	217	# we could go ballistic and use e.g. AnyEvent here, or IO::AIO,
…		…
64	while (my $socket = $listener->accept) {	219	while (my $socket = $listener->accept) {
65	# do sth. with new socket	220	# do sth. with new socket
66	}	221	}
67	}	222	}
68		223
69	=head1 DESCRIPTION	224	=head2 use AnyEvent::Fork as a faster fork+exec
70		225
71	This module allows you to create new processes, without actually forking	226	This runs C</bin/echo hi>, with stdandard output redirected to /tmp/log
72	them from your current process (avoiding the problems of forking), but	227	and standard error redirected to the communications socket. It is usually
73	preserving most of the advantages of fork.	228	faster than fork+exec, but still lets you prepare the environment.
74		229
75	It can be used to create new worker processes or new independent	230	open my $output, ">/tmp/log" or die "$!";
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		231
81	Special care has been taken to make this module useful from other modules,	232	AnyEvent::Fork
82	while still supporting specialised environments such as L<App::Staticperl>	233	->new
83	or L<PAR::Packer>.	234	->eval ('
		235	sub run {
		236	my ($fh, $output, @cmd) = @_;
84		237
85	=head1 PROBLEM STATEMENT	238	# perl will clear close-on-exec on STDOUT/STDERR
		239	open STDOUT, ">&", $output or die;
		240	open STDERR, ">&", $fh or die;
86		241
87	There are two ways to implement parallel processing on UNIX like operating	242	exec @cmd;
88	systems - fork and process, and fork+exec and process. They have different	243	}
89	advantages and disadvantages that I describe below, together with how this	244	')
90	module tries to mitigate the disadvantages.	245	->send_fh ($output)
		246	->send_arg ("/bin/echo", "hi")
		247	->run ("run", my $cv = AE::cv);
91		248
92	=over 4	249	my $stderr = $cv->recv;
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		250
149	=head1 CONCEPTS	251	=head1 CONCEPTS
150		252
151	This module can create new processes either by executing a new perl	253	This module can create new processes either by executing a new perl
152	process, or by forking from an existing "template" process.	254	process, or by forking from an existing "template" process.
…		…
169	needed the first time. Forking from this process shares the memory used	271	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	272	for the perl interpreter with the new process, but loading modules takes
171	time, and the memory is not shared with anything else.	273	time, and the memory is not shared with anything else.
172		274
173	This is ideal for when you only need one extra process of a kind, with the	275	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.	276	option of starting and stopping it on demand.
175		277
176	Example:	278	Example:
177		279
178	AnyEvent::Fork	280	AnyEvent::Fork
179	->new	281	->new
…		…
194	modules you loaded) is shared between the processes, and each new process	296	modules you loaded) is shared between the processes, and each new process
195	consumes relatively little memory of its own.	297	consumes relatively little memory of its own.
196		298
197	The disadvantage of this approach is that you need to create a template	299	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	300	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	301	only need a fixed number of processes you can create them, and then destroy
200	the template process.	302	the template process.
201		303
202	Example:	304	Example:
203		305
204	my $template = AnyEvent::Fork->new->require ("Some::Module");	306	my $template = AnyEvent::Fork->new->require ("Some::Module");
…		…
231	my ($fork_fh) = @_;	333	my ($fork_fh) = @_;
232	});	334	});
233		335
234	=back	336	=back
235		337
236	=head1 FUNCTIONS	338	=head1 THE C<AnyEvent::Fork> CLASS
		339
		340	This module exports nothing, and only implements a single class -
		341	C<AnyEvent::Fork>.
		342
		343	There are two class constructors that both create new processes - C<new>
		344	and C<new_exec>. The C<fork> method creates a new process by forking an
		345	existing one and could be considered a third constructor.
		346
		347	Most of the remaining methods deal with preparing the new process, by
		348	loading code, evaluating code and sending data to the new process. They
		349	usually return the process object, so you can chain method calls.
		350
		351	If a process object is destroyed before calling its C<run> method, then
		352	the process simply exits. After C<run> is called, all responsibility is
		353	passed to the specified function.
		354
		355	As long as there is any outstanding work to be done, process objects
		356	resist being destroyed, so there is no reason to store them unless you
		357	need them later - configure and forget works just fine.
