ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/IO-AIO/AIO.pm

Comparing IO-AIO/AIO.pm (file contents):
Revision 1.34 by root, Thu Aug 18 16:32:10 2005 UTC vs.
Revision 1.120 by root, Sun Dec 2 21:51:36 2007 UTC

…		…
5	=head1 SYNOPSIS	5	=head1 SYNOPSIS
6		6
7	use IO::AIO;	7	use IO::AIO;
8		8
9	aio_open "/etc/passwd", O_RDONLY, 0, sub {	9	aio_open "/etc/passwd", O_RDONLY, 0, sub {
10	my ($fh) = @_;	10	my $fh = shift
		11	or die "/etc/passwd: $!";
11	...	12	...
12	};	13	};
13		14
14	aio_unlink "/tmp/file", sub { };	15	aio_unlink "/tmp/file", sub { };
15		16
16	aio_read $fh, 30000, 1024, $buffer, 0, sub {	17	aio_read $fh, 30000, 1024, $buffer, 0, sub {
17	$_[0] > 0 or die "read error: $!";	18	$_[0] > 0 or die "read error: $!";
18	};	19	};
19		20
20	# Event	21	# version 2+ has request and group objects
		22	use IO::AIO 2;
		23
		24	aioreq_pri 4; # give next request a very high priority
		25	my $req = aio_unlink "/tmp/file", sub { };
		26	$req->cancel; # cancel request if still in queue
		27
		28	my $grp = aio_group sub { print "all stats done\n" };
		29	add $grp aio_stat "..." for ...;
		30
		31	# AnyEvent integration (EV, Event, Glib, Tk, urxvt, pureperl...)
		32	open my $fh, "<&=" . IO::AIO::poll_fileno or die "$!";
		33	my $w = AnyEvent->io (fh => $fh, poll => 'r', cb => sub { IO::AIO::poll_cb });
		34
		35	# EV integration
		36	my $w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
		37
		38	# Event integration
21	Event->io (fd => IO::AIO::poll_fileno,	39	Event->io (fd => IO::AIO::poll_fileno,
22	poll => 'r',	40	poll => 'r',
23	cb => \&IO::AIO::poll_cb);	41	cb => \&IO::AIO::poll_cb);
24		42
25	# Glib/Gtk2	43	# Glib/Gtk2 integration
26	add_watch Glib::IO IO::AIO::poll_fileno,	44	add_watch Glib::IO IO::AIO::poll_fileno,
27	in => sub { IO::AIO::poll_cb; 1 };	45	in => sub { IO::AIO::poll_cb; 1 };
28		46
29	# Tk	47	# Tk integration
30	Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",	48	Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",
31	readable => \&IO::AIO::poll_cb);	49	readable => \&IO::AIO::poll_cb);
32		50
33	# Danga::Socket	51	# Danga::Socket integration
34	Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>	52	Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
35	\&IO::AIO::poll_cb);	53	\&IO::AIO::poll_cb);
36		54
37
38	=head1 DESCRIPTION	55	=head1 DESCRIPTION
39		56
40	This module implements asynchronous I/O using whatever means your	57	This module implements asynchronous I/O using whatever means your
41	operating system supports.	58	operating system supports.
42		59
		60	Asynchronous means that operations that can normally block your program
		61	(e.g. reading from disk) will be done asynchronously: the operation
		62	will still block, but you can do something else in the meantime. This
		63	is extremely useful for programs that need to stay interactive even
		64	when doing heavy I/O (GUI programs, high performance network servers
		65	etc.), but can also be used to easily do operations in parallel that are
		66	normally done sequentially, e.g. stat'ing many files, which is much faster
		67	on a RAID volume or over NFS when you do a number of stat operations
		68	concurrently.
		69
		70	While most of this works on all types of file descriptors (for
		71	example sockets), using these functions on file descriptors that
		72	support nonblocking operation (again, sockets, pipes etc.) is very
		73	inefficient. Use an event loop for that (such as the L<Event|Event>
		74	module): IO::AIO will naturally fit into such an event loop itself.
		75
43	Currently, a number of threads are started that execute your read/writes	76	In this version, a number of threads are started that execute your
44	and signal their completion. You don't need thread support in your libc or	77	requests and signal their completion. You don't need thread support
45	perl, and the threads created by this module will not be visible to the	78	in perl, and the threads created by this module will not be visible
46	pthreads library. In the future, this module might make use of the native	79	to perl. In the future, this module might make use of the native aio
47	aio functions available on many operating systems. However, they are often	80	functions available on many operating systems. However, they are often
48	not well-supported (Linux doesn't allow them on normal files currently,	81	not well-supported or restricted (GNU/Linux doesn't allow them on normal
49	for example), and they would only support aio_read and aio_write, so the	82	files currently, for example), and they would only support aio_read and
50	remaining functionality would have to be implemented using threads anyway.	83	aio_write, so the remaining functionality would have to be implemented
		84	using threads anyway.
51		85
52	Although the module will work with in the presence of other threads, it is	86	Although the module will work in the presence of other (Perl-) threads,
53	currently not reentrant, so use appropriate locking yourself, always call	87	it is currently not reentrant in any way, so use appropriate locking
54	C<poll_cb> from within the same thread, or never call C<poll_cb> (or other	88	yourself, always call C<poll_cb> from within the same thread, or never
55	C<aio_> functions) recursively.	89	call C<poll_cb> (or other C<aio_> functions) recursively.
		90
		91	=head2 EXAMPLE
		92
		93	This is a simple example that uses the Event module and loads
		94	F</etc/passwd> asynchronously:
		95
		96	use Fcntl;
		97	use Event;
		98	use IO::AIO;
		99
		100	# register the IO::AIO callback with Event
		101	Event->io (fd => IO::AIO::poll_fileno,
		102	poll => 'r',
		103	cb => \&IO::AIO::poll_cb);
		104
		105	# queue the request to open /etc/passwd
		106	aio_open "/etc/passwd", O_RDONLY, 0, sub {
		107	my $fh = shift
		108	or die "error while opening: $!";
		109
		110	# stat'ing filehandles is generally non-blocking
		111	my $size = -s $fh;
		112
		113	# queue a request to read the file
		114	my $contents;
		115	aio_read $fh, 0, $size, $contents, 0, sub {
		116	$_[0] == $size
		117	or die "short read: $!";
		118
		119	close $fh;
		120
		121	# file contents now in $contents
		122	print $contents;
		123
		124	# exit event loop and program
		125	Event::unloop;
		126	};
		127	};
		128
		129	# possibly queue up other requests, or open GUI windows,
		130	# check for sockets etc. etc.
		131
		132	# process events as long as there are some:
		133	Event::loop;
		134
		135	=head1 REQUEST ANATOMY AND LIFETIME
		136
		137	Every C<aio_*> function creates a request. which is a C data structure not
		138	directly visible to Perl.
		139
		140	If called in non-void context, every request function returns a Perl
		141	object representing the request. In void context, nothing is returned,
		142	which saves a bit of memory.
		143
		144	The perl object is a fairly standard ref-to-hash object. The hash contents
		145	are not used by IO::AIO so you are free to store anything you like in it.
		146
		147	During their existance, aio requests travel through the following states,
		148	in order:
		149
		150	=over 4
		151
		152	=item ready
		153
		154	Immediately after a request is created it is put into the ready state,
		155	waiting for a thread to execute it.
		156
		157	=item execute
		158
		159	A thread has accepted the request for processing and is currently
		160	executing it (e.g. blocking in read).
		161
		162	=item pending
		163
		164	The request has been executed and is waiting for result processing.
		165
		166	While request submission and execution is fully asynchronous, result
		167	processing is not and relies on the perl interpreter calling C<poll_cb>
		168	(or another function with the same effect).
