ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/IO-AIO/AIO.pm

Comparing IO-AIO/AIO.pm (file contents):
Revision 1.39 by root, Sun Aug 28 11:05:50 2005 UTC vs.
Revision 1.278 by root, Sun Oct 1 07:24:34 2017 UTC

1	=head1 NAME	1	=head1 NAME
2		2
3	IO::AIO - Asynchronous Input/Output	3	IO::AIO - Asynchronous/Advanced Input/Output
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 {	9	aio_open "/etc/passwd", IO::AIO::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
21	Event->io (fd => IO::AIO::poll_fileno,	22	use IO::AIO 2;
22	poll => 'r',
23	cb => \&IO::AIO::poll_cb);
24		23
25	# Glib/Gtk2	24	aioreq_pri 4; # give next request a very high priority
26	add_watch Glib::IO IO::AIO::poll_fileno,	25	my $req = aio_unlink "/tmp/file", sub { };
27	in => sub { IO::AIO::poll_cb; 1 };	26	$req->cancel; # cancel request if still in queue
28		27
29	# Tk	28	my $grp = aio_group sub { print "all stats done\n" };
30	Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",	29	add $grp aio_stat "..." for ...;
31	readable => \&IO::AIO::poll_cb);
32
33	# Danga::Socket
34	Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
35	\&IO::AIO::poll_cb);
36
37		30
38	=head1 DESCRIPTION	31	=head1 DESCRIPTION
39		32
40	This module implements asynchronous I/O using whatever means your	33	This module implements asynchronous I/O using whatever means your
41	operating system supports.	34	operating system supports. It is implemented as an interface to C<libeio>
		35	(L<http://software.schmorp.de/pkg/libeio.html>).
42		36
		37	Asynchronous means that operations that can normally block your program
		38	(e.g. reading from disk) will be done asynchronously: the operation
		39	will still block, but you can do something else in the meantime. This
		40	is extremely useful for programs that need to stay interactive even
		41	when doing heavy I/O (GUI programs, high performance network servers
		42	etc.), but can also be used to easily do operations in parallel that are
		43	normally done sequentially, e.g. stat'ing many files, which is much faster
		44	on a RAID volume or over NFS when you do a number of stat operations
		45	concurrently.
		46
		47	While most of this works on all types of file descriptors (for
		48	example sockets), using these functions on file descriptors that
		49	support nonblocking operation (again, sockets, pipes etc.) is
		50	very inefficient. Use an event loop for that (such as the L<EV>
		51	module): IO::AIO will naturally fit into such an event loop itself.
		52
43	Currently, a number of threads are started that execute your read/writes	53	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	54	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	55	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	56	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	57	functions available on many operating systems. However, they are often
48	not well-supported (Linux doesn't allow them on normal files currently,	58	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	59	files currently, for example), and they would only support aio_read and
50	remaining functionality would have to be implemented using threads anyway.	60	aio_write, so the remaining functionality would have to be implemented
		61	using threads anyway.
51		62
		63	In addition to asynchronous I/O, this module also exports some rather
		64	arcane interfaces, such as C<madvise> or linux's C<splice> system call,
		65	which is why the C<A> in C<AIO> can also mean I<advanced>.
		66
52	Although the module will work with in the presence of other threads, it is	67	Although the module will work in the presence of other (Perl-) threads,
53	currently not reentrant, so use appropriate locking yourself, always call	68	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	69	yourself, always call C<poll_cb> from within the same thread, or never
55	C<aio_> functions) recursively.	70	call C<poll_cb> (or other C<aio_> functions) recursively.
		71
		72	=head2 EXAMPLE
		73
		74	This is a simple example that uses the EV module and loads
		75	F</etc/passwd> asynchronously:
		76
		77	use EV;
		78	use IO::AIO;
		79
		80	# register the IO::AIO callback with EV
		81	my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
		82
		83	# queue the request to open /etc/passwd
		84	aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub {
		85	my $fh = shift
		86	or die "error while opening: $!";
		87
		88	# stat'ing filehandles is generally non-blocking
		89	my $size = -s $fh;
		90
		91	# queue a request to read the file
		92	my $contents;
		93	aio_read $fh, 0, $size, $contents, 0, sub {
		94	$_[0] == $size
		95	or die "short read: $!";
		96
		97	close $fh;
		98
		99	# file contents now in $contents
		100	print $contents;
		101
		102	# exit event loop and program
		103	EV::break;
		104	};
		105	};
		106
		107	# possibly queue up other requests, or open GUI windows,
		108	# check for sockets etc. etc.
		109
		110	# process events as long as there are some:
		111	EV::run;
		112
		113	=head1 REQUEST ANATOMY AND LIFETIME
		114
		115	Every C<aio_*> function creates a request. which is a C data structure not
		116	directly visible to Perl.
		117
		118	If called in non-void context, every request function returns a Perl
		119	object representing the request. In void context, nothing is returned,
		120	which saves a bit of memory.
		121
		122	The perl object is a fairly standard ref-to-hash object. The hash contents
		123	are not used by IO::AIO so you are free to store anything you like in it.
		124
		125	During their existance, aio requests travel through the following states,
		126	in order:
		127
		128	=over 4
		129
		130	=item ready
		131
		132	Immediately after a request is created it is put into the ready state,
		133	waiting for a thread to execute it.
		134
		135	=item execute
		136
		137	A thread has accepted the request for processing and is currently
		138	executing it (e.g. blocking in read).
		139
		140	=item pending
		141
		142	The request has been executed and is waiting for result processing.
		143
		144	While request submission and execution is fully asynchronous, result
		145	processing is not and relies on the perl interpreter calling C<poll_cb>
		146	(or another function with the same effect).
		147
		148	=item result
		149
		150	The request results are processed synchronously by C<poll_cb>.
		151
		152	The C<poll_cb> function will process all outstanding aio requests by
		153	calling their callbacks, freeing memory associated with them and managing
		154	any groups they are contained in.
		155
		156	=item done
		157
		158	Request has reached the end of its lifetime and holds no resources anymore
		159	(except possibly for the Perl object, but its connection to the actual
		160	aio request is severed and calling its methods will either do nothing or
		161	result in a runtime error).
		162
		163	=back
56		164
57	=cut	165	=cut
58		166
59	package IO::AIO;	167	package IO::AIO;
60		168
61	no warnings;	169	use Carp ();
		170
		171	use common::sense;
62		172
63	use base 'Exporter';	173	use base 'Exporter';
64		174
65	use Fcntl ();
66
67	BEGIN {	175	BEGIN {
68	$VERSION = 1.6;	176	our $VERSION = 4.35;
69		177
70	@EXPORT = qw(aio_sendfile aio_read aio_write aio_open aio_close aio_stat	178	our @AIO_REQ = qw(aio_sendfile aio_seek aio_read aio_write aio_open aio_close
71	aio_lstat aio_unlink aio_rmdir aio_readdir aio_symlink	179	aio_stat aio_lstat aio_unlink aio_rmdir aio_readdir aio_readdirx
72	aio_fsync aio_fdatasync aio_readahead);	180	aio_scandir aio_symlink aio_readlink aio_realpath aio_fcntl aio_ioctl
73	@EXPORT_OK = qw(poll_fileno poll_cb min_parallel max_parallel	181	aio_sync aio_fsync aio_syncfs aio_fdatasync aio_sync_file_range
74	max_outstanding nreqs);	182	aio_pathsync aio_readahead aio_fiemap aio_allocate
		183	aio_rename aio_rename2 aio_link aio_move aio_copy aio_group
		184	aio_nop aio_mknod aio_load aio_rmtree aio_mkdir aio_chown
		185	aio_chmod aio_utime aio_truncate
		186	aio_msync aio_mtouch aio_mlock aio_mlockall
		187	aio_statvfs
		188	aio_wd);
		189
		190	our @EXPORT = (@AIO_REQ, qw(aioreq_pri aioreq_nice));
		191	our @EXPORT_OK = qw(poll_fileno poll_cb poll_wait flush
		192	min_parallel max_parallel max_idle idle_timeout
		193	nreqs nready npending nthreads
		194	max_poll_time max_poll_reqs
		195	sendfile fadvise madvise
		196	mmap munmap munlock munlockall);
		197
		198	push @AIO_REQ, qw(aio_busy); # not exported
		199
		200	@IO::AIO::GRP::ISA = 'IO::AIO::REQ';
75		201
76	require XSLoader;	202	require XSLoader;
77	XSLoader::load IO::AIO, $VERSION;	203	XSLoader::load ("IO::AIO", $VERSION);
78	}	204	}
79		205
80	=head1 FUNCTIONS	206	=head1 FUNCTIONS
81		207
82	=head2 AIO FUNCTIONS	208	=head2 QUICK OVERVIEW
		209
		210	This section simply lists the prototypes most of the functions for
		211	quick reference. See the following sections for function-by-function
		212	documentation.
		213
		214	aio_wd $pathname, $callback->($wd)
		215	aio_open $pathname, $flags, $mode, $callback->($fh)
		216	aio_close $fh, $callback->($status)
		217	aio_seek $fh,$offset,$whence, $callback->($offs)
		218	aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
		219	aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
		220	aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
		221	aio_readahead $fh,$offset,$length, $callback->($retval)
		222	aio_stat $fh_or_path, $callback->($status)
		223	aio_lstat $fh, $callback->($status)
		224	aio_statvfs $fh_or_path, $callback->($statvfs)
		225	aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
		226	aio_chown $fh_or_path, $uid, $gid, $callback->($status)
		227	aio_chmod $fh_or_path, $mode, $callback->($status)
		228	aio_truncate $fh_or_path, $offset, $callback->($status)
		229	aio_allocate $fh, $mode, $offset, $len, $callback->($status)
		230	aio_fiemap $fh, $start, $length, $flags, $count, $cb->(\@extents)
		231	aio_unlink $pathname, $callback->($status)
		232	aio_mknod $pathname, $mode, $dev, $callback->($status)
		233	aio_link $srcpath, $dstpath, $callback->($status)
		234	aio_symlink $srcpath, $dstpath, $callback->($status)
		235	aio_readlink $pathname, $callback->($link)
		236	aio_realpath $pathname, $callback->($path)
		237	aio_rename $srcpath, $dstpath, $callback->($status)
		238	aio_rename2 $srcpath, $dstpath, $flags, $callback->($status)
		239	aio_mkdir $pathname, $mode, $callback->($status)
		240	aio_rmdir $pathname, $callback->($status)
		241	aio_readdir $pathname, $callback->($entries)
		242	aio_readdirx $pathname, $flags, $callback->($entries, $flags)
		243	IO::AIO::READDIR_DENTS IO::AIO::READDIR_DIRS_FIRST
		244	IO::AIO::READDIR_STAT_ORDER IO::AIO::READDIR_FOUND_UNKNOWN
		245	aio_scandir $pathname, $maxreq, $callback->($dirs, $nondirs)
		246	aio_load $pathname, $data, $callback->($status)
		247	aio_copy $srcpath, $dstpath, $callback->($status)
		248	aio_move $srcpath, $dstpath, $callback->($status)
		249	aio_rmtree $pathname, $callback->($status)
		250	aio_fcntl $fh, $cmd, $arg, $callback->($status)
		251	aio_ioctl $fh, $request, $buf, $callback->($status)
		252	aio_sync $callback->($status)
		253	aio_syncfs $fh, $callback->($status)
		254	aio_fsync $fh, $callback->($status)
		255	aio_fdatasync $fh, $callback->($status)
		256	aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
		257	aio_pathsync $pathname, $callback->($status)
		258	aio_msync $scalar, $offset = 0, $length = undef, flags = MS_SYNC, $callback->($status)
		259	aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		260	aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status)
		261	aio_mlockall $flags, $callback->($status)
		262	aio_group $callback->(...)
		263	aio_nop $callback->()
		264
		265	$prev_pri = aioreq_pri [$pri]
		266	aioreq_nice $pri_adjust
		267
		268	IO::AIO::poll_wait
		269	IO::AIO::poll_cb
		270	IO::AIO::poll
		271	IO::AIO::flush
		272	IO::AIO::max_poll_reqs $nreqs
		273	IO::AIO::max_poll_time $seconds
		274	IO::AIO::min_parallel $nthreads
		275	IO::AIO::max_parallel $nthreads
		276	IO::AIO::max_idle $nthreads
		277	IO::AIO::idle_timeout $seconds
		278	IO::AIO::max_outstanding $maxreqs
		279	IO::AIO::nreqs
		280	IO::AIO::nready
		281	IO::AIO::npending
		282	$nfd = IO::AIO::get_fdlimit [EXPERIMENTAL]
		283	IO::AIO::min_fdlimit $nfd [EXPERIMENTAL]
		284
		285	IO::AIO::sendfile $ofh, $ifh, $offset, $count
		286	IO::AIO::fadvise $fh, $offset, $len, $advice
		287	IO::AIO::mmap $scalar, $length, $prot, $flags[, $fh[, $offset]]
		288	IO::AIO::munmap $scalar
		289	IO::AIO::madvise $scalar, $offset, $length, $advice
		290	IO::AIO::mprotect $scalar, $offset, $length, $protect
		291	IO::AIO::munlock $scalar, $offset = 0, $length = undef
		292	IO::AIO::munlockall
		293
		294	=head2 API NOTES
83		295
84	All the C<aio_*> calls are more or less thin wrappers around the syscall	296	All the C<aio_*> calls are more or less thin wrappers around the syscall
85	with the same name (sans C<aio_>). The arguments are similar or identical,	297	with the same name (sans C<aio_>). The arguments are similar or identical,
86	and they all accept an additional (and optional) C<$callback> argument	298	and they all accept an additional (and optional) C<$callback> argument
87	which must be a code reference. This code reference will get called with	299	which must be a code reference. This code reference will be called after
88	the syscall return code (e.g. most syscalls return C<-1> on error, unlike	300	the syscall has been executed in an asynchronous fashion. The results
89	perl, which usually delivers "false") as it's sole argument when the given	301	of the request will be passed as arguments to the callback (and, if an
90	syscall has been executed asynchronously.	302	error occured, in C<$!>) - for most requests the syscall return code (e.g.
