ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/IO-AIO/AIO.pm

Comparing IO-AIO/AIO.pm (file contents):
Revision 1.9 by root, Sun Jul 10 22:20:55 2005 UTC vs.
Revision 1.317 by root, Sun Sep 25 16:30:50 2022 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	\&IO::AIO::poll_cb;	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		30
33	=head1 DESCRIPTION	31	=head1 DESCRIPTION
34		32
35	This module implements asynchronous I/O using whatever means your	33	This module implements asynchronous I/O using whatever means your
36	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>).
37		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
38	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
39	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
40	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
41	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
42	aio functions available on many operating systems. However, they are often	57	functions available on many operating systems. However, they are often
43	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
44	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
45	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.
46		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
47	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,
48	currently not reentrant, so use appropriate locking yourself.	68	it is currently not reentrant in any way, so use appropriate locking
		69	yourself, always call C<poll_cb> from within the same thread, or never
		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
49		164
50	=cut	165	=cut
51		166
52	package IO::AIO;	167	package IO::AIO;
53		168
		169	use Carp ();
		170
		171	use common::sense;
		172
54	use base 'Exporter';	173	use base 'Exporter';
55		174
56	use Fcntl ();
57
58	BEGIN {	175	BEGIN {
59	$VERSION = 0.2;	176	our $VERSION = 4.79;
60		177
61	@EXPORT = qw(aio_read aio_write aio_open aio_close aio_stat aio_lstat aio_unlink	178	our @AIO_REQ = qw(aio_sendfile aio_seek aio_read aio_write aio_open aio_close
62	aio_fsync aio_fdatasync aio_readahead);	179	aio_stat aio_lstat aio_unlink aio_rmdir aio_readdir aio_readdirx
63	@EXPORT_OK = qw(poll_fileno poll_cb min_parallel max_parallel max_outstanding nreqs);	180	aio_scandir aio_symlink aio_readlink aio_realpath aio_fcntl aio_ioctl
		181	aio_sync aio_fsync aio_syncfs aio_fdatasync aio_sync_file_range
		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_slurp
		189	aio_wd);
		190
		191	our @EXPORT = (@AIO_REQ, qw(aioreq_pri aioreq_nice));
		192	our @EXPORT_OK = qw(poll_fileno poll_cb poll_wait flush
		193	min_parallel max_parallel max_idle idle_timeout
		194	nreqs nready npending nthreads
		195	max_poll_time max_poll_reqs
		196	sendfile fadvise madvise
		197	mmap munmap mremap munlock munlockall
		198
		199	accept4 tee splice pipe2 pipesize
		200	fexecve mount umount memfd_create eventfd
		201	timerfd_create timerfd_settime timerfd_gettime
		202	pidfd_open pidfd_send_signal pidfd_getfd);
		203
		204	push @AIO_REQ, qw(aio_busy); # not exported
		205
		206	@IO::AIO::GRP::ISA = 'IO::AIO::REQ';
64		207
65	require XSLoader;	208	require XSLoader;
66	XSLoader::load IO::AIO, $VERSION;	209	XSLoader::load ("IO::AIO", $VERSION);
67	}	210	}
68		211
69	=head1 FUNCTIONS	212	=head1 FUNCTIONS
70		213
71	=head2 AIO FUNCTIONS	214	=head2 QUICK OVERVIEW
		215
		216	This section simply lists the prototypes most of the functions for
		217	quick reference. See the following sections for function-by-function
		218	documentation.
		219
		220	aio_wd $pathname, $callback->($wd)
		221	aio_open $pathname, $flags, $mode, $callback->($fh)
		222	aio_close $fh, $callback->($status)
		223	aio_seek $fh,$offset,$whence, $callback->($offs)
		224	aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
		225	aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
		226	aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
		227	aio_readahead $fh,$offset,$length, $callback->($retval)
		228	aio_stat $fh_or_path, $callback->($status)
		229	aio_lstat $fh, $callback->($status)
		230	aio_statvfs $fh_or_path, $callback->($statvfs)
		231	aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
		232	aio_chown $fh_or_path, $uid, $gid, $callback->($status)
		233	aio_chmod $fh_or_path, $mode, $callback->($status)
		234	aio_truncate $fh_or_path, $offset, $callback->($status)
		235	aio_allocate $fh, $mode, $offset, $len, $callback->($status)
		236	aio_fiemap $fh, $start, $length, $flags, $count, $cb->(\@extents)
		237	aio_unlink $pathname, $callback->($status)
		238	aio_mknod $pathname, $mode, $dev, $callback->($status)
		239	aio_link $srcpath, $dstpath, $callback->($status)
		240	aio_symlink $srcpath, $dstpath, $callback->($status)
		241	aio_readlink $pathname, $callback->($link)
		242	aio_realpath $pathname, $callback->($path)
		243	aio_rename $srcpath, $dstpath, $callback->($status)
		244	aio_rename2 $srcpath, $dstpath, $flags, $callback->($status)
		245	aio_mkdir $pathname, $mode, $callback->($status)
		246	aio_rmdir $pathname, $callback->($status)
		247	aio_readdir $pathname, $callback->($entries)
		248	aio_readdirx $pathname, $flags, $callback->($entries, $flags)
		249	IO::AIO::READDIR_DENTS IO::AIO::READDIR_DIRS_FIRST
		250	IO::AIO::READDIR_STAT_ORDER IO::AIO::READDIR_FOUND_UNKNOWN
		251	aio_scandir $pathname, $maxreq, $callback->($dirs, $nondirs)
		252	aio_load $pathname, $data, $callback->($status)
		253	aio_copy $srcpath, $dstpath, $callback->($status)
		254	aio_move $srcpath, $dstpath, $callback->($status)
		255	aio_rmtree $pathname, $callback->($status)
		256	aio_fcntl $fh, $cmd, $arg, $callback->($status)
		257	aio_ioctl $fh, $request, $buf, $callback->($status)
		258	aio_sync $callback->($status)
		259	aio_syncfs $fh, $callback->($status)
		260	aio_fsync $fh, $callback->($status)
		261	aio_fdatasync $fh, $callback->($status)
		262	aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
		263	aio_pathsync $pathname, $callback->($status)
		264	aio_msync $scalar, $offset = 0, $length = undef, flags = MS_SYNC, $callback->($status)
		265	aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		266	aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status)
		267	aio_mlockall $flags, $callback->($status)
		268	aio_group $callback->(...)
		269	aio_nop $callback->()
		270
		271	$prev_pri = aioreq_pri [$pri]
		272	aioreq_nice $pri_adjust
		273
		274	IO::AIO::poll_wait
		275	IO::AIO::poll_cb
		276	IO::AIO::poll
		277	IO::AIO::flush
		278	IO::AIO::max_poll_reqs $nreqs
		279	IO::AIO::max_poll_time $seconds
		280	IO::AIO::min_parallel $nthreads
		281	IO::AIO::max_parallel $nthreads
		282	IO::AIO::max_idle $nthreads
		283	IO::AIO::idle_timeout $seconds
		284	IO::AIO::max_outstanding $maxreqs
		285	IO::AIO::nreqs
		286	IO::AIO::nready
		287	IO::AIO::npending
		288	IO::AIO::reinit
		289
		290	$nfd = IO::AIO::get_fdlimit
		291	IO::AIO::min_fdlimit $nfd
		292
		293	IO::AIO::sendfile $ofh, $ifh, $offset, $count
		294	IO::AIO::fadvise $fh, $offset, $len, $advice
		295	IO::AIO::fexecve $fh, $argv, $envp
		296
		297	IO::AIO::mmap $scalar, $length, $prot, $flags[, $fh[, $offset]]
		298	IO::AIO::munmap $scalar
		299	IO::AIO::mremap $scalar, $new_length, $flags[, $new_address]
		300	IO::AIO::madvise $scalar, $offset, $length, $advice
		301	IO::AIO::mprotect $scalar, $offset, $length, $protect
		302	IO::AIO::munlock $scalar, $offset = 0, $length = undef
		303	IO::AIO::munlockall
		304
		305	# stat extensions
		306	$counter = IO::AIO::st_gen
		307	$seconds = IO::AIO::st_atime, IO::AIO::st_mtime, IO::AIO::st_ctime, IO::AIO::st_btime
		308	($atime, $mtime, $ctime, $btime, ...) = IO::AIO::st_xtime
		309	$nanoseconds = IO::AIO::st_atimensec, IO::AIO::st_mtimensec, IO::AIO::st_ctimensec, IO::AIO::st_btimensec
		310	$seconds = IO::AIO::st_btimesec
		311	($atime, $mtime, $ctime, $btime, ...) = IO::AIO::st_xtimensec
		312
		313	# very much unportable syscalls
		314	IO::AIO::accept4 $r_fh, $sockaddr, $sockaddr_len, $flags
		315	IO::AIO::splice $r_fh, $r_off, $w_fh, $w_off, $length, $flags
		316	IO::AIO::tee $r_fh, $w_fh, $length, $flags
		317
		318	$actual_size = IO::AIO::pipesize $r_fh[, $new_size]
		319	($rfh, $wfh) = IO::AIO::pipe2 [$flags]
		320
		321	$fh = IO::AIO::eventfd [$initval, [$flags]]
		322	$fh = IO::AIO::memfd_create $pathname[, $flags]
		323
		324	$fh = IO::AIO::timerfd_create $clockid[, $flags]
		325	($cur_interval, $cur_value) = IO::AIO::timerfd_settime $fh, $flags, $new_interval, $nbw_value
		326	($cur_interval, $cur_value) = IO::AIO::timerfd_gettime $fh
		327
		328	$fh = IO::AIO::pidfd_open $pid[, $flags]
		329	$status = IO::AIO::pidfd_send_signal $pidfh, $signal[, $siginfo[, $flags]]
		330	$fh = IO::AIO::pidfd_getfd $pidfh, $targetfd[, $flags]
		331
		332	$retval = IO::AIO::mount $special, $path, $fstype, $flags = 0, $data = undef
		333	$retval = IO::AIO::umount $path, $flags = 0
		334
		335	=head2 API NOTES
72		336
73	All the C<aio_*> calls are more or less thin wrappers around the syscall	337	All the C<aio_*> calls are more or less thin wrappers around the syscall
74	with the same name (sans C<aio_>). The arguments are similar or identical,	338	with the same name (sans C<aio_>). The arguments are similar or identical,
75	and they all accept an additional C<$callback> argument which must be	339	and they all accept an additional (and optional) C<$callback> argument
76	a code reference. This code reference will get called with the syscall	340	which must be a code reference. This code reference will be called after
		341	the syscall has been executed in an asynchronous fashion. The results
		342	of the request will be passed as arguments to the callback (and, if an
		343	error occured, in C<$!>) - for most requests the syscall return code (e.g.
77	return code (e.g. most syscalls return C<-1> on error, unlike perl, which	344	most syscalls return C<-1> on error, unlike perl, which usually delivers
78	usually delivers "false") as it's sole argument when the given syscall has	345	"false").
79	been executed asynchronously.
80		346
81	All functions that expect a filehandle will also accept a file descriptor.	347	Some requests (such as C<aio_readdir>) pass the actual results and
		348	communicate failures by passing C<undef>.
82		349
		350	All functions expecting a filehandle keep a copy of the filehandle
		351	internally until the request has finished.
		352
		353	All functions return request objects of type L<IO::AIO::REQ> that allow
		354	further manipulation of those requests while they are in-flight.
		355
83	The filenames you pass to these routines I<must> be absolute. The reason	356	The pathnames you pass to these routines I<should> be absolute. The
84	is that at the time the request is being executed, the current working	357	reason for this is that at the time the request is being executed, the
85	directory could have changed. Alternatively, you can make sure that you	358	current working directory could have changed. Alternatively, you can
86	never change the current working directory.	359	make sure that you never change the current working directory anywhere
		360	in the program and then use relative paths. You can also take advantage
		361	of IO::AIOs working directory abstraction, that lets you specify paths
		362	relative to some previously-opened "working directory object" - see the
		363	description of the C<IO::AIO::WD> class later in this document.
		364
		365	To encode pathnames as octets, either make sure you either: a) always pass
		366	in filenames you got from outside (command line, readdir etc.) without
		367	tinkering, b) are in your native filesystem encoding, c) use the Encode
		368	module and encode your pathnames to the locale (or other) encoding in
		369	effect in the user environment, d) use Glib::filename_from_unicode on
		370	unicode filenames or e) use something else to ensure your scalar has the
		371	correct contents.
		372
		373	This works, btw. independent of the internal UTF-8 bit, which IO::AIO
		374	handles correctly whether it is set or not.
		375
		376	=head2 AIO REQUEST FUNCTIONS
87		377
88	=over 4	378	=over 4
89		379
		380	=item $prev_pri = aioreq_pri [$pri]
		381
		382	Returns the priority value that would be used for the next request and, if
		383	C<$pri> is given, sets the priority for the next aio request.
		384
		385	The default priority is C<0>, the minimum and maximum priorities are C<-4>
		386	and C<4>, respectively. Requests with higher priority will be serviced
		387	first.
		388
		389	The priority will be reset to C<0> after each call to one of the C<aio_*>
		390	functions.
		391
		392	Example: open a file with low priority, then read something from it with
		393	higher priority so the read request is serviced before other low priority
		394	open requests (potentially spamming the cache):
		395
		396	aioreq_pri -3;
		397	aio_open ..., sub {
		398	return unless $_[0];
		399
		400	aioreq_pri -2;
		401	aio_read $_[0], ..., sub {
		402	...
		403	};
		404	};
		405
		406
		407	=item aioreq_nice $pri_adjust
		408
		409	Similar to C<aioreq_pri>, but subtracts the given value from the current
		410	priority, so the effect is cumulative.
		411
		412
90	=item aio_open $pathname, $flags, $mode, $callback	413	=item aio_open $pathname, $flags, $mode, $callback->($fh)
91		414
92	Asynchronously open or create a file and call the callback with a newly	415	Asynchronously open or create a file and call the callback with a newly
93	created filehandle for the file.	416	created filehandle for the file (or C<undef> in case of an error).
94		417
95	The pathname passed to C<aio_open> must be absolute. See API NOTES, above,	418	The pathname passed to C<aio_open> must be absolute. See API NOTES, above,
96	for an explanation.	419	for an explanation.
97		420
98	The C<$mode> argument is a bitmask. See the C<Fcntl> module for a	421	The C<$flags> argument is a bitmask. See the C<Fcntl> module for a
99	list. They are the same as used in C<sysopen>.	422	list. They are the same as used by C<sysopen>.
		423
		424	Likewise, C<$mode> specifies the mode of the newly created file, if it
		425	didn't exist and C<O_CREAT> has been given, just like perl's C<sysopen>,
		426	except that it is mandatory (i.e. use C<0> if you don't create new files,
		427	and C<0666> or C<0777> if you do). Note that the C<$mode> will be modified
		428	by the umask in effect then the request is being executed, so better never
		429	change the umask.
100		430
101	Example:	431	Example:
102		432
103	aio_open "/etc/passwd", O_RDONLY, 0, sub {	433	aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub {
104	if ($_[0]) {	434	if ($_[0]) {
105	print "open successful, fh is $_[0]\n";	435	print "open successful, fh is $_[0]\n";
106	...	436	...
107	} else {	437	} else {
108	die "open failed: $!\n";	438	die "open failed: $!\n";
109	}	439	}
110	};	440	};
111		441
		442	In addition to all the common open modes/flags (C<O_RDONLY>, C<O_WRONLY>,
		443	C<O_RDWR>, C<O_CREAT>, C<O_TRUNC>, C<O_EXCL> and C<O_APPEND>), the
		444	following POSIX and non-POSIX constants are available (missing ones on
		445	your system are, as usual, C<0>):
		446
		447	C<O_ASYNC>, C<O_DIRECT>, C<O_NOATIME>, C<O_CLOEXEC>, C<O_NOCTTY>, C<O_NOFOLLOW>,
		448	C<O_NONBLOCK>, C<O_EXEC>, C<O_SEARCH>, C<O_DIRECTORY>, C<O_DSYNC>,
		449	C<O_RSYNC>, C<O_SYNC>, C<O_PATH>, C<O_TMPFILE>, C<O_TTY_INIT> and C<O_ACCMODE>.
		450
		451
112	=item aio_close $fh, $callback	452	=item aio_close $fh, $callback->($status)
113		453
114	Asynchronously close a file and call the callback with the result	454	Asynchronously close a file and call the callback with the result
115	code. I<WARNING:> although accepted, you should not pass in a perl	455	code.