		169
		170	=item result
		171
		172	The request results are processed synchronously by C<poll_cb>.
		173
		174	The C<poll_cb> function will process all outstanding aio requests by
		175	calling their callbacks, freeing memory associated with them and managing
		176	any groups they are contained in.
		177
		178	=item done
		179
		180	Request has reached the end of its lifetime and holds no resources anymore
		181	(except possibly for the Perl object, but its connection to the actual
		182	aio request is severed and calling its methods will either do nothing or
		183	result in a runtime error).
		184
		185	=back
56		186
57	=cut	187	=cut
58		188
59	package IO::AIO;	189	package IO::AIO;
60		190
		191	use Carp ();
		192
61	no warnings;	193	no warnings;
		194	use strict 'vars';
62		195
63	use base 'Exporter';	196	use base 'Exporter';
64		197
65	use Fcntl ();
66
67	BEGIN {	198	BEGIN {
68	$VERSION = 1.5;	199	our $VERSION = '2.6';
69		200
70	@EXPORT = qw(aio_read aio_write aio_open aio_close aio_stat aio_lstat aio_unlink	201	our @AIO_REQ = qw(aio_sendfile aio_read aio_write aio_open aio_close
71	aio_rmdir aio_symlink aio_fsync aio_fdatasync aio_readahead);	202	aio_stat aio_lstat aio_unlink aio_rmdir aio_readdir
72	@EXPORT_OK = qw(poll_fileno poll_cb min_parallel max_parallel max_outstanding nreqs);	203	aio_scandir aio_symlink aio_readlink aio_sync aio_fsync
		204	aio_fdatasync aio_pathsync aio_readahead
		205	aio_rename aio_link aio_move aio_copy aio_group
		206	aio_nop aio_mknod aio_load aio_rmtree aio_mkdir aio_chown
		207	aio_chmod aio_utime aio_truncate);
		208
		209	our @EXPORT = (@AIO_REQ, qw(aioreq_pri aioreq_nice aio_block));
		210	our @EXPORT_OK = qw(poll_fileno poll_cb poll_wait flush
		211	min_parallel max_parallel max_idle
		212	nreqs nready npending nthreads
		213	max_poll_time max_poll_reqs);
		214
		215	@IO::AIO::GRP::ISA = 'IO::AIO::REQ';
73		216
74	require XSLoader;	217	require XSLoader;
75	XSLoader::load IO::AIO, $VERSION;	218	XSLoader::load ("IO::AIO", $VERSION);
76	}	219	}
77		220
78	=head1 FUNCTIONS	221	=head1 FUNCTIONS
79		222
80	=head2 AIO FUNCTIONS	223	=head2 AIO REQUEST FUNCTIONS
81		224
82	All the C<aio_*> calls are more or less thin wrappers around the syscall	225	All the C<aio_*> calls are more or less thin wrappers around the syscall
83	with the same name (sans C<aio_>). The arguments are similar or identical,	226	with the same name (sans C<aio_>). The arguments are similar or identical,
84	and they all accept an additional (and optional) C<$callback> argument	227	and they all accept an additional (and optional) C<$callback> argument
85	which must be a code reference. This code reference will get called with	228	which must be a code reference. This code reference will get called with
…		…
88	syscall has been executed asynchronously.	231	syscall has been executed asynchronously.
89		232
90	All functions expecting a filehandle keep a copy of the filehandle	233	All functions expecting a filehandle keep a copy of the filehandle
91	internally until the request has finished.	234	internally until the request has finished.
92		235
		236	All functions return request objects of type L<IO::AIO::REQ> that allow
		237	further manipulation of those requests while they are in-flight.
		238
93	The pathnames you pass to these routines I<must> be absolute and	239	The pathnames you pass to these routines I<must> be absolute and
94	encoded in byte form. The reason for the former is that at the time the	240	encoded as octets. The reason for the former is that at the time the
95	request is being executed, the current working directory could have	241	request is being executed, the current working directory could have
96	changed. Alternatively, you can make sure that you never change the	242	changed. Alternatively, you can make sure that you never change the
97	current working directory.	243	current working directory anywhere in the program and then use relative
		244	paths.
98		245
99	To encode pathnames to byte form, either make sure you either: a)	246	To encode pathnames as octets, either make sure you either: a) always pass
100	always pass in filenames you got from outside (command line, readdir	247	in filenames you got from outside (command line, readdir etc.) without
101	etc.), b) are ASCII or ISO 8859-1, c) use the Encode module and encode	248	tinkering, b) are ASCII or ISO 8859-1, c) use the Encode module and encode
102	your pathnames to the locale (or other) encoding in effect in the user	249	your pathnames to the locale (or other) encoding in effect in the user
103	environment, d) use Glib::filename_from_unicode on unicode filenames or e)	250	environment, d) use Glib::filename_from_unicode on unicode filenames or e)
104	use something else.	251	use something else to ensure your scalar has the correct contents.
		252
		253	This works, btw. independent of the internal UTF-8 bit, which IO::AIO
		254	handles correctly wether it is set or not.
105		255
106	=over 4	256	=over 4
107		257
		258	=item $prev_pri = aioreq_pri [$pri]
		259
		260	Returns the priority value that would be used for the next request and, if
		261	C<$pri> is given, sets the priority for the next aio request.
		262
		263	The default priority is C<0>, the minimum and maximum priorities are C<-4>
		264	and C<4>, respectively. Requests with higher priority will be serviced
		265	first.
		266
		267	The priority will be reset to C<0> after each call to one of the C<aio_*>
		268	functions.
		269
		270	Example: open a file with low priority, then read something from it with
		271	higher priority so the read request is serviced before other low priority
		272	open requests (potentially spamming the cache):
		273
		274	aioreq_pri -3;
		275	aio_open ..., sub {
		276	return unless $_[0];
		277
		278	aioreq_pri -2;
		279	aio_read $_[0], ..., sub {
		280	...
		281	};
		282	};
		283
		284
		285	=item aioreq_nice $pri_adjust
		286
		287	Similar to C<aioreq_pri>, but subtracts the given value from the current
		288	priority, so the effect is cumulative.
		289
		290
108	=item aio_open $pathname, $flags, $mode, $callback	291	=item aio_open $pathname, $flags, $mode, $callback->($fh)
109		292
110	Asynchronously open or create a file and call the callback with a newly	293	Asynchronously open or create a file and call the callback with a newly
111	created filehandle for the file.	294	created filehandle for the file.
112		295
113	The pathname passed to C<aio_open> must be absolute. See API NOTES, above,	296	The pathname passed to C<aio_open> must be absolute. See API NOTES, above,
…		…
117	list. They are the same as used by C<sysopen>.	300	list. They are the same as used by C<sysopen>.
118		301
119	Likewise, C<$mode> specifies the mode of the newly created file, if it	302	Likewise, C<$mode> specifies the mode of the newly created file, if it
120	didn't exist and C<O_CREAT> has been given, just like perl's C<sysopen>,	303	didn't exist and C<O_CREAT> has been given, just like perl's C<sysopen>,
121	except that it is mandatory (i.e. use C<0> if you don't create new files,	304	except that it is mandatory (i.e. use C<0> if you don't create new files,
122	and C<0666> or C<0777> if you do).	305	and C<0666> or C<0777> if you do). Note that the C<$mode> will be modified
		306	by the umask in effect then the request is being executed, so better never
		307	change the umask.