		303	most syscalls return C<-1> on error, unlike perl, which usually delivers
		304	"false").
		305
		306	Some requests (such as C<aio_readdir>) pass the actual results and
		307	communicate failures by passing C<undef>.
91		308
92	All functions expecting a filehandle keep a copy of the filehandle	309	All functions expecting a filehandle keep a copy of the filehandle
93	internally until the request has finished.	310	internally until the request has finished.
94		311
		312	All functions return request objects of type L<IO::AIO::REQ> that allow
		313	further manipulation of those requests while they are in-flight.
		314
95	The pathnames you pass to these routines I<must> be absolute and	315	The pathnames you pass to these routines I<should> be absolute. The
96	encoded in byte form. The reason for the former is that at the time the	316	reason for this is that at the time the request is being executed, the
97	request is being executed, the current working directory could have	317	current working directory could have changed. Alternatively, you can
98	changed. Alternatively, you can make sure that you never change the	318	make sure that you never change the current working directory anywhere
99	current working directory.	319	in the program and then use relative paths. You can also take advantage
		320	of IO::AIOs working directory abstraction, that lets you specify paths
		321	relative to some previously-opened "working directory object" - see the
		322	description of the C<IO::AIO::WD> class later in this document.
100		323
101	To encode pathnames to byte form, either make sure you either: a)	324	To encode pathnames as octets, either make sure you either: a) always pass
102	always pass in filenames you got from outside (command line, readdir	325	in filenames you got from outside (command line, readdir etc.) without
103	etc.), b) are ASCII or ISO 8859-1, c) use the Encode module and encode	326	tinkering, b) are in your native filesystem encoding, c) use the Encode
104	your pathnames to the locale (or other) encoding in effect in the user	327	module and encode your pathnames to the locale (or other) encoding in
105	environment, d) use Glib::filename_from_unicode on unicode filenames or e)	328	effect in the user environment, d) use Glib::filename_from_unicode on
106	use something else.	329	unicode filenames or e) use something else to ensure your scalar has the
		330	correct contents.
		331
		332	This works, btw. independent of the internal UTF-8 bit, which IO::AIO
		333	handles correctly whether it is set or not.
		334
		335	=head2 AIO REQUEST FUNCTIONS
107		336
108	=over 4	337	=over 4
109		338
		339	=item $prev_pri = aioreq_pri [$pri]
		340
		341	Returns the priority value that would be used for the next request and, if
		342	C<$pri> is given, sets the priority for the next aio request.
		343
		344	The default priority is C<0>, the minimum and maximum priorities are C<-4>
		345	and C<4>, respectively. Requests with higher priority will be serviced
		346	first.
		347
		348	The priority will be reset to C<0> after each call to one of the C<aio_*>
		349	functions.
		350
		351	Example: open a file with low priority, then read something from it with
		352	higher priority so the read request is serviced before other low priority
		353	open requests (potentially spamming the cache):
		354
		355	aioreq_pri -3;
		356	aio_open ..., sub {
		357	return unless $_[0];
		358
		359	aioreq_pri -2;
		360	aio_read $_[0], ..., sub {
		361	...
		362	};
		363	};
		364
		365
		366	=item aioreq_nice $pri_adjust
		367
		368	Similar to C<aioreq_pri>, but subtracts the given value from the current
		369	priority, so the effect is cumulative.
		370
		371
110	=item aio_open $pathname, $flags, $mode, $callback	372	=item aio_open $pathname, $flags, $mode, $callback->($fh)
111		373
112	Asynchronously open or create a file and call the callback with a newly	374	Asynchronously open or create a file and call the callback with a newly
113	created filehandle for the file.	375	created filehandle for the file (or C<undef> in case of an error).
114		376
115	The pathname passed to C<aio_open> must be absolute. See API NOTES, above,	377	The pathname passed to C<aio_open> must be absolute. See API NOTES, above,
116	for an explanation.	378	for an explanation.
117		379
118	The C<$flags> argument is a bitmask. See the C<Fcntl> module for a	380	The C<$flags> argument is a bitmask. See the C<Fcntl> module for a
119	list. They are the same as used by C<sysopen>.	381	list. They are the same as used by C<sysopen>.
120		382
121	Likewise, C<$mode> specifies the mode of the newly created file, if it	383	Likewise, C<$mode> specifies the mode of the newly created file, if it
122	didn't exist and C<O_CREAT> has been given, just like perl's C<sysopen>,	384	didn't exist and C<O_CREAT> has been given, just like perl's C<sysopen>,
123	except that it is mandatory (i.e. use C<0> if you don't create new files,	385	except that it is mandatory (i.e. use C<0> if you don't create new files,
124	and C<0666> or C<0777> if you do).	386	and C<0666> or C<0777> if you do). Note that the C<$mode> will be modified
		387	by the umask in effect then the request is being executed, so better never
		388	change the umask.
125		389
126	Example:	390	Example:
127		391
128	aio_open "/etc/passwd", O_RDONLY, 0, sub {	392	aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub {
129	if ($_[0]) {	393	if ($_[0]) {
130	print "open successful, fh is $_[0]\n";	394	print "open successful, fh is $_[0]\n";
131	...	395	...
132	} else {	396	} else {
133	die "open failed: $!\n";	397	die "open failed: $!\n";
134	}	398	}
135	};	399	};
136		400
		401	In addition to all the common open modes/flags (C<O_RDONLY>, C<O_WRONLY>,
		402	C<O_RDWR>, C<O_CREAT>, C<O_TRUNC>, C<O_EXCL> and C<O_APPEND>), the
		403	following POSIX and non-POSIX constants are available (missing ones on
		404	your system are, as usual, C<0>):
		405
		406	C<O_ASYNC>, C<O_DIRECT>, C<O_NOATIME>, C<O_CLOEXEC>, C<O_NOCTTY>, C<O_NOFOLLOW>,
		407	C<O_NONBLOCK>, C<O_EXEC>, C<O_SEARCH>, C<O_DIRECTORY>, C<O_DSYNC>,
		408	C<O_RSYNC>, C<O_SYNC>, C<O_PATH>, C<O_TMPFILE>, and C<O_TTY_INIT>.
		409
		410
137	=item aio_close $fh, $callback	411	=item aio_close $fh, $callback->($status)
138		412
139	Asynchronously close a file and call the callback with the result	413	Asynchronously close a file and call the callback with the result
140	code. I<WARNING:> although accepted, you should not pass in a perl	414	code.
141	filehandle here, as perl will likely close the file descriptor another
142	time when the filehandle is destroyed. Normally, you can safely call perls
143	C<close> or just let filehandles go out of scope.
144		415
145	This is supposed to be a bug in the API, so that might change. It's	416	Unfortunately, you can't do this to perl. Perl I<insists> very strongly on
146	therefore best to avoid this function.	417	closing the file descriptor associated with the filehandle itself.
147		418
		419	Therefore, C<aio_close> will not close the filehandle - instead it will
		420	use dup2 to overwrite the file descriptor with the write-end of a pipe
		421	(the pipe fd will be created on demand and will be cached).
		422
		423	Or in other words: the file descriptor will be closed, but it will not be
		424	free for reuse until the perl filehandle is closed.
		425
		426	=cut
		427
		428	=item aio_seek $fh, $offset, $whence, $callback->($offs)
		429
		430	Seeks the filehandle to the new C<$offset>, similarly to perl's
		431	C<sysseek>. The C<$whence> can use the traditional values (C<0> for
		432	C<IO::AIO::SEEK_SET>, C<1> for C<IO::AIO::SEEK_CUR> or C<2> for
		433	C<IO::AIO::SEEK_END>).
		434
		435	The resulting absolute offset will be passed to the callback, or C<-1> in
		436	case of an error.
		437
		438	In theory, the C<$whence> constants could be different than the
		439	corresponding values from L<Fcntl>, but perl guarantees they are the same,
		440	so don't panic.
		441
		442	As a GNU/Linux (and maybe Solaris) extension, also the constants
		443	C<IO::AIO::SEEK_DATA> and C<IO::AIO::SEEK_HOLE> are available, if they
		444	could be found. No guarantees about suitability for use in C<aio_seek> or
		445	Perl's C<sysseek> can be made though, although I would naively assume they
		446	"just work".
		447
148	=item aio_read $fh,$offset,$length, $data,$dataoffset,$callback	448	=item aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
149		449
150	=item aio_write $fh,$offset,$length, $data,$dataoffset,$callback	450	=item aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
151		451
152	Reads or writes C<length> bytes from the specified C<fh> and C<offset>	452	Reads or writes C<$length> bytes from or to the specified C<$fh> and
153	into the scalar given by C<data> and offset C<dataoffset> and calls the	453	C<$offset> into the scalar given by C<$data> and offset C<$dataoffset> and
154	callback without the actual number of bytes read (or -1 on error, just	454	calls the callback with the actual number of bytes transferred (or -1 on
155	like the syscall).	455	error, just like the syscall).
		456
		457	C<aio_read> will, like C<sysread>, shrink or grow the C<$data> scalar to
		458	offset plus the actual number of bytes read.
		459
		460	If C<$offset> is undefined, then the current file descriptor offset will
		461	be used (and updated), otherwise the file descriptor offset will not be
		462	changed by these calls.
		463
		464	If C<$length> is undefined in C<aio_write>, use the remaining length of
		465	C<$data>.
		466
		467	If C<$dataoffset> is less than zero, it will be counted from the end of
		468	C<$data>.
156		469
157	The C<$data> scalar I<MUST NOT> be modified in any way while the request	470	The C<$data> scalar I<MUST NOT> be modified in any way while the request
158	is outstanding. Modifying it can result in segfaults or WW3 (if the	471	is outstanding. Modifying it can result in segfaults or World War III (if
159	necessary/optional hardware is installed).	472	the necessary/optional hardware is installed).
160		473
161	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at	474	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at
162	offset C<0> within the scalar:	475	offset C<0> within the scalar:
163		476
164	aio_read $fh, 7, 15, $buffer, 0, sub {	477	aio_read $fh, 7, 15, $buffer, 0, sub {
165	$_[0] > 0 or die "read error: $!";	478	$_[0] > 0 or die "read error: $!";
166	print "read $_[0] bytes: <$buffer>\n";	479	print "read $_[0] bytes: <$buffer>\n";
167	};	480	};
168		481
		482
169	=item aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback	483	=item aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
170		484
171	Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts	485	Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts
172	reading at byte offset C<$in_offset>, and starts writing at the current	486	reading at byte offset C<$in_offset>, and starts writing at the current
173	file offset of C<$out_fh>. Because of that, it is not safe to issue more	487	file offset of C<$out_fh>. Because of that, it is not safe to issue more
174	than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each	488	than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each
175	other.	489	other. The same C<$in_fh> works fine though, as this function does not
		490	move or use the file offset of C<$in_fh>.
176		491
		492	Please note that C<aio_sendfile> can read more bytes from C<$in_fh> than
		493	are written, and there is no way to find out how many more bytes have been
		494	read from C<aio_sendfile> alone, as C<aio_sendfile> only provides the
		495	number of bytes written to C<$out_fh>. Only if the result value equals
		496	C<$length> one can assume that C<$length> bytes have been read.
		497
		498	Unlike with other C<aio_> functions, it makes a lot of sense to use
		499	C<aio_sendfile> on non-blocking sockets, as long as one end (typically
		500	the C<$in_fh>) is a file - the file I/O will then be asynchronous, while
		501	the socket I/O will be non-blocking. Note, however, that you can run
		502	into a trap where C<aio_sendfile> reads some data with readahead, then
		503	fails to write all data, and when the socket is ready the next time, the
		504	data in the cache is already lost, forcing C<aio_sendfile> to again hit
		505	the disk. Explicit C<aio_read> + C<aio_write> let's you better control
		506	resource usage.
		507
177	This call tries to make use of a native C<sendfile> syscall to provide	508	This call tries to make use of a native C<sendfile>-like syscall to
178	zero-copy operation. For this to work, C<$out_fh> should refer to a	509	provide zero-copy operation. For this to work, C<$out_fh> should refer to
179	socket, and C<$in_fh> should refer to mmap'able file.	510	a socket, and C<$in_fh> should refer to an mmap'able file.
180		511
181	If the native sendfile call fails or is not implemented, it will be	512	If a native sendfile cannot be found or it fails with C<ENOSYS>,
182	emulated, so you can call C<aio_sendfile> on any type of filehandle	513	C<EINVAL>, C<ENOTSUP>, C<EOPNOTSUPP>, C<EAFNOSUPPORT>, C<EPROTOTYPE> or
		514	C<ENOTSOCK>, it will be emulated, so you can call C<aio_sendfile> on any
183	regardless of the limitations of the operating system.	515	type of filehandle regardless of the limitations of the operating system.
184		516
185	Please note, however, that C<aio_sendfile> can read more bytes from	517	As native sendfile syscalls (as practically any non-POSIX interface hacked
186	C<$in_fh> than are written, and there is no way to find out how many	518	together in a hurry to improve benchmark numbers) tend to be rather buggy
187	bytes have been read from C<aio_sendfile> alone, as C<aio_sendfile> only	519	on many systems, this implementation tries to work around some known bugs
188	provides the number of bytes written to C<$out_fh>. Only if the result	520	in Linux and FreeBSD kernels (probably others, too), but that might fail,
189	value equals C<$length> one can assume that C<$length> bytes have been	521	so you really really should check the return value of C<aio_sendfile> -
190	read.	522	fewer bytes than expected might have been transferred.
