ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/IO-AIO/AIO.pm

Comparing IO-AIO/AIO.pm (file contents):
Revision 1.1 by root, Sun Jul 10 17:07:44 2005 UTC vs.
Revision 1.164 by root, Tue Aug 18 03:26:02 2009 UTC

…		…
4		4
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 {
		10	my $fh = shift
		11	or die "/etc/passwd: $!";
		12	...
		13	};
		14
		15	aio_unlink "/tmp/file", sub { };
		16
		17	aio_read $fh, 30000, 1024, $buffer, 0, sub {
		18	$_[0] > 0 or die "read error: $!";
		19	};
		20
		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, POE, urxvt, pureperl...)
		32	use AnyEvent::AIO;
		33
		34	# EV integration
		35	my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
		36
		37	# Event integration
		38	Event->io (fd => IO::AIO::poll_fileno,
		39	poll => 'r',
		40	cb => \&IO::AIO::poll_cb);
		41
		42	# Glib/Gtk2 integration
		43	add_watch Glib::IO IO::AIO::poll_fileno,
		44	in => sub { IO::AIO::poll_cb; 1 };
		45
		46	# Tk integration
		47	Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",
		48	readable => \&IO::AIO::poll_cb);
		49
		50	# Danga::Socket integration
		51	Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
		52	\&IO::AIO::poll_cb);
		53
9	=head1 DESCRIPTION	54	=head1 DESCRIPTION
10		55
11	This module implements asynchronous I/O using whatever means your	56	This module implements asynchronous I/O using whatever means your
12	operating system supports. Currently, it falls back to Linux::AIO if that	57	operating system supports. It is implemented as an interface to C<libeio>
13	module is available, or uses pthreads to emulato aio functionality.	58	(L<http://software.schmorp.de/pkg/libeio.html>).
14		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
		73	very inefficient. Use an event loop for that (such as the L<EV>
		74	module): IO::AIO will naturally fit into such an event loop itself.
		75
15	Currently, in this module a number of threads are started that execute	76	In this version, a number of threads are started that execute your
16	your read/writes and signal their completion. You don't need thread	77	requests and signal their completion. You don't need thread support
17	support in your libc or perl, and the threads created by this module will	78	in perl, and the threads created by this module will not be visible
18	not be visible to the pthreads library.	79	to perl. In the future, this module might make use of the native aio
		80	functions available on many operating systems. However, they are often
		81	not well-supported or restricted (GNU/Linux doesn't allow them on normal
		82	files currently, for example), and they would only support aio_read and
		83	aio_write, so the remaining functionality would have to be implemented
		84	using threads anyway.
19		85
20	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,
21	not reentrant, so use appropriate locking yourself.	87	it is currently not reentrant in any way, so use appropriate locking
		88	yourself, always call C<poll_cb> from within the same thread, or never
		89	call C<poll_cb> (or other C<aio_> functions) recursively.
22		90
23	=head2 API NOTES	91	=head2 EXAMPLE
		92
		93	This is a simple example that uses the EV module and loads
		94	F</etc/passwd> asynchronously:
		95
		96	use Fcntl;
		97	use EV;
		98	use IO::AIO;
		99
		100	# register the IO::AIO callback with EV
		101	my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
		102
		103	# queue the request to open /etc/passwd
		104	aio_open "/etc/passwd", O_RDONLY, 0, sub {
		105	my $fh = shift
		106	or die "error while opening: $!";
		107
		108	# stat'ing filehandles is generally non-blocking
		109	my $size = -s $fh;
		110
		111	# queue a request to read the file
		112	my $contents;
		113	aio_read $fh, 0, $size, $contents, 0, sub {
		114	$_[0] == $size
		115	or die "short read: $!";
		116
		117	close $fh;
		118
		119	# file contents now in $contents
		120	print $contents;
		121
		122	# exit event loop and program
		123	EV::unloop;
		124	};
		125	};
		126
		127	# possibly queue up other requests, or open GUI windows,
		128	# check for sockets etc. etc.
		129
		130	# process events as long as there are some:
		131	EV::loop;
		132
		133	=head1 REQUEST ANATOMY AND LIFETIME
		134
		135	Every C<aio_*> function creates a request. which is a C data structure not
		136	directly visible to Perl.
		137
		138	If called in non-void context, every request function returns a Perl
		139	object representing the request. In void context, nothing is returned,
		140	which saves a bit of memory.
		141
		142	The perl object is a fairly standard ref-to-hash object. The hash contents
		143	are not used by IO::AIO so you are free to store anything you like in it.
		144
		145	During their existance, aio requests travel through the following states,
		146	in order:
		147
		148	=over 4
		149
		150	=item ready
		151
		152	Immediately after a request is created it is put into the ready state,
		153	waiting for a thread to execute it.
		154
		155	=item execute
		156
		157	A thread has accepted the request for processing and is currently
		158	executing it (e.g. blocking in read).
		159
		160	=item pending
		161
		162	The request has been executed and is waiting for result processing.
		163
		164	While request submission and execution is fully asynchronous, result
		165	processing is not and relies on the perl interpreter calling C<poll_cb>
		166	(or another function with the same effect).
		167
		168	=item result
		169
		170	The request results are processed synchronously by C<poll_cb>.
		171
		172	The C<poll_cb> function will process all outstanding aio requests by
		173	calling their callbacks, freeing memory associated with them and managing
		174	any groups they are contained in.
		175
		176	=item done
		177
		178	Request has reached the end of its lifetime and holds no resources anymore
		179	(except possibly for the Perl object, but its connection to the actual
		180	aio request is severed and calling its methods will either do nothing or
		181	result in a runtime error).
		182
		183	=back
		184
		185	=cut
		186
		187	package IO::AIO;
		188
		189	use Carp ();
		190
		191	use common::sense;
		192
		193	use base 'Exporter';
		194
		195	BEGIN {
		196	our $VERSION = '3.3';
		197
		198	our @AIO_REQ = qw(aio_sendfile aio_read aio_write aio_open aio_close
		199	aio_stat aio_lstat aio_unlink aio_rmdir aio_readdir aio_readdirx
		200	aio_scandir aio_symlink aio_readlink aio_sync aio_fsync
		201	aio_fdatasync aio_sync_file_range aio_pathsync aio_readahead
		202	aio_rename aio_link aio_move aio_copy aio_group
		203	aio_nop aio_mknod aio_load aio_rmtree aio_mkdir aio_chown
		204	aio_chmod aio_utime aio_truncate);
		205
		206	our @EXPORT = (@AIO_REQ, qw(aioreq_pri aioreq_nice));
		207	our @EXPORT_OK = qw(poll_fileno poll_cb poll_wait flush
		208	min_parallel max_parallel max_idle
		209	nreqs nready npending nthreads
		210	max_poll_time max_poll_reqs
		211	sendfile fadvise);
		212
		213	push @AIO_REQ, qw(aio_busy); # not exported
		214
		215	@IO::AIO::GRP::ISA = 'IO::AIO::REQ';
		216
		217	require XSLoader;
		218	XSLoader::load ("IO::AIO", $VERSION);
		219	}
		220
		221	=head1 FUNCTIONS
		222
		223	=head2 AIO REQUEST FUNCTIONS
24		224
25	All the C<aio_*> calls are more or less thin wrappers around the syscall	225	All the C<aio_*> calls are more or less thin wrappers around the syscall
26	with the same name (sans C<aio_>). The arguments are similar or identical,	226	with the same name (sans C<aio_>). The arguments are similar or identical,
27	and they all accept an additional C<$callback> argument which must be	227	and they all accept an additional (and optional) C<$callback> argument
28	a code reference. This code reference will get called with the syscall	228	which must be a code reference. This code reference will get called with
29	return code (e.g. most syscalls return C<-1> on error, unlike perl, which	229	the syscall return code (e.g. most syscalls return C<-1> on error, unlike
30	usually delivers "false") as it's sole argument when the given syscall has	230	perl, which usually delivers "false") as its sole argument after the given
31	been executed asynchronously.	231	syscall has been executed asynchronously.
