ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/IO-AIO/AIO.pm

Comparing IO-AIO/AIO.pm (file contents):
Revision 1.6 by root, Sun Jul 10 22:19:48 2005 UTC vs.
Revision 1.264 by root, Mon Jul 18 07:48:01 2016 UTC

…		…
4		4
5	=head1 SYNOPSIS	5	=head1 SYNOPSIS
6		6
7	use IO::AIO;	7	use IO::AIO;
8		8
9	aio_open "/etc/passwd", O_RDONLY, 0, sub {	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', async => 1,
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
47	Although the module will work with in the presence of other threads, it is	63	Although the module will work in the presence of other (Perl-) threads,
48	currently not reentrant, so use appropriate locking yourself.	64	it is currently not reentrant in any way, so use appropriate locking
		65	yourself, always call C<poll_cb> from within the same thread, or never
		66	call C<poll_cb> (or other C<aio_> functions) recursively.
		67
		68	=head2 EXAMPLE
		69
		70	This is a simple example that uses the EV module and loads
		71	F</etc/passwd> asynchronously:
		72
		73	use EV;
		74	use IO::AIO;
		75
		76	# register the IO::AIO callback with EV
		77	my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
		78
		79	# queue the request to open /etc/passwd
		80	aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub {
		81	my $fh = shift
		82	or die "error while opening: $!";
		83
		84	# stat'ing filehandles is generally non-blocking
		85	my $size = -s $fh;
		86
		87	# queue a request to read the file
		88	my $contents;
		89	aio_read $fh, 0, $size, $contents, 0, sub {
		90	$_[0] == $size
		91	or die "short read: $!";
		92
		93	close $fh;
		94
		95	# file contents now in $contents
		96	print $contents;
		97
		98	# exit event loop and program
		99	EV::break;
		100	};
		101	};
		102
		103	# possibly queue up other requests, or open GUI windows,
		104	# check for sockets etc. etc.
		105
		106	# process events as long as there are some:
		107	EV::run;
		108
		109	=head1 REQUEST ANATOMY AND LIFETIME
		110
		111	Every C<aio_*> function creates a request. which is a C data structure not
		112	directly visible to Perl.
		113
		114	If called in non-void context, every request function returns a Perl
		115	object representing the request. In void context, nothing is returned,
		116	which saves a bit of memory.
		117
		118	The perl object is a fairly standard ref-to-hash object. The hash contents
		119	are not used by IO::AIO so you are free to store anything you like in it.
		120
		121	During their existance, aio requests travel through the following states,
		122	in order:
		123
		124	=over 4
		125
		126	=item ready
		127
		128	Immediately after a request is created it is put into the ready state,
		129	waiting for a thread to execute it.
		130
		131	=item execute
		132
		133	A thread has accepted the request for processing and is currently
		134	executing it (e.g. blocking in read).
		135
		136	=item pending
		137
		138	The request has been executed and is waiting for result processing.
		139
		140	While request submission and execution is fully asynchronous, result
		141	processing is not and relies on the perl interpreter calling C<poll_cb>
		142	(or another function with the same effect).
		143
		144	=item result
		145
		146	The request results are processed synchronously by C<poll_cb>.
		147
		148	The C<poll_cb> function will process all outstanding aio requests by
		149	calling their callbacks, freeing memory associated with them and managing
		150	any groups they are contained in.
		151
		152	=item done
		153
		154	Request has reached the end of its lifetime and holds no resources anymore
		155	(except possibly for the Perl object, but its connection to the actual
		156	aio request is severed and calling its methods will either do nothing or
		157	result in a runtime error).
		158
		159	=back
49		160
50	=cut	161	=cut
51		162
52	package IO::AIO;	163	package IO::AIO;
53		164
		165	use Carp ();
		166
		167	use common::sense;
		168
54	use base 'Exporter';	169	use base 'Exporter';
55		170
56	use Fcntl ();
57
58	BEGIN {	171	BEGIN {
59	$VERSION = 0.2;	172	our $VERSION = 4.34;
60		173
61	@EXPORT = qw(aio_read aio_write aio_open aio_close aio_stat aio_lstat aio_unlink	174	our @AIO_REQ = qw(aio_sendfile aio_seek aio_read aio_write aio_open aio_close
62	aio_fsync aio_fdatasync aio_readahead);	175	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);	176	aio_scandir aio_symlink aio_readlink aio_realpath aio_fcntl aio_ioctl
		177	aio_sync aio_fsync aio_syncfs aio_fdatasync aio_sync_file_range
		178	aio_pathsync aio_readahead aio_fiemap aio_allocate
		179	aio_rename aio_link aio_move aio_copy aio_group
		180	aio_nop aio_mknod aio_load aio_rmtree aio_mkdir aio_chown
		181	aio_chmod aio_utime aio_truncate
		182	aio_msync aio_mtouch aio_mlock aio_mlockall
		183	aio_statvfs
		184	aio_wd);
		185
		186	our @EXPORT = (@AIO_REQ, qw(aioreq_pri aioreq_nice));
		187	our @EXPORT_OK = qw(poll_fileno poll_cb poll_wait flush
		188	min_parallel max_parallel max_idle idle_timeout
		189	nreqs nready npending nthreads
		190	max_poll_time max_poll_reqs
		191	sendfile fadvise madvise
		192	mmap munmap munlock munlockall);
		193
		194	push @AIO_REQ, qw(aio_busy); # not exported
		195
		196	@IO::AIO::GRP::ISA = 'IO::AIO::REQ';
64		197
65	require XSLoader;	198	require XSLoader;
66	XSLoader::load IO::AIO, $VERSION;	199	XSLoader::load ("IO::AIO", $VERSION);
67	}	200	}
68		201
69	=head1 FUNCTIONS	202	=head1 FUNCTIONS
70		203
71	=head2 AIO FUNCTIONS	204	=head2 QUICK OVERVIEW
		205
		206	This section simply lists the prototypes most of the functions for
		207	quick reference. See the following sections for function-by-function
		208	documentation.
		209
		210	aio_wd $pathname, $callback->($wd)
		211	aio_open $pathname, $flags, $mode, $callback->($fh)
		212	aio_close $fh, $callback->($status)
		213	aio_seek $fh,$offset,$whence, $callback->($offs)
		214	aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
		215	aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
		216	aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
		217	aio_readahead $fh,$offset,$length, $callback->($retval)
		218	aio_stat $fh_or_path, $callback->($status)
		219	aio_lstat $fh, $callback->($status)
		220	aio_statvfs $fh_or_path, $callback->($statvfs)
		221	aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
		222	aio_chown $fh_or_path, $uid, $gid, $callback->($status)
		223	aio_chmod $fh_or_path, $mode, $callback->($status)
		224	aio_truncate $fh_or_path, $offset, $callback->($status)
		225	aio_allocate $fh, $mode, $offset, $len, $callback->($status)
		226	aio_fiemap $fh, $start, $length, $flags, $count, $cb->(\@extents)
		227	aio_unlink $pathname, $callback->($status)
		228	aio_mknod $pathname, $mode, $dev, $callback->($status)
		229	aio_link $srcpath, $dstpath, $callback->($status)
		230	aio_symlink $srcpath, $dstpath, $callback->($status)
		231	aio_readlink $pathname, $callback->($link)
		232	aio_realpath $pathname, $callback->($path)
		233	aio_rename $srcpath, $dstpath, $callback->($status)
		234	aio_mkdir $pathname, $mode, $callback->($status)
		235	aio_rmdir $pathname, $callback->($status)
		236	aio_readdir $pathname, $callback->($entries)
		237	aio_readdirx $pathname, $flags, $callback->($entries, $flags)
		238	IO::AIO::READDIR_DENTS IO::AIO::READDIR_DIRS_FIRST
		239	IO::AIO::READDIR_STAT_ORDER IO::AIO::READDIR_FOUND_UNKNOWN
		240	aio_scandir $pathname, $maxreq, $callback->($dirs, $nondirs)
		241	aio_load $pathname, $data, $callback->($status)
		242	aio_copy $srcpath, $dstpath, $callback->($status)
		243	aio_move $srcpath, $dstpath, $callback->($status)
		244	aio_rmtree $pathname, $callback->($status)
		245	aio_fcntl $fh, $cmd, $arg, $callback->($status)
		246	aio_ioctl $fh, $request, $buf, $callback->($status)
		247	aio_sync $callback->($status)
		248	aio_syncfs $fh, $callback->($status)
		249	aio_fsync $fh, $callback->($status)
		250	aio_fdatasync $fh, $callback->($status)
		251	aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
		252	aio_pathsync $pathname, $callback->($status)
		253	aio_msync $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		254	aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		255	aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status)
		256	aio_mlockall $flags, $callback->($status)
		257	aio_group $callback->(...)