116	filehandle here, as perl will likely close the file descriptor itself when
117	the filehandle is destroyed. Normally, you can safely call perls C<close>
118	or just let filehandles go out of scope.
119		456
		457	Unfortunately, you can't do this to perl. Perl I<insists> very strongly on
		458	closing the file descriptor associated with the filehandle itself.
		459
		460	Therefore, C<aio_close> will not close the filehandle - instead it will
		461	use dup2 to overwrite the file descriptor with the write-end of a pipe
		462	(the pipe fd will be created on demand and will be cached).
		463
		464	Or in other words: the file descriptor will be closed, but it will not be
		465	free for reuse until the perl filehandle is closed.
		466
		467	=cut
		468
		469	=item aio_seek $fh, $offset, $whence, $callback->($offs)
		470
		471	Seeks the filehandle to the new C<$offset>, similarly to perl's
		472	C<sysseek>. The C<$whence> can use the traditional values (C<0> for
		473	C<IO::AIO::SEEK_SET>, C<1> for C<IO::AIO::SEEK_CUR> or C<2> for
		474	C<IO::AIO::SEEK_END>).
		475
		476	The resulting absolute offset will be passed to the callback, or C<-1> in
		477	case of an error.
		478
		479	In theory, the C<$whence> constants could be different than the
		480	corresponding values from L<Fcntl>, but perl guarantees they are the same,
		481	so don't panic.
		482
		483	As a GNU/Linux (and maybe Solaris) extension, also the constants
		484	C<IO::AIO::SEEK_DATA> and C<IO::AIO::SEEK_HOLE> are available, if they
		485	could be found. No guarantees about suitability for use in C<aio_seek> or
		486	Perl's C<sysseek> can be made though, although I would naively assume they
		487	"just work".
		488
120	=item aio_read $fh,$offset,$length, $data,$dataoffset,$callback	489	=item aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
121		490
122	=item aio_write $fh,$offset,$length, $data,$dataoffset,$callback	491	=item aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
123		492
124	Reads or writes C<length> bytes from the specified C<fh> and C<offset>	493	Reads or writes C<$length> bytes from or to the specified C<$fh> and
125	into the scalar given by C<data> and offset C<dataoffset> and calls the	494	C<$offset> into the scalar given by C<$data> and offset C<$dataoffset> and
126	callback without the actual number of bytes read (or -1 on error, just	495	calls the callback with the actual number of bytes transferred (or -1 on
127	like the syscall).	496	error, just like the syscall).
128		497
		498	C<aio_read> will, like C<sysread>, shrink or grow the C<$data> scalar to
		499	offset plus the actual number of bytes read.
		500
		501	If C<$offset> is undefined, then the current file descriptor offset will
		502	be used (and updated), otherwise the file descriptor offset will not be
		503	changed by these calls.
		504
		505	If C<$length> is undefined in C<aio_write>, use the remaining length of
		506	C<$data>.
		507
		508	If C<$dataoffset> is less than zero, it will be counted from the end of
		509	C<$data>.
		510
		511	The C<$data> scalar I<MUST NOT> be modified in any way while the request
		512	is outstanding. Modifying it can result in segfaults or World War III (if
		513	the necessary/optional hardware is installed).
		514
129	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, strating at	515	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at
130	offset C<0> within the scalar:	516	offset C<0> within the scalar:
131		517
132	aio_read $fh, 7, 15, $buffer, 0, sub {	518	aio_read $fh, 7, 15, $buffer, 0, sub {
133	$_[0] > 0 or die "read error: $!";	519	$_[0] > 0 or die "read error: $!";
134	print "read $_[0] bytes: <$buffer>\n";	520	print "read $_[0] bytes: <$buffer>\n";
135	};	521	};
136		522
		523
		524	=item aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
		525
		526	Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts
		527	reading at byte offset C<$in_offset>, and starts writing at the current
		528	file offset of C<$out_fh>. Because of that, it is not safe to issue more
		529	than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each
		530	other. The same C<$in_fh> works fine though, as this function does not
		531	move or use the file offset of C<$in_fh>.
		532
		533	Please note that C<aio_sendfile> can read more bytes from C<$in_fh> than
		534	are written, and there is no way to find out how many more bytes have been
		535	read from C<aio_sendfile> alone, as C<aio_sendfile> only provides the
		536	number of bytes written to C<$out_fh>. Only if the result value equals
		537	C<$length> one can assume that C<$length> bytes have been read.
		538
		539	Unlike with other C<aio_> functions, it makes a lot of sense to use
		540	C<aio_sendfile> on non-blocking sockets, as long as one end (typically
		541	the C<$in_fh>) is a file - the file I/O will then be asynchronous, while
		542	the socket I/O will be non-blocking. Note, however, that you can run
		543	into a trap where C<aio_sendfile> reads some data with readahead, then
		544	fails to write all data, and when the socket is ready the next time, the
		545	data in the cache is already lost, forcing C<aio_sendfile> to again hit
		546	the disk. Explicit C<aio_read> + C<aio_write> let's you better control
		547	resource usage.
		548
		549	This call tries to make use of a native C<sendfile>-like syscall to
		550	provide zero-copy operation. For this to work, C<$out_fh> should refer to
		551	a socket, and C<$in_fh> should refer to an mmap'able file.
		552
		553	If a native sendfile cannot be found or it fails with C<ENOSYS>,
		554	C<EINVAL>, C<ENOTSUP>, C<EOPNOTSUPP>, C<EAFNOSUPPORT>, C<EPROTOTYPE> or
		555	C<ENOTSOCK>, it will be emulated, so you can call C<aio_sendfile> on any
		556	type of filehandle regardless of the limitations of the operating system.
		557
		558	As native sendfile syscalls (as practically any non-POSIX interface hacked
		559	together in a hurry to improve benchmark numbers) tend to be rather buggy
		560	on many systems, this implementation tries to work around some known bugs
		561	in Linux and FreeBSD kernels (probably others, too), but that might fail,
		562	so you really really should check the return value of C<aio_sendfile> -
		563	fewer bytes than expected might have been transferred.
		564
		565
137	=item aio_readahead $fh,$offset,$length, $callback	566	=item aio_readahead $fh,$offset,$length, $callback->($retval)
138		567
139	Asynchronously reads the specified byte range into the page cache, using
140	the C<readahead> syscall. If that syscall doesn't exist the status will be
141	C<-1> and C<$!> is set to ENOSYS.
142
143	readahead() populates the page cache with data from a file so that	568	C<aio_readahead> populates the page cache with data from a file so that
144	subsequent reads from that file will not block on disk I/O. The C<$offset>	569	subsequent reads from that file will not block on disk I/O. The C<$offset>
145	argument specifies the starting point from which data is to be read and	570	argument specifies the starting point from which data is to be read and
146	C<$length> specifies the number of bytes to be read. I/O is performed in	571	C<$length> specifies the number of bytes to be read. I/O is performed in
147	whole pages, so that offset is effectively rounded down to a page boundary	572	whole pages, so that offset is effectively rounded down to a page boundary
148	and bytes are read up to the next page boundary greater than or equal to	573	and bytes are read up to the next page boundary greater than or equal to
149	(off-set+length). aio_readahead() does not read beyond the end of the	574	(off-set+length). C<aio_readahead> does not read beyond the end of the
150	file. The current file offset of the file is left unchanged.	575	file. The current file offset of the file is left unchanged.
151		576
		577	If that syscall doesn't exist (likely if your kernel isn't Linux) it will
		578	be emulated by simply reading the data, which would have a similar effect.
		579
		580
152	=item aio_stat $fh_or_path, $callback	581	=item aio_stat $fh_or_path, $callback->($status)
153		582
154	=item aio_lstat $fh, $callback	583	=item aio_lstat $fh, $callback->($status)
155		584
156	Works like perl's C<stat> or C<lstat> in void context. The callback will	585	Works almost exactly like perl's C<stat> or C<lstat> in void context. The
157	be called after the stat and the results will be available using C<stat _>	586	callback will be called after the stat and the results will be available
158	or C<-s _> etc...	587	using C<stat _> or C<-s _> and other tests (with the exception of C<-B>
		588	and C<-T>).
159		589
160	The pathname passed to C<aio_stat> must be absolute. See API NOTES, above,	590	The pathname passed to C<aio_stat> must be absolute. See API NOTES, above,
161	for an explanation.	591	for an explanation.
162		592
163	Currently, the stats are always 64-bit-stats, i.e. instead of returning an	593	Currently, the stats are always 64-bit-stats, i.e. instead of returning an
164	error when stat'ing a large file, the results will be silently truncated	594	error when stat'ing a large file, the results will be silently truncated
165	unless perl itself is compiled with large file support.	595	unless perl itself is compiled with large file support.
		596
		597	To help interpret the mode and dev/rdev stat values, IO::AIO offers the
		598	following constants and functions (if not implemented, the constants will
		599	be C<0> and the functions will either C<croak> or fall back on traditional
		600	behaviour).
		601
		602	C<S_IFMT>, C<S_IFIFO>, C<S_IFCHR>, C<S_IFBLK>, C<S_IFLNK>, C<S_IFREG>,
		603	C<S_IFDIR>, C<S_IFWHT>, C<S_IFSOCK>, C<IO::AIO::major $dev_t>,
		604	C<IO::AIO::minor $dev_t>, C<IO::AIO::makedev $major, $minor>.
		605
		606	To access higher resolution stat timestamps, see L<SUBSECOND STAT TIME
		607	ACCESS>.
166		608
167	Example: Print the length of F</etc/passwd>:	609	Example: Print the length of F</etc/passwd>:
168		610
169	aio_stat "/etc/passwd", sub {	611	aio_stat "/etc/passwd", sub {
170	$_[0] and die "stat failed: $!";	612	$_[0] and die "stat failed: $!";
171	print "size is ", -s _, "\n";	613	print "size is ", -s _, "\n";
172	};	614	};
173		615
		616
		617	=item aio_statvfs $fh_or_path, $callback->($statvfs)
		618
		619	Works like the POSIX C<statvfs> or C<fstatvfs> syscalls, depending on
		620	whether a file handle or path was passed.
		621
		622	On success, the callback is passed a hash reference with the following
		623	members: C<bsize>, C<frsize>, C<blocks>, C<bfree>, C<bavail>, C<files>,
		624	C<ffree>, C<favail>, C<fsid>, C<flag> and C<namemax>. On failure, C<undef>
		625	is passed.
		626
		627	The following POSIX IO::AIO::ST_* constants are defined: C<ST_RDONLY> and
		628	C<ST_NOSUID>.
		629
		630	The following non-POSIX IO::AIO::ST_* flag masks are defined to
		631	their correct value when available, or to C<0> on systems that do
		632	not support them: C<ST_NODEV>, C<ST_NOEXEC>, C<ST_SYNCHRONOUS>,
		633	C<ST_MANDLOCK>, C<ST_WRITE>, C<ST_APPEND>, C<ST_IMMUTABLE>, C<ST_NOATIME>,
		634	C<ST_NODIRATIME> and C<ST_RELATIME>.
		635
		636	Example: stat C</wd> and dump out the data if successful.
		637
		638	aio_statvfs "/wd", sub {
		639	my $f = $_[0]
		640	or die "statvfs: $!";
		641
		642	use Data::Dumper;
		643	say Dumper $f;
		644	};
		645
		646	# result:
		647	{
		648	bsize => 1024,
		649	bfree => 4333064312,
		650	blocks => 10253828096,
		651	files => 2050765568,
		652	flag => 4096,
		653	favail => 2042092649,
		654	bavail => 4333064312,
		655	ffree => 2042092649,
		656	namemax => 255,
		657	frsize => 1024,
		658	fsid => 1810
		659	}
		660
		661	=item aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
		662
		663	Works like perl's C<utime> function (including the special case of $atime
		664	and $mtime being undef). Fractional times are supported if the underlying
		665	syscalls support them.
		666
		667	When called with a pathname, uses utimensat(2) or utimes(2) if available,
		668	otherwise utime(2). If called on a file descriptor, uses futimens(2)
		669	or futimes(2) if available, otherwise returns ENOSYS, so this is not
		670	portable.
		671
		672	Examples:
		673
		674	# set atime and mtime to current time (basically touch(1)):
		675	aio_utime "path", undef, undef;
		676	# set atime to current time and mtime to beginning of the epoch:
		677	aio_utime "path", time, undef; # undef==0
		678
		679
		680	=item aio_chown $fh_or_path, $uid, $gid, $callback->($status)
		681
		682	Works like perl's C<chown> function, except that C<undef> for either $uid
		683	or $gid is being interpreted as "do not change" (but -1 can also be used).
		684
		685	Examples:
		686
		687	# same as "chown root path" in the shell:
		688	aio_chown "path", 0, -1;
		689	# same as above:
		690	aio_chown "path", 0, undef;
		691
		692
		693	=item aio_truncate $fh_or_path, $offset, $callback->($status)
		694
		695	Works like truncate(2) or ftruncate(2).
		696
		697
		698	=item aio_allocate $fh, $mode, $offset, $len, $callback->($status)
		699
		700	Allocates or frees disk space according to the C<$mode> argument. See the
		701	linux C<fallocate> documentation for details.
		702
		703	C<$mode> is usually C<0> or C<IO::AIO::FALLOC_FL_KEEP_SIZE> to allocate
		704	space, or C<IO::AIO::FALLOC_FL_PUNCH_HOLE | IO::AIO::FALLOC_FL_KEEP_SIZE>,
		705	to deallocate a file range.
		706
		707	IO::AIO also supports C<FALLOC_FL_COLLAPSE_RANGE>, to remove a range
		708	(without leaving a hole), C<FALLOC_FL_ZERO_RANGE>, to zero a range,
		709	C<FALLOC_FL_INSERT_RANGE> to insert a range and C<FALLOC_FL_UNSHARE_RANGE>
		710	to unshare shared blocks (see your L<fallocate(2)> manpage).
		711
		712	The file system block size used by C<fallocate> is presumably the
		713	C<f_bsize> returned by C<statvfs>, but different filesystems and filetypes
		714	can dictate other limitations.
		715
		716	If C<fallocate> isn't available or cannot be emulated (currently no
		717	emulation will be attempted), passes C<-1> and sets C<$!> to C<ENOSYS>.
		718
		719
		720	=item aio_chmod $fh_or_path, $mode, $callback->($status)
		721
		722	Works like perl's C<chmod> function.
		723
		724
174	=item aio_unlink $pathname, $callback	725	=item aio_unlink $pathname, $callback->($status)
175		726
176	Asynchronously unlink (delete) a file and call the callback with the	727	Asynchronously unlink (delete) a file and call the callback with the
177	result code.	728	result code.
178		729
		730
		731	=item aio_mknod $pathname, $mode, $dev, $callback->($status)
		732
		733	[EXPERIMENTAL]
		734
		735	Asynchronously create a device node (or fifo). See mknod(2).
		736
		737	The only (POSIX-) portable way of calling this function is:
		738
		739	aio_mknod $pathname, IO::AIO::S_IFIFO | $mode, 0, sub { ...
		740
		741	See C<aio_stat> for info about some potentially helpful extra constants
		742	and functions.
		743
		744	=item aio_link $srcpath, $dstpath, $callback->($status)
		745
		746	Asynchronously create a new link to the existing object at C<$srcpath> at
		747	the path C<$dstpath> and call the callback with the result code.