32		232
33	All functions that expect a filehandle will also accept a file descriptor.	233	All functions expecting a filehandle keep a copy of the filehandle
		234	internally until the request has finished.
34		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
35	The filenames you pass to these routines I<must> be absolute. The reason	239	The pathnames you pass to these routines I<must> be absolute and
36	is that at the time the request is being executed, the current working	240	encoded as octets. The reason for the former is that at the time the
37	directory could have changed. Alternatively, you can make sure that you	241	request is being executed, the current working directory could have
38	never change the current working directory.	242	changed. Alternatively, you can make sure that you never change the
		243	current working directory anywhere in the program and then use relative
		244	paths.
		245
		246	To encode pathnames as octets, either make sure you either: a) always pass
		247	in filenames you got from outside (command line, readdir etc.) without
		248	tinkering, b) are ASCII or ISO 8859-1, c) use the Encode module and encode
		249	your pathnames to the locale (or other) encoding in effect in the user
		250	environment, d) use Glib::filename_from_unicode on unicode filenames or e)
		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 whether it is set or not.
39		255
40	=over 4	256	=over 4
41		257
42	=cut	258	=item $prev_pri = aioreq_pri [$pri]
43		259
44	package IO::AIO;	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.
45		262
46	use base 'Exporter';	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.
47		266
48	BEGIN {	267	The priority will be reset to C<0> after each call to one of the C<aio_*>
49	$VERSION = 0.1;	268	functions.
50		269
51	@EXPORT = qw(aio_read aio_write aio_open aio_close aio_stat aio_lstat aio_unlink	270	Example: open a file with low priority, then read something from it with
52	aio_fsync aio_fdatasync aio_readahead);	271	higher priority so the read request is serviced before other low priority
53	@EXPORT_OK = qw(poll_fileno poll_cb min_parallel max_parallel nreqs);	272	open requests (potentially spamming the cache):
54		273
55	require XSLoader;	274	aioreq_pri -3;
56	XSLoader::load IO::AIO, $VERSION;	275	aio_open ..., sub {
57	}	276	return unless $_[0];
58		277
59	=item IO::AIO::min_parallel $nthreads	278	aioreq_pri -2;
		279	aio_read $_[0], ..., sub {
		280	...
		281	};
		282	};
60		283
61	Set the minimum number of AIO threads to C<$nthreads>. The default is
62	C<1>, which means a single asynchronous operation can be done at one time
63	(the number of outstanding operations, however, is unlimited).
64		284
65	It is recommended to keep the number of threads low, as some linux	285	=item aioreq_nice $pri_adjust
66	kernel versions will scale negatively with the number of threads (higher
67	parallelity => MUCH higher latency).
68		286
69	Under normal circumstances you don't need to call this function, as this	287	Similar to C<aioreq_pri>, but subtracts the given value from the current
70	module automatically starts a single async thread.	288	priority, so the effect is cumulative.
71		289
72	=item IO::AIO::max_parallel $nthreads
73		290
74	Sets the maximum number of AIO threads to C<$nthreads>. If more than
75	the specified number of threads are currently running, kill them. This
76	function blocks until the limit is reached.
77
78	This module automatically runs C<max_parallel 0> at program end, to ensure
79	that all threads are killed and that there are no outstanding requests.
80
81	Under normal circumstances you don't need to call this function.
82
83	=item $fileno = IO::AIO::poll_fileno
84
85	Return the I<request result pipe filehandle>. This filehandle must be
86	polled for reading by some mechanism outside this module (e.g. Event
87	or select, see below). If the pipe becomes readable you have to call
88	C<poll_cb> to check the results.
89
90	See C<poll_cb> for an example.
91
92	=item IO::AIO::poll_cb
93
94	Process all outstanding events on the result pipe. You have to call this
95	regularly. Returns the number of events processed. Returns immediately
96	when no events are outstanding.
97
98	You can use Event to multiplex, e.g.:
99
100	Event->io (fd => IO::AIO::poll_fileno,
101	poll => 'r', async => 1,
102	cb => \&IO::AIO::poll_cb);
103
104	=item IO::AIO::poll_wait
105
106	Wait till the result filehandle becomes ready for reading (simply does a
107	select on the filehandle. This is useful if you want to synchronously wait
108	for some requests to finish).
109
110	See C<nreqs> for an example.
111
112	=item IO::AIO::nreqs
113
114	Returns the number of requests currently outstanding.
115
116	Example: wait till there are no outstanding requests anymore:
117
118	IO::AIO::poll_wait, IO::AIO::poll_cb
119	while IO::AIO::nreqs;
120
121	=item aio_open $pathname, $flags, $mode, $callback	291	=item aio_open $pathname, $flags, $mode, $callback->($fh)
122		292
123	Asynchronously open or create a file and call the callback with the	293	Asynchronously open or create a file and call the callback with a newly
124	filedescriptor (NOT a perl filehandle, sorry for that, but watch out, this	294	created filehandle for the file.
125	might change in the future).
126		295
127	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,
128	for an explanation.	297	for an explanation.
129		298
130	The C<$mode> argument is a bitmask. See the C<Fcntl> module for a	299	The C<$flags> argument is a bitmask. See the C<Fcntl> module for a
131	list. They are the same as used in C<sysopen>.	300	list. They are the same as used by C<sysopen>.
		301
		302	Likewise, C<$mode> specifies the mode of the newly created file, if it
		303	didn't exist and C<O_CREAT> has been given, just like perl's C<sysopen>,
		304	except that it is mandatory (i.e. use C<0> if you don't create new files,
		305	and C<0666> or C<0777> if you do). Note that the C<$mode> will be modified
		306	by the umask in effect then the request is being executed, so better never
		307	change the umask.
132		308
133	Example:	309	Example:
134		310
135	aio_open "/etc/passwd", O_RDONLY, 0, sub {	311	aio_open "/etc/passwd", O_RDONLY, 0, sub {
136	if ($_[0] >= 0) {	312	if ($_[0]) {
137	open my $fh, "<&$_[0]"; # create a copy for perl
138	aio_close $_[0], sub { }; # close the aio handle
139	print "open successful, fh is $fh\n";	313	print "open successful, fh is $_[0]\n";
140	...	314	...
141	} else {	315	} else {
142	die "open failed: $!\n";	316	die "open failed: $!\n";
143	}	317	}
144	};	318	};
145		319
		320
146	=item aio_close $fh, $callback	321	=item aio_close $fh, $callback->($status)
147		322
148	Asynchronously close a file and call the callback with the result code.	323	Asynchronously close a file and call the callback with the result
		324	code.
149		325
		326	Unfortunately, you can't do this to perl. Perl I<insists> very strongly on
		327	closing the file descriptor associated with the filehandle itself.
		328
		329	Therefore, C<aio_close> will not close the filehandle - instead it will
		330	use dup2 to overwrite the file descriptor with the write-end of a pipe
		331	(the pipe fd will be created on demand and will be cached).
		332
		333	Or in other words: the file descriptor will be closed, but it will not be
		334	free for reuse until the perl filehandle is closed.