191		523
		524
192	=item aio_readahead $fh,$offset,$length, $callback	525	=item aio_readahead $fh,$offset,$length, $callback->($retval)
193		526
194	C<aio_readahead> populates the page cache with data from a file so that	527	C<aio_readahead> populates the page cache with data from a file so that
195	subsequent reads from that file will not block on disk I/O. The C<$offset>	528	subsequent reads from that file will not block on disk I/O. The C<$offset>
196	argument specifies the starting point from which data is to be read and	529	argument specifies the starting point from which data is to be read and
197	C<$length> specifies the number of bytes to be read. I/O is performed in	530	C<$length> specifies the number of bytes to be read. I/O is performed in
198	whole pages, so that offset is effectively rounded down to a page boundary	531	whole pages, so that offset is effectively rounded down to a page boundary
199	and bytes are read up to the next page boundary greater than or equal to	532	and bytes are read up to the next page boundary greater than or equal to
200	(off-set+length). C<aio_readahead> does not read beyond the end of the	533	(off-set+length). C<aio_readahead> does not read beyond the end of the
201	file. The current file offset of the file is left unchanged.	534	file. The current file offset of the file is left unchanged.
202		535
203	If that syscall doesn't exist (likely if your OS isn't Linux) it will be	536	If that syscall doesn't exist (likely if your kernel isn't Linux) it will
204	emulated by simply reading the data, which would have a similar effect.	537	be emulated by simply reading the data, which would have a similar effect.
205		538
		539
206	=item aio_stat $fh_or_path, $callback	540	=item aio_stat $fh_or_path, $callback->($status)
207		541
208	=item aio_lstat $fh, $callback	542	=item aio_lstat $fh, $callback->($status)
209		543
210	Works like perl's C<stat> or C<lstat> in void context. The callback will	544	Works like perl's C<stat> or C<lstat> in void context. The callback will
211	be called after the stat and the results will be available using C<stat _>	545	be called after the stat and the results will be available using C<stat _>
212	or C<-s _> etc...	546	or C<-s _> etc...
213		547
…		…
215	for an explanation.	549	for an explanation.
216		550
217	Currently, the stats are always 64-bit-stats, i.e. instead of returning an	551	Currently, the stats are always 64-bit-stats, i.e. instead of returning an
218	error when stat'ing a large file, the results will be silently truncated	552	error when stat'ing a large file, the results will be silently truncated
219	unless perl itself is compiled with large file support.	553	unless perl itself is compiled with large file support.
		554
		555	To help interpret the mode and dev/rdev stat values, IO::AIO offers the
		556	following constants and functions (if not implemented, the constants will
		557	be C<0> and the functions will either C<croak> or fall back on traditional
		558	behaviour).
		559
		560	C<S_IFMT>, C<S_IFIFO>, C<S_IFCHR>, C<S_IFBLK>, C<S_IFLNK>, C<S_IFREG>,
		561	C<S_IFDIR>, C<S_IFWHT>, C<S_IFSOCK>, C<IO::AIO::major $dev_t>,
		562	C<IO::AIO::minor $dev_t>, C<IO::AIO::makedev $major, $minor>.
220		563
221	Example: Print the length of F</etc/passwd>:	564	Example: Print the length of F</etc/passwd>:
222		565
223	aio_stat "/etc/passwd", sub {	566	aio_stat "/etc/passwd", sub {
224	$_[0] and die "stat failed: $!";	567	$_[0] and die "stat failed: $!";
225	print "size is ", -s _, "\n";	568	print "size is ", -s _, "\n";
226	};	569	};
227		570
		571
		572	=item aio_statvfs $fh_or_path, $callback->($statvfs)
		573
		574	Works like the POSIX C<statvfs> or C<fstatvfs> syscalls, depending on
		575	whether a file handle or path was passed.
		576
		577	On success, the callback is passed a hash reference with the following
		578	members: C<bsize>, C<frsize>, C<blocks>, C<bfree>, C<bavail>, C<files>,
		579	C<ffree>, C<favail>, C<fsid>, C<flag> and C<namemax>. On failure, C<undef>
		580	is passed.
		581
		582	The following POSIX IO::AIO::ST_* constants are defined: C<ST_RDONLY> and
		583	C<ST_NOSUID>.
		584
		585	The following non-POSIX IO::AIO::ST_* flag masks are defined to
		586	their correct value when available, or to C<0> on systems that do
		587	not support them: C<ST_NODEV>, C<ST_NOEXEC>, C<ST_SYNCHRONOUS>,
		588	C<ST_MANDLOCK>, C<ST_WRITE>, C<ST_APPEND>, C<ST_IMMUTABLE>, C<ST_NOATIME>,
		589	C<ST_NODIRATIME> and C<ST_RELATIME>.
		590
		591	Example: stat C</wd> and dump out the data if successful.
		592
		593	aio_statvfs "/wd", sub {
		594	my $f = $_[0]
		595	or die "statvfs: $!";
		596
		597	use Data::Dumper;
		598	say Dumper $f;
		599	};
		600
		601	# result:
		602	{
		603	bsize => 1024,
		604	bfree => 4333064312,
		605	blocks => 10253828096,
		606	files => 2050765568,
		607	flag => 4096,
		608	favail => 2042092649,
		609	bavail => 4333064312,
		610	ffree => 2042092649,
		611	namemax => 255,
		612	frsize => 1024,
		613	fsid => 1810
		614	}
		615
		616	Here is a (likely partial - send me updates!) list of fsid values used by
		617	Linux - it is safe to hardcode these when C<$^O> is C<linux>:
		618
		619	0x0000adf5 adfs
		620	0x0000adff affs
		621	0x5346414f afs
		622	0x09041934 anon-inode filesystem
		623	0x00000187 autofs
		624	0x42465331 befs
		625	0x1badface bfs
		626	0x42494e4d binfmt_misc
		627	0x9123683e btrfs
		628	0x0027e0eb cgroupfs
		629	0xff534d42 cifs
		630	0x73757245 coda
		631	0x012ff7b7 coh
		632	0x28cd3d45 cramfs
		633	0x453dcd28 cramfs-wend (wrong endianness)
		634	0x64626720 debugfs
		635	0x00001373 devfs
		636	0x00001cd1 devpts
		637	0x0000f15f ecryptfs
		638	0x00414a53 efs
		639	0x0000137d ext
		640	0x0000ef53 ext2/ext3/ext4
		641	0x0000ef51 ext2
		642	0xf2f52010 f2fs
		643	0x00004006 fat
		644	0x65735546 fuseblk
		645	0x65735543 fusectl
		646	0x0bad1dea futexfs
		647	0x01161970 gfs2
		648	0x47504653 gpfs
		649	0x00004244 hfs
		650	0xf995e849 hpfs
		651	0x00c0ffee hostfs
		652	0x958458f6 hugetlbfs
		653	0x2bad1dea inotifyfs
		654	0x00009660 isofs
		655	0x000072b6 jffs2
		656	0x3153464a jfs
		657	0x6b414653 k-afs
		658	0x0bd00bd0 lustre
		659	0x0000137f minix
		660	0x0000138f minix 30 char names
		661	0x00002468 minix v2
		662	0x00002478 minix v2 30 char names
		663	0x00004d5a minix v3
		664	0x19800202 mqueue
		665	0x00004d44 msdos
		666	0x0000564c novell
		667	0x00006969 nfs
		668	0x6e667364 nfsd
		669	0x00003434 nilfs
		670	0x5346544e ntfs
		671	0x00009fa1 openprom
		672	0x7461636F ocfs2
		673	0x00009fa0 proc
		674	0x6165676c pstorefs
		675	0x0000002f qnx4
		676	0x68191122 qnx6
		677	0x858458f6 ramfs
		678	0x52654973 reiserfs
		679	0x00007275 romfs
		680	0x67596969 rpc_pipefs
		681	0x73636673 securityfs
		682	0xf97cff8c selinux
		683	0x0000517b smb
		684	0x534f434b sockfs
		685	0x73717368 squashfs
		686	0x62656572 sysfs
		687	0x012ff7b6 sysv2
		688	0x012ff7b5 sysv4
		689	0x01021994 tmpfs
		690	0x15013346 udf
		691	0x00011954 ufs
		692	0x54190100 ufs byteswapped
		693	0x00009fa2 usbdevfs
		694	0x01021997 v9fs
		695	0xa501fcf5 vxfs
		696	0xabba1974 xenfs
		697	0x012ff7b4 xenix
		698	0x58465342 xfs
		699	0x012fd16d xia
		700
		701	=item aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
		702
		703	Works like perl's C<utime> function (including the special case of $atime
		704	and $mtime being undef). Fractional times are supported if the underlying
		705	syscalls support them.
		706
		707	When called with a pathname, uses utimes(2) if available, otherwise
		708	utime(2). If called on a file descriptor, uses futimes(2) if available,
		709	otherwise returns ENOSYS, so this is not portable.
		710
		711	Examples:
		712
		713	# set atime and mtime to current time (basically touch(1)):
		714	aio_utime "path", undef, undef;
		715	# set atime to current time and mtime to beginning of the epoch:
		716	aio_utime "path", time, undef; # undef==0
		717
		718
		719	=item aio_chown $fh_or_path, $uid, $gid, $callback->($status)
		720
		721	Works like perl's C<chown> function, except that C<undef> for either $uid
		722	or $gid is being interpreted as "do not change" (but -1 can also be used).
		723
		724	Examples:
		725
		726	# same as "chown root path" in the shell:
		727	aio_chown "path", 0, -1;
		728	# same as above:
		729	aio_chown "path", 0, undef;
		730
		731
		732	=item aio_truncate $fh_or_path, $offset, $callback->($status)
		733
		734	Works like truncate(2) or ftruncate(2).
		735
		736
		737	=item aio_allocate $fh, $mode, $offset, $len, $callback->($status)
		738
		739	Allocates or frees disk space according to the C<$mode> argument. See the
		740	linux C<fallocate> documentation for details.
		741
		742	C<$mode> is usually C<0> or C<IO::AIO::FALLOC_FL_KEEP_SIZE> to allocate
		743	space, or C<IO::AIO::FALLOC_FL_PUNCH_HOLE | IO::AIO::FALLOC_FL_KEEP_SIZE>,
		744	to deallocate a file range.
		745
		746	IO::AIO also supports C<FALLOC_FL_COLLAPSE_RANGE>, to remove a range
		747	(without leaving a hole), C<FALLOC_FL_ZERO_RANGE>, to zero a range,
		748	C<FALLOC_FL_INSERT_RANGE> to insert a range and C<FALLOC_FL_UNSHARE_RANGE>
		749	to unshare shared blocks (see your L<fallocate(2)> manpage).
		750
		751	The file system block size used by C<fallocate> is presumably the
		752	C<f_bsize> returned by C<statvfs>, but different filesystems and filetypes
		753	can dictate other limitations.
		754
		755	If C<fallocate> isn't available or cannot be emulated (currently no
		756	emulation will be attempted), passes C<-1> and sets C<$!> to C<ENOSYS>.
		757
		758
		759	=item aio_chmod $fh_or_path, $mode, $callback->($status)
		760
		761	Works like perl's C<chmod> function.
		762
		763
228	=item aio_unlink $pathname, $callback	764	=item aio_unlink $pathname, $callback->($status)
229		765
230	Asynchronously unlink (delete) a file and call the callback with the	766	Asynchronously unlink (delete) a file and call the callback with the
231	result code.	767	result code.
232		768
		769
		770	=item aio_mknod $pathname, $mode, $dev, $callback->($status)
		771
		772	[EXPERIMENTAL]
		773
		774	Asynchronously create a device node (or fifo). See mknod(2).
		775
		776	The only (POSIX-) portable way of calling this function is:
		777
		778	aio_mknod $pathname, IO::AIO::S_IFIFO | $mode, 0, sub { ...
		779
		780	See C<aio_stat> for info about some potentially helpful extra constants
		781	and functions.
		782
		783	=item aio_link $srcpath, $dstpath, $callback->($status)
		784
		785	Asynchronously create a new link to the existing object at C<$srcpath> at
		786	the path C<$dstpath> and call the callback with the result code.
		787
		788
		789	=item aio_symlink $srcpath, $dstpath, $callback->($status)
		790
		791	Asynchronously create a new symbolic link to the existing object at C<$srcpath> at
		792	the path C<$dstpath> and call the callback with the result code.
		793
		794
		795	=item aio_readlink $pathname, $callback->($link)
		796
		797	Asynchronously read the symlink specified by C<$path> and pass it to
		798	the callback. If an error occurs, nothing or undef gets passed to the
		799	callback.
		800
		801
		802	=item aio_realpath $pathname, $callback->($path)
		803
		804	Asynchronously make the path absolute and resolve any symlinks in
		805	C<$path>. The resulting path only consists of directories (same as
		806	L<Cwd::realpath>).
		807
		808	This request can be used to get the absolute path of the current working
		809	directory by passing it a path of F<.> (a single dot).
		810
		811
		812	=item aio_rename $srcpath, $dstpath, $callback->($status)
		813
		814	Asynchronously rename the object at C<$srcpath> to C<$dstpath>, just as
		815	rename(2) and call the callback with the result code.
		816
		817	On systems that support the AIO::WD working directory abstraction
		818	natively, the case C<[$wd, "."]> as C<$srcpath> is specialcased - instead
		819	of failing, C<rename> is called on the absolute path of C<$wd>.
		820
		821
		822	=item aio_rename2 $srcpath, $dstpath, $flags, $callback->($status)
		823
		824	Basically a version of C<aio_rename> with an additional C<$flags>
		825	argument. Calling this with C<$flags=0> is the same as calling
		826	C<aio_rename>.
		827
		828	Non-zero flags are currently only supported on GNU/Linux systems that
		829	support renameat2. Other systems fail with C<ENOSYS> in this case.
		830
		831	The following constants are available (missing ones are, as usual C<0>),
		832	see renameat2(2) for details:
		833
		834	C<IO::AIO::RENAME_NOREPLACE>, C<IO::AIO::RENAME_EXCHANGE>
		835	and C<IO::AIO::RENAME_WHITEOUT>.
		836
		837
		838	=item aio_mkdir $pathname, $mode, $callback->($status)
		839
		840	Asynchronously mkdir (create) a directory and call the callback with
		841	the result code. C<$mode> will be modified by the umask at the time the
		842	request is executed, so do not change your umask.
		843
		844
233	=item aio_rmdir $pathname, $callback	845	=item aio_rmdir $pathname, $callback->($status)
234		846
235	Asynchronously rmdir (delete) a directory and call the callback with the	847	Asynchronously rmdir (delete) a directory and call the callback with the
236	result code.	848	result code.