		748
		749
		750	=item aio_symlink $srcpath, $dstpath, $callback->($status)
		751
		752	Asynchronously create a new symbolic link to the existing object at C<$srcpath> at
		753	the path C<$dstpath> and call the callback with the result code.
		754
		755
		756	=item aio_readlink $pathname, $callback->($link)
		757
		758	Asynchronously read the symlink specified by C<$path> and pass it to
		759	the callback. If an error occurs, nothing or undef gets passed to the
		760	callback.
		761
		762
		763	=item aio_realpath $pathname, $callback->($path)
		764
		765	Asynchronously make the path absolute and resolve any symlinks in
		766	C<$path>. The resulting path only consists of directories (same as
		767	L<Cwd::realpath>).
		768
		769	This request can be used to get the absolute path of the current working
		770	directory by passing it a path of F<.> (a single dot).
		771
		772
		773	=item aio_rename $srcpath, $dstpath, $callback->($status)
		774
		775	Asynchronously rename the object at C<$srcpath> to C<$dstpath>, just as
		776	rename(2) and call the callback with the result code.
		777
		778	On systems that support the AIO::WD working directory abstraction
		779	natively, the case C<[$wd, "."]> as C<$srcpath> is specialcased - instead
		780	of failing, C<rename> is called on the absolute path of C<$wd>.
		781
		782
		783	=item aio_rename2 $srcpath, $dstpath, $flags, $callback->($status)
		784
		785	Basically a version of C<aio_rename> with an additional C<$flags>
		786	argument. Calling this with C<$flags=0> is the same as calling
		787	C<aio_rename>.
		788
		789	Non-zero flags are currently only supported on GNU/Linux systems that
		790	support renameat2. Other systems fail with C<ENOSYS> in this case.
		791
		792	The following constants are available (missing ones are, as usual C<0>),
		793	see renameat2(2) for details:
		794
		795	C<IO::AIO::RENAME_NOREPLACE>, C<IO::AIO::RENAME_EXCHANGE>
		796	and C<IO::AIO::RENAME_WHITEOUT>.
		797
		798
		799	=item aio_mkdir $pathname, $mode, $callback->($status)
		800
		801	Asynchronously mkdir (create) a directory and call the callback with
		802	the result code. C<$mode> will be modified by the umask at the time the
		803	request is executed, so do not change your umask.
		804
		805
		806	=item aio_rmdir $pathname, $callback->($status)
		807
		808	Asynchronously rmdir (delete) a directory and call the callback with the
		809	result code.
		810
		811	On systems that support the AIO::WD working directory abstraction
		812	natively, the case C<[$wd, "."]> is specialcased - instead of failing,
		813	C<rmdir> is called on the absolute path of C<$wd>.
		814
		815
		816	=item aio_readdir $pathname, $callback->($entries)
		817
		818	Unlike the POSIX call of the same name, C<aio_readdir> reads an entire
		819	directory (i.e. opendir + readdir + closedir). The entries will not be
		820	sorted, and will B<NOT> include the C<.> and C<..> entries.
		821
		822	The callback is passed a single argument which is either C<undef> or an
		823	array-ref with the filenames.
		824
		825
		826	=item aio_readdirx $pathname, $flags, $callback->($entries, $flags)
		827
		828	Quite similar to C<aio_readdir>, but the C<$flags> argument allows one to
		829	tune behaviour and output format. In case of an error, C<$entries> will be
		830	C<undef>.
		831
		832	The flags are a combination of the following constants, ORed together (the
		833	flags will also be passed to the callback, possibly modified):
		834
		835	=over 4
		836
		837	=item IO::AIO::READDIR_DENTS
		838
		839	Normally the callback gets an arrayref consisting of names only (as
		840	with C<aio_readdir>). If this flag is set, then the callback gets an
		841	arrayref with C<[$name, $type, $inode]> arrayrefs, each describing a
		842	single directory entry in more detail:
		843
		844	C<$name> is the name of the entry.
		845
		846	C<$type> is one of the C<IO::AIO::DT_xxx> constants:
		847
		848	C<IO::AIO::DT_UNKNOWN>, C<IO::AIO::DT_FIFO>, C<IO::AIO::DT_CHR>, C<IO::AIO::DT_DIR>,
		849	C<IO::AIO::DT_BLK>, C<IO::AIO::DT_REG>, C<IO::AIO::DT_LNK>, C<IO::AIO::DT_SOCK>,
		850	C<IO::AIO::DT_WHT>.
		851
		852	C<IO::AIO::DT_UNKNOWN> means just that: readdir does not know. If you need
		853	to know, you have to run stat yourself. Also, for speed/memory reasons,
		854	the C<$type> scalars are read-only: you must not modify them.
		855
		856	C<$inode> is the inode number (which might not be exact on systems with 64
		857	bit inode numbers and 32 bit perls). This field has unspecified content on
		858	systems that do not deliver the inode information.
		859
		860	=item IO::AIO::READDIR_DIRS_FIRST
		861
		862	When this flag is set, then the names will be returned in an order where
		863	likely directories come first, in optimal stat order. This is useful when
		864	you need to quickly find directories, or you want to find all directories
		865	while avoiding to stat() each entry.
		866
		867	If the system returns type information in readdir, then this is used
		868	to find directories directly. Otherwise, likely directories are names
		869	beginning with ".", or otherwise names with no dots, of which names with
		870	short names are tried first.
		871
		872	=item IO::AIO::READDIR_STAT_ORDER
		873
		874	When this flag is set, then the names will be returned in an order
		875	suitable for stat()'ing each one. That is, when you plan to stat() most or
		876	all files in the given directory, then the returned order will likely be
		877	faster.
		878
		879	If both this flag and C<IO::AIO::READDIR_DIRS_FIRST> are specified,
		880	then the likely dirs come first, resulting in a less optimal stat order
		881	for stat'ing all entries, but likely a more optimal order for finding
		882	subdirectories.
		883
		884	=item IO::AIO::READDIR_FOUND_UNKNOWN
		885
		886	This flag should not be set when calling C<aio_readdirx>. Instead, it
		887	is being set by C<aio_readdirx>, when any of the C<$type>'s found were
		888	C<IO::AIO::DT_UNKNOWN>. The absence of this flag therefore indicates that all
		889	C<$type>'s are known, which can be used to speed up some algorithms.
		890
		891	=back
		892
		893
		894	=item aio_slurp $pathname, $offset, $length, $data, $callback->($status)
		895
		896	Opens, reads and closes the given file. The data is put into C<$data>,
		897	which is resized as required.
		898
		899	If C<$offset> is negative, then it is counted from the end of the file.
		900
		901	If C<$length> is zero, then the remaining length of the file is
		902	used. Also, in this case, the same limitations to modifying C<$data> apply
		903	as when IO::AIO::mmap is used, i.e. it must only be modified in-place
		904	with C<substr>. If the size of the file is known, specifying a non-zero
		905	C<$length> results in a performance advantage.
		906
		907	This request is similar to the older C<aio_load> request, but since it is
		908	a single request, it might be more efficient to use.
		909
		910	Example: load F</etc/passwd> into C<$passwd>.
		911
		912	my $passwd;
		913	aio_slurp "/etc/passwd", 0, 0, $passwd, sub {
		914	$_[0] >= 0
		915	or die "/etc/passwd: $!\n";
		916
		917	printf "/etc/passwd is %d bytes long, and contains:\n", length $passwd;
		918	print $passwd;
		919	};
		920	IO::AIO::flush;
		921
		922
		923	=item aio_load $pathname, $data, $callback->($status)
		924
		925	This is a composite request that tries to fully load the given file into
		926	memory. Status is the same as with aio_read.
		927
		928	Using C<aio_slurp> might be more efficient, as it is a single request.
		929
		930	=cut
		931
		932	sub aio_load($$;$) {
		933	my ($path, undef, $cb) = @_;
		934	my $data = \$_[1];
		935
		936	my $pri = aioreq_pri;
		937	my $grp = aio_group $cb;
		938
		939	aioreq_pri $pri;
		940	add $grp aio_open $path, O_RDONLY, 0, sub {
		941	my $fh = shift
		942	or return $grp->result (-1);
		943
		944	aioreq_pri $pri;
		945	add $grp aio_read $fh, 0, (-s $fh), $$data, 0, sub {
		946	$grp->result ($_[0]);
		947	};
		948	};
		949
		950	$grp
		951	}
		952
		953	=item aio_copy $srcpath, $dstpath, $callback->($status)
		954
		955	Try to copy the I<file> (directories not supported as either source or
		956	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		957	a status of C<0> (ok) or C<-1> (error, see C<$!>).
		958
		959	Existing destination files will be truncated.
		960
		961	This is a composite request that creates the destination file with
		962	mode 0200 and copies the contents of the source file into it using
		963	C<aio_sendfile>, followed by restoring atime, mtime, access mode and
		964	uid/gid, in that order.
		965
		966	If an error occurs, the partial destination file will be unlinked, if
		967	possible, except when setting atime, mtime, access mode and uid/gid, where
		968	errors are being ignored.
		969
		970	=cut
		971
		972	sub aio_copy($$;$) {
		973	my ($src, $dst, $cb) = @_;
		974
		975	my $pri = aioreq_pri;
		976	my $grp = aio_group $cb;
		977
		978	aioreq_pri $pri;
		979	add $grp aio_open $src, O_RDONLY, 0, sub {
		980	if (my $src_fh = $_[0]) {
		981	my @stat = stat $src_fh; # hmm, might block over nfs?
		982
		983	aioreq_pri $pri;
		984	add $grp aio_open $dst, O_CREAT | O_WRONLY | O_TRUNC, 0200, sub {
		985	if (my $dst_fh = $_[0]) {
		986	aioreq_pri $pri;
		987	add $grp aio_sendfile $dst_fh, $src_fh, 0, $stat[7], sub {
		988	if ($_[0] == $stat[7]) {
		989	$grp->result (0);
		990	close $src_fh;
		991
		992	my $ch = sub {
		993	aioreq_pri $pri;
		994	add $grp aio_chmod $dst_fh, $stat[2] & 07777, sub {
		995	aioreq_pri $pri;
		996	add $grp aio_chown $dst_fh, $stat[4], $stat[5], sub {
		997	aioreq_pri $pri;
		998	add $grp aio_close $dst_fh;
		999	}
		1000	};
		1001	};
		1002
		1003	aioreq_pri $pri;
		1004	add $grp aio_utime $dst_fh, $stat[8], $stat[9], sub {
		1005	if ($_[0] < 0 && $! == ENOSYS) {
		1006	aioreq_pri $pri;
		1007	add $grp aio_utime $dst, $stat[8], $stat[9], $ch;
		1008	} else {
		1009	$ch->();
		1010	}
		1011	};
		1012	} else {
		1013	$grp->result (-1);
		1014	close $src_fh;
		1015	close $dst_fh;
		1016
		1017	aioreq $pri;
		1018	add $grp aio_unlink $dst;
		1019	}
		1020	};
		1021	} else {
		1022	$grp->result (-1);
		1023	}
		1024	},
		1025
		1026	} else {
		1027	$grp->result (-1);
		1028	}
		1029	};
		1030
		1031	$grp
		1032	}
		1033
		1034	=item aio_move $srcpath, $dstpath, $callback->($status)
		1035
		1036	Try to move the I<file> (directories not supported as either source or
		1037	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		1038	a status of C<0> (ok) or C<-1> (error, see C<$!>).
		1039
		1040	This is a composite request that tries to rename(2) the file first; if
		1041	rename fails with C<EXDEV>, it copies the file with C<aio_copy> and, if
		1042	that is successful, unlinks the C<$srcpath>.
		1043
		1044	=cut
		1045
		1046	sub aio_move($$;$) {
		1047	my ($src, $dst, $cb) = @_;
		1048
		1049	my $pri = aioreq_pri;
		1050	my $grp = aio_group $cb;
		1051
		1052	aioreq_pri $pri;
		1053	add $grp aio_rename $src, $dst, sub {
		1054	if ($_[0] && $! == EXDEV) {
		1055	aioreq_pri $pri;
		1056	add $grp aio_copy $src, $dst, sub {
		1057	$grp->result ($_[0]);
		1058
		1059	unless ($_[0]) {
		1060	aioreq_pri $pri;
		1061	add $grp aio_unlink $src;
		1062	}
		1063	};
		1064	} else {
		1065	$grp->result ($_[0]);
		1066	}
		1067	};
		1068
		1069	$grp
		1070	}
		1071
		1072	=item aio_scandir $pathname, $maxreq, $callback->($dirs, $nondirs)
		1073
		1074	Scans a directory (similar to C<aio_readdir>) but additionally tries to
		1075	efficiently separate the entries of directory C<$path> into two sets of
		1076	names, directories you can recurse into (directories), and ones you cannot
		1077	recurse into (everything else, including symlinks to directories).
		1078
		1079	C<aio_scandir> is a composite request that generates many sub requests.
		1080	C<$maxreq> specifies the maximum number of outstanding aio requests that
		1081	this function generates. If it is C<< <= 0 >>, then a suitable default
		1082	will be chosen (currently 4).
		1083
		1084	On error, the callback is called without arguments, otherwise it receives
		1085	two array-refs with path-relative entry names.
		1086
		1087	Example:
		1088
		1089	aio_scandir $dir, 0, sub {
		1090	my ($dirs, $nondirs) = @_;
		1091	print "real directories: @$dirs\n";
		1092	print "everything else: @$nondirs\n";
		1093	};
		1094
		1095	Implementation notes.
		1096
		1097	The C<aio_readdir> cannot be avoided, but C<stat()>'ing every entry can.
		1098
		1099	If readdir returns file type information, then this is used directly to
		1100	find directories.
		1101
		1102	Otherwise, after reading the directory, the modification time, size etc.
		1103	of the directory before and after the readdir is checked, and if they
		1104	match (and isn't the current time), the link count will be used to decide
		1105	how many entries are directories (if >= 2). Otherwise, no knowledge of the
		1106	number of subdirectories will be assumed.
		1107
		1108	Then entries will be sorted into likely directories a non-initial dot
		1109	currently) and likely non-directories (see C<aio_readdirx>). Then every
		1110	entry plus an appended C</.> will be C<stat>'ed, likely directories first,
		1111	in order of their inode numbers. If that succeeds, it assumes that the
		1112	entry is a directory or a symlink to directory (which will be checked
		1113	separately). This is often faster than stat'ing the entry itself because
		1114	filesystems might detect the type of the entry without reading the inode
		1115	data (e.g. ext2fs filetype feature), even on systems that cannot return
		1116	the filetype information on readdir.
		1117
		1118	If the known number of directories (link count - 2) has been reached, the
		1119	rest of the entries is assumed to be non-directories.
		1120
		1121	This only works with certainty on POSIX (= UNIX) filesystems, which
		1122	fortunately are the vast majority of filesystems around.
		1123
		1124	It will also likely work on non-POSIX filesystems with reduced efficiency
		1125	as those tend to return 0 or 1 as link counts, which disables the
		1126	directory counting heuristic.
		1127
		1128	=cut
		1129
		1130	sub aio_scandir($$;$) {
		1131	my ($path, $maxreq, $cb) = @_;
		1132
		1133	my $pri = aioreq_pri;
		1134
		1135	my $grp = aio_group $cb;
		1136
		1137	$maxreq = 4 if $maxreq <= 0;
		1138
		1139	# get a wd object
		1140	aioreq_pri $pri;
		1141	add $grp aio_wd $path, sub {
		1142	$_[0]
		1143	or return $grp->result ();
		1144
		1145	my $wd = [shift, "."];
		1146
		1147	# stat once
		1148	aioreq_pri $pri;
		1149	add $grp aio_stat $wd, sub {
		1150	return $grp->result () if $_[0];
		1151	my $now = time;
		1152	my $hash1 = join ":", (stat _)[0,1,3,7,9];
		1153	my $rdxflags = READDIR_DIRS_FIRST;
		1154
		1155	if ((stat _)[3] < 2) {
		1156	# at least one non-POSIX filesystem exists
		1157	# that returns useful DT_type values: btrfs,
		1158	# so optimise for this here by requesting dents
		1159	$rdxflags |= READDIR_DENTS;
		1160	}
		1161
		1162	# read the directory entries
		1163	aioreq_pri $pri;
		1164	add $grp aio_readdirx $wd, $rdxflags, sub {
		1165	my ($entries, $flags) = @_
		1166	or return $grp->result ();
		1167
		1168	if ($rdxflags & READDIR_DENTS) {
		1169	# if we requested type values, see if we can use them directly.