237		358
238	=over 4	359	=over 4
239		360
240	=cut	361	=cut
241		362
242	package AnyEvent::Fork;	363	package AnyEvent::Fork;
243		364
244	use common::sense;	365	use common::sense;
245		366
246	use Socket ();	367	use Errno ();
247		368
248	use AnyEvent;	369	use AnyEvent;
249	use AnyEvent::Fork::Util;
250	use AnyEvent::Util ();	370	use AnyEvent::Util ();
251		371
		372	use IO::FDPass;
		373
		374	our $VERSION = 0.5;
		375
252	our $PERL; # the path to the perl interpreter, deduces with various forms of magic	376	our $PERL; # the path to the perl interpreter, deduces with various forms of magic
253
254	=item my $pool = new AnyEvent::Fork key => value...
255
256	Create a new process pool. The following named parameters are supported:
257		377
258	=over 4	378	=over 4
259		379
260	=back	380	=back
261		381
…		…
268	our $TEMPLATE;	388	our $TEMPLATE;
269		389
270	sub _cmd {	390	sub _cmd {
271	my $self = shift;	391	my $self = shift;
272		392
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	393	# 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.	394	# 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)*", @_;	395	# it.
		396	push @{ $self->[2] }, pack "a L/a*", $_[0], $_[1];
278		397
279	$self->[3] \|\|= AE::io $self->[1], 1, sub {	398	$self->[3] \|\|= AE::io $self->[1], 1, sub {
		399	do {
280	# send the next "thing" in the queue - either a reference to an fh,	400	# send the next "thing" in the queue - either a reference to an fh,
281	# or a plain string.	401	# or a plain string.
282		402
283	if (ref $self->[2][0]) {	403	if (ref $self->[2][0]) {
284	# send fh	404	# send fh
285	AnyEvent::Fork::Util::fd_send fileno $self->[1], fileno ${ $self->[2][0] }	405	unless (IO::FDPass::send fileno $self->[1], fileno ${ $self->[2][0] }) {
		406	return if $! == Errno::EAGAIN \|\| $! == Errno::EWOULDBLOCK;
		407	undef $self->[3];
		408	die "AnyEvent::Fork: file descriptor send failure: $!";
		409	}
		410
286	and shift @{ $self->[2] };	411	shift @{ $self->[2] };
287		412
288	} else {	413	} else {
289	# send string	414	# send string
290	my $len = syswrite $self->[1], $self->[2][0]	415	my $len = syswrite $self->[1], $self->[2][0];
		416
		417	unless ($len) {
		418	return if $! == Errno::EAGAIN \|\| $! == Errno::EWOULDBLOCK;
		419	undef $self->[3];
291	or do { undef $self->[3]; die "AnyEvent::Fork: command write failure: $!" };	420	die "AnyEvent::Fork: command write failure: $!";
		421	}
292		422
293	substr $self->[2][0], 0, $len, "";	423	substr $self->[2][0], 0, $len, "";
294	shift @{ $self->[2] } unless length $self->[2][0];	424	shift @{ $self->[2] } unless length $self->[2][0];
295	}	425	}
		426	} while @{ $self->[2] };
296		427
297	unless (@{ $self->[2] }) {	428	# everything written
298	undef $self->[3];	429	undef $self->[3];
		430
299	# invoke run callback	431	# invoke run callback, if any
300	$self->[0]->($self->[1]) if $self->[0];	432	$self->[4]->($self->[1]) if $self->[4];
301	}
302	};	433	};
		434
		435	() # make sure we don't leak the watcher
303	}	436	}
304		437
305	sub _new {	438	sub _new {
306	my ($self, $fh) = @_;	439	my ($self, $fh, $pid) = @_;
307		440
308	AnyEvent::Util::fh_nonblocking $fh, 1;	441	AnyEvent::Util::fh_nonblocking $fh, 1;
309		442
310	$self = bless [	443	$self = bless [
311	undef, # run callback	444	$pid,
312	$fh,	445	$fh,
313	[], # write queue - strings or fd's	446	[], # write queue - strings or fd's
314	undef, # AE watcher	447	undef, # AE watcher
315	], $self;	448	], $self;
316		449
…		…
327	if ($pid eq 0) {	460	if ($pid eq 0) {
328	require AnyEvent::Fork::Serve;	461	require AnyEvent::Fork::Serve;
329	$AnyEvent::Fork::Serve::OWNER = $parent;	462	$AnyEvent::Fork::Serve::OWNER = $parent;
330	close $fh;	463	close $fh;
331	$0 = "$_[1] of $parent";	464	$0 = "$_[1] of $parent";
		465	$SIG{CHLD} = 'IGNORE';
332	AnyEvent::Fork::Serve::serve ($slave);	466	AnyEvent::Fork::Serve::serve ($slave);
333	AnyEvent::Fork::Util::_exit 0;	467	exit 0;
334	} elsif (!$pid) {	468	} elsif (!$pid) {
335	die "AnyEvent::Fork::Early/Template: unable to fork template process: $!";	469	die "AnyEvent::Fork::Early/Template: unable to fork template process: $!";
336	}	470	}
337		471
338	AnyEvent::Fork->_new ($fh)	472	AnyEvent::Fork->_new ($fh, $pid)
339	}	473	}
340		474
341	=item my $proc = new AnyEvent::Fork	475	=item my $proc = new AnyEvent::Fork
342		476
343	Create a new "empty" perl interpreter process and returns its process	477	Create a new "empty" perl interpreter process and returns its process
344	object for further manipulation.	478	object for further manipulation.