123		308
124	Example:	309	Example:
125		310
126	aio_open "/etc/passwd", O_RDONLY, 0, sub {	311	aio_open "/etc/passwd", O_RDONLY, 0, sub {
127	if ($_[0]) {	312	if ($_[0]) {
…		…
130	} else {	315	} else {
131	die "open failed: $!\n";	316	die "open failed: $!\n";
132	}	317	}
133	};	318	};
134		319
		320
135	=item aio_close $fh, $callback	321	=item aio_close $fh, $callback->($status)
136		322
137	Asynchronously close a file and call the callback with the result	323	Asynchronously close a file and call the callback with the result
138	code. I<WARNING:> although accepted, you should not pass in a perl	324	code.
139	filehandle here, as perl will likely close the file descriptor another
140	time when the filehandle is destroyed. Normally, you can safely call perls
141	C<close> or just let filehandles go out of scope.
142		325
143	This is supposed to be a bug in the API, so that might change. It's	326	Unfortunately, you can't do this to perl. Perl I<insists> very strongly on
144	therefore best to avoid this function.	327	closing the file descriptor associated with the filehandle itself. Here is
		328	what aio_close will try:
145		329
		330	1. dup()licate the fd
		331	2. asynchronously close() the duplicated fd
		332	3. dup()licate the fd once more
		333	4. let perl close() the filehandle
		334	5. asynchronously close the duplicated fd
		335
		336	The idea is that the first close() flushes stuff to disk that closing an
		337	fd will flush, so when perl closes the fd, nothing much will need to be
		338	flushed. The second async. close() will then flush stuff to disk that
		339	closing the last fd to the file will flush.
		340
		341	Just FYI, SuSv3 has this to say on close:
		342
		343	All outstanding record locks owned by the process on the file
		344	associated with the file descriptor shall be removed.
		345
		346	If fildes refers to a socket, close() shall cause the socket to be
		347	destroyed. ... close() shall block for up to the current linger
		348	interval until all data is transmitted.
		349	[this actually sounds like a specification bug, but who knows]
		350
		351	And at least Linux additionally actually flushes stuff on every close,
		352	even when the file itself is still open.
		353
		354	Sounds enourmously inefficient and complicated? Yes... please show me how
		355	to nuke perl's fd out of existence...
		356
		357	=cut
		358
		359	sub aio_close($;$) {
		360	aio_block {
		361	my ($fh, $cb) = @_;
		362
		363	my $pri = aioreq_pri;
		364	my $grp = aio_group $cb;
		365
		366	my $fd = fileno $fh;
		367
		368	defined $fd or Carp::croak "aio_close called with fd-less filehandle";
		369
		370	# if the dups fail we will simply get EBADF
		371	my $fd2 = _dup $fd;
		372	aioreq_pri $pri;
		373	add $grp _aio_close $fd2, sub {
		374	my $fd2 = _dup $fd;
		375	close $fh;
		376	aioreq_pri $pri;
		377	add $grp _aio_close $fd2, sub {
		378	$grp->result ($_[0]);
		379	};
		380	};
		381
		382	$grp
		383	}
		384	}
		385
		386
146	=item aio_read $fh,$offset,$length, $data,$dataoffset,$callback	387	=item aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
147		388
148	=item aio_write $fh,$offset,$length, $data,$dataoffset,$callback	389	=item aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
149		390
150	Reads or writes C<length> bytes from the specified C<fh> and C<offset>	391	Reads or writes C<$length> bytes from the specified C<$fh> and C<$offset>
151	into the scalar given by C<data> and offset C<dataoffset> and calls the	392	into the scalar given by C<$data> and offset C<$dataoffset> and calls the
152	callback without the actual number of bytes read (or -1 on error, just	393	callback without the actual number of bytes read (or -1 on error, just
153	like the syscall).	394	like the syscall).
154		395
		396	If C<$offset> is undefined, then the current file descriptor offset will
		397	be used (and updated), otherwise the file descriptor offset will not be
		398	changed by these calls.
		399
		400	If C<$length> is undefined in C<aio_write>, use the remaining length of C<$data>.
		401
		402	If C<$dataoffset> is less than zero, it will be counted from the end of
		403	C<$data>.
		404
155	The C<$data> scalar I<MUST NOT> be modified in any way while the request	405	The C<$data> scalar I<MUST NOT> be modified in any way while the request
156	is outstanding. Modifying it can result in segfaults or WW3 (if the	406	is outstanding. Modifying it can result in segfaults or World War III (if
157	necessary/optional hardware is installed).	407	the necessary/optional hardware is installed).
158		408
159	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at	409	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at
160	offset C<0> within the scalar:	410	offset C<0> within the scalar:
161		411
162	aio_read $fh, 7, 15, $buffer, 0, sub {	412	aio_read $fh, 7, 15, $buffer, 0, sub {
163	$_[0] > 0 or die "read error: $!";	413	$_[0] > 0 or die "read error: $!";
164	print "read $_[0] bytes: <$buffer>\n";	414	print "read $_[0] bytes: <$buffer>\n";
165	};	415	};
166		416
		417
		418	=item aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
		419
		420	Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts
		421	reading at byte offset C<$in_offset>, and starts writing at the current
		422	file offset of C<$out_fh>. Because of that, it is not safe to issue more
		423	than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each
		424	other.
		425
		426	This call tries to make use of a native C<sendfile> syscall to provide
		427	zero-copy operation. For this to work, C<$out_fh> should refer to a
		428	socket, and C<$in_fh> should refer to mmap'able file.
		429
		430	If the native sendfile call fails or is not implemented, it will be
		431	emulated, so you can call C<aio_sendfile> on any type of filehandle
		432	regardless of the limitations of the operating system.
		433
		434	Please note, however, that C<aio_sendfile> can read more bytes from
		435	C<$in_fh> than are written, and there is no way to find out how many
		436	bytes have been read from C<aio_sendfile> alone, as C<aio_sendfile> only
		437	provides the number of bytes written to C<$out_fh>. Only if the result
		438	value equals C<$length> one can assume that C<$length> bytes have been
		439	read.
		440
		441
167	=item aio_readahead $fh,$offset,$length, $callback	442	=item aio_readahead $fh,$offset,$length, $callback->($retval)
168		443
169	C<aio_readahead> populates the page cache with data from a file so that	444	C<aio_readahead> populates the page cache with data from a file so that
170	subsequent reads from that file will not block on disk I/O. The C<$offset>	445	subsequent reads from that file will not block on disk I/O. The C<$offset>
171	argument specifies the starting point from which data is to be read and	446	argument specifies the starting point from which data is to be read and
172	C<$length> specifies the number of bytes to be read. I/O is performed in	447	C<$length> specifies the number of bytes to be read. I/O is performed in
…		…
176	file. The current file offset of the file is left unchanged.	451	file. The current file offset of the file is left unchanged.
177		452
178	If that syscall doesn't exist (likely if your OS isn't Linux) it will be	453	If that syscall doesn't exist (likely if your OS isn't Linux) it will be
179	emulated by simply reading the data, which would have a similar effect.	454	emulated by simply reading the data, which would have a similar effect.
180		455
		456
181	=item aio_stat $fh_or_path, $callback	457	=item aio_stat $fh_or_path, $callback->($status)
182		458
183	=item aio_lstat $fh, $callback	459	=item aio_lstat $fh, $callback->($status)
184		460
185	Works like perl's C<stat> or C<lstat> in void context. The callback will	461	Works like perl's C<stat> or C<lstat> in void context. The callback will
186	be called after the stat and the results will be available using C<stat _>	462	be called after the stat and the results will be available using C<stat _>
187	or C<-s _> etc...	463	or C<-s _> etc...