		335
		336	=cut
		337
150	=item aio_read $fh,$offset,$length, $data,$dataoffset,$callback	338	=item aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
151		339
152	=item aio_write $fh,$offset,$length, $data,$dataoffset,$callback	340	=item aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
153		341
154	Reads or writes C<length> bytes from the specified C<fh> and C<offset>	342	Reads or writes C<$length> bytes from or to the specified C<$fh> and
155	into the scalar given by C<data> and offset C<dataoffset> and calls the	343	C<$offset> into the scalar given by C<$data> and offset C<$dataoffset>
156	callback without the actual number of bytes read (or -1 on error, just	344	and calls the callback without the actual number of bytes read (or -1 on
157	like the syscall).	345	error, just like the syscall).
158		346
		347	C<aio_read> will, like C<sysread>, shrink or grow the C<$data> scalar to
		348	offset plus the actual number of bytes read.
		349
		350	If C<$offset> is undefined, then the current file descriptor offset will
		351	be used (and updated), otherwise the file descriptor offset will not be
		352	changed by these calls.
		353
		354	If C<$length> is undefined in C<aio_write>, use the remaining length of
		355	C<$data>.
		356
		357	If C<$dataoffset> is less than zero, it will be counted from the end of
		358	C<$data>.
		359
		360	The C<$data> scalar I<MUST NOT> be modified in any way while the request
		361	is outstanding. Modifying it can result in segfaults or World War III (if
		362	the necessary/optional hardware is installed).
		363
159	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, strating at	364	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at
160	offset C<0> within the scalar:	365	offset C<0> within the scalar:
161		366
162	aio_read $fh, 7, 15, $buffer, 0, sub {	367	aio_read $fh, 7, 15, $buffer, 0, sub {
163	$_[0] >= 0 or die "read error: $!";	368	$_[0] > 0 or die "read error: $!";
164	print "read <$buffer>\n";	369	print "read $_[0] bytes: <$buffer>\n";
165	};	370	};
166		371
		372
		373	=item aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
		374
		375	Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts
		376	reading at byte offset C<$in_offset>, and starts writing at the current
		377	file offset of C<$out_fh>. Because of that, it is not safe to issue more
		378	than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each
		379	other.
		380
		381	This call tries to make use of a native C<sendfile> syscall to provide
		382	zero-copy operation. For this to work, C<$out_fh> should refer to a
		383	socket, and C<$in_fh> should refer to mmap'able file.
		384
		385	If the native sendfile call fails or is not implemented, it will be
		386	emulated, so you can call C<aio_sendfile> on any type of filehandle
		387	regardless of the limitations of the operating system.
		388
		389	Please note, however, that C<aio_sendfile> can read more bytes from
		390	C<$in_fh> than are written, and there is no way to find out how many
		391	bytes have been read from C<aio_sendfile> alone, as C<aio_sendfile> only
		392	provides the number of bytes written to C<$out_fh>. Only if the result
		393	value equals C<$length> one can assume that C<$length> bytes have been
		394	read.
		395
		396
167	=item aio_readahead $fh,$offset,$length, $callback	397	=item aio_readahead $fh,$offset,$length, $callback->($retval)
168		398
169	Asynchronously reads the specified byte range into the page cache, using
170	the C<readahead> syscall. If that syscall doesn't exist the status will be
171	C<-1> and C<$!> is set to ENOSYS.
172
173	readahead() populates the page cache with data from a file so that	399	C<aio_readahead> populates the page cache with data from a file so that
174	subsequent reads from that file will not block on disk I/O. The C<$offset>	400	subsequent reads from that file will not block on disk I/O. The C<$offset>
175	argument specifies the starting point from which data is to be read and	401	argument specifies the starting point from which data is to be read and
176	C<$length> specifies the number of bytes to be read. I/O is performed in	402	C<$length> specifies the number of bytes to be read. I/O is performed in
177	whole pages, so that offset is effectively rounded down to a page boundary	403	whole pages, so that offset is effectively rounded down to a page boundary
178	and bytes are read up to the next page boundary greater than or equal to	404	and bytes are read up to the next page boundary greater than or equal to
179	(off-set+length). aio_readahead() does not read beyond the end of the	405	(off-set+length). C<aio_readahead> does not read beyond the end of the
180	file. The current file offset of the file is left unchanged.	406	file. The current file offset of the file is left unchanged.
181		407
		408	If that syscall doesn't exist (likely if your OS isn't Linux) it will be
		409	emulated by simply reading the data, which would have a similar effect.
		410
		411
182	=item aio_stat $fh_or_path, $callback	412	=item aio_stat $fh_or_path, $callback->($status)
183		413
184	=item aio_lstat $fh, $callback	414	=item aio_lstat $fh, $callback->($status)
185		415
186	Works like perl's C<stat> or C<lstat> in void context. The callback will	416	Works like perl's C<stat> or C<lstat> in void context. The callback will
187	be called after the stat and the results will be available using C<stat _>	417	be called after the stat and the results will be available using C<stat _>
188	or C<-s _> etc...	418	or C<-s _> etc...
189		419
…		…
199	aio_stat "/etc/passwd", sub {	429	aio_stat "/etc/passwd", sub {
200	$_[0] and die "stat failed: $!";	430	$_[0] and die "stat failed: $!";
201	print "size is ", -s _, "\n";	431	print "size is ", -s _, "\n";
202	};	432	};
203		433
		434
		435	=item aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
		436
		437	Works like perl's C<utime> function (including the special case of $atime
		438	and $mtime being undef). Fractional times are supported if the underlying
		439	syscalls support them.
		440
		441	When called with a pathname, uses utimes(2) if available, otherwise
		442	utime(2). If called on a file descriptor, uses futimes(2) if available,
		443	otherwise returns ENOSYS, so this is not portable.
		444
		445	Examples:
		446
		447	# set atime and mtime to current time (basically touch(1)):
		448	aio_utime "path", undef, undef;
		449	# set atime to current time and mtime to beginning of the epoch:
		450	aio_utime "path", time, undef; # undef==0
		451
		452
		453	=item aio_chown $fh_or_path, $uid, $gid, $callback->($status)
		454
		455	Works like perl's C<chown> function, except that C<undef> for either $uid
		456	or $gid is being interpreted as "do not change" (but -1 can also be used).
		457
		458	Examples:
		459
		460	# same as "chown root path" in the shell:
		461	aio_chown "path", 0, -1;
		462	# same as above:
		463	aio_chown "path", 0, undef;
		464
		465
		466	=item aio_truncate $fh_or_path, $offset, $callback->($status)
		467
		468	Works like truncate(2) or ftruncate(2).
		469
		470
		471	=item aio_chmod $fh_or_path, $mode, $callback->($status)
		472
		473	Works like perl's C<chmod> function.
		474
		475
204	=item aio_unlink $pathname, $callback	476	=item aio_unlink $pathname, $callback->($status)
205		477
206	Asynchronously unlink (delete) a file and call the callback with the	478	Asynchronously unlink (delete) a file and call the callback with the
207	result code.	479	result code.
208		480
		481
		482	=item aio_mknod $path, $mode, $dev, $callback->($status)
		483
		484	[EXPERIMENTAL]
		485
		486	Asynchronously create a device node (or fifo). See mknod(2).
		487
		488	The only (POSIX-) portable way of calling this function is:
		489
		490	aio_mknod $path, IO::AIO::S_IFIFO | $mode, 0, sub { ...
		491
		492
		493	=item aio_link $srcpath, $dstpath, $callback->($status)
		494
		495	Asynchronously create a new link to the existing object at C<$srcpath> at
		496	the path C<$dstpath> and call the callback with the result code.
		497
		498
		499	=item aio_symlink $srcpath, $dstpath, $callback->($status)
		500
		501	Asynchronously create a new symbolic link to the existing object at C<$srcpath> at
		502	the path C<$dstpath> and call the callback with the result code.