237		849
		850	On systems that support the AIO::WD working directory abstraction
		851	natively, the case C<[$wd, "."]> is specialcased - instead of failing,
		852	C<rmdir> is called on the absolute path of C<$wd>.
		853
		854
238	=item aio_readdir $pathname $callback	855	=item aio_readdir $pathname, $callback->($entries)
239		856
240	Unlike the POSIX call of the same name, C<aio_readdir> reads an entire	857	Unlike the POSIX call of the same name, C<aio_readdir> reads an entire
241	directory (i.e. opendir + readdir + closedir). The entries will not be	858	directory (i.e. opendir + readdir + closedir). The entries will not be
242	sorted, and will B<NOT> include the C<.> and C<..> entries.	859	sorted, and will B<NOT> include the C<.> and C<..> entries.
243		860
244	The callback a single argument which is either C<undef> or an array-ref	861	The callback is passed a single argument which is either C<undef> or an
245	with the filenames.	862	array-ref with the filenames.
246		863
		864
		865	=item aio_readdirx $pathname, $flags, $callback->($entries, $flags)
		866
		867	Quite similar to C<aio_readdir>, but the C<$flags> argument allows one to
		868	tune behaviour and output format. In case of an error, C<$entries> will be
		869	C<undef>.
		870
		871	The flags are a combination of the following constants, ORed together (the
		872	flags will also be passed to the callback, possibly modified):
		873
		874	=over 4
		875
		876	=item IO::AIO::READDIR_DENTS
		877
		878	When this flag is off, then the callback gets an arrayref consisting of
		879	names only (as with C<aio_readdir>), otherwise it gets an arrayref with
		880	C<[$name, $type, $inode]> arrayrefs, each describing a single directory
		881	entry in more detail.
		882
		883	C<$name> is the name of the entry.
		884
		885	C<$type> is one of the C<IO::AIO::DT_xxx> constants:
		886
		887	C<IO::AIO::DT_UNKNOWN>, C<IO::AIO::DT_FIFO>, C<IO::AIO::DT_CHR>, C<IO::AIO::DT_DIR>,
		888	C<IO::AIO::DT_BLK>, C<IO::AIO::DT_REG>, C<IO::AIO::DT_LNK>, C<IO::AIO::DT_SOCK>,
		889	C<IO::AIO::DT_WHT>.
		890
		891	C<IO::AIO::DT_UNKNOWN> means just that: readdir does not know. If you need to
		892	know, you have to run stat yourself. Also, for speed reasons, the C<$type>
		893	scalars are read-only: you can not modify them.
		894
		895	C<$inode> is the inode number (which might not be exact on systems with 64
		896	bit inode numbers and 32 bit perls). This field has unspecified content on
		897	systems that do not deliver the inode information.
		898
		899	=item IO::AIO::READDIR_DIRS_FIRST
		900
		901	When this flag is set, then the names will be returned in an order where
		902	likely directories come first, in optimal stat order. This is useful when
		903	you need to quickly find directories, or you want to find all directories
		904	while avoiding to stat() each entry.
		905
		906	If the system returns type information in readdir, then this is used
		907	to find directories directly. Otherwise, likely directories are names
		908	beginning with ".", or otherwise names with no dots, of which names with
		909	short names are tried first.
		910
		911	=item IO::AIO::READDIR_STAT_ORDER
		912
		913	When this flag is set, then the names will be returned in an order
		914	suitable for stat()'ing each one. That is, when you plan to stat()
		915	all files in the given directory, then the returned order will likely
		916	be fastest.
		917
		918	If both this flag and C<IO::AIO::READDIR_DIRS_FIRST> are specified, then
		919	the likely dirs come first, resulting in a less optimal stat order.
		920
		921	=item IO::AIO::READDIR_FOUND_UNKNOWN
		922
		923	This flag should not be set when calling C<aio_readdirx>. Instead, it
		924	is being set by C<aio_readdirx>, when any of the C<$type>'s found were
		925	C<IO::AIO::DT_UNKNOWN>. The absence of this flag therefore indicates that all
		926	C<$type>'s are known, which can be used to speed up some algorithms.
		927
		928	=back
		929
		930
		931	=item aio_load $pathname, $data, $callback->($status)
		932
		933	This is a composite request that tries to fully load the given file into
		934	memory. Status is the same as with aio_read.
		935
		936	=cut
		937
		938	sub aio_load($$;$) {
		939	my ($path, undef, $cb) = @_;
		940	my $data = \$_[1];
		941
		942	my $pri = aioreq_pri;
		943	my $grp = aio_group $cb;
		944
		945	aioreq_pri $pri;
		946	add $grp aio_open $path, O_RDONLY, 0, sub {
		947	my $fh = shift
		948	or return $grp->result (-1);
		949
		950	aioreq_pri $pri;
		951	add $grp aio_read $fh, 0, (-s $fh), $$data, 0, sub {
		952	$grp->result ($_[0]);
		953	};
		954	};
		955
		956	$grp
		957	}
		958
		959	=item aio_copy $srcpath, $dstpath, $callback->($status)
		960
		961	Try to copy the I<file> (directories not supported as either source or
		962	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		963	a status of C<0> (ok) or C<-1> (error, see C<$!>).
		964
		965	Existing destination files will be truncated.
		966
		967	This is a composite request that creates the destination file with
		968	mode 0200 and copies the contents of the source file into it using
		969	C<aio_sendfile>, followed by restoring atime, mtime, access mode and
		970	uid/gid, in that order.
		971
		972	If an error occurs, the partial destination file will be unlinked, if
		973	possible, except when setting atime, mtime, access mode and uid/gid, where
		974	errors are being ignored.
		975
		976	=cut
		977
		978	sub aio_copy($$;$) {
		979	my ($src, $dst, $cb) = @_;
		980
		981	my $pri = aioreq_pri;
		982	my $grp = aio_group $cb;
		983
		984	aioreq_pri $pri;
		985	add $grp aio_open $src, O_RDONLY, 0, sub {
		986	if (my $src_fh = $_[0]) {
		987	my @stat = stat $src_fh; # hmm, might block over nfs?
		988
		989	aioreq_pri $pri;
		990	add $grp aio_open $dst, O_CREAT | O_WRONLY | O_TRUNC, 0200, sub {
		991	if (my $dst_fh = $_[0]) {
		992	aioreq_pri $pri;
		993	add $grp aio_sendfile $dst_fh, $src_fh, 0, $stat[7], sub {
		994	if ($_[0] == $stat[7]) {
		995	$grp->result (0);
		996	close $src_fh;
		997
		998	my $ch = sub {
		999	aioreq_pri $pri;
		1000	add $grp aio_chmod $dst_fh, $stat[2] & 07777, sub {
		1001	aioreq_pri $pri;
		1002	add $grp aio_chown $dst_fh, $stat[4], $stat[5], sub {
		1003	aioreq_pri $pri;
		1004	add $grp aio_close $dst_fh;
		1005	}
		1006	};
		1007	};
		1008
		1009	aioreq_pri $pri;
		1010	add $grp aio_utime $dst_fh, $stat[8], $stat[9], sub {
		1011	if ($_[0] < 0 && $! == ENOSYS) {
		1012	aioreq_pri $pri;
		1013	add $grp aio_utime $dst, $stat[8], $stat[9], $ch;
		1014	} else {
		1015	$ch->();
		1016	}
		1017	};
		1018	} else {
		1019	$grp->result (-1);
		1020	close $src_fh;
		1021	close $dst_fh;
		1022
		1023	aioreq $pri;
		1024	add $grp aio_unlink $dst;
		1025	}
		1026	};
		1027	} else {
		1028	$grp->result (-1);
		1029	}
		1030	},
		1031
		1032	} else {
		1033	$grp->result (-1);
		1034	}
		1035	};
		1036
		1037	$grp
		1038	}
		1039
		1040	=item aio_move $srcpath, $dstpath, $callback->($status)
		1041
		1042	Try to move the I<file> (directories not supported as either source or
		1043	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		1044	a status of C<0> (ok) or C<-1> (error, see C<$!>).
		1045
		1046	This is a composite request that tries to rename(2) the file first; if
		1047	rename fails with C<EXDEV>, it copies the file with C<aio_copy> and, if
		1048	that is successful, unlinks the C<$srcpath>.
		1049
		1050	=cut
		1051
		1052	sub aio_move($$;$) {
		1053	my ($src, $dst, $cb) = @_;
		1054
		1055	my $pri = aioreq_pri;
		1056	my $grp = aio_group $cb;
		1057
		1058	aioreq_pri $pri;
		1059	add $grp aio_rename $src, $dst, sub {
		1060	if ($_[0] && $! == EXDEV) {
		1061	aioreq_pri $pri;
		1062	add $grp aio_copy $src, $dst, sub {
		1063	$grp->result ($_[0]);
		1064
		1065	unless ($_[0]) {
		1066	aioreq_pri $pri;
		1067	add $grp aio_unlink $src;
		1068	}
		1069	};
		1070	} else {
		1071	$grp->result ($_[0]);
		1072	}
		1073	};
		1074
		1075	$grp
		1076	}
		1077
		1078	=item aio_scandir $pathname, $maxreq, $callback->($dirs, $nondirs)
		1079
		1080	Scans a directory (similar to C<aio_readdir>) but additionally tries to
		1081	efficiently separate the entries of directory C<$path> into two sets of
		1082	names, directories you can recurse into (directories), and ones you cannot
		1083	recurse into (everything else, including symlinks to directories).
		1084
		1085	C<aio_scandir> is a composite request that generates many sub requests.
		1086	C<$maxreq> specifies the maximum number of outstanding aio requests that
		1087	this function generates. If it is C<< <= 0 >>, then a suitable default
		1088	will be chosen (currently 4).
		1089
		1090	On error, the callback is called without arguments, otherwise it receives
		1091	two array-refs with path-relative entry names.
		1092
		1093	Example:
		1094
		1095	aio_scandir $dir, 0, sub {
		1096	my ($dirs, $nondirs) = @_;
		1097	print "real directories: @$dirs\n";
		1098	print "everything else: @$nondirs\n";
		1099	};
		1100
		1101	Implementation notes.
		1102
		1103	The C<aio_readdir> cannot be avoided, but C<stat()>'ing every entry can.
		1104
		1105	If readdir returns file type information, then this is used directly to
		1106	find directories.
		1107
		1108	Otherwise, after reading the directory, the modification time, size etc.
		1109	of the directory before and after the readdir is checked, and if they
		1110	match (and isn't the current time), the link count will be used to decide
		1111	how many entries are directories (if >= 2). Otherwise, no knowledge of the
		1112	number of subdirectories will be assumed.
		1113
		1114	Then entries will be sorted into likely directories a non-initial dot
		1115	currently) and likely non-directories (see C<aio_readdirx>). Then every
		1116	entry plus an appended C</.> will be C<stat>'ed, likely directories first,
		1117	in order of their inode numbers. If that succeeds, it assumes that the
		1118	entry is a directory or a symlink to directory (which will be checked
		1119	separately). This is often faster than stat'ing the entry itself because
		1120	filesystems might detect the type of the entry without reading the inode
		1121	data (e.g. ext2fs filetype feature), even on systems that cannot return
		1122	the filetype information on readdir.
		1123
		1124	If the known number of directories (link count - 2) has been reached, the
		1125	rest of the entries is assumed to be non-directories.
		1126
		1127	This only works with certainty on POSIX (= UNIX) filesystems, which
		1128	fortunately are the vast majority of filesystems around.
		1129
		1130	It will also likely work on non-POSIX filesystems with reduced efficiency
		1131	as those tend to return 0 or 1 as link counts, which disables the
		1132	directory counting heuristic.
		1133
		1134	=cut
		1135
		1136	sub aio_scandir($$;$) {
		1137	my ($path, $maxreq, $cb) = @_;
		1138
		1139	my $pri = aioreq_pri;
		1140
		1141	my $grp = aio_group $cb;
		1142
		1143	$maxreq = 4 if $maxreq <= 0;
		1144
		1145	# get a wd object
		1146	aioreq_pri $pri;
		1147	add $grp aio_wd $path, sub {
		1148	$_[0]
		1149	or return $grp->result ();
		1150
		1151	my $wd = [shift, "."];
		1152
		1153	# stat once
		1154	aioreq_pri $pri;
		1155	add $grp aio_stat $wd, sub {
		1156	return $grp->result () if $_[0];
		1157	my $now = time;
		1158	my $hash1 = join ":", (stat _)[0,1,3,7,9];
		1159
		1160	# read the directory entries
		1161	aioreq_pri $pri;
		1162	add $grp aio_readdirx $wd, READDIR_DIRS_FIRST, sub {
		1163	my $entries = shift
		1164	or return $grp->result ();
		1165
		1166	# stat the dir another time
		1167	aioreq_pri $pri;
		1168	add $grp aio_stat $wd, sub {
		1169	my $hash2 = join ":", (stat _)[0,1,3,7,9];
		1170
		1171	my $ndirs;
		1172
		1173	# take the slow route if anything looks fishy
		1174	if ($hash1 ne $hash2 or (stat _)[9] == $now) {
		1175	$ndirs = -1;
		1176	} else {
		1177	# if nlink == 2, we are finished
		1178	# for non-posix-fs's, we rely on nlink < 2
		1179	$ndirs = (stat _)[3] - 2
		1180	or return $grp->result ([], $entries);
		1181	}
		1182
		1183	my (@dirs, @nondirs);
		1184
		1185	my $statgrp = add $grp aio_group sub {
		1186	$grp->result (\@dirs, \@nondirs);
		1187	};
		1188
		1189	limit $statgrp $maxreq;
		1190	feed $statgrp sub {
		1191	return unless @$entries;
		1192	my $entry = shift @$entries;
		1193
		1194	aioreq_pri $pri;
		1195	$wd->[1] = "$entry/.";
		1196	add $statgrp aio_stat $wd, sub {
		1197	if ($_[0] < 0) {
		1198	push @nondirs, $entry;
		1199	} else {
		1200	# need to check for real directory
		1201	aioreq_pri $pri;
		1202	$wd->[1] = $entry;
		1203	add $statgrp aio_lstat $wd, sub {
		1204	if (-d _) {
		1205	push @dirs, $entry;
		1206
		1207	unless (--$ndirs) {
		1208	push @nondirs, @$entries;
		1209	feed $statgrp;
		1210	}
		1211	} else {
		1212	push @nondirs, $entry;
		1213	}
		1214	}
		1215	}
		1216	};
		1217	};
		1218	};
		1219	};
		1220	};
		1221	};
		1222
		1223	$grp
		1224	}
		1225
		1226	=item aio_rmtree $pathname, $callback->($status)
		1227
		1228	Delete a directory tree starting (and including) C<$path>, return the
		1229	status of the final C<rmdir> only. This is a composite request that
		1230	uses C<aio_scandir> to recurse into and rmdir directories, and unlink
		1231	everything else.