345		479
346	The new process is forked from a template process that is kept around	480	The new process is forked from a template process that is kept around
347	for this purpose. When it doesn't exist yet, it is created by a call to	481	for this purpose. When it doesn't exist yet, it is created by a call to
348	C<new_exec> and kept around for future calls.	482	C<new_exec> first and then stays around for future calls.
349
350	When the process object is destroyed, it will release the file handle
351	that connects it with the new process. When the new process has not yet
352	called C<run>, then the process will exit. Otherwise, what happens depends
353	entirely on the code that is executed.
354		483
355	=cut	484	=cut
356		485
357	sub new {	486	sub new {
358	my $class = shift;	487	my $class = shift;
…		…
394	reduces the amount of memory sharing that is possible, and is also slower.	523	reduces the amount of memory sharing that is possible, and is also slower.
395		524
396	You should use C<new> whenever possible, except when having a template	525	You should use C<new> whenever possible, except when having a template
397	process around is unacceptable.	526	process around is unacceptable.
398		527
399	The path to the perl interpreter is divined usign various methods - first	528	The path to the perl interpreter is divined using various methods - first
400	C<$^X> is investigated to see if the path ends with something that sounds	529	C<$^X> is investigated to see if the path ends with something that sounds
401	as if it were the perl interpreter. Failing this, the module falls back to	530	as if it were the perl interpreter. Failing this, the module falls back to
402	using C<$Config::Config{perlpath}>.	531	using C<$Config::Config{perlpath}>.
403		532
404	=cut	533	=cut
…		…
413	my $perl = $;	542	my $perl = $;
414		543
415	# first we try $^X, but the path must be absolute (always on win32), and end in sth.	544	# 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	545	# that looks like perl. this obviously only works for posix and win32
417	unless (	546	unless (
418	(AnyEvent::Fork::Util::WIN32 \|\| $perl =~ m%^/%)	547	($^O eq "MSWin32" \|\| $perl =~ m%^/%)
419	&& $perl =~ m%[/\\]perl(?:[0-9]+(\.[0-9]+)+)?(\.exe)?$%i	548	&& $perl =~ m%[/\\]perl(?:[0-9]+(\.[0-9]+)+)?(\.exe)?$%i
420	) {	549	) {
421	# if it doesn't look perlish enough, try Config	550	# if it doesn't look perlish enough, try Config
422	require Config;	551	require Config;
423	$perl = $Config::Config{perlpath};	552	$perl = $Config::Config{perlpath};
…		…
427	require Proc::FastSpawn;	556	require Proc::FastSpawn;
428		557
429	my ($fh, $slave) = AnyEvent::Util::portable_socketpair;	558	my ($fh, $slave) = AnyEvent::Util::portable_socketpair;
430	Proc::FastSpawn::fd_inherit (fileno $slave);	559	Proc::FastSpawn::fd_inherit (fileno $slave);
431		560
		561	# new fh's should always be set cloexec (due to $^F),
		562	# but hey, not on win32, so we always clear the inherit flag.