		1170
		1171	# if there were any DT_UNKNOWN entries then we assume we
		1172	# don't know. alternatively, we could assume that if we get
		1173	# one DT_DIR, then all directories are indeed marked with
		1174	# DT_DIR, but this seems not required for btrfs, and this
		1175	# is basically the "btrfs can't get it's act together" code
		1176	# branch.
		1177	unless ($flags & READDIR_FOUND_UNKNOWN) {
		1178	# now we have valid DT_ information for all entries,
		1179	# so use it as an optimisation without further stat's.
		1180	# they must also all be at the beginning of @$entries
		1181	# by now.
		1182
		1183	my $dirs;
		1184
		1185	if (@$entries) {
		1186	for (0 .. $#$entries) {
		1187	if ($entries->[$_][1] != DT_DIR) {
		1188	# splice out directories
		1189	$dirs = [splice @$entries, 0, $_];
		1190	last;
		1191	}
		1192	}
		1193
		1194	# if we didn't find any non-dir, then all entries are dirs
		1195	unless ($dirs) {
		1196	($dirs, $entries) = ($entries, []);
		1197	}
		1198	} else {
		1199	# directory is empty, so there are no sbdirs
		1200	$dirs = [];
		1201	}
		1202
		1203	# either splice'd the directories out or the dir was empty.
		1204	# convert dents to filenames
		1205	$_ = $_->[0] for @$dirs;
		1206	$_ = $_->[0] for @$entries;
		1207
		1208	return $grp->result ($dirs, $entries);
		1209	}
		1210
		1211	# cannot use, so return to our old ways
		1212	# by pretending we only scanned for names.
		1213	$_ = $_->[0] for @$entries;
		1214	}
		1215
		1216	# stat the dir another time
		1217	aioreq_pri $pri;
		1218	add $grp aio_stat $wd, sub {
		1219	my $hash2 = join ":", (stat _)[0,1,3,7,9];
		1220
		1221	my $ndirs;
		1222
		1223	# take the slow route if anything looks fishy
		1224	if ($hash1 ne $hash2 or (stat _)[9] == $now) {
		1225	$ndirs = -1;
		1226	} else {
		1227	# if nlink == 2, we are finished
		1228	# for non-posix-fs's, we rely on nlink < 2
		1229	$ndirs = (stat _)[3] - 2
		1230	or return $grp->result ([], $entries);
		1231	}
		1232
		1233	my (@dirs, @nondirs);
		1234
		1235	my $statgrp = add $grp aio_group sub {
		1236	$grp->result (\@dirs, \@nondirs);
		1237	};
		1238
		1239	limit $statgrp $maxreq;
		1240	feed $statgrp sub {
		1241	return unless @$entries;
		1242	my $entry = shift @$entries;
		1243
		1244	aioreq_pri $pri;
		1245	$wd->[1] = "$entry/.";
		1246	add $statgrp aio_stat $wd, sub {
		1247	if ($_[0] < 0) {
		1248	push @nondirs, $entry;
		1249	} else {
		1250	# need to check for real directory
		1251	aioreq_pri $pri;
		1252	$wd->[1] = $entry;
		1253	add $statgrp aio_lstat $wd, sub {
		1254	if (-d _) {
		1255	push @dirs, $entry;
		1256
		1257	unless (--$ndirs) {
		1258	push @nondirs, @$entries;
		1259	feed $statgrp;
		1260	}
		1261	} else {
		1262	push @nondirs, $entry;
		1263	}
		1264	}
		1265	}
		1266	};
		1267	};
		1268	};
		1269	};
		1270	};
		1271	};
		1272
		1273	$grp
		1274	}
		1275
		1276	=item aio_rmtree $pathname, $callback->($status)
		1277
		1278	Delete a directory tree starting (and including) C<$path>, return the
		1279	status of the final C<rmdir> only. This is a composite request that
		1280	uses C<aio_scandir> to recurse into and rmdir directories, and unlink
		1281	everything else.
		1282
		1283	=cut
		1284
		1285	sub aio_rmtree;
		1286	sub aio_rmtree($;$) {
		1287	my ($path, $cb) = @_;
		1288
		1289	my $pri = aioreq_pri;
		1290	my $grp = aio_group $cb;
		1291
		1292	aioreq_pri $pri;
		1293	add $grp aio_scandir $path, 0, sub {
		1294	my ($dirs, $nondirs) = @_;
		1295
		1296	my $dirgrp = aio_group sub {
		1297	add $grp aio_rmdir $path, sub {
		1298	$grp->result ($_[0]);
		1299	};
		1300	};
		1301
		1302	(aioreq_pri $pri), add $dirgrp aio_rmtree "$path/$_" for @$dirs;
		1303	(aioreq_pri $pri), add $dirgrp aio_unlink "$path/$_" for @$nondirs;
		1304
		1305	add $grp $dirgrp;
		1306	};
		1307
		1308	$grp
		1309	}
		1310
		1311	=item aio_fcntl $fh, $cmd, $arg, $callback->($status)
		1312
		1313	=item aio_ioctl $fh, $request, $buf, $callback->($status)
		1314
		1315	These work just like the C<fcntl> and C<ioctl> built-in functions, except
		1316	they execute asynchronously and pass the return value to the callback.
		1317
		1318	Both calls can be used for a lot of things, some of which make more sense
		1319	to run asynchronously in their own thread, while some others make less
		1320	sense. For example, calls that block waiting for external events, such
		1321	as locking, will also lock down an I/O thread while it is waiting, which
		1322	can deadlock the whole I/O system. At the same time, there might be no
		1323	alternative to using a thread to wait.
		1324
		1325	So in general, you should only use these calls for things that do
		1326	(filesystem) I/O, not for things that wait for other events (network,
		1327	other processes), although if you are careful and know what you are doing,
		1328	you still can.
		1329
		1330	The following constants are available and can be used for normal C<ioctl>
		1331	and C<fcntl> as well (missing ones are, as usual C<0>):
		1332
		1333	C<F_DUPFD_CLOEXEC>,
		1334
		1335	C<F_OFD_GETLK>, C<F_OFD_SETLK>, C<F_OFD_GETLKW>,
		1336
		1337	C<FIFREEZE>, C<FITHAW>, C<FITRIM>, C<FICLONE>, C<FICLONERANGE>, C<FIDEDUPERANGE>.
		1338
		1339	C<F_ADD_SEALS>, C<F_GET_SEALS>, C<F_SEAL_SEAL>, C<F_SEAL_SHRINK>, C<F_SEAL_GROW> and
		1340	C<F_SEAL_WRITE>.
		1341
		1342	C<FS_IOC_GETFLAGS>, C<FS_IOC_SETFLAGS>, C<FS_IOC_GETVERSION>, C<FS_IOC_SETVERSION>,
		1343	C<FS_IOC_FIEMAP>.
		1344
		1345	C<FS_IOC_FSGETXATTR>, C<FS_IOC_FSSETXATTR>, C<FS_IOC_SET_ENCRYPTION_POLICY>,
		1346	C<FS_IOC_GET_ENCRYPTION_PWSALT>, C<FS_IOC_GET_ENCRYPTION_POLICY>, C<FS_KEY_DESCRIPTOR_SIZE>.
		1347
		1348	C<FS_SECRM_FL>, C<FS_UNRM_FL>, C<FS_COMPR_FL>, C<FS_SYNC_FL>, C<FS_IMMUTABLE_FL>,
		1349	C<FS_APPEND_FL>, C<FS_NODUMP_FL>, C<FS_NOATIME_FL>, C<FS_DIRTY_FL>,
		1350	C<FS_COMPRBLK_FL>, C<FS_NOCOMP_FL>, C<FS_ENCRYPT_FL>, C<FS_BTREE_FL>,
		1351	C<FS_INDEX_FL>, C<FS_JOURNAL_DATA_FL>, C<FS_NOTAIL_FL>, C<FS_DIRSYNC_FL>, C<FS_TOPDIR_FL>,
		1352	C<FS_FL_USER_MODIFIABLE>.
		1353
		1354	C<FS_XFLAG_REALTIME>, C<FS_XFLAG_PREALLOC>, C<FS_XFLAG_IMMUTABLE>, C<FS_XFLAG_APPEND>,
		1355	C<FS_XFLAG_SYNC>, C<FS_XFLAG_NOATIME>, C<FS_XFLAG_NODUMP>, C<FS_XFLAG_RTINHERIT>,
		1356	C<FS_XFLAG_PROJINHERIT>, C<FS_XFLAG_NOSYMLINKS>, C<FS_XFLAG_EXTSIZE>, C<FS_XFLAG_EXTSZINHERIT>,
		1357	C<FS_XFLAG_NODEFRAG>, C<FS_XFLAG_FILESTREAM>, C<FS_XFLAG_DAX>, C<FS_XFLAG_HASATTR>,
		1358
		1359	C<BLKROSET>, C<BLKROGET>, C<BLKRRPART>, C<BLKGETSIZE>, C<BLKFLSBUF>, C<BLKRASET>,
		1360	C<BLKRAGET>, C<BLKFRASET>, C<BLKFRAGET>, C<BLKSECTSET>, C<BLKSECTGET>, C<BLKSSZGET>,
		1361	C<BLKBSZGET>, C<BLKBSZSET>, C<BLKGETSIZE64>,
		1362
		1363
		1364	=item aio_sync $callback->($status)
		1365
		1366	Asynchronously call sync and call the callback when finished.
		1367
179	=item aio_fsync $fh, $callback	1368	=item aio_fsync $fh, $callback->($status)
180		1369
181	Asynchronously call fsync on the given filehandle and call the callback	1370	Asynchronously call fsync on the given filehandle and call the callback
182	with the fsync result code.	1371	with the fsync result code.
183		1372
184	=item aio_fdatasync $fh, $callback	1373	=item aio_fdatasync $fh, $callback->($status)
185		1374
186	Asynchronously call fdatasync on the given filehandle and call the	1375	Asynchronously call fdatasync on the given filehandle and call the
187	callback with the fdatasync result code.	1376	callback with the fdatasync result code.
188		1377
		1378	If this call isn't available because your OS lacks it or it couldn't be
		1379	detected, it will be emulated by calling C<fsync> instead.
		1380
		1381	=item aio_syncfs $fh, $callback->($status)
		1382
		1383	Asynchronously call the syncfs syscall to sync the filesystem associated
		1384	to the given filehandle and call the callback with the syncfs result
		1385	code. If syncfs is not available, calls sync(), but returns C<-1> and sets
		1386	errno to C<ENOSYS> nevertheless.
		1387
		1388	=item aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
		1389
		1390	Sync the data portion of the file specified by C<$offset> and C<$length>
		1391	to disk (but NOT the metadata), by calling the Linux-specific
		1392	sync_file_range call. If sync_file_range is not available or it returns
		1393	ENOSYS, then fdatasync or fsync is being substituted.
		1394
		1395	C<$flags> can be a combination of C<IO::AIO::SYNC_FILE_RANGE_WAIT_BEFORE>,
		1396	C<IO::AIO::SYNC_FILE_RANGE_WRITE> and
		1397	C<IO::AIO::SYNC_FILE_RANGE_WAIT_AFTER>: refer to the sync_file_range
		1398	manpage for details.
		1399
		1400	=item aio_pathsync $pathname, $callback->($status)
		1401
		1402	This request tries to open, fsync and close the given path. This is a
		1403	composite request intended to sync directories after directory operations
		1404	(E.g. rename). This might not work on all operating systems or have any
		1405	specific effect, but usually it makes sure that directory changes get
		1406	written to disc. It works for anything that can be opened for read-only,
		1407	not just directories.
		1408
		1409	Future versions of this function might fall back to other methods when
		1410	C<fsync> on the directory fails (such as calling C<sync>).
		1411
		1412	Passes C<0> when everything went ok, and C<-1> on error.
		1413
		1414	=cut
		1415
		1416	sub aio_pathsync($;$) {
		1417	my ($path, $cb) = @_;
		1418
		1419	my $pri = aioreq_pri;
		1420	my $grp = aio_group $cb;
		1421
		1422	aioreq_pri $pri;
		1423	add $grp aio_open $path, O_RDONLY, 0, sub {
		1424	my ($fh) = @_;
		1425	if ($fh) {
		1426	aioreq_pri $pri;
		1427	add $grp aio_fsync $fh, sub {
		1428	$grp->result ($_[0]);
		1429
		1430	aioreq_pri $pri;
		1431	add $grp aio_close $fh;
		1432	};
		1433	} else {
		1434	$grp->result (-1);
		1435	}
		1436	};
		1437
		1438	$grp
		1439	}
		1440
		1441	=item aio_msync $scalar, $offset = 0, $length = undef, flags = MS_SYNC, $callback->($status)
		1442
		1443	This is a rather advanced IO::AIO call, which only works on mmap(2)ed
		1444	scalars (see the C<IO::AIO::mmap> function, although it also works on data
		1445	scalars managed by the L<Sys::Mmap> or L<Mmap> modules, note that the
		1446	scalar must only be modified in-place while an aio operation is pending on
		1447	it).
		1448
		1449	It calls the C<msync> function of your OS, if available, with the memory
		1450	area starting at C<$offset> in the string and ending C<$length> bytes
		1451	later. If C<$length> is negative, counts from the end, and if C<$length>
		1452	is C<undef>, then it goes till the end of the string. The flags can be
		1453	either C<IO::AIO::MS_ASYNC> or C<IO::AIO::MS_SYNC>, plus an optional
		1454	C<IO::AIO::MS_INVALIDATE>.
		1455
		1456	=item aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		1457
		1458	This is a rather advanced IO::AIO call, which works best on mmap(2)ed
		1459	scalars.
		1460
		1461	It touches (reads or writes) all memory pages in the specified
		1462	range inside the scalar. All caveats and parameters are the same
		1463	as for C<aio_msync>, above, except for flags, which must be either
		1464	C<0> (which reads all pages and ensures they are instantiated) or
		1465	C<IO::AIO::MT_MODIFY>, which modifies the memory pages (by reading and
		1466	writing an octet from it, which dirties the page).
		1467
		1468	=item aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status)
		1469
		1470	This is a rather advanced IO::AIO call, which works best on mmap(2)ed
		1471	scalars.
		1472
		1473	It reads in all the pages of the underlying storage into memory (if any)
		1474	and locks them, so they are not getting swapped/paged out or removed.
		1475
		1476	If C<$length> is undefined, then the scalar will be locked till the end.
		1477
		1478	On systems that do not implement C<mlock>, this function returns C<-1>
		1479	and sets errno to C<ENOSYS>.
		1480
		1481	Note that the corresponding C<munlock> is synchronous and is
		1482	documented under L<MISCELLANEOUS FUNCTIONS>.
		1483
		1484	Example: open a file, mmap and mlock it - both will be undone when
		1485	C<$data> gets destroyed.
		1486
		1487	open my $fh, "<", $path or die "$path: $!";
		1488	my $data;
		1489	IO::AIO::mmap $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh;
		1490	aio_mlock $data; # mlock in background
		1491
		1492	=item aio_mlockall $flags, $callback->($status)
		1493
		1494	Calls the C<mlockall> function with the given C<$flags> (a
		1495	combination of C<IO::AIO::MCL_CURRENT>, C<IO::AIO::MCL_FUTURE> and
		1496	C<IO::AIO::MCL_ONFAULT>).
		1497
		1498	On systems that do not implement C<mlockall>, this function returns C<-1>
		1499	and sets errno to C<ENOSYS>. Similarly, flag combinations not supported
		1500	by the system result in a return value of C<-1> with errno being set to
		1501	C<EINVAL>.
		1502
		1503	Note that the corresponding C<munlockall> is synchronous and is
		1504	documented under L<MISCELLANEOUS FUNCTIONS>.