		258	aio_nop $callback->()
		259
		260	$prev_pri = aioreq_pri [$pri]
		261	aioreq_nice $pri_adjust
		262
		263	IO::AIO::poll_wait
		264	IO::AIO::poll_cb
		265	IO::AIO::poll
		266	IO::AIO::flush
		267	IO::AIO::max_poll_reqs $nreqs
		268	IO::AIO::max_poll_time $seconds
		269	IO::AIO::min_parallel $nthreads
		270	IO::AIO::max_parallel $nthreads
		271	IO::AIO::max_idle $nthreads
		272	IO::AIO::idle_timeout $seconds
		273	IO::AIO::max_outstanding $maxreqs
		274	IO::AIO::nreqs
		275	IO::AIO::nready
		276	IO::AIO::npending
		277
		278	IO::AIO::sendfile $ofh, $ifh, $offset, $count
		279	IO::AIO::fadvise $fh, $offset, $len, $advice
		280	IO::AIO::mmap $scalar, $length, $prot, $flags[, $fh[, $offset]]
		281	IO::AIO::munmap $scalar
		282	IO::AIO::madvise $scalar, $offset, $length, $advice
		283	IO::AIO::mprotect $scalar, $offset, $length, $protect
		284	IO::AIO::munlock $scalar, $offset = 0, $length = undef
		285	IO::AIO::munlockall
		286
		287	=head2 API NOTES
72		288
73	All the C<aio_*> calls are more or less thin wrappers around the syscall	289	All the C<aio_*> calls are more or less thin wrappers around the syscall
74	with the same name (sans C<aio_>). The arguments are similar or identical,	290	with the same name (sans C<aio_>). The arguments are similar or identical,
75	and they all accept an additional C<$callback> argument which must be	291	and they all accept an additional (and optional) C<$callback> argument
76	a code reference. This code reference will get called with the syscall	292	which must be a code reference. This code reference will be called after
		293	the syscall has been executed in an asynchronous fashion. The results
		294	of the request will be passed as arguments to the callback (and, if an
		295	error occured, in C<$!>) - for most requests the syscall return code (e.g.
77	return code (e.g. most syscalls return C<-1> on error, unlike perl, which	296	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	297	"false").
79	been executed asynchronously.
80		298
81	All functions that expect a filehandle will also accept a file descriptor.	299	Some requests (such as C<aio_readdir>) pass the actual results and
		300	communicate failures by passing C<undef>.
82		301
		302	All functions expecting a filehandle keep a copy of the filehandle
		303	internally until the request has finished.
		304
		305	All functions return request objects of type L<IO::AIO::REQ> that allow
		306	further manipulation of those requests while they are in-flight.
		307
83	The filenames you pass to these routines I<must> be absolute. The reason	308	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	309	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	310	current working directory could have changed. Alternatively, you can
86	never change the current working directory.	311	make sure that you never change the current working directory anywhere
		312	in the program and then use relative paths. You can also take advantage
		313	of IO::AIOs working directory abstraction, that lets you specify paths
		314	relative to some previously-opened "working directory object" - see the
		315	description of the C<IO::AIO::WD> class later in this document.
		316
		317	To encode pathnames as octets, either make sure you either: a) always pass
		318	in filenames you got from outside (command line, readdir etc.) without
		319	tinkering, b) are in your native filesystem encoding, c) use the Encode
		320	module and encode your pathnames to the locale (or other) encoding in
		321	effect in the user environment, d) use Glib::filename_from_unicode on
		322	unicode filenames or e) use something else to ensure your scalar has the
		323	correct contents.
		324
		325	This works, btw. independent of the internal UTF-8 bit, which IO::AIO
		326	handles correctly whether it is set or not.
		327
		328	=head2 AIO REQUEST FUNCTIONS
87		329
88	=over 4	330	=over 4
89		331
		332	=item $prev_pri = aioreq_pri [$pri]
		333
		334	Returns the priority value that would be used for the next request and, if
		335	C<$pri> is given, sets the priority for the next aio request.
		336
		337	The default priority is C<0>, the minimum and maximum priorities are C<-4>
		338	and C<4>, respectively. Requests with higher priority will be serviced
		339	first.
		340
		341	The priority will be reset to C<0> after each call to one of the C<aio_*>
		342	functions.
		343
		344	Example: open a file with low priority, then read something from it with
		345	higher priority so the read request is serviced before other low priority
		346	open requests (potentially spamming the cache):
		347
		348	aioreq_pri -3;
		349	aio_open ..., sub {
		350	return unless $_[0];
		351
		352	aioreq_pri -2;
		353	aio_read $_[0], ..., sub {
		354	...
		355	};
		356	};
		357
		358
		359	=item aioreq_nice $pri_adjust
		360
		361	Similar to C<aioreq_pri>, but subtracts the given value from the current
		362	priority, so the effect is cumulative.
		363
		364
90	=item aio_open $pathname, $flags, $mode, $callback	365	=item aio_open $pathname, $flags, $mode, $callback->($fh)
91		366
92	Asynchronously open or create a file and call the callback with a newly	367	Asynchronously open or create a file and call the callback with a newly
93	created filehandle for the file.	368	created filehandle for the file (or C<undef> in case of an error).
94		369
95	The pathname passed to C<aio_open> must be absolute. See API NOTES, above,	370	The pathname passed to C<aio_open> must be absolute. See API NOTES, above,
96	for an explanation.	371	for an explanation.
97		372
98	The C<$mode> argument is a bitmask. See the C<Fcntl> module for a	373	The C<$flags> argument is a bitmask. See the C<Fcntl> module for a
99	list. They are the same as used in C<sysopen>.	374	list. They are the same as used by C<sysopen>.
		375
		376	Likewise, C<$mode> specifies the mode of the newly created file, if it
		377	didn't exist and C<O_CREAT> has been given, just like perl's C<sysopen>,
		378	except that it is mandatory (i.e. use C<0> if you don't create new files,
		379	and C<0666> or C<0777> if you do). Note that the C<$mode> will be modified
		380	by the umask in effect then the request is being executed, so better never
		381	change the umask.
100		382
101	Example:	383	Example:
102		384
103	aio_open "/etc/passwd", O_RDONLY, 0, sub {	385	aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub {
104	if ($_[0]) {	386	if ($_[0]) {
105	print "open successful, fh is $_[0]\n";	387	print "open successful, fh is $_[0]\n";
106	...	388	...
107	} else {	389	} else {
108	die "open failed: $!\n";	390	die "open failed: $!\n";
109	}	391	}
110	};	392	};
111		393
		394	In addition to all the common open modes/flags (C<O_RDONLY>, C<O_WRONLY>,
		395	C<O_RDWR>, C<O_CREAT>, C<O_TRUNC>, C<O_EXCL> and C<O_APPEND>), the
		396	following POSIX and non-POSIX constants are available (missing ones on
		397	your system are, as usual, C<0>):
		398
		399	C<O_ASYNC>, C<O_DIRECT>, C<O_NOATIME>, C<O_CLOEXEC>, C<O_NOCTTY>, C<O_NOFOLLOW>,
		400	C<O_NONBLOCK>, C<O_EXEC>, C<O_SEARCH>, C<O_DIRECTORY>, C<O_DSYNC>,
		401	C<O_RSYNC>, C<O_SYNC>, C<O_PATH>, C<O_TMPFILE>, and C<O_TTY_INIT>.
		402
		403
112	=item aio_close $fh, $callback	404	=item aio_close $fh, $callback->($status)
113		405
114	Asynchronously close a file and call the callback with the result	406	Asynchronously close a file and call the callback with the result
115	code. I<WARNING:> although accepted, you should not pass in a perl	407	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		408
		409	Unfortunately, you can't do this to perl. Perl I<insists> very strongly on
		410	closing the file descriptor associated with the filehandle itself.