		1232
		1233	=cut
		1234
		1235	sub aio_rmtree;
		1236	sub aio_rmtree($;$) {
		1237	my ($path, $cb) = @_;
		1238
		1239	my $pri = aioreq_pri;
		1240	my $grp = aio_group $cb;
		1241
		1242	aioreq_pri $pri;
		1243	add $grp aio_scandir $path, 0, sub {
		1244	my ($dirs, $nondirs) = @_;
		1245
		1246	my $dirgrp = aio_group sub {
		1247	add $grp aio_rmdir $path, sub {
		1248	$grp->result ($_[0]);
		1249	};
		1250	};
		1251
		1252	(aioreq_pri $pri), add $dirgrp aio_rmtree "$path/$_" for @$dirs;
		1253	(aioreq_pri $pri), add $dirgrp aio_unlink "$path/$_" for @$nondirs;
		1254
		1255	add $grp $dirgrp;
		1256	};
		1257
		1258	$grp
		1259	}
		1260
		1261	=item aio_fcntl $fh, $cmd, $arg, $callback->($status)
		1262
		1263	=item aio_ioctl $fh, $request, $buf, $callback->($status)
		1264
		1265	These work just like the C<fcntl> and C<ioctl> built-in functions, except
		1266	they execute asynchronously and pass the return value to the callback.
		1267
		1268	Both calls can be used for a lot of things, some of which make more sense
		1269	to run asynchronously in their own thread, while some others make less
		1270	sense. For example, calls that block waiting for external events, such
		1271	as locking, will also lock down an I/O thread while it is waiting, which
		1272	can deadlock the whole I/O system. At the same time, there might be no
		1273	alternative to using a thread to wait.
		1274
		1275	So in general, you should only use these calls for things that do
		1276	(filesystem) I/O, not for things that wait for other events (network,
		1277	other processes), although if you are careful and know what you are doing,
		1278	you still can.
		1279
		1280	The following constants are available (missing ones are, as usual C<0>):
		1281
		1282	C<F_DUPFD_CLOEXEC>,
		1283
		1284	C<F_OFD_GETLK>, C<F_OFD_SETLK>, C<F_OFD_GETLKW>,
		1285
		1286	C<FIFREEZE>, C<FITHAW>, C<FITRIM>, C<FICLONE>, C<FICLONERANGE>, C<FIDEDUPERANGE>.
		1287
		1288	C<FS_IOC_GETFLAGS>, C<FS_IOC_SETFLAGS>, C<FS_IOC_GETVERSION>, C<FS_IOC_SETVERSION>,
		1289	C<FS_IOC_FIEMAP>.
		1290
		1291	C<FS_IOC_FSGETXATTR>, C<FS_IOC_FSSETXATTR>, C<FS_IOC_SET_ENCRYPTION_POLICY>,
		1292	C<FS_IOC_GET_ENCRYPTION_PWSALT>, C<FS_IOC_GET_ENCRYPTION_POLICY>, C<FS_KEY_DESCRIPTOR_SIZE>.
		1293
		1294	C<FS_SECRM_FL>, C<FS_UNRM_FL>, C<FS_COMPR_FL>, C<FS_SYNC_FL>, C<FS_IMMUTABLE_FL>,
		1295	C<FS_APPEND_FL>, C<FS_NODUMP_FL>, C<FS_NOATIME_FL>, C<FS_DIRTY_FL>,
		1296	C<FS_COMPRBLK_FL>, C<FS_NOCOMP_FL>, C<FS_ENCRYPT_FL>, C<FS_BTREE_FL>,
		1297	C<FS_INDEX_FL>, C<FS_JOURNAL_DATA_FL>, C<FS_NOTAIL_FL>, C<FS_DIRSYNC_FL>, C<FS_TOPDIR_FL>,
		1298	C<FS_FL_USER_MODIFIABLE>.
		1299
		1300	C<FS_XFLAG_REALTIME>, C<FS_XFLAG_PREALLOC>, C<FS_XFLAG_IMMUTABLE>, C<FS_XFLAG_APPEND>,
		1301	C<FS_XFLAG_SYNC>, C<FS_XFLAG_NOATIME>, C<FS_XFLAG_NODUMP>, C<FS_XFLAG_RTINHERIT>,
		1302	C<FS_XFLAG_PROJINHERIT>, C<FS_XFLAG_NOSYMLINKS>, C<FS_XFLAG_EXTSIZE>, C<FS_XFLAG_EXTSZINHERIT>,
		1303	C<FS_XFLAG_NODEFRAG>, C<FS_XFLAG_FILESTREAM>, C<FS_XFLAG_DAX>, C<FS_XFLAG_HASATTR>,
		1304
		1305	=item aio_sync $callback->($status)
		1306
		1307	Asynchronously call sync and call the callback when finished.
		1308
247	=item aio_fsync $fh, $callback	1309	=item aio_fsync $fh, $callback->($status)
248		1310
249	Asynchronously call fsync on the given filehandle and call the callback	1311	Asynchronously call fsync on the given filehandle and call the callback
250	with the fsync result code.	1312	with the fsync result code.
251		1313
252	=item aio_fdatasync $fh, $callback	1314	=item aio_fdatasync $fh, $callback->($status)
253		1315
254	Asynchronously call fdatasync on the given filehandle and call the	1316	Asynchronously call fdatasync on the given filehandle and call the
255	callback with the fdatasync result code.	1317	callback with the fdatasync result code.
256		1318
257	If this call isn't available because your OS lacks it or it couldn't be	1319	If this call isn't available because your OS lacks it or it couldn't be
258	detected, it will be emulated by calling C<fsync> instead.	1320	detected, it will be emulated by calling C<fsync> instead.
259		1321
		1322	=item aio_syncfs $fh, $callback->($status)
		1323
		1324	Asynchronously call the syncfs syscall to sync the filesystem associated
		1325	to the given filehandle and call the callback with the syncfs result
		1326	code. If syncfs is not available, calls sync(), but returns C<-1> and sets
		1327	errno to C<ENOSYS> nevertheless.
		1328
		1329	=item aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
		1330
		1331	Sync the data portion of the file specified by C<$offset> and C<$length>
		1332	to disk (but NOT the metadata), by calling the Linux-specific
		1333	sync_file_range call. If sync_file_range is not available or it returns
		1334	ENOSYS, then fdatasync or fsync is being substituted.
		1335
		1336	C<$flags> can be a combination of C<IO::AIO::SYNC_FILE_RANGE_WAIT_BEFORE>,
		1337	C<IO::AIO::SYNC_FILE_RANGE_WRITE> and
		1338	C<IO::AIO::SYNC_FILE_RANGE_WAIT_AFTER>: refer to the sync_file_range
		1339	manpage for details.
		1340
		1341	=item aio_pathsync $pathname, $callback->($status)
		1342
		1343	This request tries to open, fsync and close the given path. This is a
		1344	composite request intended to sync directories after directory operations
		1345	(E.g. rename). This might not work on all operating systems or have any
		1346	specific effect, but usually it makes sure that directory changes get
		1347	written to disc. It works for anything that can be opened for read-only,
		1348	not just directories.
		1349
		1350	Future versions of this function might fall back to other methods when
		1351	C<fsync> on the directory fails (such as calling C<sync>).
		1352
		1353	Passes C<0> when everything went ok, and C<-1> on error.
		1354
		1355	=cut
		1356
		1357	sub aio_pathsync($;$) {
		1358	my ($path, $cb) = @_;
		1359
		1360	my $pri = aioreq_pri;
		1361	my $grp = aio_group $cb;
		1362
		1363	aioreq_pri $pri;
		1364	add $grp aio_open $path, O_RDONLY, 0, sub {
		1365	my ($fh) = @_;
		1366	if ($fh) {
		1367	aioreq_pri $pri;
		1368	add $grp aio_fsync $fh, sub {
		1369	$grp->result ($_[0]);
		1370
		1371	aioreq_pri $pri;
		1372	add $grp aio_close $fh;
		1373	};
		1374	} else {
		1375	$grp->result (-1);
		1376	}
		1377	};
		1378
		1379	$grp
		1380	}
		1381
		1382	=item aio_msync $scalar, $offset = 0, $length = undef, flags = MS_SYNC, $callback->($status)
		1383
		1384	This is a rather advanced IO::AIO call, which only works on mmap(2)ed
		1385	scalars (see the C<IO::AIO::mmap> function, although it also works on data
		1386	scalars managed by the L<Sys::Mmap> or L<Mmap> modules, note that the
		1387	scalar must only be modified in-place while an aio operation is pending on
		1388	it).
		1389
		1390	It calls the C<msync> function of your OS, if available, with the memory
		1391	area starting at C<$offset> in the string and ending C<$length> bytes
		1392	later. If C<$length> is negative, counts from the end, and if C<$length>
		1393	is C<undef>, then it goes till the end of the string. The flags can be
		1394	either C<IO::AIO::MS_ASYNC> or C<IO::AIO::MS_SYNC>, plus an optional
		1395	C<IO::AIO::MS_INVALIDATE>.
		1396
		1397	=item aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		1398
		1399	This is a rather advanced IO::AIO call, which works best on mmap(2)ed
		1400	scalars.
		1401
		1402	It touches (reads or writes) all memory pages in the specified
		1403	range inside the scalar. All caveats and parameters are the same
		1404	as for C<aio_msync>, above, except for flags, which must be either
		1405	C<0> (which reads all pages and ensures they are instantiated) or
		1406	C<IO::AIO::MT_MODIFY>, which modifies the memory pages (by reading and
		1407	writing an octet from it, which dirties the page).
		1408
		1409	=item aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status)
		1410
		1411	This is a rather advanced IO::AIO call, which works best on mmap(2)ed
		1412	scalars.
		1413
		1414	It reads in all the pages of the underlying storage into memory (if any)
		1415	and locks them, so they are not getting swapped/paged out or removed.
		1416
		1417	If C<$length> is undefined, then the scalar will be locked till the end.
		1418
		1419	On systems that do not implement C<mlock>, this function returns C<-1>
		1420	and sets errno to C<ENOSYS>.
		1421
		1422	Note that the corresponding C<munlock> is synchronous and is
		1423	documented under L<MISCELLANEOUS FUNCTIONS>.
		1424
		1425	Example: open a file, mmap and mlock it - both will be undone when
		1426	C<$data> gets destroyed.
		1427
		1428	open my $fh, "<", $path or die "$path: $!";
		1429	my $data;
		1430	IO::AIO::mmap $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh;
		1431	aio_mlock $data; # mlock in background
		1432
		1433	=item aio_mlockall $flags, $callback->($status)
		1434
		1435	Calls the C<mlockall> function with the given C<$flags> (a combination of
		1436	C<IO::AIO::MCL_CURRENT> and C<IO::AIO::MCL_FUTURE>).
		1437
		1438	On systems that do not implement C<mlockall>, this function returns C<-1>
		1439	and sets errno to C<ENOSYS>.
		1440
		1441	Note that the corresponding C<munlockall> is synchronous and is
		1442	documented under L<MISCELLANEOUS FUNCTIONS>.
		1443
		1444	Example: asynchronously lock all current and future pages into memory.
		1445
		1446	aio_mlockall IO::AIO::MCL_FUTURE;
		1447
		1448	=item aio_fiemap $fh, $start, $length, $flags, $count, $cb->(\@extents)
		1449
		1450	Queries the extents of the given file (by calling the Linux C<FIEMAP>
		1451	ioctl, see L<http://cvs.schmorp.de/IO-AIO/doc/fiemap.txt> for details). If
		1452	the ioctl is not available on your OS, then this request will fail with
		1453	C<ENOSYS>.
		1454
		1455	C<$start> is the starting offset to query extents for, C<$length> is the
		1456	size of the range to query - if it is C<undef>, then the whole file will
		1457	be queried.
		1458
		1459	C<$flags> is a combination of flags (C<IO::AIO::FIEMAP_FLAG_SYNC> or
		1460	C<IO::AIO::FIEMAP_FLAG_XATTR> - C<IO::AIO::FIEMAP_FLAGS_COMPAT> is also
		1461	exported), and is normally C<0> or C<IO::AIO::FIEMAP_FLAG_SYNC> to query
		1462	the data portion.
		1463
		1464	C<$count> is the maximum number of extent records to return. If it is
		1465	C<undef>, then IO::AIO queries all extents of the range. As a very special
		1466	case, if it is C<0>, then the callback receives the number of extents
		1467	instead of the extents themselves (which is unreliable, see below).