		503
		504
		505	=item aio_readlink $path, $callback->($link)
		506
		507	Asynchronously read the symlink specified by C<$path> and pass it to
		508	the callback. If an error occurs, nothing or undef gets passed to the
		509	callback.
		510
		511
		512	=item aio_rename $srcpath, $dstpath, $callback->($status)
		513
		514	Asynchronously rename the object at C<$srcpath> to C<$dstpath>, just as
		515	rename(2) and call the callback with the result code.
		516
		517
		518	=item aio_mkdir $pathname, $mode, $callback->($status)
		519
		520	Asynchronously mkdir (create) a directory and call the callback with
		521	the result code. C<$mode> will be modified by the umask at the time the
		522	request is executed, so do not change your umask.
		523
		524
		525	=item aio_rmdir $pathname, $callback->($status)
		526
		527	Asynchronously rmdir (delete) a directory and call the callback with the
		528	result code.
		529
		530
		531	=item aio_readdir $pathname, $callback->($entries)
		532
		533	Unlike the POSIX call of the same name, C<aio_readdir> reads an entire
		534	directory (i.e. opendir + readdir + closedir). The entries will not be
		535	sorted, and will B<NOT> include the C<.> and C<..> entries.
		536
		537	The callback is passed a single argument which is either C<undef> or an
		538	array-ref with the filenames.
		539
		540
		541	=item aio_readdirx $pathname, $flags, $callback->($entries, $flags)
		542
		543	Quite similar to C<aio_readdir>, but the C<$flags> argument allows to tune
		544	behaviour and output format. In case of an error, C<$entries> will be
		545	C<undef>.
		546
		547	The flags are a combination of the following constants, ORed together (the
		548	flags will also be passed to the callback, possibly modified):
		549
		550	=over 4
		551
		552	=item IO::AIO::READDIR_DENTS
		553
		554	When this flag is off, then the callback gets an arrayref with of names
		555	only (as with C<aio_readdir>), otherwise it gets an arrayref with
		556	C<[$name, $type, $inode]> arrayrefs, each describing a single directory
		557	entry in more detail.
		558
		559	C<$name> is the name of the entry.
		560
		561	C<$type> is one of the C<IO::AIO::DT_xxx> constants:
		562
		563	C<IO::AIO::DT_UNKNOWN>, C<IO::AIO::DT_FIFO>, C<IO::AIO::DT_CHR>, C<IO::AIO::DT_DIR>,
		564	C<IO::AIO::DT_BLK>, C<IO::AIO::DT_REG>, C<IO::AIO::DT_LNK>, C<IO::AIO::DT_SOCK>,
		565	C<IO::AIO::DT_WHT>.
		566
		567	C<IO::AIO::DT_UNKNOWN> means just that: readdir does not know. If you need to
		568	know, you have to run stat yourself. Also, for speed reasons, the C<$type>
		569	scalars are read-only: you can not modify them.
		570
		571	C<$inode> is the inode number (which might not be exact on systems with 64
		572	bit inode numbers and 32 bit perls). This field has unspecified content on
		573	systems that do not deliver the inode information.
		574
		575	=item IO::AIO::READDIR_DIRS_FIRST
		576
		577	When this flag is set, then the names will be returned in an order where
		578	likely directories come first. This is useful when you need to quickly
		579	find directories, or you want to find all directories while avoiding to
		580	stat() each entry.
		581
		582	If the system returns type information in readdir, then this is used
		583	to find directories directly. Otherwise, likely directories are files
		584	beginning with ".", or otherwise files with no dots, of which files with
		585	short names are tried first.
		586
		587	=item IO::AIO::READDIR_STAT_ORDER
		588
		589	When this flag is set, then the names will be returned in an order
		590	suitable for stat()'ing each one. That is, when you plan to stat()
		591	all files in the given directory, then the returned order will likely
		592	be fastest.
		593
		594	If both this flag and C<IO::AIO::READDIR_DIRS_FIRST> are specified, then
		595	the likely dirs come first, resulting in a less optimal stat order.
		596
		597	=item IO::AIO::READDIR_FOUND_UNKNOWN
		598
		599	This flag should not be set when calling C<aio_readdirx>. Instead, it
		600	is being set by C<aio_readdirx>, when any of the C<$type>'s found were
		601	C<IO::AIO::DT_UNKNOWN>. The absense of this flag therefore indicates that all
		602	C<$type>'s are known, which can be used to speed up some algorithms.
		603
		604	=back
		605
		606
		607	=item aio_load $path, $data, $callback->($status)
		608
		609	This is a composite request that tries to fully load the given file into
		610	memory. Status is the same as with aio_read.
		611
		612	=cut
		613
		614	sub aio_load($$;$) {
		615	my ($path, undef, $cb) = @_;
		616	my $data = \$_[1];
		617
		618	my $pri = aioreq_pri;
		619	my $grp = aio_group $cb;
		620
		621	aioreq_pri $pri;
		622	add $grp aio_open $path, O_RDONLY, 0, sub {
		623	my $fh = shift
		624	or return $grp->result (-1);
		625
		626	aioreq_pri $pri;
		627	add $grp aio_read $fh, 0, (-s $fh), $$data, 0, sub {
		628	$grp->result ($_[0]);
		629	};
		630	};
		631
		632	$grp
		633	}
		634
		635	=item aio_copy $srcpath, $dstpath, $callback->($status)
		636
		637	Try to copy the I<file> (directories not supported as either source or
		638	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		639	a status of C<0> (error) or C<-1> (ok).
		640
		641	This is a composite request that creates the destination file with
		642	mode 0200 and copies the contents of the source file into it using
		643	C<aio_sendfile>, followed by restoring atime, mtime, access mode and
		644	uid/gid, in that order.
		645
		646	If an error occurs, the partial destination file will be unlinked, if
		647	possible, except when setting atime, mtime, access mode and uid/gid, where
		648	errors are being ignored.
		649
		650	=cut
		651
		652	sub aio_copy($$;$) {
		653	my ($src, $dst, $cb) = @_;
		654
		655	my $pri = aioreq_pri;
		656	my $grp = aio_group $cb;
		657
		658	aioreq_pri $pri;
		659	add $grp aio_open $src, O_RDONLY, 0, sub {
		660	if (my $src_fh = $_[0]) {
		661	my @stat = stat $src_fh; # hmm, might bock over nfs?
		662
		663	aioreq_pri $pri;
		664	add $grp aio_open $dst, O_CREAT | O_WRONLY | O_TRUNC, 0200, sub {
		665	if (my $dst_fh = $_[0]) {
		666	aioreq_pri $pri;
		667	add $grp aio_sendfile $dst_fh, $src_fh, 0, $stat[7], sub {
		668	if ($_[0] == $stat[7]) {
		669	$grp->result (0);
		670	close $src_fh;
		671
		672	my $ch = sub {
		673	aioreq_pri $pri;
		674	add $grp aio_chmod $dst_fh, $stat[2] & 07777, sub {
		675	aioreq_pri $pri;
		676	add $grp aio_chown $dst_fh, $stat[4], $stat[5], sub {
		677	aioreq_pri $pri;
		678	add $grp aio_close $dst_fh;
		679	}
		680	};
		681	};
		682
		683	aioreq_pri $pri;
		684	add $grp aio_utime $dst_fh, $stat[8], $stat[9], sub {
		685	if ($_[0] < 0 && $! == ENOSYS) {
		686	aioreq_pri $pri;
		687	add $grp aio_utime $dst, $stat[8], $stat[9], $ch;
		688	} else {
		689	$ch->();
		690	}
		691	};
		692	} else {
		693	$grp->result (-1);
		694	close $src_fh;
		695	close $dst_fh;
		696
		697	aioreq $pri;
		698	add $grp aio_unlink $dst;
		699	}
		700	};
		701	} else {
		702	$grp->result (-1);
		703	}
		704	},
		705
		706	} else {
		707	$grp->result (-1);
		708	}
		709	};
		710
		711	$grp
		712	}
		713
		714	=item aio_move $srcpath, $dstpath, $callback->($status)
		715
		716	Try to move the I<file> (directories not supported as either source or
		717	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		718	the status C<0> (error) or C<-1> (ok).