		411
		412	Therefore, C<aio_close> will not close the filehandle - instead it will
		413	use dup2 to overwrite the file descriptor with the write-end of a pipe
		414	(the pipe fd will be created on demand and will be cached).
		415
		416	Or in other words: the file descriptor will be closed, but it will not be
		417	free for reuse until the perl filehandle is closed.
		418
		419	=cut
		420
		421	=item aio_seek $fh, $offset, $whence, $callback->($offs)
		422
		423	Seeks the filehandle to the new C<$offset>, similarly to perl's
		424	C<sysseek>. The C<$whence> can use the traditional values (C<0> for
		425	C<IO::AIO::SEEK_SET>, C<1> for C<IO::AIO::SEEK_CUR> or C<2> for
		426	C<IO::AIO::SEEK_END>).
		427
		428	The resulting absolute offset will be passed to the callback, or C<-1> in
		429	case of an error.
		430
		431	In theory, the C<$whence> constants could be different than the
		432	corresponding values from L<Fcntl>, but perl guarantees they are the same,
		433	so don't panic.
		434
		435	As a GNU/Linux (and maybe Solaris) extension, also the constants
		436	C<IO::AIO::SEEK_DATA> and C<IO::AIO::SEEK_HOLE> are available, if they
		437	could be found. No guarantees about suitability for use in C<aio_seek> or
		438	Perl's C<sysseek> can be made though, although I would naively assume they
		439	"just work".
		440
120	=item aio_read $fh,$offset,$length, $data,$dataoffset,$callback	441	=item aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
121		442
122	=item aio_write $fh,$offset,$length, $data,$dataoffset,$callback	443	=item aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
123		444
124	Reads or writes C<length> bytes from the specified C<fh> and C<offset>	445	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	446	C<$offset> into the scalar given by C<$data> and offset C<$dataoffset>
126	callback without the actual number of bytes read (or -1 on error, just	447	and calls the callback without the actual number of bytes read (or -1 on
127	like the syscall).	448	error, just like the syscall).
128		449
		450	C<aio_read> will, like C<sysread>, shrink or grow the C<$data> scalar to
		451	offset plus the actual number of bytes read.
		452
		453	If C<$offset> is undefined, then the current file descriptor offset will
		454	be used (and updated), otherwise the file descriptor offset will not be
		455	changed by these calls.
		456
		457	If C<$length> is undefined in C<aio_write>, use the remaining length of
		458	C<$data>.
		459
		460	If C<$dataoffset> is less than zero, it will be counted from the end of
		461	C<$data>.
		462
		463	The C<$data> scalar I<MUST NOT> be modified in any way while the request
		464	is outstanding. Modifying it can result in segfaults or World War III (if
		465	the necessary/optional hardware is installed).
		466
129	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, strating at	467	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at
130	offset C<0> within the scalar:	468	offset C<0> within the scalar:
131		469
132	aio_read $fh, 7, 15, $buffer, 0, sub {	470	aio_read $fh, 7, 15, $buffer, 0, sub {
133	$_[0] >= 0 or die "read error: $!";	471	$_[0] > 0 or die "read error: $!";
134	print "read <$buffer>\n";	472	print "read $_[0] bytes: <$buffer>\n";
135	};	473	};
136		474
		475
		476	=item aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
		477
		478	Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts
		479	reading at byte offset C<$in_offset>, and starts writing at the current
		480	file offset of C<$out_fh>. Because of that, it is not safe to issue more
		481	than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each
		482	other. The same C<$in_fh> works fine though, as this function does not
		483	move or use the file offset of C<$in_fh>.
		484
		485	Please note that C<aio_sendfile> can read more bytes from C<$in_fh> than
		486	are written, and there is no way to find out how many more bytes have been
		487	read from C<aio_sendfile> alone, as C<aio_sendfile> only provides the
		488	number of bytes written to C<$out_fh>. Only if the result value equals
		489	C<$length> one can assume that C<$length> bytes have been read.
		490
		491	Unlike with other C<aio_> functions, it makes a lot of sense to use
		492	C<aio_sendfile> on non-blocking sockets, as long as one end (typically
		493	the C<$in_fh>) is a file - the file I/O will then be asynchronous, while
		494	the socket I/O will be non-blocking. Note, however, that you can run
		495	into a trap where C<aio_sendfile> reads some data with readahead, then
		496	fails to write all data, and when the socket is ready the next time, the
		497	data in the cache is already lost, forcing C<aio_sendfile> to again hit
		498	the disk. Explicit C<aio_read> + C<aio_write> let's you better control
		499	resource usage.
		500
		501	This call tries to make use of a native C<sendfile>-like syscall to
		502	provide zero-copy operation. For this to work, C<$out_fh> should refer to
		503	a socket, and C<$in_fh> should refer to an mmap'able file.
		504
		505	If a native sendfile cannot be found or it fails with C<ENOSYS>,
		506	C<EINVAL>, C<ENOTSUP>, C<EOPNOTSUPP>, C<EAFNOSUPPORT>, C<EPROTOTYPE> or
		507	C<ENOTSOCK>, it will be emulated, so you can call C<aio_sendfile> on any
		508	type of filehandle regardless of the limitations of the operating system.
		509
		510	As native sendfile syscalls (as practically any non-POSIX interface hacked
		511	together in a hurry to improve benchmark numbers) tend to be rather buggy
		512	on many systems, this implementation tries to work around some known bugs
		513	in Linux and FreeBSD kernels (probably others, too), but that might fail,
		514	so you really really should check the return value of C<aio_sendfile> -
		515	fewer bytes than expected might have been transferred.
		516
		517
137	=item aio_readahead $fh,$offset,$length, $callback	518	=item aio_readahead $fh,$offset,$length, $callback->($retval)
138		519
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<$!> is set to ENOSYS.
142
143	readahead() populates the page cache with data from a file so that	520	C<aio_readahead> populates the page cache with data from a file so that
144	subsequent reads from that file will not block on disk I/O. The C<$offset>	521	subsequent reads from that file will not block on disk I/O. The C<$offset>
145	argument specifies the starting point from which data is to be read and	522	argument specifies the starting point from which data is to be read and
146	C<$length> specifies the number of bytes to be read. I/O is performed in	523	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	524	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	525	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	526	(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.	527	file. The current file offset of the file is left unchanged.
151		528
		529	If that syscall doesn't exist (likely if your kernel isn't Linux) it will
		530	be emulated by simply reading the data, which would have a similar effect.
		531
		532
152	=item aio_stat $fh_or_path, $callback	533	=item aio_stat $fh_or_path, $callback->($status)
153		534
154	=item aio_lstat $fh, $callback	535	=item aio_lstat $fh, $callback->($status)
155		536
156	Works like perl's C<stat> or C<lstat> in void context. The callback will	537	Works like perl's C<stat> or C<lstat> in void context. The callback will
157	be called after the stat and the results will be available using C<stat _>	538	be called after the stat and the results will be available using C<stat _>
158	or C<-s _> etc...	539	or C<-s _> etc...
159		540
…		…
161	for an explanation.	542	for an explanation.
162		543
163	Currently, the stats are always 64-bit-stats, i.e. instead of returning an	544	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	545	error when stat'ing a large file, the results will be silently truncated
165	unless perl itself is compiled with large file support.	546	unless perl itself is compiled with large file support.