		1468
		1469	If an error occurs, the callback receives no arguments. The special
		1470	C<errno> value C<IO::AIO::EBADR> is available to test for flag errors.
		1471
		1472	Otherwise, the callback receives an array reference with extent
		1473	structures. Each extent structure is an array reference itself, with the
		1474	following members:
		1475
		1476	[$logical, $physical, $length, $flags]
		1477
		1478	Flags is any combination of the following flag values (typically either C<0>
		1479	or C<IO::AIO::FIEMAP_EXTENT_LAST> (1)):
		1480
		1481	C<IO::AIO::FIEMAP_EXTENT_LAST>, C<IO::AIO::FIEMAP_EXTENT_UNKNOWN>,
		1482	C<IO::AIO::FIEMAP_EXTENT_DELALLOC>, C<IO::AIO::FIEMAP_EXTENT_ENCODED>,
		1483	C<IO::AIO::FIEMAP_EXTENT_DATA_ENCRYPTED>, C<IO::AIO::FIEMAP_EXTENT_NOT_ALIGNED>,
		1484	C<IO::AIO::FIEMAP_EXTENT_DATA_INLINE>, C<IO::AIO::FIEMAP_EXTENT_DATA_TAIL>,
		1485	C<IO::AIO::FIEMAP_EXTENT_UNWRITTEN>, C<IO::AIO::FIEMAP_EXTENT_MERGED> or
		1486	C<IO::AIO::FIEMAP_EXTENT_SHARED>.
		1487
		1488	At the time of this writing (Linux 3.2), this request is unreliable unless
		1489	C<$count> is C<undef>, as the kernel has all sorts of bugs preventing
		1490	it to return all extents of a range for files with a large number of
		1491	extents. The code (only) works around all these issues if C<$count> is
		1492	C<undef>.
		1493
		1494	=item aio_group $callback->(...)
		1495
		1496	This is a very special aio request: Instead of doing something, it is a
		1497	container for other aio requests, which is useful if you want to bundle
		1498	many requests into a single, composite, request with a definite callback
		1499	and the ability to cancel the whole request with its subrequests.
		1500
		1501	Returns an object of class L<IO::AIO::GRP>. See its documentation below
		1502	for more info.
		1503
		1504	Example:
		1505
		1506	my $grp = aio_group sub {
		1507	print "all stats done\n";
		1508	};
		1509
		1510	add $grp
		1511	(aio_stat ...),
		1512	(aio_stat ...),
		1513	...;
		1514
		1515	=item aio_nop $callback->()
		1516
		1517	This is a special request - it does nothing in itself and is only used for
		1518	side effects, such as when you want to add a dummy request to a group so
		1519	that finishing the requests in the group depends on executing the given
		1520	code.
		1521
		1522	While this request does nothing, it still goes through the execution
		1523	phase and still requires a worker thread. Thus, the callback will not
		1524	be executed immediately but only after other requests in the queue have
		1525	entered their execution phase. This can be used to measure request
		1526	latency.
		1527
		1528	=item IO::AIO::aio_busy $fractional_seconds, $callback->() *NOT EXPORTED*
		1529
		1530	Mainly used for debugging and benchmarking, this aio request puts one of
		1531	the request workers to sleep for the given time.
		1532
		1533	While it is theoretically handy to have simple I/O scheduling requests
		1534	like sleep and file handle readable/writable, the overhead this creates is
		1535	immense (it blocks a thread for a long time) so do not use this function
		1536	except to put your application under artificial I/O pressure.
		1537
260	=back	1538	=back
261		1539
		1540
		1541	=head2 IO::AIO::WD - multiple working directories
		1542
		1543	Your process only has one current working directory, which is used by all
		1544	threads. This makes it hard to use relative paths (some other component
		1545	could call C<chdir> at any time, and it is hard to control when the path
		1546	will be used by IO::AIO).
		1547
		1548	One solution for this is to always use absolute paths. This usually works,
		1549	but can be quite slow (the kernel has to walk the whole path on every
		1550	access), and can also be a hassle to implement.
		1551
		1552	Newer POSIX systems have a number of functions (openat, fdopendir,
		1553	futimensat and so on) that make it possible to specify working directories
		1554	per operation.
		1555
		1556	For portability, and because the clowns who "designed", or shall I write,
		1557	perpetrated this new interface were obviously half-drunk, this abstraction
		1558	cannot be perfect, though.
		1559
		1560	IO::AIO allows you to convert directory paths into a so-called IO::AIO::WD
		1561	object. This object stores the canonicalised, absolute version of the
		1562	path, and on systems that allow it, also a directory file descriptor.
		1563
		1564	Everywhere where a pathname is accepted by IO::AIO (e.g. in C<aio_stat>
		1565	or C<aio_unlink>), one can specify an array reference with an IO::AIO::WD
		1566	object and a pathname instead (or the IO::AIO::WD object alone, which
		1567	gets interpreted as C<[$wd, "."]>). If the pathname is absolute, the
		1568	IO::AIO::WD object is ignored, otherwise the pathname is resolved relative
		1569	to that IO::AIO::WD object.
		1570
		1571	For example, to get a wd object for F</etc> and then stat F<passwd>
		1572	inside, you would write:
		1573
		1574	aio_wd "/etc", sub {
		1575	my $etcdir = shift;
		1576
		1577	# although $etcdir can be undef on error, there is generally no reason
		1578	# to check for errors here, as aio_stat will fail with ENOENT
		1579	# when $etcdir is undef.
		1580
		1581	aio_stat [$etcdir, "passwd"], sub {
		1582	# yay
		1583	};
		1584	};
		1585
		1586	The fact that C<aio_wd> is a request and not a normal function shows that
		1587	creating an IO::AIO::WD object is itself a potentially blocking operation,
		1588	which is why it is done asynchronously.
		1589
		1590	To stat the directory obtained with C<aio_wd> above, one could write
		1591	either of the following three request calls:
		1592
		1593	aio_lstat "/etc" , sub { ... # pathname as normal string
		1594	aio_lstat [$wd, "."], sub { ... # "." relative to $wd (i.e. $wd itself)
		1595	aio_lstat $wd , sub { ... # shorthand for the previous
		1596
		1597	As with normal pathnames, IO::AIO keeps a copy of the working directory
		1598	object and the pathname string, so you could write the following without
		1599	causing any issues due to C<$path> getting reused:
		1600
		1601	my $path = [$wd, undef];
		1602
		1603	for my $name (qw(abc def ghi)) {
		1604	$path->[1] = $name;
		1605	aio_stat $path, sub {
		1606	# ...
		1607	};
		1608	}
		1609
		1610	There are some caveats: when directories get renamed (or deleted), the
		1611	pathname string doesn't change, so will point to the new directory (or
		1612	nowhere at all), while the directory fd, if available on the system,
		1613	will still point to the original directory. Most functions accepting a
		1614	pathname will use the directory fd on newer systems, and the string on
		1615	older systems. Some functions (such as C<aio_realpath>) will always rely on
		1616	the string form of the pathname.
		1617
		1618	So this functionality is mainly useful to get some protection against
		1619	C<chdir>, to easily get an absolute path out of a relative path for future
		1620	reference, and to speed up doing many operations in the same directory
		1621	(e.g. when stat'ing all files in a directory).
		1622
		1623	The following functions implement this working directory abstraction:
		1624
		1625	=over 4
		1626
		1627	=item aio_wd $pathname, $callback->($wd)
		1628
		1629	Asynchonously canonicalise the given pathname and convert it to an
		1630	IO::AIO::WD object representing it. If possible and supported on the
		1631	system, also open a directory fd to speed up pathname resolution relative
		1632	to this working directory.
		1633
		1634	If something goes wrong, then C<undef> is passwd to the callback instead
		1635	of a working directory object and C<$!> is set appropriately. Since
		1636	passing C<undef> as working directory component of a pathname fails the
		1637	request with C<ENOENT>, there is often no need for error checking in the
		1638	C<aio_wd> callback, as future requests using the value will fail in the
		1639	expected way.
		1640
		1641	=item IO::AIO::CWD
		1642
		1643	This is a compiletime constant (object) that represents the process
		1644	current working directory.
		1645
		1646	Specifying this object as working directory object for a pathname is as if
		1647	the pathname would be specified directly, without a directory object. For
		1648	example, these calls are functionally identical:
		1649
		1650	aio_stat "somefile", sub { ... };
		1651	aio_stat [IO::AIO::CWD, "somefile"], sub { ... };
		1652
		1653	=back
		1654
		1655	To recover the path associated with an IO::AIO::WD object, you can use
		1656	C<aio_realpath>:
		1657
		1658	aio_realpath $wd, sub {
		1659	warn "path is $_[0]\n";
		1660	};
		1661
		1662	Currently, C<aio_statvfs> always, and C<aio_rename> and C<aio_rmdir>
		1663	sometimes, fall back to using an absolue path.
		1664
		1665	=head2 IO::AIO::REQ CLASS
		1666
		1667	All non-aggregate C<aio_*> functions return an object of this class when
		1668	called in non-void context.
		1669
		1670	=over 4
		1671
		1672	=item cancel $req
		1673
		1674	Cancels the request, if possible. Has the effect of skipping execution
		1675	when entering the B<execute> state and skipping calling the callback when
		1676	entering the the B<result> state, but will leave the request otherwise
		1677	untouched (with the exception of readdir). That means that requests that
		1678	currently execute will not be stopped and resources held by the request
		1679	will not be freed prematurely.
		1680
		1681	=item cb $req $callback->(...)
		1682
		1683	Replace (or simply set) the callback registered to the request.
		1684
		1685	=back
		1686
		1687	=head2 IO::AIO::GRP CLASS
		1688
		1689	This class is a subclass of L<IO::AIO::REQ>, so all its methods apply to
		1690	objects of this class, too.
		1691
		1692	A IO::AIO::GRP object is a special request that can contain multiple other
		1693	aio requests.
		1694
		1695	You create one by calling the C<aio_group> constructing function with a
		1696	callback that will be called when all contained requests have entered the
		1697	C<done> state:
		1698
		1699	my $grp = aio_group sub {
		1700	print "all requests are done\n";
		1701	};
		1702
		1703	You add requests by calling the C<add> method with one or more
		1704	C<IO::AIO::REQ> objects:
		1705
		1706	$grp->add (aio_unlink "...");
		1707
		1708	add $grp aio_stat "...", sub {
		1709	$_[0] or return $grp->result ("error");
		1710
		1711	# add another request dynamically, if first succeeded
		1712	add $grp aio_open "...", sub {
		1713	$grp->result ("ok");
		1714	};
		1715	};
		1716
		1717	This makes it very easy to create composite requests (see the source of
		1718	C<aio_move> for an application) that work and feel like simple requests.
		1719
		1720	=over 4
		1721
		1722	=item * The IO::AIO::GRP objects will be cleaned up during calls to
		1723	C<IO::AIO::poll_cb>, just like any other request.
		1724
		1725	=item * They can be canceled like any other request. Canceling will cancel not
		1726	only the request itself, but also all requests it contains.
		1727
		1728	=item * They can also can also be added to other IO::AIO::GRP objects.
		1729
		1730	=item * You must not add requests to a group from within the group callback (or
		1731	any later time).
		1732
		1733	=back
		1734
		1735	Their lifetime, simplified, looks like this: when they are empty, they
		1736	will finish very quickly. If they contain only requests that are in the
		1737	C<done> state, they will also finish. Otherwise they will continue to
		1738	exist.
		1739
		1740	That means after creating a group you have some time to add requests
		1741	(precisely before the callback has been invoked, which is only done within
		1742	the C<poll_cb>). And in the callbacks of those requests, you can add
		1743	further requests to the group. And only when all those requests have
		1744	finished will the the group itself finish.
		1745
		1746	=over 4
		1747
		1748	=item add $grp ...
		1749
		1750	=item $grp->add (...)
		1751
		1752	Add one or more requests to the group. Any type of L<IO::AIO::REQ> can
		1753	be added, including other groups, as long as you do not create circular
		1754	dependencies.
		1755
		1756	Returns all its arguments.
		1757
		1758	=item $grp->cancel_subs
		1759
		1760	Cancel all subrequests and clears any feeder, but not the group request
		1761	itself. Useful when you queued a lot of events but got a result early.
		1762
		1763	The group request will finish normally (you cannot add requests to the
		1764	group).
		1765
		1766	=item $grp->result (...)
		1767
		1768	Set the result value(s) that will be passed to the group callback when all
		1769	subrequests have finished and set the groups errno to the current value
		1770	of errno (just like calling C<errno> without an error number). By default,
		1771	no argument will be passed and errno is zero.
		1772
		1773	=item $grp->errno ([$errno])
		1774
		1775	Sets the group errno value to C<$errno>, or the current value of errno
		1776	when the argument is missing.
		1777
		1778	Every aio request has an associated errno value that is restored when
		1779	the callback is invoked. This method lets you change this value from its
		1780	default (0).
		1781
		1782	Calling C<result> will also set errno, so make sure you either set C<$!>
		1783	before the call to C<result>, or call c<errno> after it.
		1784
		1785	=item feed $grp $callback->($grp)
		1786
		1787	Sets a feeder/generator on this group: every group can have an attached
		1788	generator that generates requests if idle. The idea behind this is that,
		1789	although you could just queue as many requests as you want in a group,
		1790	this might starve other requests for a potentially long time. For example,
		1791	C<aio_scandir> might generate hundreds of thousands of C<aio_stat>
		1792	requests, delaying any later requests for a long time.
		1793
		1794	To avoid this, and allow incremental generation of requests, you can
		1795	instead a group and set a feeder on it that generates those requests. The
		1796	feed callback will be called whenever there are few enough (see C<limit>,
		1797	below) requests active in the group itself and is expected to queue more
		1798	requests.
		1799
		1800	The feed callback can queue as many requests as it likes (i.e. C<add> does
		1801	not impose any limits).
		1802
		1803	If the feed does not queue more requests when called, it will be
		1804	automatically removed from the group.
		1805
		1806	If the feed limit is C<0> when this method is called, it will be set to
		1807	C<2> automatically.
		1808
		1809	Example:
		1810
		1811	# stat all files in @files, but only ever use four aio requests concurrently:
		1812
		1813	my $grp = aio_group sub { print "finished\n" };
		1814	limit $grp 4;
		1815	feed $grp sub {
		1816	my $file = pop @files
		1817	or return;
		1818
		1819	add $grp aio_stat $file, sub { ... };
		1820	};
		1821
		1822	=item limit $grp $num
		1823
		1824	Sets the feeder limit for the group: The feeder will be called whenever
		1825	the group contains less than this many requests.
		1826
		1827	Setting the limit to C<0> will pause the feeding process.
		1828
		1829	The default value for the limit is C<0>, but note that setting a feeder
		1830	automatically bumps it up to C<2>.