		719
		720	This is a composite request that tries to rename(2) the file first; if
		721	rename fails with C<EXDEV>, it copies the file with C<aio_copy> and, if
		722	that is successful, unlinks the C<$srcpath>.
		723
		724	=cut
		725
		726	sub aio_move($$;$) {
		727	my ($src, $dst, $cb) = @_;
		728
		729	my $pri = aioreq_pri;
		730	my $grp = aio_group $cb;
		731
		732	aioreq_pri $pri;
		733	add $grp aio_rename $src, $dst, sub {
		734	if ($_[0] && $! == EXDEV) {
		735	aioreq_pri $pri;
		736	add $grp aio_copy $src, $dst, sub {
		737	$grp->result ($_[0]);
		738
		739	if (!$_[0]) {
		740	aioreq_pri $pri;
		741	add $grp aio_unlink $src;
		742	}
		743	};
		744	} else {
		745	$grp->result ($_[0]);
		746	}
		747	};
		748
		749	$grp
		750	}
		751
		752	=item aio_scandir $path, $maxreq, $callback->($dirs, $nondirs)
		753
		754	Scans a directory (similar to C<aio_readdir>) but additionally tries to
		755	efficiently separate the entries of directory C<$path> into two sets of
		756	names, directories you can recurse into (directories), and ones you cannot
		757	recurse into (everything else, including symlinks to directories).
		758
		759	C<aio_scandir> is a composite request that creates of many sub requests_
		760	C<$maxreq> specifies the maximum number of outstanding aio requests that
		761	this function generates. If it is C<< <= 0 >>, then a suitable default
		762	will be chosen (currently 4).
		763
		764	On error, the callback is called without arguments, otherwise it receives
		765	two array-refs with path-relative entry names.
		766
		767	Example:
		768
		769	aio_scandir $dir, 0, sub {
		770	my ($dirs, $nondirs) = @_;
		771	print "real directories: @$dirs\n";
		772	print "everything else: @$nondirs\n";
		773	};
		774
		775	Implementation notes.
		776
		777	The C<aio_readdir> cannot be avoided, but C<stat()>'ing every entry can.
		778
		779	If readdir returns file type information, then this is used directly to
		780	find directories.
		781
		782	Otherwise, after reading the directory, the modification time, size etc.
		783	of the directory before and after the readdir is checked, and if they
		784	match (and isn't the current time), the link count will be used to decide
		785	how many entries are directories (if >= 2). Otherwise, no knowledge of the
		786	number of subdirectories will be assumed.
		787
		788	Then entries will be sorted into likely directories a non-initial dot
		789	currently) and likely non-directories (see C<aio_readdirx>). Then every
		790	entry plus an appended C</.> will be C<stat>'ed, likely directories first,
		791	in order of their inode numbers. If that succeeds, it assumes that the
		792	entry is a directory or a symlink to directory (which will be checked
		793	seperately). This is often faster than stat'ing the entry itself because
		794	filesystems might detect the type of the entry without reading the inode
		795	data (e.g. ext2fs filetype feature), even on systems that cannot return
		796	the filetype information on readdir.
		797
		798	If the known number of directories (link count - 2) has been reached, the
		799	rest of the entries is assumed to be non-directories.
		800
		801	This only works with certainty on POSIX (= UNIX) filesystems, which
		802	fortunately are the vast majority of filesystems around.
		803
		804	It will also likely work on non-POSIX filesystems with reduced efficiency
		805	as those tend to return 0 or 1 as link counts, which disables the
		806	directory counting heuristic.
		807
		808	=cut
		809
		810	sub aio_scandir($$;$) {
		811	my ($path, $maxreq, $cb) = @_;
		812
		813	my $pri = aioreq_pri;
		814
		815	my $grp = aio_group $cb;
		816
		817	$maxreq = 4 if $maxreq <= 0;
		818
		819	# stat once
		820	aioreq_pri $pri;
		821	add $grp aio_stat $path, sub {
		822	return $grp->result () if $_[0];
		823	my $now = time;
		824	my $hash1 = join ":", (stat _)[0,1,3,7,9];
		825
		826	# read the directory entries
		827	aioreq_pri $pri;
		828	add $grp aio_readdirx $path, READDIR_DIRS_FIRST, sub {
		829	my $entries = shift
		830	or return $grp->result ();
		831
		832	# stat the dir another time
		833	aioreq_pri $pri;
		834	add $grp aio_stat $path, sub {
		835	my $hash2 = join ":", (stat _)[0,1,3,7,9];
		836
		837	my $ndirs;
		838
		839	# take the slow route if anything looks fishy
		840	if ($hash1 ne $hash2 or (stat _)[9] == $now) {
		841	$ndirs = -1;
		842	} else {
		843	# if nlink == 2, we are finished
		844	# for non-posix-fs's, we rely on nlink < 2
		845	$ndirs = (stat _)[3] - 2
		846	or return $grp->result ([], $entries);
		847	}
		848
		849	my (@dirs, @nondirs);
		850
		851	my $statgrp = add $grp aio_group sub {
		852	$grp->result (\@dirs, \@nondirs);
		853	};
		854
		855	limit $statgrp $maxreq;
		856	feed $statgrp sub {
		857	return unless @$entries;
		858	my $entry = shift @$entries;
		859
		860	aioreq_pri $pri;
		861	add $statgrp aio_stat "$path/$entry/.", sub {
		862	if ($_[0] < 0) {
		863	push @nondirs, $entry;
		864	} else {
		865	# need to check for real directory
		866	aioreq_pri $pri;
		867	add $statgrp aio_lstat "$path/$entry", sub {
		868	if (-d _) {
		869	push @dirs, $entry;
		870
		871	unless (--$ndirs) {
		872	push @nondirs, @$entries;
		873	feed $statgrp;
		874	}
		875	} else {
		876	push @nondirs, $entry;
		877	}
		878	}
		879	}
		880	};
		881	};
		882	};
		883	};
		884	};
		885
		886	$grp
		887	}
		888
		889	=item aio_rmtree $path, $callback->($status)
		890
		891	Delete a directory tree starting (and including) C<$path>, return the
		892	status of the final C<rmdir> only. This is a composite request that
		893	uses C<aio_scandir> to recurse into and rmdir directories, and unlink
		894	everything else.
		895
		896	=cut
		897
		898	sub aio_rmtree;
		899	sub aio_rmtree($;$) {
		900	my ($path, $cb) = @_;
		901
		902	my $pri = aioreq_pri;
		903	my $grp = aio_group $cb;
		904
		905	aioreq_pri $pri;
		906	add $grp aio_scandir $path, 0, sub {
		907	my ($dirs, $nondirs) = @_;
		908
		909	my $dirgrp = aio_group sub {
		910	add $grp aio_rmdir $path, sub {
		911	$grp->result ($_[0]);
		912	};
		913	};
		914
		915	(aioreq_pri $pri), add $dirgrp aio_rmtree "$path/$_" for @$dirs;
		916	(aioreq_pri $pri), add $dirgrp aio_unlink "$path/$_" for @$nondirs;
		917
		918	add $grp $dirgrp;
		919	};
		920
		921	$grp
		922	}
		923
		924	=item aio_sync $callback->($status)
		925
		926	Asynchronously call sync and call the callback when finished.
		927
209	=item aio_fsync $fh, $callback	928	=item aio_fsync $fh, $callback->($status)
210		929
211	Asynchronously call fsync on the given filehandle and call the callback	930	Asynchronously call fsync on the given filehandle and call the callback
212	with the fsync result code.	931	with the fsync result code.