188		464
…		…
198	aio_stat "/etc/passwd", sub {	474	aio_stat "/etc/passwd", sub {
199	$_[0] and die "stat failed: $!";	475	$_[0] and die "stat failed: $!";
200	print "size is ", -s _, "\n";	476	print "size is ", -s _, "\n";
201	};	477	};
202		478
		479
		480	=item aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
		481
		482	Works like perl's C<utime> function (including the special case of $atime
		483	and $mtime being undef). Fractional times are supported if the underlying
		484	syscalls support them.
		485
		486	When called with a pathname, uses utimes(2) if available, otherwise
		487	utime(2). If called on a file descriptor, uses futimes(2) if available,
		488	otherwise returns ENOSYS, so this is not portable.
		489
		490	Examples:
		491
		492	# set atime and mtime to current time (basically touch(1)):
		493	aio_utime "path", undef, undef;
		494	# set atime to current time and mtime to beginning of the epoch:
		495	aio_utime "path", time, undef; # undef==0
		496
		497
		498	=item aio_chown $fh_or_path, $uid, $gid, $callback->($status)
		499
		500	Works like perl's C<chown> function, except that C<undef> for either $uid
		501	or $gid is being interpreted as "do not change" (but -1 can also be used).
		502
		503	Examples:
		504
		505	# same as "chown root path" in the shell:
		506	aio_chown "path", 0, -1;
		507	# same as above:
		508	aio_chown "path", 0, undef;
		509
		510
		511	=item aio_truncate $fh_or_path, $offset, $callback->($status)
		512
		513	Works like truncate(2) or ftruncate(2).
		514
		515
		516	=item aio_chmod $fh_or_path, $mode, $callback->($status)
		517
		518	Works like perl's C<chmod> function.
		519
		520
203	=item aio_unlink $pathname, $callback	521	=item aio_unlink $pathname, $callback->($status)
204		522
205	Asynchronously unlink (delete) a file and call the callback with the	523	Asynchronously unlink (delete) a file and call the callback with the
206	result code.	524	result code.
207		525
		526
		527	=item aio_mknod $path, $mode, $dev, $callback->($status)
		528
		529	[EXPERIMENTAL]
		530
		531	Asynchronously create a device node (or fifo). See mknod(2).
		532
		533	The only (POSIX-) portable way of calling this function is:
		534
		535	aio_mknod $path, IO::AIO::S_IFIFO | $mode, 0, sub { ...
		536
		537
		538	=item aio_link $srcpath, $dstpath, $callback->($status)
		539
		540	Asynchronously create a new link to the existing object at C<$srcpath> at
		541	the path C<$dstpath> and call the callback with the result code.
		542
		543
		544	=item aio_symlink $srcpath, $dstpath, $callback->($status)
		545
		546	Asynchronously create a new symbolic link to the existing object at C<$srcpath> at
		547	the path C<$dstpath> and call the callback with the result code.
		548
		549
		550	=item aio_readlink $path, $callback->($link)
		551
		552	Asynchronously read the symlink specified by C<$path> and pass it to
		553	the callback. If an error occurs, nothing or undef gets passed to the
		554	callback.
		555
		556
		557	=item aio_rename $srcpath, $dstpath, $callback->($status)
		558
		559	Asynchronously rename the object at C<$srcpath> to C<$dstpath>, just as
		560	rename(2) and call the callback with the result code.
		561
		562
		563	=item aio_mkdir $pathname, $mode, $callback->($status)
		564
		565	Asynchronously mkdir (create) a directory and call the callback with
		566	the result code. C<$mode> will be modified by the umask at the time the
		567	request is executed, so do not change your umask.
		568
		569
208	=item aio_rmdir $pathname, $callback	570	=item aio_rmdir $pathname, $callback->($status)
209		571
210	Asynchronously rmdir (delete) a directory and call the callback with the	572	Asynchronously rmdir (delete) a directory and call the callback with the
211	result code.	573	result code.
212		574
		575
		576	=item aio_readdir $pathname, $callback->($entries)
		577
		578	Unlike the POSIX call of the same name, C<aio_readdir> reads an entire
		579	directory (i.e. opendir + readdir + closedir). The entries will not be
		580	sorted, and will B<NOT> include the C<.> and C<..> entries.
		581
		582	The callback a single argument which is either C<undef> or an array-ref
		583	with the filenames.
		584
		585
		586	=item aio_load $path, $data, $callback->($status)
		587
		588	This is a composite request that tries to fully load the given file into
		589	memory. Status is the same as with aio_read.
		590
		591	=cut
		592
		593	sub aio_load($$;$) {
		594	aio_block {
		595	my ($path, undef, $cb) = @_;
		596	my $data = \$_[1];
		597
		598	my $pri = aioreq_pri;
		599	my $grp = aio_group $cb;
		600
		601	aioreq_pri $pri;
		602	add $grp aio_open $path, O_RDONLY, 0, sub {
		603	my $fh = shift
		604	or return $grp->result (-1);
		605
		606	aioreq_pri $pri;
		607	add $grp aio_read $fh, 0, (-s $fh), $$data, 0, sub {
		608	$grp->result ($_[0]);
		609	};
		610	};
		611
		612	$grp
		613	}
		614	}
		615
		616	=item aio_copy $srcpath, $dstpath, $callback->($status)
		617
		618	Try to copy the I<file> (directories not supported as either source or
		619	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		620	the C<0> (error) or C<-1> ok.
		621
		622	This is a composite request that it creates the destination file with
		623	mode 0200 and copies the contents of the source file into it using
		624	C<aio_sendfile>, followed by restoring atime, mtime, access mode and
		625	uid/gid, in that order.
		626
		627	If an error occurs, the partial destination file will be unlinked, if
		628	possible, except when setting atime, mtime, access mode and uid/gid, where
		629	errors are being ignored.
		630
		631	=cut
		632
		633	sub aio_copy($$;$) {
		634	aio_block {
		635	my ($src, $dst, $cb) = @_;
		636
		637	my $pri = aioreq_pri;
		638	my $grp = aio_group $cb;
		639
		640	aioreq_pri $pri;
		641	add $grp aio_open $src, O_RDONLY, 0, sub {
		642	if (my $src_fh = $_[0]) {
		643	my @stat = stat $src_fh;
		644
		645	aioreq_pri $pri;
		646	add $grp aio_open $dst, O_CREAT | O_WRONLY | O_TRUNC, 0200, sub {
		647	if (my $dst_fh = $_[0]) {
		648	aioreq_pri $pri;
		649	add $grp aio_sendfile $dst_fh, $src_fh, 0, $stat[7], sub {
		650	if ($_[0] == $stat[7]) {
		651	$grp->result (0);
		652	close $src_fh;
		653
		654	# those should not normally block. should. should.
		655	utime $stat[8], $stat[9], $dst;
		656	chmod $stat[2] & 07777, $dst_fh;
		657	chown $stat[4], $stat[5], $dst_fh;
		658
		659	aioreq_pri $pri;
		660	add $grp aio_close $dst_fh;
		661	} else {
		662	$grp->result (-1);
		663	close $src_fh;
		664	close $dst_fh;
		665
		666	aioreq $pri;
		667	add $grp aio_unlink $dst;
		668	}
		669	};
		670	} else {
		671	$grp->result (-1);
		672	}
		673	},
		674
		675	} else {
		676	$grp->result (-1);
		677	}
		678	};
		679
		680	$grp
		681	}
		682	}
		683
		684	=item aio_move $srcpath, $dstpath, $callback->($status)
		685
		686	Try to move the I<file> (directories not supported as either source or
		687	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		688	the C<0> (error) or C<-1> ok.