		547
		548	To help interpret the mode and dev/rdev stat values, IO::AIO offers the
		549	following constants and functions (if not implemented, the constants will
		550	be C<0> and the functions will either C<croak> or fall back on traditional
		551	behaviour).
		552
		553	C<S_IFMT>, C<S_IFIFO>, C<S_IFCHR>, C<S_IFBLK>, C<S_IFLNK>, C<S_IFREG>,
		554	C<S_IFDIR>, C<S_IFWHT>, C<S_IFSOCK>, C<IO::AIO::major $dev_t>,
		555	C<IO::AIO::minor $dev_t>, C<IO::AIO::makedev $major, $minor>.
166		556
167	Example: Print the length of F</etc/passwd>:	557	Example: Print the length of F</etc/passwd>:
168		558
169	aio_stat "/etc/passwd", sub {	559	aio_stat "/etc/passwd", sub {
170	$_[0] and die "stat failed: $!";	560	$_[0] and die "stat failed: $!";
171	print "size is ", -s _, "\n";	561	print "size is ", -s _, "\n";
172	};	562	};
173		563
		564
		565	=item aio_statvfs $fh_or_path, $callback->($statvfs)
		566
		567	Works like the POSIX C<statvfs> or C<fstatvfs> syscalls, depending on
		568	whether a file handle or path was passed.
		569
		570	On success, the callback is passed a hash reference with the following
		571	members: C<bsize>, C<frsize>, C<blocks>, C<bfree>, C<bavail>, C<files>,
		572	C<ffree>, C<favail>, C<fsid>, C<flag> and C<namemax>. On failure, C<undef>
		573	is passed.
		574
		575	The following POSIX IO::AIO::ST_* constants are defined: C<ST_RDONLY> and
		576	C<ST_NOSUID>.
		577
		578	The following non-POSIX IO::AIO::ST_* flag masks are defined to
		579	their correct value when available, or to C<0> on systems that do
		580	not support them: C<ST_NODEV>, C<ST_NOEXEC>, C<ST_SYNCHRONOUS>,
		581	C<ST_MANDLOCK>, C<ST_WRITE>, C<ST_APPEND>, C<ST_IMMUTABLE>, C<ST_NOATIME>,
		582	C<ST_NODIRATIME> and C<ST_RELATIME>.
		583
		584	Example: stat C</wd> and dump out the data if successful.
		585
		586	aio_statvfs "/wd", sub {
		587	my $f = $_[0]
		588	or die "statvfs: $!";
		589
		590	use Data::Dumper;
		591	say Dumper $f;
		592	};
		593
		594	# result:
		595	{
		596	bsize => 1024,
		597	bfree => 4333064312,
		598	blocks => 10253828096,
		599	files => 2050765568,
		600	flag => 4096,
		601	favail => 2042092649,
		602	bavail => 4333064312,
		603	ffree => 2042092649,
		604	namemax => 255,
		605	frsize => 1024,
		606	fsid => 1810
		607	}
		608
		609	Here is a (likely partial - send me updates!) list of fsid values used by
		610	Linux - it is safe to hardcode these when C<$^O> is C<linux>:
		611
		612	0x0000adf5 adfs
		613	0x0000adff affs
		614	0x5346414f afs
		615	0x09041934 anon-inode filesystem
		616	0x00000187 autofs
		617	0x42465331 befs
		618	0x1badface bfs
		619	0x42494e4d binfmt_misc
		620	0x9123683e btrfs
		621	0x0027e0eb cgroupfs
		622	0xff534d42 cifs
		623	0x73757245 coda
		624	0x012ff7b7 coh
		625	0x28cd3d45 cramfs
		626	0x453dcd28 cramfs-wend (wrong endianness)
		627	0x64626720 debugfs
		628	0x00001373 devfs
		629	0x00001cd1 devpts
		630	0x0000f15f ecryptfs
		631	0x00414a53 efs
		632	0x0000137d ext
		633	0x0000ef53 ext2/ext3/ext4
		634	0x0000ef51 ext2
		635	0xf2f52010 f2fs
		636	0x00004006 fat
		637	0x65735546 fuseblk
		638	0x65735543 fusectl
		639	0x0bad1dea futexfs
		640	0x01161970 gfs2
		641	0x47504653 gpfs
		642	0x00004244 hfs
		643	0xf995e849 hpfs
		644	0x00c0ffee hostfs
		645	0x958458f6 hugetlbfs
		646	0x2bad1dea inotifyfs
		647	0x00009660 isofs
		648	0x000072b6 jffs2
		649	0x3153464a jfs
		650	0x6b414653 k-afs
		651	0x0bd00bd0 lustre
		652	0x0000137f minix
		653	0x0000138f minix 30 char names
		654	0x00002468 minix v2
		655	0x00002478 minix v2 30 char names
		656	0x00004d5a minix v3
		657	0x19800202 mqueue
		658	0x00004d44 msdos
		659	0x0000564c novell
		660	0x00006969 nfs
		661	0x6e667364 nfsd
		662	0x00003434 nilfs
		663	0x5346544e ntfs
		664	0x00009fa1 openprom
		665	0x7461636F ocfs2
		666	0x00009fa0 proc
		667	0x6165676c pstorefs
		668	0x0000002f qnx4
		669	0x68191122 qnx6
		670	0x858458f6 ramfs
		671	0x52654973 reiserfs
		672	0x00007275 romfs
		673	0x67596969 rpc_pipefs
		674	0x73636673 securityfs
		675	0xf97cff8c selinux
		676	0x0000517b smb
		677	0x534f434b sockfs
		678	0x73717368 squashfs
		679	0x62656572 sysfs
		680	0x012ff7b6 sysv2
		681	0x012ff7b5 sysv4
		682	0x01021994 tmpfs
		683	0x15013346 udf
		684	0x00011954 ufs
		685	0x54190100 ufs byteswapped
		686	0x00009fa2 usbdevfs
		687	0x01021997 v9fs
		688	0xa501fcf5 vxfs
		689	0xabba1974 xenfs
		690	0x012ff7b4 xenix
		691	0x58465342 xfs
		692	0x012fd16d xia
		693
		694	=item aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
		695
		696	Works like perl's C<utime> function (including the special case of $atime
		697	and $mtime being undef). Fractional times are supported if the underlying
		698	syscalls support them.
		699
		700	When called with a pathname, uses utimes(2) if available, otherwise
		701	utime(2). If called on a file descriptor, uses futimes(2) if available,
		702	otherwise returns ENOSYS, so this is not portable.
		703
		704	Examples:
		705
		706	# set atime and mtime to current time (basically touch(1)):
		707	aio_utime "path", undef, undef;
		708	# set atime to current time and mtime to beginning of the epoch:
		709	aio_utime "path", time, undef; # undef==0
		710
		711
		712	=item aio_chown $fh_or_path, $uid, $gid, $callback->($status)
		713
		714	Works like perl's C<chown> function, except that C<undef> for either $uid
		715	or $gid is being interpreted as "do not change" (but -1 can also be used).
		716
		717	Examples:
		718
		719	# same as "chown root path" in the shell:
		720	aio_chown "path", 0, -1;
		721	# same as above:
		722	aio_chown "path", 0, undef;
		723
		724
		725	=item aio_truncate $fh_or_path, $offset, $callback->($status)
		726
		727	Works like truncate(2) or ftruncate(2).
		728
		729
		730	=item aio_allocate $fh, $mode, $offset, $len, $callback->($status)
		731
		732	Allocates or frees disk space according to the C<$mode> argument. See the
		733	linux C<fallocate> documentation for details.
		734
		735	C<$mode> is usually C<0> or C<IO::AIO::FALLOC_FL_KEEP_SIZE> to allocate
		736	space, or C<IO::AIO::FALLOC_FL_PUNCH_HOLE | IO::AIO::FALLOC_FL_KEEP_SIZE>,
		737	to deallocate a file range.
		738
		739	IO::AIO also supports C<FALLOC_FL_COLLAPSE_RANGE>, to remove a range
		740	(without leaving a hole) and C<FALLOC_FL_ZERO_RANGE>, to zero a range (see
		741	your L<fallocate(2)> manpage).