213		932
214	=item aio_fdatasync $fh, $callback	933	=item aio_fdatasync $fh, $callback->($status)
215		934
216	Asynchronously call fdatasync on the given filehandle and call the	935	Asynchronously call fdatasync on the given filehandle and call the
217	callback with the fdatasync result code.	936	callback with the fdatasync result code.
218		937
		938	If this call isn't available because your OS lacks it or it couldn't be
		939	detected, it will be emulated by calling C<fsync> instead.
		940
		941	=item aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
		942
		943	Sync the data portion of the file specified by C<$offset> and C<$length>
		944	to disk (but NOT the metadata), by calling the Linux-specific
		945	sync_file_range call. If sync_file_range is not available or it returns
		946	ENOSYS, then fdatasync or fsync is being substituted.
		947
		948	C<$flags> can be a combination of C<IO::AIO::SYNC_FILE_RANGE_WAIT_BEFORE>,
		949	C<IO::AIO::SYNC_FILE_RANGE_WRITE> and
		950	C<IO::AIO::SYNC_FILE_RANGE_WAIT_AFTER>: refer to the sync_file_range
		951	manpage for details.
		952
		953	=item aio_pathsync $path, $callback->($status)
		954
		955	This request tries to open, fsync and close the given path. This is a
		956	composite request intended to sync directories after directory operations
		957	(E.g. rename). This might not work on all operating systems or have any
		958	specific effect, but usually it makes sure that directory changes get
		959	written to disc. It works for anything that can be opened for read-only,
		960	not just directories.
		961
		962	Future versions of this function might fall back to other methods when
		963	C<fsync> on the directory fails (such as calling C<sync>).
		964
		965	Passes C<0> when everything went ok, and C<-1> on error.
		966
219	=cut	967	=cut
220		968
221	min_parallel 4;	969	sub aio_pathsync($;$) {
		970	my ($path, $cb) = @_;
222		971
223	END {	972	my $pri = aioreq_pri;
224	max_parallel 0;	973	my $grp = aio_group $cb;
		974
		975	aioreq_pri $pri;
		976	add $grp aio_open $path, O_RDONLY, 0, sub {
		977	my ($fh) = @_;
		978	if ($fh) {
		979	aioreq_pri $pri;
		980	add $grp aio_fsync $fh, sub {
		981	$grp->result ($_[0]);
		982
		983	aioreq_pri $pri;
		984	add $grp aio_close $fh;
		985	};
		986	} else {
		987	$grp->result (-1);
		988	}
		989	};
		990
		991	$grp
225	}	992	}
226		993
		994	=item aio_group $callback->(...)
		995
		996	This is a very special aio request: Instead of doing something, it is a
		997	container for other aio requests, which is useful if you want to bundle
		998	many requests into a single, composite, request with a definite callback
		999	and the ability to cancel the whole request with its subrequests.
		1000
		1001	Returns an object of class L<IO::AIO::GRP>. See its documentation below
		1002	for more info.
		1003
		1004	Example:
		1005
		1006	my $grp = aio_group sub {
		1007	print "all stats done\n";
		1008	};
		1009
		1010	add $grp
		1011	(aio_stat ...),
		1012	(aio_stat ...),
		1013	...;
		1014
		1015	=item aio_nop $callback->()
		1016
		1017	This is a special request - it does nothing in itself and is only used for
		1018	side effects, such as when you want to add a dummy request to a group so
		1019	that finishing the requests in the group depends on executing the given
		1020	code.
		1021
		1022	While this request does nothing, it still goes through the execution
		1023	phase and still requires a worker thread. Thus, the callback will not
		1024	be executed immediately but only after other requests in the queue have
		1025	entered their execution phase. This can be used to measure request
		1026	latency.
		1027
		1028	=item IO::AIO::aio_busy $fractional_seconds, $callback->() *NOT EXPORTED*
		1029
		1030	Mainly used for debugging and benchmarking, this aio request puts one of
		1031	the request workers to sleep for the given time.
		1032
		1033	While it is theoretically handy to have simple I/O scheduling requests
		1034	like sleep and file handle readable/writable, the overhead this creates is
		1035	immense (it blocks a thread for a long time) so do not use this function
		1036	except to put your application under artificial I/O pressure.
		1037
		1038	=back
		1039
		1040	=head2 IO::AIO::REQ CLASS
		1041
		1042	All non-aggregate C<aio_*> functions return an object of this class when
		1043	called in non-void context.
		1044
		1045	=over 4
		1046
		1047	=item cancel $req
		1048
		1049	Cancels the request, if possible. Has the effect of skipping execution
		1050	when entering the B<execute> state and skipping calling the callback when
		1051	entering the the B<result> state, but will leave the request otherwise
		1052	untouched (with the exception of readdir). That means that requests that
		1053	currently execute will not be stopped and resources held by the request
		1054	will not be freed prematurely.
		1055
		1056	=item cb $req $callback->(...)
		1057
		1058	Replace (or simply set) the callback registered to the request.
		1059
		1060	=back
		1061
		1062	=head2 IO::AIO::GRP CLASS
		1063
		1064	This class is a subclass of L<IO::AIO::REQ>, so all its methods apply to
		1065	objects of this class, too.
		1066
		1067	A IO::AIO::GRP object is a special request that can contain multiple other
		1068	aio requests.
		1069
		1070	You create one by calling the C<aio_group> constructing function with a
		1071	callback that will be called when all contained requests have entered the
		1072	C<done> state:
		1073
		1074	my $grp = aio_group sub {
		1075	print "all requests are done\n";
		1076	};
		1077
		1078	You add requests by calling the C<add> method with one or more
		1079	C<IO::AIO::REQ> objects:
		1080
		1081	$grp->add (aio_unlink "...");
		1082
		1083	add $grp aio_stat "...", sub {
		1084	$_[0] or return $grp->result ("error");
		1085
		1086	# add another request dynamically, if first succeeded
		1087	add $grp aio_open "...", sub {
		1088	$grp->result ("ok");
		1089	};
		1090	};
		1091
		1092	This makes it very easy to create composite requests (see the source of
		1093	C<aio_move> for an application) that work and feel like simple requests.
		1094
		1095	=over 4
		1096
		1097	=item * The IO::AIO::GRP objects will be cleaned up during calls to
		1098	C<IO::AIO::poll_cb>, just like any other request.
		1099
		1100	=item * They can be canceled like any other request. Canceling will cancel not
		1101	only the request itself, but also all requests it contains.
		1102
		1103	=item * They can also can also be added to other IO::AIO::GRP objects.
		1104
		1105	=item * You must not add requests to a group from within the group callback (or
		1106	any later time).
		1107
		1108	=back
		1109
		1110	Their lifetime, simplified, looks like this: when they are empty, they
		1111	will finish very quickly. If they contain only requests that are in the
		1112	C<done> state, they will also finish. Otherwise they will continue to
		1113	exist.
		1114
		1115	That means after creating a group you have some time to add requests
		1116	(precisely before the callback has been invoked, which is only done within
		1117	the C<poll_cb>). And in the callbacks of those requests, you can add
		1118	further requests to the group. And only when all those requests have
		1119	finished will the the group itself finish.