		1505
		1506	Example: asynchronously lock all current and future pages into memory.
		1507
		1508	aio_mlockall IO::AIO::MCL_FUTURE;
		1509
		1510	=item aio_fiemap $fh, $start, $length, $flags, $count, $cb->(\@extents)
		1511
		1512	Queries the extents of the given file (by calling the Linux C<FIEMAP>
		1513	ioctl, see L<http://cvs.schmorp.de/IO-AIO/doc/fiemap.txt> for details). If
		1514	the ioctl is not available on your OS, then this request will fail with
		1515	C<ENOSYS>.
		1516
		1517	C<$start> is the starting offset to query extents for, C<$length> is the
		1518	size of the range to query - if it is C<undef>, then the whole file will
		1519	be queried.
		1520
		1521	C<$flags> is a combination of flags (C<IO::AIO::FIEMAP_FLAG_SYNC> or
		1522	C<IO::AIO::FIEMAP_FLAG_XATTR> - C<IO::AIO::FIEMAP_FLAGS_COMPAT> is also
		1523	exported), and is normally C<0> or C<IO::AIO::FIEMAP_FLAG_SYNC> to query
		1524	the data portion.
		1525
		1526	C<$count> is the maximum number of extent records to return. If it is
		1527	C<undef>, then IO::AIO queries all extents of the range. As a very special
		1528	case, if it is C<0>, then the callback receives the number of extents
		1529	instead of the extents themselves (which is unreliable, see below).
		1530
		1531	If an error occurs, the callback receives no arguments. The special
		1532	C<errno> value C<IO::AIO::EBADR> is available to test for flag errors.
		1533
		1534	Otherwise, the callback receives an array reference with extent
		1535	structures. Each extent structure is an array reference itself, with the
		1536	following members:
		1537
		1538	[$logical, $physical, $length, $flags]
		1539
		1540	Flags is any combination of the following flag values (typically either C<0>
		1541	or C<IO::AIO::FIEMAP_EXTENT_LAST> (1)):
		1542
		1543	C<IO::AIO::FIEMAP_EXTENT_LAST>, C<IO::AIO::FIEMAP_EXTENT_UNKNOWN>,
		1544	C<IO::AIO::FIEMAP_EXTENT_DELALLOC>, C<IO::AIO::FIEMAP_EXTENT_ENCODED>,
		1545	C<IO::AIO::FIEMAP_EXTENT_DATA_ENCRYPTED>, C<IO::AIO::FIEMAP_EXTENT_NOT_ALIGNED>,
		1546	C<IO::AIO::FIEMAP_EXTENT_DATA_INLINE>, C<IO::AIO::FIEMAP_EXTENT_DATA_TAIL>,
		1547	C<IO::AIO::FIEMAP_EXTENT_UNWRITTEN>, C<IO::AIO::FIEMAP_EXTENT_MERGED> or
		1548	C<IO::AIO::FIEMAP_EXTENT_SHARED>.
		1549
		1550	At the time of this writing (Linux 3.2), this request is unreliable unless
		1551	C<$count> is C<undef>, as the kernel has all sorts of bugs preventing
		1552	it to return all extents of a range for files with a large number of
		1553	extents. The code (only) works around all these issues if C<$count> is
		1554	C<undef>.
		1555
		1556	=item aio_group $callback->(...)
		1557
		1558	This is a very special aio request: Instead of doing something, it is a
		1559	container for other aio requests, which is useful if you want to bundle
		1560	many requests into a single, composite, request with a definite callback
		1561	and the ability to cancel the whole request with its subrequests.
		1562
		1563	Returns an object of class L<IO::AIO::GRP>. See its documentation below
		1564	for more info.
		1565
		1566	Example:
		1567
		1568	my $grp = aio_group sub {
		1569	print "all stats done\n";
		1570	};
		1571
		1572	add $grp
		1573	(aio_stat ...),
		1574	(aio_stat ...),
		1575	...;
		1576
		1577	=item aio_nop $callback->()
		1578
		1579	This is a special request - it does nothing in itself and is only used for
		1580	side effects, such as when you want to add a dummy request to a group so
		1581	that finishing the requests in the group depends on executing the given
		1582	code.
		1583
		1584	While this request does nothing, it still goes through the execution
		1585	phase and still requires a worker thread. Thus, the callback will not
		1586	be executed immediately but only after other requests in the queue have
		1587	entered their execution phase. This can be used to measure request
		1588	latency.
		1589
		1590	=item IO::AIO::aio_busy $fractional_seconds, $callback->() *NOT EXPORTED*
		1591
		1592	Mainly used for debugging and benchmarking, this aio request puts one of
		1593	the request workers to sleep for the given time.
		1594
		1595	While it is theoretically handy to have simple I/O scheduling requests
		1596	like sleep and file handle readable/writable, the overhead this creates is
		1597	immense (it blocks a thread for a long time) so do not use this function
		1598	except to put your application under artificial I/O pressure.
		1599
189	=back	1600	=back
190		1601
		1602
		1603	=head2 IO::AIO::WD - multiple working directories
		1604
		1605	Your process only has one current working directory, which is used by all
		1606	threads. This makes it hard to use relative paths (some other component
		1607	could call C<chdir> at any time, and it is hard to control when the path
		1608	will be used by IO::AIO).
		1609
		1610	One solution for this is to always use absolute paths. This usually works,
		1611	but can be quite slow (the kernel has to walk the whole path on every
		1612	access), and can also be a hassle to implement.
		1613
		1614	Newer POSIX systems have a number of functions (openat, fdopendir,
		1615	futimensat and so on) that make it possible to specify working directories
		1616	per operation.
		1617
		1618	For portability, and because the clowns who "designed", or shall I write,
		1619	perpetrated this new interface were obviously half-drunk, this abstraction
		1620	cannot be perfect, though.
		1621
		1622	IO::AIO allows you to convert directory paths into a so-called IO::AIO::WD
		1623	object. This object stores the canonicalised, absolute version of the
		1624	path, and on systems that allow it, also a directory file descriptor.
		1625
		1626	Everywhere where a pathname is accepted by IO::AIO (e.g. in C<aio_stat>
		1627	or C<aio_unlink>), one can specify an array reference with an IO::AIO::WD
		1628	object and a pathname instead (or the IO::AIO::WD object alone, which
		1629	gets interpreted as C<[$wd, "."]>). If the pathname is absolute, the
		1630	IO::AIO::WD object is ignored, otherwise the pathname is resolved relative
		1631	to that IO::AIO::WD object.
		1632
		1633	For example, to get a wd object for F</etc> and then stat F<passwd>
		1634	inside, you would write:
		1635
		1636	aio_wd "/etc", sub {
		1637	my $etcdir = shift;
		1638
		1639	# although $etcdir can be undef on error, there is generally no reason
		1640	# to check for errors here, as aio_stat will fail with ENOENT
		1641	# when $etcdir is undef.
		1642
		1643	aio_stat [$etcdir, "passwd"], sub {
		1644	# yay
		1645	};
		1646	};
		1647
		1648	The fact that C<aio_wd> is a request and not a normal function shows that
		1649	creating an IO::AIO::WD object is itself a potentially blocking operation,
		1650	which is why it is done asynchronously.
		1651
		1652	To stat the directory obtained with C<aio_wd> above, one could write
		1653	either of the following three request calls:
		1654
		1655	aio_lstat "/etc" , sub { ... # pathname as normal string
		1656	aio_lstat [$wd, "."], sub { ... # "." relative to $wd (i.e. $wd itself)
		1657	aio_lstat $wd , sub { ... # shorthand for the previous
		1658
		1659	As with normal pathnames, IO::AIO keeps a copy of the working directory
		1660	object and the pathname string, so you could write the following without
		1661	causing any issues due to C<$path> getting reused:
		1662
		1663	my $path = [$wd, undef];
		1664
		1665	for my $name (qw(abc def ghi)) {
		1666	$path->[1] = $name;
		1667	aio_stat $path, sub {
		1668	# ...
		1669	};
		1670	}
		1671
		1672	There are some caveats: when directories get renamed (or deleted), the
		1673	pathname string doesn't change, so will point to the new directory (or
		1674	nowhere at all), while the directory fd, if available on the system,
		1675	will still point to the original directory. Most functions accepting a
		1676	pathname will use the directory fd on newer systems, and the string on
		1677	older systems. Some functions (such as C<aio_realpath>) will always rely on
		1678	the string form of the pathname.
		1679
		1680	So this functionality is mainly useful to get some protection against
		1681	C<chdir>, to easily get an absolute path out of a relative path for future
		1682	reference, and to speed up doing many operations in the same directory
		1683	(e.g. when stat'ing all files in a directory).
		1684
		1685	The following functions implement this working directory abstraction:
		1686
		1687	=over 4
		1688
		1689	=item aio_wd $pathname, $callback->($wd)
		1690
		1691	Asynchonously canonicalise the given pathname and convert it to an
		1692	IO::AIO::WD object representing it. If possible and supported on the
		1693	system, also open a directory fd to speed up pathname resolution relative
		1694	to this working directory.
		1695
		1696	If something goes wrong, then C<undef> is passwd to the callback instead
		1697	of a working directory object and C<$!> is set appropriately. Since
		1698	passing C<undef> as working directory component of a pathname fails the
		1699	request with C<ENOENT>, there is often no need for error checking in the
		1700	C<aio_wd> callback, as future requests using the value will fail in the
		1701	expected way.
		1702
		1703	=item IO::AIO::CWD
		1704
		1705	This is a compile time constant (object) that represents the process
		1706	current working directory.
		1707
		1708	Specifying this object as working directory object for a pathname is as if
		1709	the pathname would be specified directly, without a directory object. For
		1710	example, these calls are functionally identical:
		1711
		1712	aio_stat "somefile", sub { ... };
		1713	aio_stat [IO::AIO::CWD, "somefile"], sub { ... };
		1714
		1715	=back
		1716
		1717	To recover the path associated with an IO::AIO::WD object, you can use
		1718	C<aio_realpath>:
		1719
		1720	aio_realpath $wd, sub {
		1721	warn "path is $_[0]\n";
		1722	};
		1723
		1724	Currently, C<aio_statvfs> always, and C<aio_rename> and C<aio_rmdir>
		1725	sometimes, fall back to using an absolue path.
		1726
		1727	=head2 IO::AIO::REQ CLASS
		1728
		1729	All non-aggregate C<aio_*> functions return an object of this class when
		1730	called in non-void context.
		1731
		1732	=over 4
		1733
		1734	=item cancel $req
		1735
		1736	Cancels the request, if possible. Has the effect of skipping execution
		1737	when entering the B<execute> state and skipping calling the callback when
		1738	entering the the B<result> state, but will leave the request otherwise
		1739	untouched (with the exception of readdir). That means that requests that
		1740	currently execute will not be stopped and resources held by the request
		1741	will not be freed prematurely.
		1742
		1743	=item cb $req $callback->(...)
		1744
		1745	Replace (or simply set) the callback registered to the request.
		1746
		1747	=back
		1748
		1749	=head2 IO::AIO::GRP CLASS
		1750
		1751	This class is a subclass of L<IO::AIO::REQ>, so all its methods apply to
		1752	objects of this class, too.
		1753
		1754	A IO::AIO::GRP object is a special request that can contain multiple other
		1755	aio requests.
		1756
		1757	You create one by calling the C<aio_group> constructing function with a
		1758	callback that will be called when all contained requests have entered the
		1759	C<done> state:
		1760
		1761	my $grp = aio_group sub {
		1762	print "all requests are done\n";
		1763	};
		1764
		1765	You add requests by calling the C<add> method with one or more
		1766	C<IO::AIO::REQ> objects:
		1767
		1768	$grp->add (aio_unlink "...");
		1769
		1770	add $grp aio_stat "...", sub {
		1771	$_[0] or return $grp->result ("error");
		1772
		1773	# add another request dynamically, if first succeeded
		1774	add $grp aio_open "...", sub {
		1775	$grp->result ("ok");
		1776	};
		1777	};
		1778
		1779	This makes it very easy to create composite requests (see the source of
		1780	C<aio_move> for an application) that work and feel like simple requests.
		1781
		1782	=over 4
		1783
		1784	=item * The IO::AIO::GRP objects will be cleaned up during calls to
		1785	C<IO::AIO::poll_cb>, just like any other request.
		1786
		1787	=item * They can be canceled like any other request. Canceling will cancel not
		1788	only the request itself, but also all requests it contains.
		1789
		1790	=item * They can also can also be added to other IO::AIO::GRP objects.
		1791
		1792	=item * You must not add requests to a group from within the group callback (or
		1793	any later time).
		1794
		1795	=back
		1796
		1797	Their lifetime, simplified, looks like this: when they are empty, they
		1798	will finish very quickly. If they contain only requests that are in the
		1799	C<done> state, they will also finish. Otherwise they will continue to
		1800	exist.
		1801
		1802	That means after creating a group you have some time to add requests
		1803	(precisely before the callback has been invoked, which is only done within
		1804	the C<poll_cb>). And in the callbacks of those requests, you can add
		1805	further requests to the group. And only when all those requests have
		1806	finished will the the group itself finish.
		1807
		1808	=over 4
		1809
		1810	=item add $grp ...
		1811
		1812	=item $grp->add (...)
		1813
		1814	Add one or more requests to the group. Any type of L<IO::AIO::REQ> can
		1815	be added, including other groups, as long as you do not create circular
		1816	dependencies.
		1817
		1818	Returns all its arguments.
		1819
		1820	=item $grp->cancel_subs
		1821
		1822	Cancel all subrequests and clears any feeder, but not the group request
		1823	itself. Useful when you queued a lot of events but got a result early.
		1824
		1825	The group request will finish normally (you cannot add requests to the
		1826	group).
		1827
		1828	=item $grp->result (...)
		1829
		1830	Set the result value(s) that will be passed to the group callback when all
		1831	subrequests have finished and set the groups errno to the current value
		1832	of errno (just like calling C<errno> without an error number). By default,
		1833	no argument will be passed and errno is zero.
		1834
		1835	=item $grp->errno ([$errno])
		1836
		1837	Sets the group errno value to C<$errno>, or the current value of errno
		1838	when the argument is missing.
		1839
		1840	Every aio request has an associated errno value that is restored when
		1841	the callback is invoked. This method lets you change this value from its
		1842	default (0).
		1843
		1844	Calling C<result> will also set errno, so make sure you either set C<$!>
		1845	before the call to C<result>, or call c<errno> after it.
		1846
		1847	=item feed $grp $callback->($grp)
		1848
		1849	Sets a feeder/generator on this group: every group can have an attached
		1850	generator that generates requests if idle. The idea behind this is that,
		1851	although you could just queue as many requests as you want in a group,
		1852	this might starve other requests for a potentially long time. For example,
		1853	C<aio_scandir> might generate hundreds of thousands of C<aio_stat>
		1854	requests, delaying any later requests for a long time.
		1855
		1856	To avoid this, and allow incremental generation of requests, you can
		1857	instead a group and set a feeder on it that generates those requests. The
		1858	feed callback will be called whenever there are few enough (see C<limit>,
		1859	below) requests active in the group itself and is expected to queue more
		1860	requests.
		1861
		1862	The feed callback can queue as many requests as it likes (i.e. C<add> does
		1863	not impose any limits).
		1864
		1865	If the feed does not queue more requests when called, it will be
		1866	automatically removed from the group.
		1867
		1868	If the feed limit is C<0> when this method is called, it will be set to
		1869	C<2> automatically.
		1870
		1871	Example:
		1872
		1873	# stat all files in @files, but only ever use four aio requests concurrently:
		1874
		1875	my $grp = aio_group sub { print "finished\n" };
		1876	limit $grp 4;
		1877	feed $grp sub {
		1878	my $file = pop @files
		1879	or return;
		1880
		1881	add $grp aio_stat $file, sub { ... };
		1882	};
		1883
		1884	=item limit $grp $num
		1885
		1886	Sets the feeder limit for the group: The feeder will be called whenever
		1887	the group contains less than this many requests.