		742
		743	The file system block size used by C<fallocate> is presumably the
		744	C<f_bsize> returned by C<statvfs>.
		745
		746	If C<fallocate> isn't available or cannot be emulated (currently no
		747	emulation will be attempted), passes C<-1> and sets C<$!> to C<ENOSYS>.
		748
		749
		750	=item aio_chmod $fh_or_path, $mode, $callback->($status)
		751
		752	Works like perl's C<chmod> function.
		753
		754
174	=item aio_unlink $pathname, $callback	755	=item aio_unlink $pathname, $callback->($status)
175		756
176	Asynchronously unlink (delete) a file and call the callback with the	757	Asynchronously unlink (delete) a file and call the callback with the
177	result code.	758	result code.
178		759
		760
		761	=item aio_mknod $pathname, $mode, $dev, $callback->($status)
		762
		763	[EXPERIMENTAL]
		764
		765	Asynchronously create a device node (or fifo). See mknod(2).
		766
		767	The only (POSIX-) portable way of calling this function is:
		768
		769	aio_mknod $pathname, IO::AIO::S_IFIFO | $mode, 0, sub { ...
		770
		771	See C<aio_stat> for info about some potentially helpful extra constants
		772	and functions.
		773
		774	=item aio_link $srcpath, $dstpath, $callback->($status)
		775
		776	Asynchronously create a new link to the existing object at C<$srcpath> at
		777	the path C<$dstpath> and call the callback with the result code.
		778
		779
		780	=item aio_symlink $srcpath, $dstpath, $callback->($status)
		781
		782	Asynchronously create a new symbolic link to the existing object at C<$srcpath> at
		783	the path C<$dstpath> and call the callback with the result code.
		784
		785
		786	=item aio_readlink $pathname, $callback->($link)
		787
		788	Asynchronously read the symlink specified by C<$path> and pass it to
		789	the callback. If an error occurs, nothing or undef gets passed to the
		790	callback.
		791
		792
		793	=item aio_realpath $pathname, $callback->($path)
		794
		795	Asynchronously make the path absolute and resolve any symlinks in
		796	C<$path>. The resulting path only consists of directories (same as
		797	L<Cwd::realpath>).
		798
		799	This request can be used to get the absolute path of the current working
		800	directory by passing it a path of F<.> (a single dot).
		801
		802
		803	=item aio_rename $srcpath, $dstpath, $callback->($status)
		804
		805	Asynchronously rename the object at C<$srcpath> to C<$dstpath>, just as
		806	rename(2) and call the callback with the result code.
		807
		808	On systems that support the AIO::WD working directory abstraction
		809	natively, the case C<[$wd, "."]> as C<$srcpath> is specialcased - instead
		810	of failing, C<rename> is called on the absolute path of C<$wd>.
		811
		812
		813	=item aio_mkdir $pathname, $mode, $callback->($status)
		814
		815	Asynchronously mkdir (create) a directory and call the callback with
		816	the result code. C<$mode> will be modified by the umask at the time the
		817	request is executed, so do not change your umask.
		818
		819
		820	=item aio_rmdir $pathname, $callback->($status)
		821
		822	Asynchronously rmdir (delete) a directory and call the callback with the
		823	result code.
		824
		825	On systems that support the AIO::WD working directory abstraction
		826	natively, the case C<[$wd, "."]> is specialcased - instead of failing,
		827	C<rmdir> is called on the absolute path of C<$wd>.
		828
		829
		830	=item aio_readdir $pathname, $callback->($entries)
		831
		832	Unlike the POSIX call of the same name, C<aio_readdir> reads an entire
		833	directory (i.e. opendir + readdir + closedir). The entries will not be
		834	sorted, and will B<NOT> include the C<.> and C<..> entries.
		835
		836	The callback is passed a single argument which is either C<undef> or an
		837	array-ref with the filenames.
		838
		839
		840	=item aio_readdirx $pathname, $flags, $callback->($entries, $flags)
		841
		842	Quite similar to C<aio_readdir>, but the C<$flags> argument allows one to
		843	tune behaviour and output format. In case of an error, C<$entries> will be
		844	C<undef>.
		845
		846	The flags are a combination of the following constants, ORed together (the
		847	flags will also be passed to the callback, possibly modified):
		848
		849	=over 4
		850
		851	=item IO::AIO::READDIR_DENTS
		852
		853	When this flag is off, then the callback gets an arrayref consisting of
		854	names only (as with C<aio_readdir>), otherwise it gets an arrayref with
		855	C<[$name, $type, $inode]> arrayrefs, each describing a single directory
		856	entry in more detail.
		857
		858	C<$name> is the name of the entry.
		859
		860	C<$type> is one of the C<IO::AIO::DT_xxx> constants:
		861
		862	C<IO::AIO::DT_UNKNOWN>, C<IO::AIO::DT_FIFO>, C<IO::AIO::DT_CHR>, C<IO::AIO::DT_DIR>,
		863	C<IO::AIO::DT_BLK>, C<IO::AIO::DT_REG>, C<IO::AIO::DT_LNK>, C<IO::AIO::DT_SOCK>,
		864	C<IO::AIO::DT_WHT>.
		865
		866	C<IO::AIO::DT_UNKNOWN> means just that: readdir does not know. If you need to
		867	know, you have to run stat yourself. Also, for speed reasons, the C<$type>
		868	scalars are read-only: you can not modify them.
		869
		870	C<$inode> is the inode number (which might not be exact on systems with 64
		871	bit inode numbers and 32 bit perls). This field has unspecified content on
		872	systems that do not deliver the inode information.
		873
		874	=item IO::AIO::READDIR_DIRS_FIRST
		875
		876	When this flag is set, then the names will be returned in an order where
		877	likely directories come first, in optimal stat order. This is useful when
		878	you need to quickly find directories, or you want to find all directories
		879	while avoiding to stat() each entry.
		880
		881	If the system returns type information in readdir, then this is used
		882	to find directories directly. Otherwise, likely directories are names
		883	beginning with ".", or otherwise names with no dots, of which names with
		884	short names are tried first.
		885
		886	=item IO::AIO::READDIR_STAT_ORDER
		887
		888	When this flag is set, then the names will be returned in an order
		889	suitable for stat()'ing each one. That is, when you plan to stat()
		890	all files in the given directory, then the returned order will likely
		891	be fastest.
		892
		893	If both this flag and C<IO::AIO::READDIR_DIRS_FIRST> are specified, then
		894	the likely dirs come first, resulting in a less optimal stat order.
		895
		896	=item IO::AIO::READDIR_FOUND_UNKNOWN
		897
		898	This flag should not be set when calling C<aio_readdirx>. Instead, it
		899	is being set by C<aio_readdirx>, when any of the C<$type>'s found were
		900	C<IO::AIO::DT_UNKNOWN>. The absence of this flag therefore indicates that all
		901	C<$type>'s are known, which can be used to speed up some algorithms.
		902
		903	=back
		904
		905
		906	=item aio_load $pathname, $data, $callback->($status)
		907
		908	This is a composite request that tries to fully load the given file into
		909	memory. Status is the same as with aio_read.
		910
		911	=cut
		912
		913	sub aio_load($$;$) {
		914	my ($path, undef, $cb) = @_;
		915	my $data = \$_[1];
		916
		917	my $pri = aioreq_pri;
		918	my $grp = aio_group $cb;
		919
		920	aioreq_pri $pri;
		921	add $grp aio_open $path, O_RDONLY, 0, sub {
		922	my $fh = shift
		923	or return $grp->result (-1);
		924
		925	aioreq_pri $pri;
		926	add $grp aio_read $fh, 0, (-s $fh), $$data, 0, sub {
		927	$grp->result ($_[0]);
		928	};
		929	};
		930
		931	$grp
		932	}
		933
		934	=item aio_copy $srcpath, $dstpath, $callback->($status)
		935
		936	Try to copy the I<file> (directories not supported as either source or
		937	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		938	a status of C<0> (ok) or C<-1> (error, see C<$!>).