		1831
		1832	=back
		1833
262	=head2 SUPPORT FUNCTIONS	1834	=head2 SUPPORT FUNCTIONS
263		1835
		1836	=head3 EVENT PROCESSING AND EVENT LOOP INTEGRATION
		1837
264	=over 4	1838	=over 4
265		1839
266	=item $fileno = IO::AIO::poll_fileno	1840	=item $fileno = IO::AIO::poll_fileno
267		1841
268	Return the I<request result pipe file descriptor>. This filehandle must be	1842	Return the I<request result pipe file descriptor>. This filehandle must be
269	polled for reading by some mechanism outside this module (e.g. Event or	1843	polled for reading by some mechanism outside this module (e.g. EV, Glib,
270	select, see below or the SYNOPSIS). If the pipe becomes readable you have	1844	select and so on, see below or the SYNOPSIS). If the pipe becomes readable
271	to call C<poll_cb> to check the results.	1845	you have to call C<poll_cb> to check the results.
272		1846
273	See C<poll_cb> for an example.	1847	See C<poll_cb> for an example.
274		1848
275	=item IO::AIO::poll_cb	1849	=item IO::AIO::poll_cb
276		1850
277	Process all outstanding events on the result pipe. You have to call this	1851	Process some requests that have reached the result phase (i.e. they have
278	regularly. Returns the number of events processed. Returns immediately	1852	been executed but the results are not yet reported). You have to call
279	when no events are outstanding.	1853	this "regularly" to finish outstanding requests.
		1854
		1855	Returns C<0> if all events could be processed (or there were no
		1856	events to process), or C<-1> if it returned earlier for whatever
		1857	reason. Returns immediately when no events are outstanding. The amount
		1858	of events processed depends on the settings of C<IO::AIO::max_poll_req>,
		1859	C<IO::AIO::max_poll_time> and C<IO::AIO::max_outstanding>.
		1860
		1861	If not all requests were processed for whatever reason, the poll file
		1862	descriptor will still be ready when C<poll_cb> returns, so normally you
		1863	don't have to do anything special to have it called later.
		1864
		1865	Apart from calling C<IO::AIO::poll_cb> when the event filehandle becomes
		1866	ready, it can be beneficial to call this function from loops which submit
		1867	a lot of requests, to make sure the results get processed when they become
		1868	available and not just when the loop is finished and the event loop takes
		1869	over again. This function returns very fast when there are no outstanding
		1870	requests.
280		1871
281	Example: Install an Event watcher that automatically calls	1872	Example: Install an Event watcher that automatically calls
282	IO::AIO::poll_cb with high priority:	1873	IO::AIO::poll_cb with high priority (more examples can be found in the
		1874	SYNOPSIS section, at the top of this document):
283		1875
284	Event->io (fd => IO::AIO::poll_fileno,	1876	Event->io (fd => IO::AIO::poll_fileno,
285	poll => 'r', async => 1,	1877	poll => 'r', async => 1,
286	cb => \&IO::AIO::poll_cb);	1878	cb => \&IO::AIO::poll_cb);
287		1879
288	=item IO::AIO::poll_wait	1880	=item IO::AIO::poll_wait
289		1881
290	Wait till the result filehandle becomes ready for reading (simply does a	1882	Wait until either at least one request is in the result phase or no
291	C<select> on the filehandle. This is useful if you want to synchronously wait	1883	requests are outstanding anymore.
292	for some requests to finish).	1884
		1885	This is useful if you want to synchronously wait for some requests to
		1886	become ready, without actually handling them.
293		1887
294	See C<nreqs> for an example.	1888	See C<nreqs> for an example.
295		1889
		1890	=item IO::AIO::poll
		1891
		1892	Waits until some requests have been handled.
		1893
		1894	Returns the number of requests processed, but is otherwise strictly
		1895	equivalent to:
		1896
		1897	IO::AIO::poll_wait, IO::AIO::poll_cb
		1898
296	=item IO::AIO::nreqs	1899	=item IO::AIO::flush
297		1900
298	Returns the number of requests currently outstanding (i.e. for which their	1901	Wait till all outstanding AIO requests have been handled.
299	callback has not been invoked yet).
300		1902
301	Example: wait till there are no outstanding requests anymore:	1903	Strictly equivalent to:
302		1904
303	IO::AIO::poll_wait, IO::AIO::poll_cb	1905	IO::AIO::poll_wait, IO::AIO::poll_cb
304	while IO::AIO::nreqs;	1906	while IO::AIO::nreqs;
305		1907
306	=item IO::AIO::flush	1908	=item IO::AIO::max_poll_reqs $nreqs
307		1909
308	Wait till all outstanding AIO requests have been handled.	1910	=item IO::AIO::max_poll_time $seconds
309		1911
310	Strictly equivalent to:	1912	These set the maximum number of requests (default C<0>, meaning infinity)
		1913	that are being processed by C<IO::AIO::poll_cb> in one call, respectively
		1914	the maximum amount of time (default C<0>, meaning infinity) spent in
		1915	C<IO::AIO::poll_cb> to process requests (more correctly the mininum amount
		1916	of time C<poll_cb> is allowed to use).
311		1917
312	IO::AIO::poll_wait, IO::AIO::poll_cb	1918	Setting C<max_poll_time> to a non-zero value creates an overhead of one
313	while IO::AIO::nreqs;	1919	syscall per request processed, which is not normally a problem unless your
		1920	callbacks are really really fast or your OS is really really slow (I am
		1921	not mentioning Solaris here). Using C<max_poll_reqs> incurs no overhead.
314		1922
315	=item IO::AIO::poll	1923	Setting these is useful if you want to ensure some level of
		1924	interactiveness when perl is not fast enough to process all requests in
		1925	time.
316		1926
317	Waits until some requests have been handled.	1927	For interactive programs, values such as C<0.01> to C<0.1> should be fine.
318		1928
319	Strictly equivalent to:	1929	Example: Install an Event watcher that automatically calls
		1930	IO::AIO::poll_cb with low priority, to ensure that other parts of the
		1931	program get the CPU sometimes even under high AIO load.
320		1932
321	IO::AIO::poll_wait, IO::AIO::poll_cb	1933	# try not to spend much more than 0.1s in poll_cb
322	if IO::AIO::nreqs;	1934	IO::AIO::max_poll_time 0.1;
		1935
		1936	# use a low priority so other tasks have priority
		1937	Event->io (fd => IO::AIO::poll_fileno,
		1938	poll => 'r', nice => 1,
		1939	cb => &IO::AIO::poll_cb);
		1940
		1941	=back
		1942
		1943	=head3 CONTROLLING THE NUMBER OF THREADS
		1944
		1945	=over
323		1946
324	=item IO::AIO::min_parallel $nthreads	1947	=item IO::AIO::min_parallel $nthreads
325		1948
326	Set the minimum number of AIO threads to C<$nthreads>. The current default	1949	Set the minimum number of AIO threads to C<$nthreads>. The current
327	is C<4>, which means four asynchronous operations can be done at one time	1950	default is C<8>, which means eight asynchronous operations can execute
328	(the number of outstanding operations, however, is unlimited).	1951	concurrently at any one time (the number of outstanding requests,
		1952	however, is unlimited).
329		1953
330	IO::AIO starts threads only on demand, when an AIO request is queued and	1954	IO::AIO starts threads only on demand, when an AIO request is queued and
331	no free thread exists.	1955	no free thread exists. Please note that queueing up a hundred requests can
		1956	create demand for a hundred threads, even if it turns out that everything
		1957	is in the cache and could have been processed faster by a single thread.
332		1958
333	It is recommended to keep the number of threads low, as some Linux	1959	It is recommended to keep the number of threads relatively low, as some
334	kernel versions will scale negatively with the number of threads (higher	1960	Linux kernel versions will scale negatively with the number of threads
335	parallelity => MUCH higher latency). With current Linux 2.6 versions, 4-32	1961	(higher parallelity => MUCH higher latency). With current Linux 2.6
336	threads should be fine.	1962	versions, 4-32 threads should be fine.
337		1963
338	Under most circumstances you don't need to call this function, as the	1964	Under most circumstances you don't need to call this function, as the
339	module selects a default that is suitable for low to moderate load.	1965	module selects a default that is suitable for low to moderate load.
340		1966
341	=item IO::AIO::max_parallel $nthreads	1967	=item IO::AIO::max_parallel $nthreads
…		…
350	This module automatically runs C<max_parallel 0> at program end, to ensure	1976	This module automatically runs C<max_parallel 0> at program end, to ensure
351	that all threads are killed and that there are no outstanding requests.	1977	that all threads are killed and that there are no outstanding requests.
352		1978
353	Under normal circumstances you don't need to call this function.	1979	Under normal circumstances you don't need to call this function.
354		1980
		1981	=item IO::AIO::max_idle $nthreads
		1982
		1983	Limit the number of threads (default: 4) that are allowed to idle
		1984	(i.e., threads that did not get a request to process within the idle
		1985	timeout (default: 10 seconds). That means if a thread becomes idle while
		1986	C<$nthreads> other threads are also idle, it will free its resources and
		1987	exit.
		1988
		1989	This is useful when you allow a large number of threads (e.g. 100 or 1000)
		1990	to allow for extremely high load situations, but want to free resources
		1991	under normal circumstances (1000 threads can easily consume 30MB of RAM).
		1992
		1993	The default is probably ok in most situations, especially if thread
		1994	creation is fast. If thread creation is very slow on your system you might
		1995	want to use larger values.
		1996
		1997	=item IO::AIO::idle_timeout $seconds
		1998
		1999	Sets the minimum idle timeout (default 10) after which worker threads are
		2000	allowed to exit. SEe C<IO::AIO::max_idle>.
		2001
355	=item $oldnreqs = IO::AIO::max_outstanding $nreqs	2002	=item IO::AIO::max_outstanding $maxreqs
356		2003
357	Sets the maximum number of outstanding requests to C<$nreqs>. If you	2004	Sets the maximum number of outstanding requests to C<$nreqs>. If
358	try to queue up more than this number of requests, the caller will block until	2005	you do queue up more than this number of requests, the next call to
359	some requests have been handled.	2006	C<IO::AIO::poll_cb> (and other functions calling C<poll_cb>, such as
		2007	C<IO::AIO::flush> or C<IO::AIO::poll>) will block until the limit is no
		2008	longer exceeded.
360		2009
361	The default is very large, so normally there is no practical limit. If you	2010	In other words, this setting does not enforce a queue limit, but can be
362	queue up many requests in a loop it often improves speed if you set	2011	used to make poll functions block if the limit is exceeded.
363	this to a relatively low number, such as C<100>.
364		2012
365	Under normal circumstances you don't need to call this function.	2013	This is a very bad function to use in interactive programs because it
		2014	blocks, and a bad way to reduce concurrency because it is inexact: Better
		2015	use an C<aio_group> together with a feed callback.
		2016
		2017	Its main use is in scripts without an event loop - when you want to stat
		2018	a lot of files, you can write something like this:
		2019
		2020	IO::AIO::max_outstanding 32;
		2021
		2022	for my $path (...) {
		2023	aio_stat $path , ...;
		2024	IO::AIO::poll_cb;
		2025	}
		2026
		2027	IO::AIO::flush;
		2028
		2029	The call to C<poll_cb> inside the loop will normally return instantly, but
		2030	as soon as more thna C<32> reqeusts are in-flight, it will block until
		2031	some requests have been handled. This keeps the loop from pushing a large
		2032	number of C<aio_stat> requests onto the queue.
		2033
		2034	The default value for C<max_outstanding> is very large, so there is no
		2035	practical limit on the number of outstanding requests.
366		2036
367	=back	2037	=back
368		2038
		2039	=head3 STATISTICAL INFORMATION
		2040
		2041	=over
		2042
		2043	=item IO::AIO::nreqs
		2044
		2045	Returns the number of requests currently in the ready, execute or pending
		2046	states (i.e. for which their callback has not been invoked yet).
		2047
		2048	Example: wait till there are no outstanding requests anymore:
		2049
		2050	IO::AIO::poll_wait, IO::AIO::poll_cb
		2051	while IO::AIO::nreqs;
		2052
		2053	=item IO::AIO::nready
		2054
		2055	Returns the number of requests currently in the ready state (not yet
		2056	executed).
		2057
		2058	=item IO::AIO::npending
		2059
		2060	Returns the number of requests currently in the pending state (executed,
		2061	but not yet processed by poll_cb).
		2062
		2063	=back
		2064
		2065	=head3 MISCELLANEOUS FUNCTIONS
		2066
		2067	IO::AIO implements some functions that are useful when you want to use
		2068	some "Advanced I/O" function not available to in Perl, without going the
		2069	"Asynchronous I/O" route. Many of these have an asynchronous C<aio_*>
		2070	counterpart.
		2071
		2072	=over 4
		2073
		2074	=item $numfd = IO::AIO::get_fdlimit
		2075
		2076	This function is I<EXPERIMENTAL> and subject to change.
		2077
		2078	Tries to find the current file descriptor limit and returns it, or
		2079	C<undef> and sets C<$!> in case of an error. The limit is one larger than
		2080	the highest valid file descriptor number.
		2081
		2082	=item IO::AIO::min_fdlimit [$numfd]
		2083
		2084	This function is I<EXPERIMENTAL> and subject to change.
		2085
		2086	Try to increase the current file descriptor limit(s) to at least C<$numfd>
		2087	by changing the soft or hard file descriptor resource limit. If C<$numfd>
		2088	is missing, it will try to set a very high limit, although this is not
		2089	recommended when you know the actual minimum that you require.
		2090
		2091	If the limit cannot be raised enough, the function makes a best-effort
		2092	attempt to increase the limit as much as possible, using various
		2093	tricks, while still failing. You can query the resulting limit using
		2094	C<IO::AIO::get_fdlimit>.
		2095
		2096	If an error occurs, returns C<undef> and sets C<$!>, otherwise returns
		2097	true.
		2098
		2099	=item IO::AIO::sendfile $ofh, $ifh, $offset, $count
		2100
		2101	Calls the C<eio_sendfile_sync> function, which is like C<aio_sendfile>,
		2102	but is blocking (this makes most sense if you know the input data is
		2103	likely cached already and the output filehandle is set to non-blocking
		2104	operations).