		1888
		1889	Setting the limit to C<0> will pause the feeding process.
		1890
		1891	The default value for the limit is C<0>, but note that setting a feeder
		1892	automatically bumps it up to C<2>.
		1893
		1894	=back
		1895
		1896
191	=head2 SUPPORT FUNCTIONS	1897	=head2 SUPPORT FUNCTIONS
192		1898
		1899	=head3 EVENT PROCESSING AND EVENT LOOP INTEGRATION
		1900
193	=over 4	1901	=over 4
194		1902
195	=item $fileno = IO::AIO::poll_fileno	1903	=item $fileno = IO::AIO::poll_fileno
196		1904
197	Return the I<request result pipe filehandle>. This filehandle must be	1905	Return the I<request result pipe file descriptor>. This filehandle must be
198	polled for reading by some mechanism outside this module (e.g. Event	1906	polled for reading by some mechanism outside this module (e.g. EV, Glib,
199	or select, see below). If the pipe becomes readable you have to call	1907	select and so on, see below or the SYNOPSIS). If the pipe becomes readable
200	C<poll_cb> to check the results.	1908	you have to call C<poll_cb> to check the results.
201		1909
202	See C<poll_cb> for an example.	1910	See C<poll_cb> for an example.
203		1911
204	=item IO::AIO::poll_cb	1912	=item IO::AIO::poll_cb
205		1913
206	Process all outstanding events on the result pipe. You have to call this	1914	Process some requests that have reached the result phase (i.e. they have
207	regularly. Returns the number of events processed. Returns immediately	1915	been executed but the results are not yet reported). You have to call
208	when no events are outstanding.	1916	this "regularly" to finish outstanding requests.
209		1917
210	You can use Event to multiplex, e.g.:	1918	Returns C<0> if all events could be processed (or there were no
		1919	events to process), or C<-1> if it returned earlier for whatever
		1920	reason. Returns immediately when no events are outstanding. The amount
		1921	of events processed depends on the settings of C<IO::AIO::max_poll_req>,
		1922	C<IO::AIO::max_poll_time> and C<IO::AIO::max_outstanding>.
		1923
		1924	If not all requests were processed for whatever reason, the poll file
		1925	descriptor will still be ready when C<poll_cb> returns, so normally you
		1926	don't have to do anything special to have it called later.
		1927
		1928	Apart from calling C<IO::AIO::poll_cb> when the event filehandle becomes
		1929	ready, it can be beneficial to call this function from loops which submit
		1930	a lot of requests, to make sure the results get processed when they become
		1931	available and not just when the loop is finished and the event loop takes
		1932	over again. This function returns very fast when there are no outstanding
		1933	requests.
		1934
		1935	Example: Install an Event watcher that automatically calls
		1936	IO::AIO::poll_cb with high priority (more examples can be found in the
		1937	SYNOPSIS section, at the top of this document):
211		1938
212	Event->io (fd => IO::AIO::poll_fileno,	1939	Event->io (fd => IO::AIO::poll_fileno,
213	poll => 'r', async => 1,	1940	poll => 'r', async => 1,
214	cb => \&IO::AIO::poll_cb);	1941	cb => \&IO::AIO::poll_cb);
215		1942
216	=item IO::AIO::poll_wait	1943	=item IO::AIO::poll_wait
217		1944
218	Wait till the result filehandle becomes ready for reading (simply does a	1945	Wait until either at least one request is in the result phase or no
219	select on the filehandle. This is useful if you want to synchronously wait	1946	requests are outstanding anymore.
220	for some requests to finish).	1947
		1948	This is useful if you want to synchronously wait for some requests to
		1949	become ready, without actually handling them.
221		1950
222	See C<nreqs> for an example.	1951	See C<nreqs> for an example.
223		1952
		1953	=item IO::AIO::poll
		1954
		1955	Waits until some requests have been handled.
		1956
		1957	Returns the number of requests processed, but is otherwise strictly
		1958	equivalent to:
		1959
		1960	IO::AIO::poll_wait, IO::AIO::poll_cb
		1961
224	=item IO::AIO::nreqs	1962	=item IO::AIO::flush
225		1963
226	Returns the number of requests currently outstanding.	1964	Wait till all outstanding AIO requests have been handled.
227		1965
228	Example: wait till there are no outstanding requests anymore:	1966	Strictly equivalent to:
229		1967
230	IO::AIO::poll_wait, IO::AIO::poll_cb	1968	IO::AIO::poll_wait, IO::AIO::poll_cb
231	while IO::AIO::nreqs;	1969	while IO::AIO::nreqs;
232		1970
		1971	This function can be useful at program aborts, to make sure outstanding
		1972	I/O has been done (C<IO::AIO> uses an C<END> block which already calls
		1973	this function on normal exits), or when you are merely using C<IO::AIO>
		1974	for its more advanced functions, rather than for async I/O, e.g.:
		1975
		1976	my ($dirs, $nondirs);
		1977	IO::AIO::aio_scandir "/tmp", 0, sub { ($dirs, $nondirs) = @_ };
		1978	IO::AIO::flush;
		1979	# $dirs, $nondirs are now set
		1980
		1981	=item IO::AIO::max_poll_reqs $nreqs
		1982
		1983	=item IO::AIO::max_poll_time $seconds
		1984
		1985	These set the maximum number of requests (default C<0>, meaning infinity)
		1986	that are being processed by C<IO::AIO::poll_cb> in one call, respectively
		1987	the maximum amount of time (default C<0>, meaning infinity) spent in
		1988	C<IO::AIO::poll_cb> to process requests (more correctly the mininum amount
		1989	of time C<poll_cb> is allowed to use).
		1990
		1991	Setting C<max_poll_time> to a non-zero value creates an overhead of one
		1992	syscall per request processed, which is not normally a problem unless your
		1993	callbacks are really really fast or your OS is really really slow (I am
		1994	not mentioning Solaris here). Using C<max_poll_reqs> incurs no overhead.
		1995
		1996	Setting these is useful if you want to ensure some level of
		1997	interactiveness when perl is not fast enough to process all requests in
		1998	time.
		1999
		2000	For interactive programs, values such as C<0.01> to C<0.1> should be fine.
		2001
		2002	Example: Install an Event watcher that automatically calls
		2003	IO::AIO::poll_cb with low priority, to ensure that other parts of the
		2004	program get the CPU sometimes even under high AIO load.
		2005
		2006	# try not to spend much more than 0.1s in poll_cb
		2007	IO::AIO::max_poll_time 0.1;
		2008
		2009	# use a low priority so other tasks have priority
		2010	Event->io (fd => IO::AIO::poll_fileno,
		2011	poll => 'r', nice => 1,
		2012	cb => &IO::AIO::poll_cb);
		2013
		2014	=back
		2015
		2016
		2017	=head3 CONTROLLING THE NUMBER OF THREADS
		2018
		2019	=over
		2020
233	=item IO::AIO::min_parallel $nthreads	2021	=item IO::AIO::min_parallel $nthreads
234		2022
235	Set the minimum number of AIO threads to C<$nthreads>. The default is	2023	Set the minimum number of AIO threads to C<$nthreads>. The current
236	C<1>, which means a single asynchronous operation can be done at one time	2024	default is C<8>, which means eight asynchronous operations can execute
237	(the number of outstanding operations, however, is unlimited).	2025	concurrently at any one time (the number of outstanding requests,
		2026	however, is unlimited).
238		2027
		2028	IO::AIO starts threads only on demand, when an AIO request is queued and
		2029	no free thread exists. Please note that queueing up a hundred requests can
		2030	create demand for a hundred threads, even if it turns out that everything
		2031	is in the cache and could have been processed faster by a single thread.
		2032
239	It is recommended to keep the number of threads low, as some Linux	2033	It is recommended to keep the number of threads relatively low, as some
240	kernel versions will scale negatively with the number of threads (higher	2034	Linux kernel versions will scale negatively with the number of threads
241	parallelity => MUCH higher latency). With current Linux 2.6 versions, 4-32	2035	(higher parallelity => MUCH higher latency). With current Linux 2.6
242	threads should be fine.	2036	versions, 4-32 threads should be fine.
243		2037
244	Under normal circumstances you don't need to call this function, as this	2038	Under most circumstances you don't need to call this function, as the
245	module automatically starts some threads (the exact number might change,	2039	module selects a default that is suitable for low to moderate load.
246	and is currently 4).
247		2040
248	=item IO::AIO::max_parallel $nthreads	2041	=item IO::AIO::max_parallel $nthreads
249		2042
250	Sets the maximum number of AIO threads to C<$nthreads>. If more than	2043	Sets the maximum number of AIO threads to C<$nthreads>. If more than the
251	the specified number of threads are currently running, kill them. This	2044	specified number of threads are currently running, this function kills
252	function blocks until the limit is reached.	2045	them. This function blocks until the limit is reached.
		2046
		2047	While C<$nthreads> are zero, aio requests get queued but not executed
		2048	until the number of threads has been increased again.
253		2049
254	This module automatically runs C<max_parallel 0> at program end, to ensure	2050	This module automatically runs C<max_parallel 0> at program end, to ensure
255	that all threads are killed and that there are no outstanding requests.	2051	that all threads are killed and that there are no outstanding requests.
256		2052
257	Under normal circumstances you don't need to call this function.	2053	Under normal circumstances you don't need to call this function.
258		2054
		2055	=item IO::AIO::max_idle $nthreads
		2056
		2057	Limit the number of threads (default: 4) that are allowed to idle
		2058	(i.e., threads that did not get a request to process within the idle
		2059	timeout (default: 10 seconds). That means if a thread becomes idle while
		2060	C<$nthreads> other threads are also idle, it will free its resources and
		2061	exit.
		2062
		2063	This is useful when you allow a large number of threads (e.g. 100 or 1000)
		2064	to allow for extremely high load situations, but want to free resources
		2065	under normal circumstances (1000 threads can easily consume 30MB of RAM).
		2066
		2067	The default is probably ok in most situations, especially if thread
		2068	creation is fast. If thread creation is very slow on your system you might
		2069	want to use larger values.
		2070
		2071	=item IO::AIO::idle_timeout $seconds
		2072
		2073	Sets the minimum idle timeout (default 10) after which worker threads are
		2074	allowed to exit. SEe C<IO::AIO::max_idle>.
		2075
259	=item $oldnreqs = IO::AIO::max_outstanding $nreqs	2076	=item IO::AIO::max_outstanding $maxreqs
260		2077
261	Sets the maximum number of outstanding requests to C<$nreqs>. If you	2078	Sets the maximum number of outstanding requests to C<$nreqs>. If
262	try to queue up more than this number of requests, the caller will block until	2079	you do queue up more than this number of requests, the next call to
263	some requests have been handled.	2080	C<IO::AIO::poll_cb> (and other functions calling C<poll_cb>, such as
		2081	C<IO::AIO::flush> or C<IO::AIO::poll>) will block until the limit is no
		2082	longer exceeded.
264		2083
265	The default is very large, so normally there is no practical limit. If you	2084	In other words, this setting does not enforce a queue limit, but can be
266	queue up many requests in a loop it it often improves speed if you set	2085	used to make poll functions block if the limit is exceeded.
267	this to a relatively low number, such as C<100>.
268		2086
269	Under normal circumstances you don't need to call this function.	2087	This is a bad function to use in interactive programs because it blocks,
		2088	and a bad way to reduce concurrency because it is inexact. If you need to
		2089	issue many requests without being able to call a poll function on demand,
		2090	it is better to use an C<aio_group> together with a feed callback.
		2091
		2092	Its main use is in scripts without an event loop - when you want to stat a
		2093	lot of files, you can write something like this:
		2094
		2095	IO::AIO::max_outstanding 32;
		2096
		2097	for my $path (...) {
		2098	aio_stat $path , ...;
		2099	IO::AIO::poll_cb;
		2100	}
		2101
		2102	IO::AIO::flush;
		2103
		2104	The call to C<poll_cb> inside the loop will normally return instantly,
		2105	allowing the loop to progress, but as soon as more than C<32> requests
		2106	are in-flight, it will block until some requests have been handled. This
		2107	keeps the loop from pushing a large number of C<aio_stat> requests onto
		2108	the queue (which, with many paths to stat, can use up a lot of memory).
		2109
		2110	The default value for C<max_outstanding> is very large, so there is no
		2111	practical limit on the number of outstanding requests.
270		2112
271	=back	2113	=back
272		2114
		2115
		2116	=head3 STATISTICAL INFORMATION
		2117
		2118	=over
		2119
		2120	=item IO::AIO::nreqs
		2121
		2122	Returns the number of requests currently in the ready, execute or pending
		2123	states (i.e. for which their callback has not been invoked yet).
		2124
		2125	Example: wait till there are no outstanding requests anymore:
		2126
		2127	IO::AIO::poll_wait, IO::AIO::poll_cb
		2128	while IO::AIO::nreqs;
		2129
		2130	=item IO::AIO::nready
		2131
		2132	Returns the number of requests currently in the ready state (not yet
		2133	executed).
		2134
		2135	=item IO::AIO::npending
		2136
		2137	Returns the number of requests currently in the pending state (executed,
		2138	but not yet processed by poll_cb).
		2139
		2140	=back
		2141
		2142
		2143	=head3 SUBSECOND STAT TIME ACCESS
		2144
		2145	Both C<aio_stat>/C<aio_lstat> and perl's C<stat>/C<lstat> functions can
		2146	generally find access/modification and change times with subsecond time
		2147	accuracy of the system supports it, but perl's built-in functions only
		2148	return the integer part.
		2149
		2150	The following functions return the timestamps of the most recent
		2151	stat with subsecond precision on most systems and work both after
		2152	C<aio_stat>/C<aio_lstat> and perl's C<stat>/C<lstat> calls. Their return
		2153	value is only meaningful after a successful C<stat>/C<lstat> call, or
		2154	during/after a successful C<aio_stat>/C<aio_lstat> callback.
		2155
		2156	This is similar to the L<Time::HiRes> C<stat> functions, but can return
		2157	full resolution without rounding and work with standard perl C<stat>,
		2158	alleviating the need to call the special C<Time::HiRes> functions, which
		2159	do not act like their perl counterparts.
		2160
		2161	On operating systems or file systems where subsecond time resolution is
		2162	not supported or could not be detected, a fractional part of C<0> is
		2163	returned, so it is always safe to call these functions.
		2164
		2165	=over 4
		2166
		2167	=item $seconds = IO::AIO::st_atime, IO::AIO::st_mtime, IO::AIO::st_ctime, IO::AIO::st_btime
		2168
		2169	Return the access, modication, change or birth time, respectively,
		2170	including fractional part. Due to the limited precision of floating point,
		2171	the accuracy on most platforms is only a bit better than milliseconds
		2172	for times around now - see the I<nsec> function family, below, for full
		2173	accuracy.
		2174
		2175	File birth time is only available when the OS and perl support it (on
		2176	FreeBSD and NetBSD at the time of this writing, although support is
		2177	adaptive, so if your OS/perl gains support, IO::AIO can take advantage of
		2178	it). On systems where it isn't available, C<0> is currently returned, but
		2179	this might change to C<undef> in a future version.
		2180
		2181	=item ($atime, $mtime, $ctime, $btime, ...) = IO::AIO::st_xtime
		2182
		2183	Returns access, modification, change and birth time all in one go, and
		2184	maybe more times in the future version.
		2185
		2186	=item $nanoseconds = IO::AIO::st_atimensec, IO::AIO::st_mtimensec, IO::AIO::st_ctimensec, IO::AIO::st_btimensec
		2187
		2188	Return the fractional access, modifcation, change or birth time, in nanoseconds,
		2189	as an integer in the range C<0> to C<999999999>.