		939
		940	This is a composite request that creates the destination file with
		941	mode 0200 and copies the contents of the source file into it using
		942	C<aio_sendfile>, followed by restoring atime, mtime, access mode and
		943	uid/gid, in that order.
		944
		945	If an error occurs, the partial destination file will be unlinked, if
		946	possible, except when setting atime, mtime, access mode and uid/gid, where
		947	errors are being ignored.
		948
		949	=cut
		950
		951	sub aio_copy($$;$) {
		952	my ($src, $dst, $cb) = @_;
		953
		954	my $pri = aioreq_pri;
		955	my $grp = aio_group $cb;
		956
		957	aioreq_pri $pri;
		958	add $grp aio_open $src, O_RDONLY, 0, sub {
		959	if (my $src_fh = $_[0]) {
		960	my @stat = stat $src_fh; # hmm, might block over nfs?
		961
		962	aioreq_pri $pri;
		963	add $grp aio_open $dst, O_CREAT | O_WRONLY | O_TRUNC, 0200, sub {
		964	if (my $dst_fh = $_[0]) {
		965	aioreq_pri $pri;
		966	add $grp aio_sendfile $dst_fh, $src_fh, 0, $stat[7], sub {
		967	if ($_[0] == $stat[7]) {
		968	$grp->result (0);
		969	close $src_fh;
		970
		971	my $ch = sub {
		972	aioreq_pri $pri;
		973	add $grp aio_chmod $dst_fh, $stat[2] & 07777, sub {
		974	aioreq_pri $pri;
		975	add $grp aio_chown $dst_fh, $stat[4], $stat[5], sub {
		976	aioreq_pri $pri;
		977	add $grp aio_close $dst_fh;
		978	}
		979	};
		980	};
		981
		982	aioreq_pri $pri;
		983	add $grp aio_utime $dst_fh, $stat[8], $stat[9], sub {
		984	if ($_[0] < 0 && $! == ENOSYS) {
		985	aioreq_pri $pri;
		986	add $grp aio_utime $dst, $stat[8], $stat[9], $ch;
		987	} else {
		988	$ch->();
		989	}
		990	};
		991	} else {
		992	$grp->result (-1);
		993	close $src_fh;
		994	close $dst_fh;
		995
		996	aioreq $pri;
		997	add $grp aio_unlink $dst;
		998	}
		999	};
		1000	} else {
		1001	$grp->result (-1);
		1002	}
		1003	},
		1004
		1005	} else {
		1006	$grp->result (-1);
		1007	}
		1008	};
		1009
		1010	$grp
		1011	}
		1012
		1013	=item aio_move $srcpath, $dstpath, $callback->($status)
		1014
		1015	Try to move the I<file> (directories not supported as either source or
		1016	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		1017	a status of C<0> (ok) or C<-1> (error, see C<$!>).
		1018
		1019	This is a composite request that tries to rename(2) the file first; if
		1020	rename fails with C<EXDEV>, it copies the file with C<aio_copy> and, if
		1021	that is successful, unlinks the C<$srcpath>.
		1022
		1023	=cut
		1024
		1025	sub aio_move($$;$) {
		1026	my ($src, $dst, $cb) = @_;
		1027
		1028	my $pri = aioreq_pri;
		1029	my $grp = aio_group $cb;
		1030
		1031	aioreq_pri $pri;
		1032	add $grp aio_rename $src, $dst, sub {
		1033	if ($_[0] && $! == EXDEV) {
		1034	aioreq_pri $pri;
		1035	add $grp aio_copy $src, $dst, sub {
		1036	$grp->result ($_[0]);
		1037
		1038	unless ($_[0]) {
		1039	aioreq_pri $pri;
		1040	add $grp aio_unlink $src;
		1041	}
		1042	};
		1043	} else {
		1044	$grp->result ($_[0]);
		1045	}
		1046	};
		1047
		1048	$grp
		1049	}
		1050
		1051	=item aio_scandir $pathname, $maxreq, $callback->($dirs, $nondirs)
		1052
		1053	Scans a directory (similar to C<aio_readdir>) but additionally tries to
		1054	efficiently separate the entries of directory C<$path> into two sets of
		1055	names, directories you can recurse into (directories), and ones you cannot
		1056	recurse into (everything else, including symlinks to directories).
		1057
		1058	C<aio_scandir> is a composite request that creates of many sub requests_
		1059	C<$maxreq> specifies the maximum number of outstanding aio requests that
		1060	this function generates. If it is C<< <= 0 >>, then a suitable default
		1061	will be chosen (currently 4).
		1062
		1063	On error, the callback is called without arguments, otherwise it receives
		1064	two array-refs with path-relative entry names.
		1065
		1066	Example:
		1067
		1068	aio_scandir $dir, 0, sub {
		1069	my ($dirs, $nondirs) = @_;
		1070	print "real directories: @$dirs\n";
		1071	print "everything else: @$nondirs\n";
		1072	};
		1073
		1074	Implementation notes.
		1075
		1076	The C<aio_readdir> cannot be avoided, but C<stat()>'ing every entry can.
		1077
		1078	If readdir returns file type information, then this is used directly to
		1079	find directories.
		1080
		1081	Otherwise, after reading the directory, the modification time, size etc.
		1082	of the directory before and after the readdir is checked, and if they
		1083	match (and isn't the current time), the link count will be used to decide
		1084	how many entries are directories (if >= 2). Otherwise, no knowledge of the
		1085	number of subdirectories will be assumed.
		1086
		1087	Then entries will be sorted into likely directories a non-initial dot
		1088	currently) and likely non-directories (see C<aio_readdirx>). Then every
		1089	entry plus an appended C</.> will be C<stat>'ed, likely directories first,
		1090	in order of their inode numbers. If that succeeds, it assumes that the
		1091	entry is a directory or a symlink to directory (which will be checked
		1092	separately). This is often faster than stat'ing the entry itself because
		1093	filesystems might detect the type of the entry without reading the inode
		1094	data (e.g. ext2fs filetype feature), even on systems that cannot return
		1095	the filetype information on readdir.
		1096
		1097	If the known number of directories (link count - 2) has been reached, the
		1098	rest of the entries is assumed to be non-directories.
		1099
		1100	This only works with certainty on POSIX (= UNIX) filesystems, which
		1101	fortunately are the vast majority of filesystems around.
		1102
		1103	It will also likely work on non-POSIX filesystems with reduced efficiency
		1104	as those tend to return 0 or 1 as link counts, which disables the
		1105	directory counting heuristic.