		689
		690	This is a composite request that tries to rename(2) the file first. If
		691	rename files with C<EXDEV>, it copies the file with C<aio_copy> and, if
		692	that is successful, unlinking the C<$srcpath>.
		693
		694	=cut
		695
		696	sub aio_move($$;$) {
		697	aio_block {
		698	my ($src, $dst, $cb) = @_;
		699
		700	my $pri = aioreq_pri;
		701	my $grp = aio_group $cb;
		702
		703	aioreq_pri $pri;
		704	add $grp aio_rename $src, $dst, sub {
		705	if ($_[0] && $! == EXDEV) {
		706	aioreq_pri $pri;
		707	add $grp aio_copy $src, $dst, sub {
		708	$grp->result ($_[0]);
		709
		710	if (!$_[0]) {
		711	aioreq_pri $pri;
		712	add $grp aio_unlink $src;
		713	}
		714	};
		715	} else {
		716	$grp->result ($_[0]);
		717	}
		718	};
		719
		720	$grp
		721	}
		722	}
		723
		724	=item aio_scandir $path, $maxreq, $callback->($dirs, $nondirs)
		725
		726	Scans a directory (similar to C<aio_readdir>) but additionally tries to
		727	efficiently separate the entries of directory C<$path> into two sets of
		728	names, directories you can recurse into (directories), and ones you cannot
		729	recurse into (everything else, including symlinks to directories).
		730
		731	C<aio_scandir> is a composite request that creates of many sub requests_
		732	C<$maxreq> specifies the maximum number of outstanding aio requests that
		733	this function generates. If it is C<< <= 0 >>, then a suitable default
		734	will be chosen (currently 4).
		735
		736	On error, the callback is called without arguments, otherwise it receives
		737	two array-refs with path-relative entry names.
		738
		739	Example:
		740
		741	aio_scandir $dir, 0, sub {
		742	my ($dirs, $nondirs) = @_;
		743	print "real directories: @$dirs\n";
		744	print "everything else: @$nondirs\n";
		745	};
		746
		747	Implementation notes.
		748
		749	The C<aio_readdir> cannot be avoided, but C<stat()>'ing every entry can.
		750
		751	After reading the directory, the modification time, size etc. of the
		752	directory before and after the readdir is checked, and if they match (and
		753	isn't the current time), the link count will be used to decide how many
		754	entries are directories (if >= 2). Otherwise, no knowledge of the number
		755	of subdirectories will be assumed.
		756
		757	Then entries will be sorted into likely directories (everything without
		758	a non-initial dot currently) and likely non-directories (everything
		759	else). Then every entry plus an appended C</.> will be C<stat>'ed,
		760	likely directories first. If that succeeds, it assumes that the entry
		761	is a directory or a symlink to directory (which will be checked
		762	seperately). This is often faster than stat'ing the entry itself because
		763	filesystems might detect the type of the entry without reading the inode
		764	data (e.g. ext2fs filetype feature).
		765
		766	If the known number of directories (link count - 2) has been reached, the
		767	rest of the entries is assumed to be non-directories.
		768
		769	This only works with certainty on POSIX (= UNIX) filesystems, which
		770	fortunately are the vast majority of filesystems around.
		771
		772	It will also likely work on non-POSIX filesystems with reduced efficiency
		773	as those tend to return 0 or 1 as link counts, which disables the
		774	directory counting heuristic.
		775
		776	=cut
		777
		778	sub aio_scandir($$;$) {
		779	aio_block {
		780	my ($path, $maxreq, $cb) = @_;
		781
		782	my $pri = aioreq_pri;
		783
		784	my $grp = aio_group $cb;
		785
		786	$maxreq = 4 if $maxreq <= 0;
		787
		788	# stat once
		789	aioreq_pri $pri;
		790	add $grp aio_stat $path, sub {
		791	return $grp->result () if $_[0];
		792	my $now = time;
		793	my $hash1 = join ":", (stat _)[0,1,3,7,9];
		794
		795	# read the directory entries
		796	aioreq_pri $pri;
		797	add $grp aio_readdir $path, sub {
		798	my $entries = shift
		799	or return $grp->result ();
		800
		801	# stat the dir another time
		802	aioreq_pri $pri;
		803	add $grp aio_stat $path, sub {
		804	my $hash2 = join ":", (stat _)[0,1,3,7,9];
		805
		806	my $ndirs;
		807
		808	# take the slow route if anything looks fishy
		809	if ($hash1 ne $hash2 or (stat _)[9] == $now) {
		810	$ndirs = -1;
		811	} else {
		812	# if nlink == 2, we are finished
		813	# on non-posix-fs's, we rely on nlink < 2
		814	$ndirs = (stat _)[3] - 2
		815	or return $grp->result ([], $entries);
		816	}
		817
		818	# sort into likely dirs and likely nondirs
		819	# dirs == files without ".", short entries first
		820	$entries = [map $_->[0],
		821	sort { $b->[1] cmp $a->[1] }
		822	map [$_, sprintf "%s%04d", (/.\./ ? "1" : "0"), length],
		823	@$entries];
		824
		825	my (@dirs, @nondirs);
		826
		827	my $statgrp = add $grp aio_group sub {
		828	$grp->result (\@dirs, \@nondirs);
		829	};
		830
		831	limit $statgrp $maxreq;
		832	feed $statgrp sub {
		833	return unless @$entries;
		834	my $entry = pop @$entries;
		835
		836	aioreq_pri $pri;
		837	add $statgrp aio_stat "$path/$entry/.", sub {
		838	if ($_[0] < 0) {
		839	push @nondirs, $entry;
		840	} else {
		841	# need to check for real directory
		842	aioreq_pri $pri;
		843	add $statgrp aio_lstat "$path/$entry", sub {
		844	if (-d _) {
		845	push @dirs, $entry;
		846
		847	unless (--$ndirs) {
		848	push @nondirs, @$entries;
		849	feed $statgrp;
		850	}
		851	} else {
		852	push @nondirs, $entry;
		853	}
		854	}
		855	}
		856	};
		857	};
		858	};
		859	};
		860	};
		861
		862	$grp
		863	}
		864	}
		865
		866	=item aio_rmtree $path, $callback->($status)
		867
		868	Delete a directory tree starting (and including) C<$path>, return the
		869	status of the final C<rmdir> only. This is a composite request that
		870	uses C<aio_scandir> to recurse into and rmdir directories, and unlink
		871	everything else.
		872
		873	=cut
		874
		875	sub aio_rmtree;
		876	sub aio_rmtree($;$) {
		877	aio_block {
		878	my ($path, $cb) = @_;
		879
		880	my $pri = aioreq_pri;
		881	my $grp = aio_group $cb;
		882
		883	aioreq_pri $pri;
		884	add $grp aio_scandir $path, 0, sub {
		885	my ($dirs, $nondirs) = @_;
		886
		887	my $dirgrp = aio_group sub {
		888	add $grp aio_rmdir $path, sub {
		889	$grp->result ($_[0]);
		890	};
		891	};
		892
		893	(aioreq_pri $pri), add $dirgrp aio_rmtree "$path/$_" for @$dirs;
		894	(aioreq_pri $pri), add $dirgrp aio_unlink "$path/$_" for @$nondirs;
		895
		896	add $grp $dirgrp;
		897	};
		898
		899	$grp
		900	}
		901	}
		902
		903	=item aio_sync $callback->($status)
		904
		905	Asynchronously call sync and call the callback when finished.
		906
213	=item aio_fsync $fh, $callback	907	=item aio_fsync $fh, $callback->($status)
214		908
215	Asynchronously call fsync on the given filehandle and call the callback	909	Asynchronously call fsync on the given filehandle and call the callback
216	with the fsync result code.	910	with the fsync result code.