		563	Proc::FastSpawn::fd_inherit (fileno $fh, 0);
		564
432	# quick. also doesn't work in win32. of course. what did you expect	565	# quick. also doesn't work in win32. of course. what did you expect
433	#local $ENV{PERL5LIB} = join ":", grep !ref, @INC;	566	#local $ENV{PERL5LIB} = join ":", grep !ref, @INC;
434	my %env = %ENV;	567	my %env = %ENV;
435	$env{PERL5LIB} = join +(AnyEvent::Fork::Util::WIN32 ? ";" : ":"), grep !ref, @INC;	568	$env{PERL5LIB} = join +($^O eq "MSWin32" ? ";" : ":"), grep !ref, @INC;
436		569
437	Proc::FastSpawn::spawn (	570	my $pid = Proc::FastSpawn::spawn (
438	$perl,	571	$perl,
439	["perl", "-MAnyEvent::Fork::Serve", "-e", "AnyEvent::Fork::Serve::me", fileno $slave, $$],	572	["perl", "-MAnyEvent::Fork::Serve", "-e", "AnyEvent::Fork::Serve::me", fileno $slave, $$],
440	[map "$_=$env{$_}", keys %env],	573	[map "$_=$env{$_}", keys %env],
441	) or die "unable to spawn AnyEvent::Fork server: $!";	574	) or die "unable to spawn AnyEvent::Fork server: $!";
442		575
443	$self->_new ($fh)	576	$self->_new ($fh, $pid)
		577	}
		578
		579	=item $pid = $proc->pid
		580
		581	Returns the process id of the process I<iff it is a direct child of the
		582	process running AnyEvent::Fork>, and C<undef> otherwise.
		583
		584	Normally, only processes created via C<< AnyEvent::Fork->new_exec >> and
		585	L<AnyEvent::Fork::Template> are direct children, and you are responsible
		586	to clean up their zombies when they die.
		587
		588	All other processes are not direct children, and will be cleaned up by
		589	AnyEvent::Fork itself.
		590
		591	=cut
		592
		593	sub pid {
		594	$_[0][0]
444	}	595	}
445		596
446	=item $proc = $proc->eval ($perlcode, @args)	597	=item $proc = $proc->eval ($perlcode, @args)
447		598
448	Evaluates the given C<$perlcode> as ... perl code, while setting C<@_> to	599	Evaluates the given C<$perlcode> as ... perl code, while setting C<@_> to
449	the strings specified by C<@args>.	600	the strings specified by C<@args>, in the "main" package.
450		601
451	This call is meant to do any custom initialisation that might be required	602	This 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	603	(for example, the C<require> method uses it). It's not supposed to be used
453	to completely take over the process, use C<run> for that.	604	to completely take over the process, use C<run> for that.
454		605
455	The code will usually be executed after this call returns, and there is no	606	The code will usually be executed after this call returns, and there is no
456	way to pass anything back to the calling process. Any evaluation errors	607	way to pass anything back to the calling process. Any evaluation errors
457	will be reported to stderr and cause the process to exit.	608	will be reported to stderr and cause the process to exit.
458		609
		610	If you want to execute some code (that isn't in a module) to take over the
		611	process, you should compile a function via C<eval> first, and then call
		612	it via C<run>. This also gives you access to any arguments passed via the
		613	C<send_xxx> methods, such as file handles. See the L<use AnyEvent::Fork as
		614	a faster fork+exec> example to see it in action.
		615
459	Returns the process object for easy chaining of method calls.	616	Returns the process object for easy chaining of method calls.
460		617
461	=cut	618	=cut
462		619
463	sub eval {	620	sub eval {
464	my ($self, $code, @args) = @_;	621	my ($self, $code, @args) = @_;
465		622
466	$self->_cmd (e => $code, @args);	623	$self->_cmd (e => pack "(w/a)", $code, @args);
467		624
468	$self	625	$self
469	}	626	}
470		627
471	=item $proc = $proc->require ($module, ...)	628	=item $proc = $proc->require ($module, ...)
…		…
488	=item $proc = $proc->send_fh ($handle, ...)	645	=item $proc = $proc->send_fh ($handle, ...)
489		646
490	Send one or more file handles (I<not> file descriptors) to the process,	647	Send one or more file handles (I<not> file descriptors) to the process,
491	to prepare a call to C<run>.	648	to prepare a call to C<run>.
492		649
493	The process object keeps a reference to the handles until this is done,	650	The process object keeps a reference to the handles until they have
494	so you must not explicitly close the handles. This is most easily	651	been passed over to the process, so you must not explicitly close the
495	accomplished by simply not storing the file handles anywhere after passing	652	handles. This is most easily accomplished by simply not storing the file
496	them to this method.	653	handles anywhere after passing them to this method - when AnyEvent::Fork
		654	is finished using them, perl will automatically close them.