		1106
		1107	=cut
		1108
		1109	sub aio_scandir($$;$) {
		1110	my ($path, $maxreq, $cb) = @_;
		1111
		1112	my $pri = aioreq_pri;
		1113
		1114	my $grp = aio_group $cb;
		1115
		1116	$maxreq = 4 if $maxreq <= 0;
		1117
		1118	# get a wd object
		1119	aioreq_pri $pri;
		1120	add $grp aio_wd $path, sub {
		1121	$_[0]
		1122	or return $grp->result ();
		1123
		1124	my $wd = [shift, "."];
		1125
		1126	# stat once
		1127	aioreq_pri $pri;
		1128	add $grp aio_stat $wd, sub {
		1129	return $grp->result () if $_[0];
		1130	my $now = time;
		1131	my $hash1 = join ":", (stat _)[0,1,3,7,9];
		1132
		1133	# read the directory entries
		1134	aioreq_pri $pri;
		1135	add $grp aio_readdirx $wd, READDIR_DIRS_FIRST, sub {
		1136	my $entries = shift
		1137	or return $grp->result ();
		1138
		1139	# stat the dir another time
		1140	aioreq_pri $pri;
		1141	add $grp aio_stat $wd, sub {
		1142	my $hash2 = join ":", (stat _)[0,1,3,7,9];
		1143
		1144	my $ndirs;
		1145
		1146	# take the slow route if anything looks fishy
		1147	if ($hash1 ne $hash2 or (stat _)[9] == $now) {
		1148	$ndirs = -1;
		1149	} else {
		1150	# if nlink == 2, we are finished
		1151	# for non-posix-fs's, we rely on nlink < 2
		1152	$ndirs = (stat _)[3] - 2
		1153	or return $grp->result ([], $entries);
		1154	}
		1155
		1156	my (@dirs, @nondirs);
		1157
		1158	my $statgrp = add $grp aio_group sub {
		1159	$grp->result (\@dirs, \@nondirs);
		1160	};
		1161
		1162	limit $statgrp $maxreq;
		1163	feed $statgrp sub {
		1164	return unless @$entries;
		1165	my $entry = shift @$entries;
		1166
		1167	aioreq_pri $pri;
		1168	$wd->[1] = "$entry/.";
		1169	add $statgrp aio_stat $wd, sub {
		1170	if ($_[0] < 0) {
		1171	push @nondirs, $entry;
		1172	} else {
		1173	# need to check for real directory
		1174	aioreq_pri $pri;
		1175	$wd->[1] = $entry;
		1176	add $statgrp aio_lstat $wd, sub {
		1177	if (-d _) {
		1178	push @dirs, $entry;
		1179
		1180	unless (--$ndirs) {
		1181	push @nondirs, @$entries;
		1182	feed $statgrp;
		1183	}
		1184	} else {
		1185	push @nondirs, $entry;
		1186	}
		1187	}
		1188	}
		1189	};
		1190	};
		1191	};
		1192	};
		1193	};
		1194	};
		1195
		1196	$grp
		1197	}
		1198
		1199	=item aio_rmtree $pathname, $callback->($status)
		1200
		1201	Delete a directory tree starting (and including) C<$path>, return the
		1202	status of the final C<rmdir> only. This is a composite request that
		1203	uses C<aio_scandir> to recurse into and rmdir directories, and unlink
		1204	everything else.
		1205
		1206	=cut
		1207
		1208	sub aio_rmtree;
		1209	sub aio_rmtree($;$) {
		1210	my ($path, $cb) = @_;
		1211
		1212	my $pri = aioreq_pri;
		1213	my $grp = aio_group $cb;
		1214
		1215	aioreq_pri $pri;
		1216	add $grp aio_scandir $path, 0, sub {
		1217	my ($dirs, $nondirs) = @_;
		1218
		1219	my $dirgrp = aio_group sub {
		1220	add $grp aio_rmdir $path, sub {
		1221	$grp->result ($_[0]);
		1222	};
		1223	};
		1224
		1225	(aioreq_pri $pri), add $dirgrp aio_rmtree "$path/$_" for @$dirs;
		1226	(aioreq_pri $pri), add $dirgrp aio_unlink "$path/$_" for @$nondirs;
		1227
		1228	add $grp $dirgrp;
		1229	};
		1230
		1231	$grp
		1232	}
		1233
		1234	=item aio_fcntl $fh, $cmd, $arg, $callback->($status)
		1235
		1236	=item aio_ioctl $fh, $request, $buf, $callback->($status)
		1237
		1238	These work just like the C<fcntl> and C<ioctl> built-in functions, except
		1239	they execute asynchronously and pass the return value to the callback.
		1240
		1241	Both calls can be used for a lot of things, some of which make more sense
		1242	to run asynchronously in their own thread, while some others make less
		1243	sense. For example, calls that block waiting for external events, such
		1244	as locking, will also lock down an I/O thread while it is waiting, which
		1245	can deadlock the whole I/O system. At the same time, there might be no
		1246	alternative to using a thread to wait.
		1247
		1248	So in general, you should only use these calls for things that do
		1249	(filesystem) I/O, not for things that wait for other events (network,
		1250	other processes), although if you are careful and know what you are doing,
		1251	you still can.
		1252
		1253	The following constants are available (missing ones are, as usual C<0>):
		1254
		1255	C<FIFREEZE>, C<FITHAW>, C<FITRIM>, C<FICLONE>, C<FICLONERANGE>, C<FIDEDUPERANGE>.
		1256
		1257	C<FS_IOC_GETFLAGS>, C<FS_IOC_SETFLAGS>, C<FS_IOC_GETVERSION>, C<FS_IOC_SETVERSION>,
		1258	C<FS_IOC_FIEMAP>.
		1259
		1260	C<FS_IOC_FSGETXATTR>, C<FS_IOC_FSSETXATTR>, C<FS_IOC_SET_ENCRYPTION_POLICY>,
		1261	C<FS_IOC_GET_ENCRYPTION_PWSALT>, C<FS_IOC_GET_ENCRYPTION_POLICY>, C<FS_KEY_DESCRIPTOR_SIZE>.
		1262
		1263	C<FS_SECRM_FL>, C<FS_UNRM_FL>, C<FS_COMPR_FL>, C<FS_SYNC_FL>, C<FS_IMMUTABLE_FL>,
		1264	C<FS_APPEND_FL>, C<FS_NODUMP_FL>, C<FS_NOATIME_FL>, C<FS_DIRTY_FL>,
		1265	C<FS_COMPRBLK_FL>, C<FS_NOCOMP_FL>, C<FS_ENCRYPT_FL>, C<FS_BTREE_FL>,
		1266	C<FS_INDEX_FL>, C<FS_JOURNAL_DATA_FL>, C<FS_NOTAIL_FL>, C<FS_DIRSYNC_FL>, C<FS_TOPDIR_FL>,
		1267	C<FS_FL_USER_MODIFIABLE>.
		1268
		1269	C<FS_XFLAG_REALTIME>, C<FS_XFLAG_PREALLOC>, C<FS_XFLAG_IMMUTABLE>, C<FS_XFLAG_APPEND>,
		1270	C<FS_XFLAG_SYNC>, C<FS_XFLAG_NOATIME>, C<FS_XFLAG_NODUMP>, C<FS_XFLAG_RTINHERIT>,
		1271	C<FS_XFLAG_PROJINHERIT>, C<FS_XFLAG_NOSYMLINKS>, C<FS_XFLAG_EXTSIZE>, C<FS_XFLAG_EXTSZINHERIT>,
		1272	C<FS_XFLAG_NODEFRAG>, C<FS_XFLAG_FILESTREAM>, C<FS_XFLAG_DAX>, C<FS_XFLAG_HASATTR>,
		1273
		1274	=item aio_sync $callback->($status)
		1275
		1276	Asynchronously call sync and call the callback when finished.
		1277
179	=item aio_fsync $fh, $callback	1278	=item aio_fsync $fh, $callback->($status)
180		1279
181	Asynchronously call fsync on the given filehandle and call the callback	1280	Asynchronously call fsync on the given filehandle and call the callback
182	with the fsync result code.	1281	with the fsync result code.
183		1282
184	=item aio_fdatasync $fh, $callback	1283	=item aio_fdatasync $fh, $callback->($status)
185		1284
186	Asynchronously call fdatasync on the given filehandle and call the	1285	Asynchronously call fdatasync on the given filehandle and call the
187	callback with the fdatasync result code.	1286	callback with the fdatasync result code.
188		1287
		1288	If this call isn't available because your OS lacks it or it couldn't be
		1289	detected, it will be emulated by calling C<fsync> instead.
		1290
		1291	=item aio_syncfs $fh, $callback->($status)
		1292
		1293	Asynchronously call the syncfs syscall to sync the filesystem associated
		1294	to the given filehandle and call the callback with the syncfs result
		1295	code. If syncfs is not available, calls sync(), but returns C<-1> and sets
		1296	errno to C<ENOSYS> nevertheless.
		1297
		1298	=item aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
		1299
		1300	Sync the data portion of the file specified by C<$offset> and C<$length>
		1301	to disk (but NOT the metadata), by calling the Linux-specific
		1302	sync_file_range call. If sync_file_range is not available or it returns
		1303	ENOSYS, then fdatasync or fsync is being substituted.
		1304
		1305	C<$flags> can be a combination of C<IO::AIO::SYNC_FILE_RANGE_WAIT_BEFORE>,
		1306	C<IO::AIO::SYNC_FILE_RANGE_WRITE> and
		1307	C<IO::AIO::SYNC_FILE_RANGE_WAIT_AFTER>: refer to the sync_file_range
		1308	manpage for details.