		2190
		2191	Note that no accessors are provided for access, modification and
		2192	change times - you need to get those from C<stat _> if required (C<int
		2193	IO::AIO::st_atime> and so on will I<not> generally give you the correct
		2194	value).
		2195
		2196	=item $seconds = IO::AIO::st_btimesec
		2197
		2198	The (integral) seconds part of the file birth time, if available.
		2199
		2200	=item ($atime, $mtime, $ctime, $btime, ...) = IO::AIO::st_xtimensec
		2201
		2202	Like the functions above, but returns all four times in one go (and maybe
		2203	more in future versions).
		2204
		2205	=item $counter = IO::AIO::st_gen
		2206
		2207	Returns the generation counter (in practice this is just a random number)
		2208	of the file. This is only available on platforms which have this member in
		2209	their C<struct stat> (most BSDs at the time of this writing) and generally
		2210	only to the root usert. If unsupported, C<0> is returned, but this might
		2211	change to C<undef> in a future version.
		2212
		2213	=back
		2214
		2215	Example: print the high resolution modification time of F</etc>, using
		2216	C<stat>, and C<IO::AIO::aio_stat>.
		2217
		2218	if (stat "/etc") {
		2219	printf "stat(/etc) mtime: %f\n", IO::AIO::st_mtime;
		2220	}
		2221
		2222	IO::AIO::aio_stat "/etc", sub {
		2223	$_[0]
		2224	and return;
		2225
		2226	printf "aio_stat(/etc) mtime: %d.%09d\n", (stat _)[9], IO::AIO::st_mtimensec;
		2227	};
		2228
		2229	IO::AIO::flush;
		2230
		2231	Output of the awbove on my system, showing reduced and full accuracy:
		2232
		2233	stat(/etc) mtime: 1534043702.020808
		2234	aio_stat(/etc) mtime: 1534043702.020807792
		2235
		2236
		2237	=head3 MISCELLANEOUS FUNCTIONS
		2238
		2239	IO::AIO implements some functions that are useful when you want to use
		2240	some "Advanced I/O" function not available to in Perl, without going the
		2241	"Asynchronous I/O" route. Many of these have an asynchronous C<aio_*>
		2242	counterpart.
		2243
		2244	=over 4
		2245
		2246	=item $retval = IO::AIO::fexecve $fh, $argv, $envp
		2247
		2248	A more-or-less direct equivalent to the POSIX C<fexecve> functions, which
		2249	allows you to specify the program to be executed via a file descriptor (or
		2250	handle). Returns C<-1> and sets errno to C<ENOSYS> if not available.
		2251
		2252	=item $retval = IO::AIO::mount $special, $path, $fstype, $flags = 0, $data = undef
		2253
		2254	Calls the GNU/Linux mount syscall with the given arguments. All except
		2255	C<$flags> are strings, and if C<$data> is C<undef>, a C<NULL> will be
		2256	passed.
		2257
		2258	The following values for C<$flags> are available:
		2259
		2260	C<IO::AIO::MS_RDONLY>, C<IO::AIO::MS_NOSUID>, C<IO::AIO::MS_NODEV>, C<IO::AIO::MS_NOEXEC>, C<IO::AIO::MS_SYNCHRONOUS>,
		2261	C<IO::AIO::MS_REMOUNT>, C<IO::AIO::MS_MANDLOCK>, C<IO::AIO::MS_DIRSYNC>, C<IO::AIO::MS_NOATIME>,
		2262	C<IO::AIO::MS_NODIRATIME>, C<IO::AIO::MS_BIND>, C<IO::AIO::MS_MOVE>, C<IO::AIO::MS_REC>, C<IO::AIO::MS_SILENT>,
		2263	C<IO::AIO::MS_POSIXACL>, C<IO::AIO::MS_UNBINDABLE>, C<IO::AIO::MS_PRIVATE>, C<IO::AIO::MS_SLAVE>, C<IO::AIO::MS_SHARED>,
		2264	C<IO::AIO::MS_RELATIME>, C<IO::AIO::MS_KERNMOUNT>, C<IO::AIO::MS_I_VERSION>, C<IO::AIO::MS_STRICTATIME>,
		2265	C<IO::AIO::MS_LAZYTIME>, C<IO::AIO::MS_ACTIVE>, C<IO::AIO::MS_NOUSER>, C<IO::AIO::MS_RMT_MASK>, C<IO::AIO::MS_MGC_VAL> and
		2266	C<IO::AIO::MS_MGC_MSK>.
		2267
		2268	=item $retval = IO::AIO::umount $path, $flags = 0
		2269
		2270	Invokes the GNU/Linux C<umount> or C<umount2> syscalls. Always calls
		2271	C<umount> if C<$flags> is C<0>, otherwqise always tries to call
		2272	C<umount2>.
		2273
		2274	The following C<$flags> are available:
		2275
		2276	C<IO::AIO::MNT_FORCE>, C<IO::AIO::MNT_DETACH>, C<IO::AIO::MNT_EXPIRE> and C<IO::AIO::UMOUNT_NOFOLLOW>.
		2277
		2278	=item $numfd = IO::AIO::get_fdlimit
		2279
		2280	Tries to find the current file descriptor limit and returns it, or
		2281	C<undef> and sets C<$!> in case of an error. The limit is one larger than
		2282	the highest valid file descriptor number.
		2283
		2284	=item IO::AIO::min_fdlimit [$numfd]
		2285
		2286	Try to increase the current file descriptor limit(s) to at least C<$numfd>
		2287	by changing the soft or hard file descriptor resource limit. If C<$numfd>
		2288	is missing, it will try to set a very high limit, although this is not
		2289	recommended when you know the actual minimum that you require.
		2290
		2291	If the limit cannot be raised enough, the function makes a best-effort
		2292	attempt to increase the limit as much as possible, using various
		2293	tricks, while still failing. You can query the resulting limit using
		2294	C<IO::AIO::get_fdlimit>.
		2295
		2296	If an error occurs, returns C<undef> and sets C<$!>, otherwise returns
		2297	true.
		2298
		2299	=item IO::AIO::sendfile $ofh, $ifh, $offset, $count
		2300
		2301	Calls the C<eio_sendfile_sync> function, which is like C<aio_sendfile>,
		2302	but is blocking (this makes most sense if you know the input data is
		2303	likely cached already and the output filehandle is set to non-blocking
		2304	operations).
		2305
		2306	Returns the number of bytes copied, or C<-1> on error.
		2307
		2308	=item IO::AIO::fadvise $fh, $offset, $len, $advice
		2309
		2310	Simply calls the C<posix_fadvise> function (see its
		2311	manpage for details). The following advice constants are
		2312	available: C<IO::AIO::FADV_NORMAL>, C<IO::AIO::FADV_SEQUENTIAL>,
		2313	C<IO::AIO::FADV_RANDOM>, C<IO::AIO::FADV_NOREUSE>,
		2314	C<IO::AIO::FADV_WILLNEED>, C<IO::AIO::FADV_DONTNEED>.
		2315
		2316	On systems that do not implement C<posix_fadvise>, this function returns
		2317	ENOSYS, otherwise the return value of C<posix_fadvise>.
		2318
		2319	=item IO::AIO::madvise $scalar, $offset, $len, $advice
		2320
		2321	Simply calls the C<posix_madvise> function (see its
		2322	manpage for details). The following advice constants are
		2323	available: C<IO::AIO::MADV_NORMAL>, C<IO::AIO::MADV_SEQUENTIAL>,
		2324	C<IO::AIO::MADV_RANDOM>, C<IO::AIO::MADV_WILLNEED>,
		2325	C<IO::AIO::MADV_DONTNEED>.
		2326
		2327	If C<$offset> is negative, counts from the end. If C<$length> is negative,
		2328	the remaining length of the C<$scalar> is used. If possible, C<$length>
		2329	will be reduced to fit into the C<$scalar>.
		2330
		2331	On systems that do not implement C<posix_madvise>, this function returns
		2332	ENOSYS, otherwise the return value of C<posix_madvise>.
		2333
		2334	=item IO::AIO::mprotect $scalar, $offset, $len, $protect
		2335
		2336	Simply calls the C<mprotect> function on the preferably AIO::mmap'ed
		2337	$scalar (see its manpage for details). The following protect
		2338	constants are available: C<IO::AIO::PROT_NONE>, C<IO::AIO::PROT_READ>,
		2339	C<IO::AIO::PROT_WRITE>, C<IO::AIO::PROT_EXEC>.
		2340
		2341	If C<$offset> is negative, counts from the end. If C<$length> is negative,
		2342	the remaining length of the C<$scalar> is used. If possible, C<$length>
		2343	will be reduced to fit into the C<$scalar>.
		2344
		2345	On systems that do not implement C<mprotect>, this function returns
		2346	ENOSYS, otherwise the return value of C<mprotect>.
		2347
		2348	=item IO::AIO::mmap $scalar, $length, $prot, $flags, $fh[, $offset]
		2349
		2350	Memory-maps a file (or anonymous memory range) and attaches it to the
		2351	given C<$scalar>, which will act like a string scalar. Returns true on
		2352	success, and false otherwise.
		2353
		2354	The scalar must exist, but its contents do not matter - this means you
		2355	cannot use a nonexistant array or hash element. When in doubt, C<undef>
		2356	the scalar first.
		2357
		2358	The only operations allowed on the mmapped scalar are C<substr>/C<vec>,
		2359	which don't change the string length, and most read-only operations such
		2360	as copying it or searching it with regexes and so on.
		2361
		2362	Anything else is unsafe and will, at best, result in memory leaks.
		2363
		2364	The memory map associated with the C<$scalar> is automatically removed
		2365	when the C<$scalar> is undef'd or destroyed, or when the C<IO::AIO::mmap>
		2366	or C<IO::AIO::munmap> functions are called on it.
		2367
		2368	This calls the C<mmap>(2) function internally. See your system's manual
		2369	page for details on the C<$length>, C<$prot> and C<$flags> parameters.
		2370
		2371	The C<$length> must be larger than zero and smaller than the actual
		2372	filesize.
		2373
		2374	C<$prot> is a combination of C<IO::AIO::PROT_NONE>, C<IO::AIO::PROT_EXEC>,
		2375	C<IO::AIO::PROT_READ> and/or C<IO::AIO::PROT_WRITE>,
		2376
		2377	C<$flags> can be a combination of
		2378	C<IO::AIO::MAP_SHARED> or
		2379	C<IO::AIO::MAP_PRIVATE>,
		2380	or a number of system-specific flags (when not available, the are C<0>):
		2381	C<IO::AIO::MAP_ANONYMOUS> (which is set to C<MAP_ANON> if your system only provides this constant),
		2382	C<IO::AIO::MAP_LOCKED>,
		2383	C<IO::AIO::MAP_NORESERVE>,
		2384	C<IO::AIO::MAP_POPULATE>,
		2385	C<IO::AIO::MAP_NONBLOCK>,
		2386	C<IO::AIO::MAP_FIXED>,
		2387	C<IO::AIO::MAP_GROWSDOWN>,
		2388	C<IO::AIO::MAP_32BIT>,
		2389	C<IO::AIO::MAP_HUGETLB>,
		2390	C<IO::AIO::MAP_STACK>,
		2391	C<IO::AIO::MAP_FIXED_NOREPLACE>,
		2392	C<IO::AIO::MAP_SHARED_VALIDATE>,
		2393	C<IO::AIO::MAP_SYNC> or
		2394	C<IO::AIO::MAP_UNINITIALIZED>.
		2395
		2396	If C<$fh> is C<undef>, then a file descriptor of C<-1> is passed.
		2397
		2398	C<$offset> is the offset from the start of the file - it generally must be
		2399	a multiple of C<IO::AIO::PAGESIZE> and defaults to C<0>.
		2400
		2401	Example:
		2402
		2403	use Digest::MD5;
		2404	use IO::AIO;
		2405
		2406	open my $fh, "<verybigfile"
		2407	or die "$!";
		2408
		2409	IO::AIO::mmap my $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh
		2410	or die "verybigfile: $!";
		2411
		2412	my $fast_md5 = md5 $data;
		2413
		2414	=item IO::AIO::munmap $scalar
		2415
		2416	Removes a previous mmap and undefines the C<$scalar>.
		2417
		2418	=item IO::AIO::mremap $scalar, $new_length, $flags = MREMAP_MAYMOVE[, $new_address = 0]
		2419
		2420	Calls the Linux-specific mremap(2) system call. The C<$scalar> must have
		2421	been mapped by C<IO::AIO::mmap>, and C<$flags> must currently either be
		2422	C<0> or C<IO::AIO::MREMAP_MAYMOVE>.
		2423
		2424	Returns true if successful, and false otherwise. If the underlying mmapped
		2425	region has changed address, then the true value has the numerical value
		2426	C<1>, otherwise it has the numerical value C<0>:
		2427
		2428	my $success = IO::AIO::mremap $mmapped, 8192, IO::AIO::MREMAP_MAYMOVE
		2429	or die "mremap: $!";
		2430
		2431	if ($success*1) {
		2432	warn "scalar has chanegd address in memory\n";
		2433	}
		2434
		2435	C<IO::AIO::MREMAP_FIXED> and the C<$new_address> argument are currently
		2436	implemented, but not supported and might go away in a future version.
		2437
		2438	On systems where this call is not supported or is not emulated, this call
		2439	returns falls and sets C<$!> to C<ENOSYS>.
		2440
		2441	=item IO::AIO::mlockall $flags
		2442
		2443	Calls the C<eio_mlockall_sync> function, which is like C<aio_mlockall>,
		2444	but is blocking.
		2445
		2446	=item IO::AIO::munlock $scalar, $offset = 0, $length = undef
		2447
		2448	Calls the C<munlock> function, undoing the effects of a previous
		2449	C<aio_mlock> call (see its description for details).
		2450
		2451	=item IO::AIO::munlockall
		2452
		2453	Calls the C<munlockall> function.
		2454
		2455	On systems that do not implement C<munlockall>, this function returns
		2456	ENOSYS, otherwise the return value of C<munlockall>.
		2457
		2458	=item $fh = IO::AIO::accept4 $r_fh, $sockaddr, $sockaddr_maxlen, $flags
		2459
		2460	Uses the GNU/Linux C<accept4(2)> syscall, if available, to accept a socket
		2461	and return the new file handle on success, or sets C<$!> and returns
		2462	C<undef> on error.
		2463
		2464	The remote name of the new socket will be stored in C<$sockaddr>, which
		2465	will be extended to allow for at least C<$sockaddr_maxlen> octets. If the
		2466	socket name does not fit into C<$sockaddr_maxlen> octets, this is signaled
		2467	by returning a longer string in C<$sockaddr>, which might or might not be
		2468	truncated.
		2469
		2470	To accept name-less sockets, use C<undef> for C<$sockaddr> and C<0> for
		2471	C<$sockaddr_maxlen>.
		2472
		2473	The main reasons to use this syscall rather than portable C<accept(2)>
		2474	are that you can specify C<SOCK_NONBLOCK> and/or C<SOCK_CLOEXEC>
		2475	flags and you can accept name-less sockets by specifying C<0> for
		2476	C<$sockaddr_maxlen>, which is sadly not possible with perl's interface to
		2477	C<accept>.
		2478
		2479	=item IO::AIO::splice $r_fh, $r_off, $w_fh, $w_off, $length, $flags
		2480
		2481	Calls the GNU/Linux C<splice(2)> syscall, if available. If C<$r_off> or
		2482	C<$w_off> are C<undef>, then C<NULL> is passed for these, otherwise they
		2483	should be the file offset.
		2484
		2485	C<$r_fh> and C<$w_fh> should not refer to the same file, as splice might
		2486	silently corrupt the data in this case.