497		655
498	Returns the process object for easy chaining of method calls.	656	Returns the process object for easy chaining of method calls.
499		657
500	Example: pass an fh to a process, and release it without closing. it will	658	Example: pass a file handle to a process, and release it without
501	be closed automatically when it is no longer used.	659	closing. It will be closed automatically when it is no longer used.
502		660
503	$proc->send_fh ($my_fh);	661	$proc->send_fh ($my_fh);
504	undef $my_fh; # free the reference if you want, but DO NOT CLOSE IT	662	undef $my_fh; # free the reference if you want, but DO NOT CLOSE IT
505		663
506	=cut	664	=cut
…		…
517	}	675	}
518		676
519	=item $proc = $proc->send_arg ($string, ...)	677	=item $proc = $proc->send_arg ($string, ...)
520		678
521	Send one or more argument strings to the process, to prepare a call to	679	Send one or more argument strings to the process, to prepare a call to
522	C<run>. The strings can be any octet string.	680	C<run>. The strings can be any octet strings.
523		681
		682	The protocol is optimised to pass a moderate number of relatively short
		683	strings - while you can pass up to 4GB of data in one go, this is more
		684	meant to pass some ID information or other startup info, not big chunks of
		685	data.
		686
524	Returns the process object for easy chaining of emthod calls.	687	Returns the process object for easy chaining of method calls.
525		688
526	=cut	689	=cut
527		690
528	sub send_arg {	691	sub send_arg {
529	my ($self, @arg) = @_;	692	my ($self, @arg) = @_;
530		693
531	$self->_cmd (a => @arg);	694	$self->_cmd (a => pack "(w/a)", @arg);
532		695
533	$self	696	$self
534	}	697	}
535		698
536	=item $proc->run ($func, $cb->($fh))	699	=item $proc->run ($func, $cb->($fh))
537		700
538	Enter the function specified by the fully qualified name in C<$func> in	701	Enter the function specified by the function name in C<$func> in the
539	the process. The function is called with the communication socket as first	702	process. The function is called with the communication socket as first
540	argument, followed by all file handles and string arguments sent earlier	703	argument, followed by all file handles and string arguments sent earlier
541	via C<send_fh> and C<send_arg> methods, in the order they were called.	704	via C<send_fh> and C<send_arg> methods, in the order they were called.
542		705
543	If the called function returns, the process exits.
544
545	Preparing the process can take time - when the process is ready, the
546	callback is invoked with the local communications socket as argument.
547
548	The process object becomes unusable on return from this function.	706	The process object becomes unusable on return from this function - any
		707	further method calls result in undefined behaviour.
		708
		709	The function name should be fully qualified, but if it isn't, it will be
		710	looked up in the C<main> package.
		711
		712	If the called function returns, doesn't exist, or any error occurs, the
		713	process exits.
		714
		715	Preparing the process is done in the background - when all commands have
		716	been sent, the callback is invoked with the local communications socket
		717	as argument. At this point you can start using the socket in any way you
		718	like.
549		719
550	If the communication socket isn't used, it should be closed on both sides,	720	If the communication socket isn't used, it should be closed on both sides,
551	to save on kernel memory.	721	to save on kernel memory.
552		722
553	The socket is non-blocking in the parent, and blocking in the newly	723	The socket is non-blocking in the parent, and blocking in the newly
554	created process. The close-on-exec flag is set on both. Even if not used	724	created process. The close-on-exec flag is set in both.
		725
555	otherwise, the socket can be a good indicator for the existance of the	726	Even if not used otherwise, the socket can be a good indicator for the
556	process - if the other process exits, you get a readable event on it,	727	existence of the process - if the other process exits, you get a readable
557	because exiting the process closes the socket (if it didn't create any	728	event on it, because exiting the process closes the socket (if it didn't
558	children using fork).	729	create any children using fork).
559		730
560	Example: create a template for a process pool, pass a few strings, some	731	Example: create a template for a process pool, pass a few strings, some
561	file handles, then fork, pass one more string, and run some code.	732	file handles, then fork, pass one more string, and run some code.
562		733
563	my $pool = AnyEvent::Fork	734	my $pool = AnyEvent::Fork
…		…
571	->send_arg ("str3")	742	->send_arg ("str3")
572	->run ("Some::function", sub {	743	->run ("Some::function", sub {
573	my ($fh) = @_;	744	my ($fh) = @_;
574		745
575	# fh is nonblocking, but we trust that the OS can accept these	746	# fh is nonblocking, but we trust that the OS can accept these
576	# extra 3 octets anyway.	747	# few octets anyway.