217		911
218	=item aio_fdatasync $fh, $callback	912	=item aio_fdatasync $fh, $callback->($status)
219		913
220	Asynchronously call fdatasync on the given filehandle and call the	914	Asynchronously call fdatasync on the given filehandle and call the
221	callback with the fdatasync result code.	915	callback with the fdatasync result code.
222		916
223	If this call isn't available because your OS lacks it or it couldn't be	917	If this call isn't available because your OS lacks it or it couldn't be
224	detected, it will be emulated by calling C<fsync> instead.	918	detected, it will be emulated by calling C<fsync> instead.
225		919
		920	=item aio_pathsync $path, $callback->($status)
		921
		922	This request tries to open, fsync and close the given path. This is a
		923	composite request intended tosync directories after directory operations
		924	(E.g. rename). This might not work on all operating systems or have any
		925	specific effect, but usually it makes sure that directory changes get
		926	written to disc. It works for anything that can be opened for read-only,
		927	not just directories.
		928
		929	Passes C<0> when everything went ok, and C<-1> on error.
		930
		931	=cut
		932
		933	sub aio_pathsync($;$) {
		934	aio_block {
		935	my ($path, $cb) = @_;
		936
		937	my $pri = aioreq_pri;
		938	my $grp = aio_group $cb;
		939
		940	aioreq_pri $pri;
		941	add $grp aio_open $path, O_RDONLY, 0, sub {
		942	my ($fh) = @_;
		943	if ($fh) {
		944	aioreq_pri $pri;
		945	add $grp aio_fsync $fh, sub {
		946	$grp->result ($_[0]);
		947
		948	aioreq_pri $pri;
		949	add $grp aio_close $fh;
		950	};
		951	} else {
		952	$grp->result (-1);
		953	}
		954	};
		955
		956	$grp
		957	}
		958	}
		959
		960	=item aio_group $callback->(...)
		961
		962	This is a very special aio request: Instead of doing something, it is a
		963	container for other aio requests, which is useful if you want to bundle
		964	many requests into a single, composite, request with a definite callback
		965	and the ability to cancel the whole request with its subrequests.
		966
		967	Returns an object of class L<IO::AIO::GRP>. See its documentation below
		968	for more info.
		969
		970	Example:
		971
		972	my $grp = aio_group sub {
		973	print "all stats done\n";
		974	};
		975
		976	add $grp
		977	(aio_stat ...),
		978	(aio_stat ...),
		979	...;
		980
		981	=item aio_nop $callback->()
		982
		983	This is a special request - it does nothing in itself and is only used for
		984	side effects, such as when you want to add a dummy request to a group so
		985	that finishing the requests in the group depends on executing the given
		986	code.
		987
		988	While this request does nothing, it still goes through the execution
		989	phase and still requires a worker thread. Thus, the callback will not
		990	be executed immediately but only after other requests in the queue have
		991	entered their execution phase. This can be used to measure request
		992	latency.
		993
		994	=item IO::AIO::aio_busy $fractional_seconds, $callback->() *NOT EXPORTED*
		995
		996	Mainly used for debugging and benchmarking, this aio request puts one of
		997	the request workers to sleep for the given time.
		998
		999	While it is theoretically handy to have simple I/O scheduling requests
		1000	like sleep and file handle readable/writable, the overhead this creates is
		1001	immense (it blocks a thread for a long time) so do not use this function
		1002	except to put your application under artificial I/O pressure.
		1003
226	=back	1004	=back
227		1005
		1006	=head2 IO::AIO::REQ CLASS
		1007
		1008	All non-aggregate C<aio_*> functions return an object of this class when
		1009	called in non-void context.
		1010
		1011	=over 4
		1012
		1013	=item cancel $req
		1014
		1015	Cancels the request, if possible. Has the effect of skipping execution
		1016	when entering the B<execute> state and skipping calling the callback when
		1017	entering the the B<result> state, but will leave the request otherwise
		1018	untouched. That means that requests that currently execute will not be
		1019	stopped and resources held by the request will not be freed prematurely.
		1020
		1021	=item cb $req $callback->(...)
		1022
		1023	Replace (or simply set) the callback registered to the request.
		1024
		1025	=back
		1026
		1027	=head2 IO::AIO::GRP CLASS
		1028
		1029	This class is a subclass of L<IO::AIO::REQ>, so all its methods apply to
		1030	objects of this class, too.
		1031
		1032	A IO::AIO::GRP object is a special request that can contain multiple other
		1033	aio requests.
		1034
		1035	You create one by calling the C<aio_group> constructing function with a
		1036	callback that will be called when all contained requests have entered the
		1037	C<done> state:
		1038
		1039	my $grp = aio_group sub {
		1040	print "all requests are done\n";
		1041	};
		1042
		1043	You add requests by calling the C<add> method with one or more
		1044	C<IO::AIO::REQ> objects:
		1045
		1046	$grp->add (aio_unlink "...");
		1047
		1048	add $grp aio_stat "...", sub {
		1049	$_[0] or return $grp->result ("error");
		1050
		1051	# add another request dynamically, if first succeeded
		1052	add $grp aio_open "...", sub {
		1053	$grp->result ("ok");
		1054	};
		1055	};
		1056
		1057	This makes it very easy to create composite requests (see the source of
		1058	C<aio_move> for an application) that work and feel like simple requests.
		1059
		1060	=over 4
		1061
		1062	=item * The IO::AIO::GRP objects will be cleaned up during calls to
		1063	C<IO::AIO::poll_cb>, just like any other request.
		1064
		1065	=item * They can be canceled like any other request. Canceling will cancel not
		1066	only the request itself, but also all requests it contains.
		1067
		1068	=item * They can also can also be added to other IO::AIO::GRP objects.
		1069
		1070	=item * You must not add requests to a group from within the group callback (or
		1071	any later time).
		1072
		1073	=back
		1074
		1075	Their lifetime, simplified, looks like this: when they are empty, they
		1076	will finish very quickly. If they contain only requests that are in the
		1077	C<done> state, they will also finish. Otherwise they will continue to
		1078	exist.
		1079
		1080	That means after creating a group you have some time to add requests. And
		1081	in the callbacks of those requests, you can add further requests to the
		1082	group. And only when all those requests have finished will the the group
		1083	itself finish.
		1084
		1085	=over 4
		1086
		1087	=item add $grp ...
		1088
		1089	=item $grp->add (...)
		1090
		1091	Add one or more requests to the group. Any type of L<IO::AIO::REQ> can
		1092	be added, including other groups, as long as you do not create circular
		1093	dependencies.
		1094
		1095	Returns all its arguments.
		1096
		1097	=item $grp->cancel_subs
		1098
		1099	Cancel all subrequests and clears any feeder, but not the group request
		1100	itself. Useful when you queued a lot of events but got a result early.
		1101
		1102	=item $grp->result (...)
		1103
		1104	Set the result value(s) that will be passed to the group callback when all
		1105	subrequests have finished and set the groups errno to the current value
		1106	of errno (just like calling C<errno> without an error number). By default,
		1107	no argument will be passed and errno is zero.
		1108
		1109	=item $grp->errno ([$errno])
		1110
		1111	Sets the group errno value to C<$errno>, or the current value of errno
		1112	when the argument is missing.
		1113
		1114	Every aio request has an associated errno value that is restored when
		1115	the callback is invoked. This method lets you change this value from its
		1116	default (0).
		1117
		1118	Calling C<result> will also set errno, so make sure you either set C<$!>
		1119	before the call to C<result>, or call c<errno> after it.