		2487
		2488	The following symbol flag values are available: C<IO::AIO::SPLICE_F_MOVE>,
		2489	C<IO::AIO::SPLICE_F_NONBLOCK>, C<IO::AIO::SPLICE_F_MORE> and
		2490	C<IO::AIO::SPLICE_F_GIFT>.
		2491
		2492	See the C<splice(2)> manpage for details.
		2493
		2494	=item IO::AIO::tee $r_fh, $w_fh, $length, $flags
		2495
		2496	Calls the GNU/Linux C<tee(2)> syscall, see its manpage and the
		2497	description for C<IO::AIO::splice> above for details.
		2498
		2499	=item $actual_size = IO::AIO::pipesize $r_fh[, $new_size]
		2500
		2501	Attempts to query or change the pipe buffer size. Obviously works only
		2502	on pipes, and currently works only on GNU/Linux systems, and fails with
		2503	C<-1>/C<ENOSYS> everywhere else. If anybody knows how to influence pipe buffer
		2504	size on other systems, drop me a note.
		2505
		2506	=item ($rfh, $wfh) = IO::AIO::pipe2 [$flags]
		2507
		2508	This is a direct interface to the Linux L<pipe2(2)> system call. If
		2509	C<$flags> is missing or C<0>, then this should be the same as a call to
		2510	perl's built-in C<pipe> function and create a new pipe, and works on
		2511	systems that lack the pipe2 syscall. On win32, this case invokes C<_pipe
		2512	(..., 4096, O_BINARY)>.
		2513
		2514	If C<$flags> is non-zero, it tries to invoke the pipe2 system call with
		2515	the given flags (Linux 2.6.27, glibc 2.9).
		2516
		2517	On success, the read and write file handles are returned.
		2518
		2519	On error, nothing will be returned. If the pipe2 syscall is missing and
		2520	C<$flags> is non-zero, fails with C<ENOSYS>.
		2521
		2522	Please refer to L<pipe2(2)> for more info on the C<$flags>, but at the
		2523	time of this writing, C<IO::AIO::O_CLOEXEC>, C<IO::AIO::O_NONBLOCK> and
		2524	C<IO::AIO::O_DIRECT> (Linux 3.4, for packet-based pipes) were supported.
		2525
		2526	Example: create a pipe race-free w.r.t. threads and fork:
		2527
		2528	my ($rfh, $wfh) = IO::AIO::pipe2 IO::AIO::O_CLOEXEC
		2529	or die "pipe2: $!\n";
		2530
		2531	=item $fh = IO::AIO::memfd_create $pathname[, $flags]
		2532
		2533	This is a direct interface to the Linux L<memfd_create(2)> system
		2534	call. The (unhelpful) default for C<$flags> is C<0>, but your default
		2535	should be C<IO::AIO::MFD_CLOEXEC>.
		2536
		2537	On success, the new memfd filehandle is returned, otherwise returns
		2538	C<undef>. If the memfd_create syscall is missing, fails with C<ENOSYS>.
		2539
		2540	Please refer to L<memfd_create(2)> for more info on this call.
		2541
		2542	The following C<$flags> values are available: C<IO::AIO::MFD_CLOEXEC>,
		2543	C<IO::AIO::MFD_ALLOW_SEALING>, C<IO::AIO::MFD_HUGETLB>,
		2544	C<IO::AIO::MFD_HUGETLB_2MB> and C<IO::AIO::MFD_HUGETLB_1GB>.
		2545
		2546	Example: create a new memfd.
		2547
		2548	my $fh = IO::AIO::memfd_create "somenameforprocfd", IO::AIO::MFD_CLOEXEC
		2549	or die "memfd_create: $!\n";
		2550
		2551	=item $fh = IO::AIO::pidfd_open $pid[, $flags]
		2552
		2553	This is an interface to the Linux L<pidfd_open(2)> system call. The
		2554	default for C<$flags> is C<0>.
		2555
		2556	On success, a new pidfd filehandle is returned (that is already set to
		2557	close-on-exec), otherwise returns C<undef>. If the syscall is missing,
		2558	fails with C<ENOSYS>.
		2559
		2560	Example: open pid 6341 as pidfd.
		2561
		2562	my $fh = IO::AIO::pidfd_open 6341
		2563	or die "pidfd_open: $!\n";
		2564
		2565	=item $status = IO::AIO::pidfd_send_signal $pidfh, $signal[, $siginfo[, $flags]]
		2566
		2567	This is an interface to the Linux L<pidfd_send_signal> system call. The
		2568	default for C<$siginfo> is C<undef> and the default for C<$flags> is C<0>.
		2569
		2570	Returns the system call status. If the syscall is missing, fails with
		2571	C<ENOSYS>.
		2572
		2573	When specified, C<$siginfo> must be a reference to a hash with one or more
		2574	of the following members:
		2575
		2576	=over
		2577
		2578	=item code - the C<si_code> member
		2579
		2580	=item pid - the C<si_pid> member
		2581
		2582	=item uid - the C<si_uid> member
		2583
		2584	=item value_int - the C<si_value.sival_int> member
		2585
		2586	=item value_ptr - the C<si_value.sival_ptr> member, specified as an integer
		2587
		2588	=back
		2589
		2590	Example: send a SIGKILL to the specified process.
		2591
		2592	my $status = IO::AIO::pidfd_send_signal $pidfh, 9, undef
		2593	and die "pidfd_send_signal: $!\n";
		2594
		2595	Example: send a SIGKILL to the specified process with extra data.
		2596
		2597	my $status = IO::AIO::pidfd_send_signal $pidfh, 9, { code => -1, value_int => 7 }
		2598	and die "pidfd_send_signal: $!\n";
		2599
		2600	=item $fh = IO::AIO::pidfd_getfd $pidfh, $targetfd[, $flags]
		2601
		2602	This is an interface to the Linux L<pidfd_getfd> system call. The default
		2603	for C<$flags> is C<0>.
		2604
		2605	On success, returns a dup'ed copy of the target file descriptor (specified
		2606	as an integer) returned (that is already set to close-on-exec), otherwise
		2607	returns C<undef>. If the syscall is missing, fails with C<ENOSYS>.
		2608
		2609	Example: get a copy of standard error of another process and print soemthing to it.
		2610
		2611	my $errfh = IO::AIO::pidfd_getfd $pidfh, 2
		2612	or die "pidfd_getfd: $!\n";
		2613	print $errfh "stderr\n";
		2614
		2615	=item $fh = IO::AIO::eventfd [$initval, [$flags]]
		2616
		2617	This is a direct interface to the Linux L<eventfd(2)> system call. The
		2618	(unhelpful) defaults for C<$initval> and C<$flags> are C<0> for both.
		2619
		2620	On success, the new eventfd filehandle is returned, otherwise returns
		2621	C<undef>. If the eventfd syscall is missing, fails with C<ENOSYS>.
		2622
		2623	Please refer to L<eventfd(2)> for more info on this call.
		2624
		2625	The following symbol flag values are available: C<IO::AIO::EFD_CLOEXEC>,
		2626	C<IO::AIO::EFD_NONBLOCK> and C<IO::AIO::EFD_SEMAPHORE> (Linux 2.6.30).
		2627
		2628	Example: create a new eventfd filehandle:
		2629
		2630	$fh = IO::AIO::eventfd 0, IO::AIO::EFD_CLOEXEC
		2631	or die "eventfd: $!\n";
		2632
		2633	=item $fh = IO::AIO::timerfd_create $clockid[, $flags]
		2634
		2635	This is a direct interface to the Linux L<timerfd_create(2)> system
		2636	call. The (unhelpful) default for C<$flags> is C<0>, but your default
		2637	should be C<IO::AIO::TFD_CLOEXEC>.
		2638
		2639	On success, the new timerfd filehandle is returned, otherwise returns
		2640	C<undef>. If the timerfd_create syscall is missing, fails with C<ENOSYS>.
		2641
		2642	Please refer to L<timerfd_create(2)> for more info on this call.
		2643
		2644	The following C<$clockid> values are
		2645	available: C<IO::AIO::CLOCK_REALTIME>, C<IO::AIO::CLOCK_MONOTONIC>
		2646	C<IO::AIO::CLOCK_CLOCK_BOOTTIME> (Linux 3.15)
		2647	C<IO::AIO::CLOCK_CLOCK_REALTIME_ALARM> (Linux 3.11) and
		2648	C<IO::AIO::CLOCK_CLOCK_BOOTTIME_ALARM> (Linux 3.11).
		2649
		2650	The following C<$flags> values are available (Linux
		2651	2.6.27): C<IO::AIO::TFD_NONBLOCK> and C<IO::AIO::TFD_CLOEXEC>.
		2652
		2653	Example: create a new timerfd and set it to one-second repeated alarms,
		2654	then wait for two alarms:
		2655
		2656	my $fh = IO::AIO::timerfd_create IO::AIO::CLOCK_BOOTTIME, IO::AIO::TFD_CLOEXEC
		2657	or die "timerfd_create: $!\n";
		2658
		2659	defined IO::AIO::timerfd_settime $fh, 0, 1, 1
		2660	or die "timerfd_settime: $!\n";
		2661
		2662	for (1..2) {
		2663	8 == sysread $fh, my $buf, 8
		2664	or die "timerfd read failure\n";
		2665
		2666	printf "number of expirations (likely 1): %d\n",
		2667	unpack "Q", $buf;
		2668	}
		2669
		2670	=item ($cur_interval, $cur_value) = IO::AIO::timerfd_settime $fh, $flags, $new_interval, $nbw_value
		2671
		2672	This is a direct interface to the Linux L<timerfd_settime(2)> system
		2673	call. Please refer to its manpage for more info on this call.
		2674
		2675	The new itimerspec is specified using two (possibly fractional) second
		2676	values, C<$new_interval> and C<$new_value>).
		2677
		2678	On success, the current interval and value are returned (as per
		2679	C<timerfd_gettime>). On failure, the empty list is returned.
		2680
		2681	The following C<$flags> values are
		2682	available: C<IO::AIO::TFD_TIMER_ABSTIME> and
		2683	C<IO::AIO::TFD_TIMER_CANCEL_ON_SET>.
		2684
		2685	See C<IO::AIO::timerfd_create> for a full example.
		2686
		2687	=item ($cur_interval, $cur_value) = IO::AIO::timerfd_gettime $fh
		2688
		2689	This is a direct interface to the Linux L<timerfd_gettime(2)> system
		2690	call. Please refer to its manpage for more info on this call.
		2691
		2692	On success, returns the current values of interval and value for the given
		2693	timerfd (as potentially fractional second values). On failure, the empty
		2694	list is returned.
		2695
		2696	=back
		2697
273	=cut	2698	=cut
274		2699
275	# support function to convert a fd into a perl filehandle
276	sub _fd2fh {
277	return undef if $_[0] < 0;
278
279	# try to be perl5.6-compatible
280	local *AIO_FH;
281	open AIO_FH, "+<&=$_[0]"
282	or return undef;
283
284	*AIO_FH
285	}
286
287	min_parallel 4;	2700	min_parallel 8;
288		2701
289	END {	2702	END { flush }
290	max_parallel 0;
291	}
292		2703
293	1;	2704	1;
294		2705
		2706	=head1 EVENT LOOP INTEGRATION
		2707
		2708	It is recommended to use L<AnyEvent::AIO> to integrate IO::AIO
		2709	automatically into many event loops:
		2710
		2711	# AnyEvent integration (EV, Event, Glib, Tk, POE, urxvt, pureperl...)
		2712	use AnyEvent::AIO;
		2713
		2714	You can also integrate IO::AIO manually into many event loops, here are
		2715	some examples of how to do this:
		2716
		2717	# EV integration
		2718	my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
		2719
		2720	# Event integration
		2721	Event->io (fd => IO::AIO::poll_fileno,
		2722	poll => 'r',
		2723	cb => \&IO::AIO::poll_cb);
		2724
		2725	# Glib/Gtk2 integration
		2726	add_watch Glib::IO IO::AIO::poll_fileno,
		2727	in => sub { IO::AIO::poll_cb; 1 };
		2728
		2729	# Tk integration
		2730	Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",
		2731	readable => \&IO::AIO::poll_cb);
		2732
		2733	# Danga::Socket integration
		2734	Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
		2735	\&IO::AIO::poll_cb);
		2736
		2737	=head2 FORK BEHAVIOUR
		2738
		2739	Usage of pthreads in a program changes the semantics of fork
		2740	considerably. Specifically, only async-safe functions can be called after
		2741	fork. Perl doesn't know about this, so in general, you cannot call fork
		2742	with defined behaviour in perl if pthreads are involved. IO::AIO uses
		2743	pthreads, so this applies, but many other extensions and (for inexplicable
		2744	reasons) perl itself often is linked against pthreads, so this limitation
		2745	applies to quite a lot of perls.
		2746
		2747	This module no longer tries to fight your OS, or POSIX. That means IO::AIO
		2748	only works in the process that loaded it. Forking is fully supported, but
		2749	using IO::AIO in the child is not.
		2750
		2751	You might get around by not I<using> IO::AIO before (or after)
		2752	forking. You could also try to call the L<IO::AIO::reinit> function in the
		2753	child:
		2754
		2755	=over 4
		2756
		2757	=item IO::AIO::reinit
		2758
		2759	Abandons all current requests and I/O threads and simply reinitialises all
		2760	data structures. This is not an operation supported by any standards, but
		2761	happens to work on GNU/Linux and some newer BSD systems.
		2762
		2763	The only reasonable use for this function is to call it after forking, if
		2764	C<IO::AIO> was used in the parent. Calling it while IO::AIO is active in
		2765	the process will result in undefined behaviour. Calling it at any time
		2766	will also result in any undefined (by POSIX) behaviour.
		2767
		2768	=back
		2769
		2770	=head2 LINUX-SPECIFIC CALLS
		2771
		2772	When a call is documented as "linux-specific" then this means it
		2773	originated on GNU/Linux. C<IO::AIO> will usually try to autodetect the
		2774	availability and compatibility of such calls regardless of the platform
		2775	it is compiled on, so platforms such as FreeBSD which often implement
		2776	these calls will work. When in doubt, call them and see if they fail wth
		2777	C<ENOSYS>.
		2778
		2779	=head2 MEMORY USAGE
		2780
		2781	Per-request usage:
		2782
		2783	Each aio request uses - depending on your architecture - around 100-200
		2784	bytes of memory. In addition, stat requests need a stat buffer (possibly
		2785	a few hundred bytes), readdir requires a result buffer and so on. Perl
		2786	scalars and other data passed into aio requests will also be locked and
		2787	will consume memory till the request has entered the done state.
		2788
		2789	This is not awfully much, so queuing lots of requests is not usually a
		2790	problem.
		2791
		2792	Per-thread usage:
		2793
		2794	In the execution phase, some aio requests require more memory for
		2795	temporary buffers, and each thread requires a stack and other data
		2796	structures (usually around 16k-128k, depending on the OS).
		2797
		2798	=head1 KNOWN BUGS
		2799
		2800	Known bugs will be fixed in the next release :)
		2801
		2802	=head1 KNOWN ISSUES
		2803
		2804	Calls that try to "import" foreign memory areas (such as C<IO::AIO::mmap>
		2805	or C<IO::AIO::aio_slurp>) do not work with generic lvalues, such as
		2806	non-created hash slots or other scalars I didn't think of. It's best to
		2807	avoid such and either use scalar variables or making sure that the scalar
		2808	exists (e.g. by storing C<undef>) and isn't "funny" (e.g. tied).
		2809
		2810	I am not sure anything can be done about this, so this is considered a
		2811	known issue, rather than a bug.
		2812
295	=head1 SEE ALSO	2813	=head1 SEE ALSO
296		2814
297	L<Coro>, L<Linux::AIO>.	2815	L<AnyEvent::AIO> for easy integration into event loops, L<Coro::AIO> for a
		2816	more natural syntax and L<IO::FDPass> for file descriptor passing.
298		2817
299	=head1 AUTHOR	2818	=head1 AUTHOR
300		2819
301	Marc Lehmann <schmorp@schmorp.de>	2820	Marc Lehmann <schmorp@schmorp.de>
302	http://home.schmorp.de/	2821	http://home.schmorp.de/

Diff Legend

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