		2105
		2106	Returns the number of bytes copied, or C<-1> on error.
		2107
		2108	=item IO::AIO::fadvise $fh, $offset, $len, $advice
		2109
		2110	Simply calls the C<posix_fadvise> function (see its
		2111	manpage for details). The following advice constants are
		2112	available: C<IO::AIO::FADV_NORMAL>, C<IO::AIO::FADV_SEQUENTIAL>,
		2113	C<IO::AIO::FADV_RANDOM>, C<IO::AIO::FADV_NOREUSE>,
		2114	C<IO::AIO::FADV_WILLNEED>, C<IO::AIO::FADV_DONTNEED>.
		2115
		2116	On systems that do not implement C<posix_fadvise>, this function returns
		2117	ENOSYS, otherwise the return value of C<posix_fadvise>.
		2118
		2119	=item IO::AIO::madvise $scalar, $offset, $len, $advice
		2120
		2121	Simply calls the C<posix_madvise> function (see its
		2122	manpage for details). The following advice constants are
		2123	available: C<IO::AIO::MADV_NORMAL>, C<IO::AIO::MADV_SEQUENTIAL>,
		2124	C<IO::AIO::MADV_RANDOM>, C<IO::AIO::MADV_WILLNEED>,
		2125	C<IO::AIO::MADV_DONTNEED>.
		2126
		2127	If C<$offset> is negative, counts from the end. If C<$length> is negative,
		2128	the remaining length of the C<$scalar> is used. If possible, C<$length>
		2129	will be reduced to fit into the C<$scalar>.
		2130
		2131	On systems that do not implement C<posix_madvise>, this function returns
		2132	ENOSYS, otherwise the return value of C<posix_madvise>.
		2133
		2134	=item IO::AIO::mprotect $scalar, $offset, $len, $protect
		2135
		2136	Simply calls the C<mprotect> function on the preferably AIO::mmap'ed
		2137	$scalar (see its manpage for details). The following protect
		2138	constants are available: C<IO::AIO::PROT_NONE>, C<IO::AIO::PROT_READ>,
		2139	C<IO::AIO::PROT_WRITE>, C<IO::AIO::PROT_EXEC>.
		2140
		2141	If C<$offset> is negative, counts from the end. If C<$length> is negative,
		2142	the remaining length of the C<$scalar> is used. If possible, C<$length>
		2143	will be reduced to fit into the C<$scalar>.
		2144
		2145	On systems that do not implement C<mprotect>, this function returns
		2146	ENOSYS, otherwise the return value of C<mprotect>.
		2147
		2148	=item IO::AIO::mmap $scalar, $length, $prot, $flags, $fh[, $offset]
		2149
		2150	Memory-maps a file (or anonymous memory range) and attaches it to the
		2151	given C<$scalar>, which will act like a string scalar. Returns true on
		2152	success, and false otherwise.
		2153
		2154	The scalar must exist, but its contents do not matter - this means you
		2155	cannot use a nonexistant array or hash element. When in doubt, C<undef>
		2156	the scalar first.
		2157
		2158	The only operations allowed on the mmapped scalar are C<substr>/C<vec>,
		2159	which don't change the string length, and most read-only operations such
		2160	as copying it or searching it with regexes and so on.
		2161
		2162	Anything else is unsafe and will, at best, result in memory leaks.
		2163
		2164	The memory map associated with the C<$scalar> is automatically removed
		2165	when the C<$scalar> is undef'd or destroyed, or when the C<IO::AIO::mmap>
		2166	or C<IO::AIO::munmap> functions are called on it.
		2167
		2168	This calls the C<mmap>(2) function internally. See your system's manual
		2169	page for details on the C<$length>, C<$prot> and C<$flags> parameters.
		2170
		2171	The C<$length> must be larger than zero and smaller than the actual
		2172	filesize.
		2173
		2174	C<$prot> is a combination of C<IO::AIO::PROT_NONE>, C<IO::AIO::PROT_EXEC>,
		2175	C<IO::AIO::PROT_READ> and/or C<IO::AIO::PROT_WRITE>,
		2176
		2177	C<$flags> can be a combination of
		2178	C<IO::AIO::MAP_SHARED> or
		2179	C<IO::AIO::MAP_PRIVATE>,
		2180	or a number of system-specific flags (when not available, the are C<0>):
		2181	C<IO::AIO::MAP_ANONYMOUS> (which is set to C<MAP_ANON> if your system only provides this constant),
		2182	C<IO::AIO::MAP_LOCKED>,
		2183	C<IO::AIO::MAP_NORESERVE>,
		2184	C<IO::AIO::MAP_POPULATE>,
		2185	C<IO::AIO::MAP_NONBLOCK>,
		2186	C<IO::AIO::MAP_FIXED>,
		2187	C<IO::AIO::MAP_GROWSDOWN>,
		2188	C<IO::AIO::MAP_32BIT>,
		2189	C<IO::AIO::MAP_HUGETLB> or
		2190	C<IO::AIO::MAP_STACK>.
		2191
		2192	If C<$fh> is C<undef>, then a file descriptor of C<-1> is passed.
		2193
		2194	C<$offset> is the offset from the start of the file - it generally must be
		2195	a multiple of C<IO::AIO::PAGESIZE> and defaults to C<0>.
		2196
		2197	Example:
		2198
		2199	use Digest::MD5;
		2200	use IO::AIO;
		2201
		2202	open my $fh, "<verybigfile"
		2203	or die "$!";
		2204
		2205	IO::AIO::mmap my $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh
		2206	or die "verybigfile: $!";
		2207
		2208	my $fast_md5 = md5 $data;
		2209
		2210	=item IO::AIO::munmap $scalar
		2211
		2212	Removes a previous mmap and undefines the C<$scalar>.
		2213
		2214	=item IO::AIO::munlock $scalar, $offset = 0, $length = undef
		2215
		2216	Calls the C<munlock> function, undoing the effects of a previous
		2217	C<aio_mlock> call (see its description for details).
		2218
		2219	=item IO::AIO::munlockall
		2220
		2221	Calls the C<munlockall> function.
		2222
		2223	On systems that do not implement C<munlockall>, this function returns
		2224	ENOSYS, otherwise the return value of C<munlockall>.
		2225
		2226	=item IO::AIO::splice $r_fh, $r_off, $w_fh, $w_off, $length, $flags
		2227
		2228	Calls the GNU/Linux C<splice(2)> syscall, if available. If C<$r_off> or
		2229	C<$w_off> are C<undef>, then C<NULL> is passed for these, otherwise they
		2230	should be the file offset.
		2231
		2232	C<$r_fh> and C<$w_fh> should not refer to the same file, as splice might
		2233	silently corrupt the data in this case.
		2234
		2235	The following symbol flag values are available: C<IO::AIO::SPLICE_F_MOVE>,
		2236	C<IO::AIO::SPLICE_F_NONBLOCK>, C<IO::AIO::SPLICE_F_MORE> and
		2237	C<IO::AIO::SPLICE_F_GIFT>.
		2238
		2239	See the C<splice(2)> manpage for details.
		2240
		2241	=item IO::AIO::tee $r_fh, $w_fh, $length, $flags
		2242
		2243	Calls the GNU/Linux C<tee(2)> syscall, see its manpage and the
		2244	description for C<IO::AIO::splice> above for details.
		2245
		2246	=item $actual_size = IO::AIO::pipesize $r_fh[, $new_size]
		2247
		2248	Attempts to query or change the pipe buffer size. Obviously works only
		2249	on pipes, and currently works only on GNU/Linux systems, and fails with
		2250	C<-1>/C<ENOSYS> everywhere else. If anybody knows how to influence pipe buffer
		2251	size on other systems, drop me a note.
		2252
		2253	=item ($rfh, $wfh) = IO::AIO::pipe2 [$flags]
		2254
		2255	This is a direct interface to the Linux L<pipe2(2)> system call. If
		2256	C<$flags> is missing or C<0>, then this should be the same as a call to
		2257	perl's built-in C<pipe> function and create a new pipe, and works on
		2258	systems that lack the pipe2 syscall. On win32, this case invokes C<_pipe
		2259	(..., 4096, O_BINARY)>.
		2260
		2261	If C<$flags> is non-zero, it tries to invoke the pipe2 system call with
		2262	the given flags (Linux 2.6.27, glibc 2.9).
		2263
		2264	On success, the read and write file handles are returned.
		2265
		2266	On error, nothing will be returned. If the pipe2 syscall is missing and
		2267	C<$flags> is non-zero, fails with C<ENOSYS>.
		2268
		2269	Please refer to L<pipe2(2)> for more info on the C<$flags>, but at the
		2270	time of this writing, C<IO::AIO::O_CLOEXEC>, C<IO::AIO::O_NONBLOCK> and
		2271	C<IO::AIO::O_DIRECT> (Linux 3.4, for packet-based pipes) were supported.
		2272
		2273	=back
		2274
369	=cut	2275	=cut
370		2276
371	# support function to convert a fd into a perl filehandle
372	sub _fd2fh {
373	return undef if $_[0] < 0;
374
375	# try to generate nice filehandles
376	my $sym = "IO::AIO::fd#$_[0]";
377	local *$sym;
378
379	open *$sym, "+<&=$_[0]" # usually works under any unix
380	or open *$sym, "<&=$_[0]" # cygwin needs this
381	or open *$sym, ">&=$_[0]" # or this
382	or return undef;
383
384	*$sym
385	}
386
387	min_parallel 4;	2277	min_parallel 8;
388		2278
389	END {	2279	END { flush }
390	max_parallel 0;
391	}
392		2280
393	1;	2281	1;
394		2282
		2283	=head1 EVENT LOOP INTEGRATION
		2284
		2285	It is recommended to use L<AnyEvent::AIO> to integrate IO::AIO
		2286	automatically into many event loops:
		2287
		2288	# AnyEvent integration (EV, Event, Glib, Tk, POE, urxvt, pureperl...)
		2289	use AnyEvent::AIO;
		2290
		2291	You can also integrate IO::AIO manually into many event loops, here are
		2292	some examples of how to do this:
		2293
		2294	# EV integration
		2295	my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
		2296
		2297	# Event integration
		2298	Event->io (fd => IO::AIO::poll_fileno,
		2299	poll => 'r',
		2300	cb => \&IO::AIO::poll_cb);
		2301
		2302	# Glib/Gtk2 integration
		2303	add_watch Glib::IO IO::AIO::poll_fileno,
		2304	in => sub { IO::AIO::poll_cb; 1 };
		2305
		2306	# Tk integration
		2307	Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",
		2308	readable => \&IO::AIO::poll_cb);
		2309
		2310	# Danga::Socket integration
		2311	Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
		2312	\&IO::AIO::poll_cb);
		2313
395	=head2 FORK BEHAVIOUR	2314	=head2 FORK BEHAVIOUR
396		2315
397	Before the fork, IO::AIO enters a quiescent state where no requests	2316	Usage of pthreads in a program changes the semantics of fork
398	can be added in other threads and no results will be processed. After	2317	considerably. Specifically, only async-safe functions can be called after
399	the fork the parent simply leaves the quiescent state and continues	2318	fork. Perl doesn't know about this, so in general, you cannot call fork
400	request/result processing, while the child clears the request/result	2319	with defined behaviour in perl if pthreads are involved. IO::AIO uses
401	queue (so the requests started before the fork will only be handled in	2320	pthreads, so this applies, but many other extensions and (for inexplicable
402	the parent). Threats will be started on demand until the limit ste in the	2321	reasons) perl itself often is linked against pthreads, so this limitation
403	parent process has been reached again.	2322	applies to quite a lot of perls.
		2323
		2324	This module no longer tries to fight your OS, or POSIX. That means IO::AIO
		2325	only works in the process that loaded it. Forking is fully supported, but
		2326	using IO::AIO in the child is not.
		2327
		2328	You might get around by not I<using> IO::AIO before (or after)
		2329	forking. You could also try to call the L<IO::AIO::reinit> function in the
		2330	child:
		2331
		2332	=over 4
		2333
		2334	=item IO::AIO::reinit
		2335
		2336	Abandons all current requests and I/O threads and simply reinitialises all
		2337	data structures. This is not an operation supported by any standards, but
		2338	happens to work on GNU/Linux and some newer BSD systems.
		2339
		2340	The only reasonable use for this function is to call it after forking, if
		2341	C<IO::AIO> was used in the parent. Calling it while IO::AIO is active in
		2342	the process will result in undefined behaviour. Calling it at any time
		2343	will also result in any undefined (by POSIX) behaviour.
		2344
		2345	=back
		2346
		2347	=head2 MEMORY USAGE
		2348
		2349	Per-request usage:
		2350
		2351	Each aio request uses - depending on your architecture - around 100-200
		2352	bytes of memory. In addition, stat requests need a stat buffer (possibly
		2353	a few hundred bytes), readdir requires a result buffer and so on. Perl
		2354	scalars and other data passed into aio requests will also be locked and
		2355	will consume memory till the request has entered the done state.
		2356
		2357	This is not awfully much, so queuing lots of requests is not usually a
		2358	problem.
		2359
		2360	Per-thread usage:
		2361
		2362	In the execution phase, some aio requests require more memory for
		2363	temporary buffers, and each thread requires a stack and other data
		2364	structures (usually around 16k-128k, depending on the OS).
		2365
		2366	=head1 KNOWN BUGS
		2367
		2368	Known bugs will be fixed in the next release.
404		2369
405	=head1 SEE ALSO	2370	=head1 SEE ALSO
406		2371
407	L<Coro>, L<Linux::AIO>.	2372	L<AnyEvent::AIO> for easy integration into event loops, L<Coro::AIO> for a
		2373	more natural syntax.
408		2374
409	=head1 AUTHOR	2375	=head1 AUTHOR
410		2376
411	Marc Lehmann <schmorp@schmorp.de>	2377	Marc Lehmann <schmorp@schmorp.de>
412	http://home.schmorp.de/	2378	http://home.schmorp.de/

Diff Legend

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