		1120
		1121	=over 4
		1122
		1123	=item add $grp ...
		1124
		1125	=item $grp->add (...)
		1126
		1127	Add one or more requests to the group. Any type of L<IO::AIO::REQ> can
		1128	be added, including other groups, as long as you do not create circular
		1129	dependencies.
		1130
		1131	Returns all its arguments.
		1132
		1133	=item $grp->cancel_subs
		1134
		1135	Cancel all subrequests and clears any feeder, but not the group request
		1136	itself. Useful when you queued a lot of events but got a result early.
		1137
		1138	=item $grp->result (...)
		1139
		1140	Set the result value(s) that will be passed to the group callback when all
		1141	subrequests have finished and set the groups errno to the current value
		1142	of errno (just like calling C<errno> without an error number). By default,
		1143	no argument will be passed and errno is zero.
		1144
		1145	=item $grp->errno ([$errno])
		1146
		1147	Sets the group errno value to C<$errno>, or the current value of errno
		1148	when the argument is missing.
		1149
		1150	Every aio request has an associated errno value that is restored when
		1151	the callback is invoked. This method lets you change this value from its
		1152	default (0).
		1153
		1154	Calling C<result> will also set errno, so make sure you either set C<$!>
		1155	before the call to C<result>, or call c<errno> after it.
		1156
		1157	=item feed $grp $callback->($grp)
		1158
		1159	Sets a feeder/generator on this group: every group can have an attached
		1160	generator that generates requests if idle. The idea behind this is that,
		1161	although you could just queue as many requests as you want in a group,
		1162	this might starve other requests for a potentially long time. For example,
		1163	C<aio_scandir> might generate hundreds of thousands C<aio_stat> requests,
		1164	delaying any later requests for a long time.
		1165
		1166	To avoid this, and allow incremental generation of requests, you can
		1167	instead a group and set a feeder on it that generates those requests. The
		1168	feed callback will be called whenever there are few enough (see C<limit>,
		1169	below) requests active in the group itself and is expected to queue more
		1170	requests.
		1171
		1172	The feed callback can queue as many requests as it likes (i.e. C<add> does
		1173	not impose any limits).
		1174
		1175	If the feed does not queue more requests when called, it will be
		1176	automatically removed from the group.
		1177
		1178	If the feed limit is C<0> when this method is called, it will be set to
		1179	C<2> automatically.
		1180
		1181	Example:
		1182
		1183	# stat all files in @files, but only ever use four aio requests concurrently:
		1184
		1185	my $grp = aio_group sub { print "finished\n" };
		1186	limit $grp 4;
		1187	feed $grp sub {
		1188	my $file = pop @files
		1189	or return;
		1190
		1191	add $grp aio_stat $file, sub { ... };
		1192	};
		1193
		1194	=item limit $grp $num
		1195
		1196	Sets the feeder limit for the group: The feeder will be called whenever
		1197	the group contains less than this many requests.
		1198
		1199	Setting the limit to C<0> will pause the feeding process.
		1200
		1201	The default value for the limit is C<0>, but note that setting a feeder
		1202	automatically bumps it up to C<2>.
		1203
		1204	=back
		1205
		1206	=head2 SUPPORT FUNCTIONS
		1207
		1208	=head3 EVENT PROCESSING AND EVENT LOOP INTEGRATION
		1209
		1210	=over 4
		1211
		1212	=item $fileno = IO::AIO::poll_fileno
		1213
		1214	Return the I<request result pipe file descriptor>. This filehandle must be
		1215	polled for reading by some mechanism outside this module (e.g. EV, Glib,
		1216	select and so on, see below or the SYNOPSIS). If the pipe becomes readable
		1217	you have to call C<poll_cb> to check the results.
		1218
		1219	See C<poll_cb> for an example.
		1220
		1221	=item IO::AIO::poll_cb
		1222
		1223	Process some outstanding events on the result pipe. You have to call this
		1224	regularly. Returns C<0> if all events could be processed, or C<-1> if it
		1225	returned earlier for whatever reason. Returns immediately when no events
		1226	are outstanding. The amount of events processed depends on the settings of
		1227	C<IO::AIO::max_poll_req> and C<IO::AIO::max_poll_time>.
		1228
		1229	If not all requests were processed for whatever reason, the filehandle
		1230	will still be ready when C<poll_cb> returns, so normally you don't have to
		1231	do anything special to have it called later.
		1232
		1233	Example: Install an Event watcher that automatically calls
		1234	IO::AIO::poll_cb with high priority (more examples can be found in the
		1235	SYNOPSIS section, at the top of this document):
		1236
		1237	Event->io (fd => IO::AIO::poll_fileno,
		1238	poll => 'r', async => 1,
		1239	cb => \&IO::AIO::poll_cb);
		1240
		1241	=item IO::AIO::max_poll_reqs $nreqs
		1242
		1243	=item IO::AIO::max_poll_time $seconds
		1244
		1245	These set the maximum number of requests (default C<0>, meaning infinity)
		1246	that are being processed by C<IO::AIO::poll_cb> in one call, respectively
		1247	the maximum amount of time (default C<0>, meaning infinity) spent in
		1248	C<IO::AIO::poll_cb> to process requests (more correctly the mininum amount
		1249	of time C<poll_cb> is allowed to use).
		1250
		1251	Setting C<max_poll_time> to a non-zero value creates an overhead of one
		1252	syscall per request processed, which is not normally a problem unless your
		1253	callbacks are really really fast or your OS is really really slow (I am
		1254	not mentioning Solaris here). Using C<max_poll_reqs> incurs no overhead.
		1255
		1256	Setting these is useful if you want to ensure some level of
		1257	interactiveness when perl is not fast enough to process all requests in
		1258	time.
		1259
		1260	For interactive programs, values such as C<0.01> to C<0.1> should be fine.
		1261
		1262	Example: Install an Event watcher that automatically calls
		1263	IO::AIO::poll_cb with low priority, to ensure that other parts of the
		1264	program get the CPU sometimes even under high AIO load.
		1265
		1266	# try not to spend much more than 0.1s in poll_cb
		1267	IO::AIO::max_poll_time 0.1;
		1268
		1269	# use a low priority so other tasks have priority
		1270	Event->io (fd => IO::AIO::poll_fileno,
		1271	poll => 'r', nice => 1,
		1272	cb => &IO::AIO::poll_cb);
		1273
		1274	=item IO::AIO::poll_wait
		1275
		1276	If there are any outstanding requests and none of them in the result
		1277	phase, wait till the result filehandle becomes ready for reading (simply
		1278	does a C<select> on the filehandle. This is useful if you want to
		1279	synchronously wait for some requests to finish).
		1280
		1281	See C<nreqs> for an example.
		1282
		1283	=item IO::AIO::poll
		1284
		1285	Waits until some requests have been handled.
		1286
		1287	Returns the number of requests processed, but is otherwise strictly
		1288	equivalent to:
		1289
		1290	IO::AIO::poll_wait, IO::AIO::poll_cb
		1291
		1292	=item IO::AIO::flush
		1293
		1294	Wait till all outstanding AIO requests have been handled.
		1295
		1296	Strictly equivalent to:
		1297
		1298	IO::AIO::poll_wait, IO::AIO::poll_cb
		1299	while IO::AIO::nreqs;
		1300
		1301	=back
		1302
		1303	=head3 CONTROLLING THE NUMBER OF THREADS
		1304
		1305	=over
		1306
		1307	=item IO::AIO::min_parallel $nthreads
		1308
		1309	Set the minimum number of AIO threads to C<$nthreads>. The current
		1310	default is C<8>, which means eight asynchronous operations can execute
		1311	concurrently at any one time (the number of outstanding requests,
		1312	however, is unlimited).