		1120
		1121	=item feed $grp $callback->($grp)
		1122
		1123	Sets a feeder/generator on this group: every group can have an attached
		1124	generator that generates requests if idle. The idea behind this is that,
		1125	although you could just queue as many requests as you want in a group,
		1126	this might starve other requests for a potentially long time. For
		1127	example, C<aio_scandir> might generate hundreds of thousands C<aio_stat>
		1128	requests, delaying any later requests for a long time.
		1129
		1130	To avoid this, and allow incremental generation of requests, you can
		1131	instead a group and set a feeder on it that generates those requests. The
		1132	feed callback will be called whenever there are few enough (see C<limit>,
		1133	below) requests active in the group itself and is expected to queue more
		1134	requests.
		1135
		1136	The feed callback can queue as many requests as it likes (i.e. C<add> does
		1137	not impose any limits).
		1138
		1139	If the feed does not queue more requests when called, it will be
		1140	automatically removed from the group.
		1141
		1142	If the feed limit is C<0>, it will be set to C<2> automatically.
		1143
		1144	Example:
		1145
		1146	# stat all files in @files, but only ever use four aio requests concurrently:
		1147
		1148	my $grp = aio_group sub { print "finished\n" };
		1149	limit $grp 4;
		1150	feed $grp sub {
		1151	my $file = pop @files
		1152	or return;
		1153
		1154	add $grp aio_stat $file, sub { ... };
		1155	};
		1156
		1157	=item limit $grp $num
		1158
		1159	Sets the feeder limit for the group: The feeder will be called whenever
		1160	the group contains less than this many requests.
		1161
		1162	Setting the limit to C<0> will pause the feeding process.
		1163
		1164	=back
		1165
228	=head2 SUPPORT FUNCTIONS	1166	=head2 SUPPORT FUNCTIONS
		1167
		1168	=head3 EVENT PROCESSING AND EVENT LOOP INTEGRATION
229		1169
230	=over 4	1170	=over 4
231		1171
232	=item $fileno = IO::AIO::poll_fileno	1172	=item $fileno = IO::AIO::poll_fileno
233		1173
…		…
238		1178
239	See C<poll_cb> for an example.	1179	See C<poll_cb> for an example.
240		1180
241	=item IO::AIO::poll_cb	1181	=item IO::AIO::poll_cb
242		1182
243	Process all outstanding events on the result pipe. You have to call this	1183	Process some outstanding events on the result pipe. You have to call this
244	regularly. Returns the number of events processed. Returns immediately	1184	regularly. Returns the number of events processed. Returns immediately
245	when no events are outstanding.	1185	when no events are outstanding. The amount of events processed depends on
		1186	the settings of C<IO::AIO::max_poll_req> and C<IO::AIO::max_poll_time>.
		1187
		1188	If not all requests were processed for whatever reason, the filehandle
		1189	will still be ready when C<poll_cb> returns.
246		1190
247	Example: Install an Event watcher that automatically calls	1191	Example: Install an Event watcher that automatically calls
248	IO::AIO::poll_cb with high priority:	1192	IO::AIO::poll_cb with high priority:
249		1193
250	Event->io (fd => IO::AIO::poll_fileno,	1194	Event->io (fd => IO::AIO::poll_fileno,
251	poll => 'r', async => 1,	1195	poll => 'r', async => 1,
252	cb => \&IO::AIO::poll_cb);	1196	cb => \&IO::AIO::poll_cb);
253		1197
		1198	=item IO::AIO::max_poll_reqs $nreqs
		1199
		1200	=item IO::AIO::max_poll_time $seconds
		1201
		1202	These set the maximum number of requests (default C<0>, meaning infinity)
		1203	that are being processed by C<IO::AIO::poll_cb> in one call, respectively
		1204	the maximum amount of time (default C<0>, meaning infinity) spent in
		1205	C<IO::AIO::poll_cb> to process requests (more correctly the mininum amount
		1206	of time C<poll_cb> is allowed to use).
		1207
		1208	Setting C<max_poll_time> to a non-zero value creates an overhead of one
		1209	syscall per request processed, which is not normally a problem unless your
		1210	callbacks are really really fast or your OS is really really slow (I am
		1211	not mentioning Solaris here). Using C<max_poll_reqs> incurs no overhead.
		1212
		1213	Setting these is useful if you want to ensure some level of
		1214	interactiveness when perl is not fast enough to process all requests in
		1215	time.
		1216
		1217	For interactive programs, values such as C<0.01> to C<0.1> should be fine.
		1218
		1219	Example: Install an Event watcher that automatically calls
		1220	IO::AIO::poll_cb with low priority, to ensure that other parts of the
		1221	program get the CPU sometimes even under high AIO load.
		1222
		1223	# try not to spend much more than 0.1s in poll_cb
		1224	IO::AIO::max_poll_time 0.1;
		1225
		1226	# use a low priority so other tasks have priority
		1227	Event->io (fd => IO::AIO::poll_fileno,
		1228	poll => 'r', nice => 1,
		1229	cb => &IO::AIO::poll_cb);
		1230
254	=item IO::AIO::poll_wait	1231	=item IO::AIO::poll_wait
255		1232
		1233	If there are any outstanding requests and none of them in the result
256	Wait till the result filehandle becomes ready for reading (simply does a	1234	phase, wait till the result filehandle becomes ready for reading (simply
257	C<select> on the filehandle. This is useful if you want to synchronously wait	1235	does a C<select> on the filehandle. This is useful if you want to
258	for some requests to finish).	1236	synchronously wait for some requests to finish).
259		1237
260	See C<nreqs> for an example.	1238	See C<nreqs> for an example.
261		1239
		1240	=item IO::AIO::poll
		1241
		1242	Waits until some requests have been handled.
		1243
		1244	Returns the number of requests processed, but is otherwise strictly
		1245	equivalent to:
		1246
		1247	IO::AIO::poll_wait, IO::AIO::poll_cb
		1248
262	=item IO::AIO::nreqs	1249	=item IO::AIO::flush
263		1250
264	Returns the number of requests currently outstanding (i.e. for which their	1251	Wait till all outstanding AIO requests have been handled.
265	callback has not been invoked yet).
266		1252
267	Example: wait till there are no outstanding requests anymore:	1253	Strictly equivalent to:
268		1254
269	IO::AIO::poll_wait, IO::AIO::poll_cb	1255	IO::AIO::poll_wait, IO::AIO::poll_cb
270	while IO::AIO::nreqs;	1256	while IO::AIO::nreqs;
271		1257
272	=item IO::AIO::flush	1258	=back
273		1259
274	Wait till all outstanding AIO requests have been handled.	1260	=head3 CONTROLLING THE NUMBER OF THREADS
275		1261
276	Strictly equivalent to:	1262	=over
277
278	IO::AIO::poll_wait, IO::AIO::poll_cb
279	while IO::AIO::nreqs;
280
281	=item IO::AIO::poll
282
283	Waits until some requests have been handled.
284
285	Strictly equivalent to:
286
287	IO::AIO::poll_wait, IO::AIO::poll_cb
288	if IO::AIO::nreqs;
289		1263
290	=item IO::AIO::min_parallel $nthreads	1264	=item IO::AIO::min_parallel $nthreads
291		1265
292	Set the minimum number of AIO threads to C<$nthreads>. The current default	1266	Set the minimum number of AIO threads to C<$nthreads>. The current
293	is C<4>, which means four asynchronous operations can be done at one time	1267	default is C<8>, which means eight asynchronous operations can execute
294	(the number of outstanding operations, however, is unlimited).	1268	concurrently at any one time (the number of outstanding requests,
		1269	however, is unlimited).