577	syswrite $fh, "hi #$_\n";	748	syswrite $fh, "hi #$_\n";
578		749
579	# $fh is being closed here, as we don't store it anywhere	750	# $fh is being closed here, as we don't store it anywhere
580	});	751	});
581	}	752	}
…		…
583	# Some::function might look like this - all parameters passed before fork	754	# Some::function might look like this - all parameters passed before fork
584	# and after will be passed, in order, after the communications socket.	755	# and after will be passed, in order, after the communications socket.
585	sub Some::function {	756	sub Some::function {
586	my ($fh, $str1, $str2, $fh1, $fh2, $str3) = @_;	757	my ($fh, $str1, $str2, $fh1, $fh2, $str3) = @_;
587		758
588	print scalar <$fh>; # prints "hi 1\n" and "hi 2\n"	759	print scalar <$fh>; # prints "hi #1\n" and "hi #2\n" in any order
589	}	760	}
590		761
591	=cut	762	=cut
592		763
593	sub run {	764	sub run {
594	my ($self, $func, $cb) = @_;	765	my ($self, $func, $cb) = @_;
595		766
596	$self->[0] = $cb;	767	$self->[4] = $cb;
597	$self->_cmd (r => $func);	768	$self->_cmd (r => $func);
598	}	769	}
599		770
600	=back	771	=back
601		772
		773	=head1 PERFORMANCE
		774
		775	Now for some unscientific benchmark numbers (all done on an amd64
		776	GNU/Linux box). These are intended to give you an idea of the relative
		777	performance you can expect, they are not meant to be absolute performance
		778	numbers.
		779
		780	OK, so, I ran a simple benchmark that creates a socket pair, forks, calls
		781	exit in the child and waits for the socket to close in the parent. I did
		782	load AnyEvent, EV and AnyEvent::Fork, for a total process size of 5100kB.
		783
		784	2079 new processes per second, using manual socketpair + fork
		785
		786	Then I did the same thing, but instead of calling fork, I called
		787	AnyEvent::Fork->new->run ("CORE::exit") and then again waited for the
		788	socket form the child to close on exit. This does the same thing as manual
		789	socket pair + fork, except that what is forked is the template process
		790	(2440kB), and the socket needs to be passed to the server at the other end
		791	of the socket first.
		792
		793	2307 new processes per second, using AnyEvent::Fork->new
		794
		795	And finally, using C<new_exec> instead C<new>, using vforks+execs to exec
		796	a new perl interpreter and compile the small server each time, I get:
		797
		798	479 vfork+execs per second, using AnyEvent::Fork->new_exec
		799
		800	So how can C<< AnyEvent->new >> be faster than a standard fork, even
		801	though it uses the same operations, but adds a lot of overhead?
		802
		803	The difference is simply the process size: forking the 5MB process takes
		804	so much longer than forking the 2.5MB template process that the extra
		805	overhead introduced is canceled out.
		806
		807	If the benchmark process grows, the normal fork becomes even slower:
		808
		809	1340 new processes, manual fork of a 20MB process
		810	731 new processes, manual fork of a 200MB process
		811	235 new processes, manual fork of a 2000MB process
		812
		813	What that means (to me) is that I can use this module without having a bad
		814	conscience because of the extra overhead required to start new processes.
		815
		816	=head1 TYPICAL PROBLEMS
		817
		818	This section lists typical problems that remain. I hope by recognising
		819	them, most can be avoided.
		820
		821	=over 4
		822
		823	=item leaked file descriptors for exec'ed processes
		824
		825	POSIX systems inherit file descriptors by default when exec'ing a new
		826	process. While perl itself laudably sets the close-on-exec flags on new
		827	file handles, most C libraries don't care, and even if all cared, it's
		828	often not possible to set the flag in a race-free manner.
		829
		830	That means some file descriptors can leak through. And since it isn't
		831	possible to know which file descriptors are "good" and "necessary" (or
		832	even to know which file descriptors are open), there is no good way to
		833	close the ones that might harm.
		834
		835	As an example of what "harm" can be done consider a web server that
		836	accepts connections and afterwards some module uses AnyEvent::Fork for the
		837	first time, causing it to fork and exec a new process, which might inherit
		838	the network socket. When the server closes the socket, it is still open
		839	in the child (which doesn't even know that) and the client might conclude
		840	that the connection is still fine.