		1313
		1314	IO::AIO starts threads only on demand, when an AIO request is queued and
		1315	no free thread exists. Please note that queueing up a hundred requests can
		1316	create demand for a hundred threads, even if it turns out that everything
		1317	is in the cache and could have been processed faster by a single thread.
		1318
		1319	It is recommended to keep the number of threads relatively low, as some
		1320	Linux kernel versions will scale negatively with the number of threads
		1321	(higher parallelity => MUCH higher latency). With current Linux 2.6
		1322	versions, 4-32 threads should be fine.
		1323
		1324	Under most circumstances you don't need to call this function, as the
		1325	module selects a default that is suitable for low to moderate load.
		1326
		1327	=item IO::AIO::max_parallel $nthreads
		1328
		1329	Sets the maximum number of AIO threads to C<$nthreads>. If more than the
		1330	specified number of threads are currently running, this function kills
		1331	them. This function blocks until the limit is reached.
		1332
		1333	While C<$nthreads> are zero, aio requests get queued but not executed
		1334	until the number of threads has been increased again.
		1335
		1336	This module automatically runs C<max_parallel 0> at program end, to ensure
		1337	that all threads are killed and that there are no outstanding requests.
		1338
		1339	Under normal circumstances you don't need to call this function.
		1340
		1341	=item IO::AIO::max_idle $nthreads
		1342
		1343	Limit the number of threads (default: 4) that are allowed to idle (i.e.,
		1344	threads that did not get a request to process within 10 seconds). That
		1345	means if a thread becomes idle while C<$nthreads> other threads are also
		1346	idle, it will free its resources and exit.
		1347
		1348	This is useful when you allow a large number of threads (e.g. 100 or 1000)
		1349	to allow for extremely high load situations, but want to free resources
		1350	under normal circumstances (1000 threads can easily consume 30MB of RAM).
		1351
		1352	The default is probably ok in most situations, especially if thread
		1353	creation is fast. If thread creation is very slow on your system you might
		1354	want to use larger values.
		1355
		1356	=item IO::AIO::max_outstanding $maxreqs
		1357
		1358	This is a very bad function to use in interactive programs because it
		1359	blocks, and a bad way to reduce concurrency because it is inexact: Better
		1360	use an C<aio_group> together with a feed callback.
		1361
		1362	Sets the maximum number of outstanding requests to C<$nreqs>. If you
		1363	do queue up more than this number of requests, the next call to the
		1364	C<poll_cb> (and C<poll_some> and other functions calling C<poll_cb>)
		1365	function will block until the limit is no longer exceeded.
		1366
		1367	The default value is very large, so there is no practical limit on the
		1368	number of outstanding requests.
		1369
		1370	You can still queue as many requests as you want. Therefore,
		1371	C<max_outstanding> is mainly useful in simple scripts (with low values) or
		1372	as a stop gap to shield against fatal memory overflow (with large values).
		1373
		1374	=back
		1375
		1376	=head3 STATISTICAL INFORMATION
		1377
		1378	=over
		1379
		1380	=item IO::AIO::nreqs
		1381
		1382	Returns the number of requests currently in the ready, execute or pending
		1383	states (i.e. for which their callback has not been invoked yet).
		1384
		1385	Example: wait till there are no outstanding requests anymore:
		1386
		1387	IO::AIO::poll_wait, IO::AIO::poll_cb
		1388	while IO::AIO::nreqs;
		1389
		1390	=item IO::AIO::nready
		1391
		1392	Returns the number of requests currently in the ready state (not yet
		1393	executed).
		1394
		1395	=item IO::AIO::npending
		1396
		1397	Returns the number of requests currently in the pending state (executed,
		1398	but not yet processed by poll_cb).
		1399
		1400	=back
		1401
		1402	=head3 MISCELLANEOUS FUNCTIONS
		1403
		1404	IO::AIO implements some functions that might be useful, but are not
		1405	asynchronous.
		1406
		1407	=over 4
		1408
		1409	=item IO::AIO::sendfile $ofh, $ifh, $offset, $count
		1410
		1411	Calls the C<eio_sendfile_sync> function, which is like C<aio_sendfile>,
		1412	but is blocking (this makes most sense if you know the input data is
		1413	likely cached already and the output filehandle is set to non-blocking
		1414	operations).
		1415
		1416	Returns the number of bytes copied, or C<-1> on error.
		1417
		1418	=item IO::AIO::fadvise $fh, $offset, $len, $advice
		1419
		1420	Simply calls the C<posix_fadvise> function (see it's
		1421	manpage for details). The following advice constants are
		1422	avaiable: C<IO::AIO::FADV_NORMAL>, C<IO::AIO::FADV_SEQUENTIAL>,
		1423	C<IO::AIO::FADV_RANDOM>, C<IO::AIO::FADV_NOREUSE>,
		1424	C<IO::AIO::FADV_WILLNEED>, C<IO::AIO::FADV_DONTNEED>.
		1425
		1426	On systems that do not implement C<posix_fadvise>, this function returns
		1427	ENOSYS, otherwise the return value of C<posix_fadvise>.
		1428
		1429	=back
		1430
		1431	=cut
		1432
		1433	min_parallel 8;
		1434
		1435	END { flush }
		1436
227	1;	1437	1;
228		1438
229	=back	1439	=head2 FORK BEHAVIOUR
230		1440
		1441	This module should do "the right thing" when the process using it forks:
		1442
		1443	Before the fork, IO::AIO enters a quiescent state where no requests
		1444	can be added in other threads and no results will be processed. After
		1445	the fork the parent simply leaves the quiescent state and continues
		1446	request/result processing, while the child frees the request/result queue
		1447	(so that the requests started before the fork will only be handled in the
		1448	parent). Threads will be started on demand until the limit set in the
		1449	parent process has been reached again.
		1450
		1451	In short: the parent will, after a short pause, continue as if fork had
		1452	not been called, while the child will act as if IO::AIO has not been used
		1453	yet.
		1454
		1455	=head2 MEMORY USAGE
		1456
		1457	Per-request usage:
		1458
		1459	Each aio request uses - depending on your architecture - around 100-200
		1460	bytes of memory. In addition, stat requests need a stat buffer (possibly
		1461	a few hundred bytes), readdir requires a result buffer and so on. Perl
		1462	scalars and other data passed into aio requests will also be locked and
		1463	will consume memory till the request has entered the done state.
		1464
		1465	This is not awfully much, so queuing lots of requests is not usually a
		1466	problem.
		1467
		1468	Per-thread usage:
		1469
		1470	In the execution phase, some aio requests require more memory for
		1471	temporary buffers, and each thread requires a stack and other data
		1472	structures (usually around 16k-128k, depending on the OS).
		1473
231	=head1 BUGS	1474	=head1 KNOWN BUGS
232		1475
233	- aio_open gives a fd, but all other functions expect a perl filehandle.	1476	Known bugs will be fixed in the next release.
234		1477
235	=head1 SEE ALSO	1478	=head1 SEE ALSO
236		1479
237	L<Coro>, L<Linux::AIO>.	1480	L<AnyEvent::AIO> for easy integration into event loops, L<Coro::AIO> for a
		1481	more natural syntax.
238		1482
239	=head1 AUTHOR	1483	=head1 AUTHOR
240		1484
241	Marc Lehmann <schmorp@schmorp.de>	1485	Marc Lehmann <schmorp@schmorp.de>
242	http://home.schmorp.de/	1486	http://home.schmorp.de/

Diff Legend

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