295		1270
296	IO::AIO starts threads only on demand, when an AIO request is queued and	1271	IO::AIO starts threads only on demand, when an AIO request is queued and
297	no free thread exists.	1272	no free thread exists. Please note that queueing up a hundred requests can
		1273	create demand for a hundred threads, even if it turns out that everything
		1274	is in the cache and could have been processed faster by a single thread.
298		1275
299	It is recommended to keep the number of threads low, as some Linux	1276	It is recommended to keep the number of threads relatively low, as some
300	kernel versions will scale negatively with the number of threads (higher	1277	Linux kernel versions will scale negatively with the number of threads
301	parallelity => MUCH higher latency). With current Linux 2.6 versions, 4-32	1278	(higher parallelity => MUCH higher latency). With current Linux 2.6
302	threads should be fine.	1279	versions, 4-32 threads should be fine.
303		1280
304	Under most circumstances you don't need to call this function, as the	1281	Under most circumstances you don't need to call this function, as the
305	module selects a default that is suitable for low to moderate load.	1282	module selects a default that is suitable for low to moderate load.
306		1283
307	=item IO::AIO::max_parallel $nthreads	1284	=item IO::AIO::max_parallel $nthreads
…		…
316	This module automatically runs C<max_parallel 0> at program end, to ensure	1293	This module automatically runs C<max_parallel 0> at program end, to ensure
317	that all threads are killed and that there are no outstanding requests.	1294	that all threads are killed and that there are no outstanding requests.
318		1295
319	Under normal circumstances you don't need to call this function.	1296	Under normal circumstances you don't need to call this function.
320		1297
		1298	=item IO::AIO::max_idle $nthreads
		1299
		1300	Limit the number of threads (default: 4) that are allowed to idle (i.e.,
		1301	threads that did not get a request to process within 10 seconds). That
		1302	means if a thread becomes idle while C<$nthreads> other threads are also
		1303	idle, it will free its resources and exit.
		1304
		1305	This is useful when you allow a large number of threads (e.g. 100 or 1000)
		1306	to allow for extremely high load situations, but want to free resources
		1307	under normal circumstances (1000 threads can easily consume 30MB of RAM).
		1308
		1309	The default is probably ok in most situations, especially if thread
		1310	creation is fast. If thread creation is very slow on your system you might
		1311	want to use larger values.
		1312
321	=item $oldnreqs = IO::AIO::max_outstanding $nreqs	1313	=item $oldmaxreqs = IO::AIO::max_outstanding $maxreqs
		1314
		1315	This is a very bad function to use in interactive programs because it
		1316	blocks, and a bad way to reduce concurrency because it is inexact: Better
		1317	use an C<aio_group> together with a feed callback.
322		1318
323	Sets the maximum number of outstanding requests to C<$nreqs>. If you	1319	Sets the maximum number of outstanding requests to C<$nreqs>. If you
324	try to queue up more than this number of requests, the caller will block until	1320	do queue up more than this number of requests, the next call to the
325	some requests have been handled.	1321	C<poll_cb> (and C<poll_some> and other functions calling C<poll_cb>)
		1322	function will block until the limit is no longer exceeded.
326		1323
327	The default is very large, so normally there is no practical limit. If you	1324	The default value is very large, so there is no practical limit on the
328	queue up many requests in a loop it often improves speed if you set	1325	number of outstanding requests.
329	this to a relatively low number, such as C<100>.
330		1326
331	Under normal circumstances you don't need to call this function.	1327	You can still queue as many requests as you want. Therefore,
		1328	C<max_oustsanding> is mainly useful in simple scripts (with low values) or
		1329	as a stop gap to shield against fatal memory overflow (with large values).
332		1330
333	=back	1331	=back
334		1332
		1333	=head3 STATISTICAL INFORMATION
		1334
		1335	=over
		1336
		1337	=item IO::AIO::nreqs
		1338
		1339	Returns the number of requests currently in the ready, execute or pending
		1340	states (i.e. for which their callback has not been invoked yet).
		1341
		1342	Example: wait till there are no outstanding requests anymore:
		1343
		1344	IO::AIO::poll_wait, IO::AIO::poll_cb
		1345	while IO::AIO::nreqs;
		1346
		1347	=item IO::AIO::nready
		1348
		1349	Returns the number of requests currently in the ready state (not yet
		1350	executed).
		1351
		1352	=item IO::AIO::npending
		1353
		1354	Returns the number of requests currently in the pending state (executed,
		1355	but not yet processed by poll_cb).
		1356
		1357	=back
		1358
335	=cut	1359	=cut
336		1360
337	# support function to convert a fd into a perl filehandle
338	sub _fd2fh {
339	return undef if $_[0] < 0;
340
341	# try to generate nice filehandles
342	my $sym = "IO::AIO::fd#$_[0]";
343	local *$sym;
344
345	open *$sym, "+<&=$_[0]" # usually works under any unix
346	or open *$sym, "<&=$_[0]" # cygwin needs this
347	or open *$sym, ">&=$_[0]" # or this
348	or return undef;
349
350	*$sym
351	}
352
353	min_parallel 4;	1361	min_parallel 8;
354		1362
355	END {	1363	END { flush }
356	max_parallel 0;
357	}
358		1364
359	1;	1365	1;
360		1366
361	=head2 FORK BEHAVIOUR	1367	=head2 FORK BEHAVIOUR
		1368
		1369	This module should do "the right thing" when the process using it forks:
362		1370
363	Before the fork, IO::AIO enters a quiescent state where no requests	1371	Before the fork, IO::AIO enters a quiescent state where no requests
364	can be added in other threads and no results will be processed. After	1372	can be added in other threads and no results will be processed. After
365	the fork the parent simply leaves the quiescent state and continues	1373	the fork the parent simply leaves the quiescent state and continues
366	request/result processing, while the child clears the request/result	1374	request/result processing, while the child frees the request/result queue
367	queue (so the requests started before the fork will only be handled in	1375	(so that the requests started before the fork will only be handled in the
368	the parent). Threats will be started on demand until the limit ste in the	1376	parent). Threads will be started on demand until the limit set in the
369	parent process has been reached again.	1377	parent process has been reached again.
370		1378
		1379	In short: the parent will, after a short pause, continue as if fork had
		1380	not been called, while the child will act as if IO::AIO has not been used
		1381	yet.
		1382
		1383	=head2 MEMORY USAGE
		1384
		1385	Per-request usage:
		1386
		1387	Each aio request uses - depending on your architecture - around 100-200
		1388	bytes of memory. In addition, stat requests need a stat buffer (possibly
		1389	a few hundred bytes), readdir requires a result buffer and so on. Perl
		1390	scalars and other data passed into aio requests will also be locked and
		1391	will consume memory till the request has entered the done state.
		1392
		1393	This is not awfully much, so queuing lots of requests is not usually a
		1394	problem.
		1395
		1396	Per-thread usage:
		1397
		1398	In the execution phase, some aio requests require more memory for
		1399	temporary buffers, and each thread requires a stack and other data
		1400	structures (usually around 16k-128k, depending on the OS).
		1401
		1402	=head1 KNOWN BUGS
		1403
		1404	Known bugs will be fixed in the next release.
		1405
371	=head1 SEE ALSO	1406	=head1 SEE ALSO
372		1407
373	L<Coro>, L<Linux::AIO>.	1408	L<Coro::AIO>.
374		1409
375	=head1 AUTHOR	1410	=head1 AUTHOR
376		1411
377	Marc Lehmann <schmorp@schmorp.de>	1412	Marc Lehmann <schmorp@schmorp.de>
378	http://home.schmorp.de/	1413	http://home.schmorp.de/

Diff Legend

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