		841
		842	For the main program, there are multiple remedies available -
		843	L<AnyEvent::Fork::Early> is one, creating a process early and not using
		844	C<new_exec> is another, as in both cases, the first process can be exec'ed
		845	well before many random file descriptors are open.
		846
		847	In general, the solution for these kind of problems is to fix the
		848	libraries or the code that leaks those file descriptors.
		849
		850	Fortunately, most of these leaked descriptors do no harm, other than
		851	sitting on some resources.
		852
		853	=item leaked file descriptors for fork'ed processes
		854
		855	Normally, L<AnyEvent::Fork> does start new processes by exec'ing them,
		856	which closes file descriptors not marked for being inherited.
		857
		858	However, L<AnyEvent::Fork::Early> and L<AnyEvent::Fork::Template> offer
		859	a way to create these processes by forking, and this leaks more file
		860	descriptors than exec'ing them, as there is no way to mark descriptors as
		861	"close on fork".
		862
		863	An example would be modules like L<EV>, L<IO::AIO> or L<Gtk2>. Both create
		864	pipes for internal uses, and L<Gtk2> might open a connection to the X
		865	server. L<EV> and L<IO::AIO> can deal with fork, but Gtk2 might have
		866	trouble with a fork.
		867
		868	The solution is to either not load these modules before use'ing
		869	L<AnyEvent::Fork::Early> or L<AnyEvent::Fork::Template>, or to delay
		870	initialising them, for example, by calling C<init Gtk2> manually.
		871
		872	=item exiting calls object destructors
		873
		874	This only applies to users of L<AnyEvent::Fork:Early> and
		875	L<AnyEvent::Fork::Template>, or when initialiasing code creates objects
		876	that reference external resources.
		877
		878	When a process created by AnyEvent::Fork exits, it might do so by calling
		879	exit, or simply letting perl reach the end of the program. At which point
		880	Perl runs all destructors.
		881
		882	Not all destructors are fork-safe - for example, an object that represents
		883	the connection to an X display might tell the X server to free resources,
		884	which is inconvenient when the "real" object in the parent still needs to
		885	use them.
		886
		887	This is obviously not a problem for L<AnyEvent::Fork::Early>, as you used
		888	it as the very first thing, right?
		889
		890	It is a problem for L<AnyEvent::Fork::Template> though - and the solution
		891	is to not create objects with nontrivial destructors that might have an
		892	effect outside of Perl.
		893
		894	=back
		895
602	=head1 PORTABILITY NOTES	896	=head1 PORTABILITY NOTES
603		897
604	Win32 is a loser - code has been written for this platform, pain has been	898	Native win32 perls are somewhat supported (AnyEvent::Fork::Early is a nop,
605	felt, but in the end, this platform is just too broken - maybe a later	899	and ::Template is not going to work), and it cost a lot of blood and sweat
606	version can do it.	900	to make it so, mostly due to the bloody broken perl that nobody seems to
		901	care about. The fork emulation is a bad joke - I have yet to see something
		902	useful that you can do with it without running into memory corruption
		903	issues or other braindamage. Hrrrr.
		904
		905	Cygwin perl is not supported at the moment due to some hilarious
		906	shortcomings of its API - see L<IO::FDPoll> for more details.
		907
		908	=head1 SEE ALSO
		909
		910	L<AnyEvent::Fork::Early> (to avoid executing a perl interpreter),
		911	L<AnyEvent::Fork::Template> (to create a process by forking the main
		912	program at a convenient time).
607		913
608	=head1 AUTHOR	914	=head1 AUTHOR
609		915
610	Marc Lehmann <schmorp@schmorp.de>	916	Marc Lehmann <schmorp@schmorp.de>
611	http://home.schmorp.de/	917	http://home.schmorp.de/

Diff Legend

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

Comparing AnyEvent-Fork/Fork.pm (file contents): Revision 1.9 by root, Thu Apr 4 03:45:12 2013 UTC vs. Revision 1.39 by root, Sat Apr 6 22:39:37 2013 UTC

Diff Legend

Comparing AnyEvent-Fork/Fork.pm (file contents):
Revision 1.9 by root, Thu Apr 4 03:45:12 2013 UTC vs.
Revision 1.39 by root, Sat Apr 6 22:39:37 2013 UTC