		1309
		1310	=item aio_pathsync $pathname, $callback->($status)
		1311
		1312	This request tries to open, fsync and close the given path. This is a
		1313	composite request intended to sync directories after directory operations
		1314	(E.g. rename). This might not work on all operating systems or have any
		1315	specific effect, but usually it makes sure that directory changes get
		1316	written to disc. It works for anything that can be opened for read-only,
		1317	not just directories.
		1318
		1319	Future versions of this function might fall back to other methods when
		1320	C<fsync> on the directory fails (such as calling C<sync>).
		1321
		1322	Passes C<0> when everything went ok, and C<-1> on error.
		1323
		1324	=cut
		1325
		1326	sub aio_pathsync($;$) {
		1327	my ($path, $cb) = @_;
		1328
		1329	my $pri = aioreq_pri;
		1330	my $grp = aio_group $cb;
		1331
		1332	aioreq_pri $pri;
		1333	add $grp aio_open $path, O_RDONLY, 0, sub {
		1334	my ($fh) = @_;
		1335	if ($fh) {
		1336	aioreq_pri $pri;
		1337	add $grp aio_fsync $fh, sub {
		1338	$grp->result ($_[0]);
		1339
		1340	aioreq_pri $pri;
		1341	add $grp aio_close $fh;
		1342	};
		1343	} else {
		1344	$grp->result (-1);
		1345	}
		1346	};
		1347
		1348	$grp
		1349	}
		1350
		1351	=item aio_msync $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		1352
		1353	This is a rather advanced IO::AIO call, which only works on mmap(2)ed
		1354	scalars (see the C<IO::AIO::mmap> function, although it also works on data
		1355	scalars managed by the L<Sys::Mmap> or L<Mmap> modules, note that the
		1356	scalar must only be modified in-place while an aio operation is pending on
		1357	it).
		1358
		1359	It calls the C<msync> function of your OS, if available, with the memory
		1360	area starting at C<$offset> in the string and ending C<$length> bytes
		1361	later. If C<$length> is negative, counts from the end, and if C<$length>
		1362	is C<undef>, then it goes till the end of the string. The flags can be
		1363	a combination of C<IO::AIO::MS_ASYNC>, C<IO::AIO::MS_INVALIDATE> and
		1364	C<IO::AIO::MS_SYNC>.
		1365
		1366	=item aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		1367
		1368	This is a rather advanced IO::AIO call, which works best on mmap(2)ed
		1369	scalars.
		1370
		1371	It touches (reads or writes) all memory pages in the specified
		1372	range inside the scalar. All caveats and parameters are the same
		1373	as for C<aio_msync>, above, except for flags, which must be either
		1374	C<0> (which reads all pages and ensures they are instantiated) or
		1375	C<IO::AIO::MT_MODIFY>, which modifies the memory pages (by reading and
		1376	writing an octet from it, which dirties the page).
		1377
		1378	=item aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status)
		1379
		1380	This is a rather advanced IO::AIO call, which works best on mmap(2)ed
		1381	scalars.
		1382
		1383	It reads in all the pages of the underlying storage into memory (if any)
		1384	and locks them, so they are not getting swapped/paged out or removed.
		1385
		1386	If C<$length> is undefined, then the scalar will be locked till the end.
		1387
		1388	On systems that do not implement C<mlock>, this function returns C<-1>
		1389	and sets errno to C<ENOSYS>.
		1390
		1391	Note that the corresponding C<munlock> is synchronous and is
		1392	documented under L<MISCELLANEOUS FUNCTIONS>.
		1393
		1394	Example: open a file, mmap and mlock it - both will be undone when
		1395	C<$data> gets destroyed.
		1396
		1397	open my $fh, "<", $path or die "$path: $!";
		1398	my $data;
		1399	IO::AIO::mmap $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh;
		1400	aio_mlock $data; # mlock in background
		1401
		1402	=item aio_mlockall $flags, $callback->($status)
		1403
		1404	Calls the C<mlockall> function with the given C<$flags> (a combination of
		1405	C<IO::AIO::MCL_CURRENT> and C<IO::AIO::MCL_FUTURE>).
		1406
		1407	On systems that do not implement C<mlockall>, this function returns C<-1>
		1408	and sets errno to C<ENOSYS>.
		1409
		1410	Note that the corresponding C<munlockall> is synchronous and is
		1411	documented under L<MISCELLANEOUS FUNCTIONS>.
		1412
		1413	Example: asynchronously lock all current and future pages into memory.
		1414
		1415	aio_mlockall IO::AIO::MCL_FUTURE;
		1416
		1417	=item aio_fiemap $fh, $start, $length, $flags, $count, $cb->(\@extents)
		1418
		1419	Queries the extents of the given file (by calling the Linux C<FIEMAP>
		1420	ioctl, see L<http://cvs.schmorp.de/IO-AIO/doc/fiemap.txt> for details). If
		1421	the ioctl is not available on your OS, then this request will fail with
		1422	C<ENOSYS>.
		1423
		1424	C<$start> is the starting offset to query extents for, C<$length> is the
		1425	size of the range to query - if it is C<undef>, then the whole file will
		1426	be queried.
		1427
		1428	C<$flags> is a combination of flags (C<IO::AIO::FIEMAP_FLAG_SYNC> or
		1429	C<IO::AIO::FIEMAP_FLAG_XATTR> - C<IO::AIO::FIEMAP_FLAGS_COMPAT> is also
		1430	exported), and is normally C<0> or C<IO::AIO::FIEMAP_FLAG_SYNC> to query
		1431	the data portion.
		1432
		1433	C<$count> is the maximum number of extent records to return. If it is
		1434	C<undef>, then IO::AIO queries all extents of the range. As a very special
		1435	case, if it is C<0>, then the callback receives the number of extents
		1436	instead of the extents themselves (which is unreliable, see below).
		1437
		1438	If an error occurs, the callback receives no arguments. The special
		1439	C<errno> value C<IO::AIO::EBADR> is available to test for flag errors.
		1440
		1441	Otherwise, the callback receives an array reference with extent
		1442	structures. Each extent structure is an array reference itself, with the
		1443	following members:
		1444
		1445	[$logical, $physical, $length, $flags]
		1446
		1447	Flags is any combination of the following flag values (typically either C<0>
		1448	or C<IO::AIO::FIEMAP_EXTENT_LAST> (1)):
		1449
		1450	C<IO::AIO::FIEMAP_EXTENT_LAST>, C<IO::AIO::FIEMAP_EXTENT_UNKNOWN>,
		1451	C<IO::AIO::FIEMAP_EXTENT_DELALLOC>, C<IO::AIO::FIEMAP_EXTENT_ENCODED>,
		1452	C<IO::AIO::FIEMAP_EXTENT_DATA_ENCRYPTED>, C<IO::AIO::FIEMAP_EXTENT_NOT_ALIGNED>,
		1453	C<IO::AIO::FIEMAP_EXTENT_DATA_INLINE>, C<IO::AIO::FIEMAP_EXTENT_DATA_TAIL>,
		1454	C<IO::AIO::FIEMAP_EXTENT_UNWRITTEN>, C<IO::AIO::FIEMAP_EXTENT_MERGED> or
		1455	C<IO::AIO::FIEMAP_EXTENT_SHARED>.
		1456
		1457	At the time of this writing (Linux 3.2), this requets is unreliable unless
		1458	C<$count> is C<undef>, as the kernel has all sorts of bugs preventing
		1459	it to return all extents of a range for files with large number of
		1460	extents. The code works around all these issues if C<$count> is undef.
		1461
		1462	=item aio_group $callback->(...)
		1463
		1464	This is a very special aio request: Instead of doing something, it is a
		1465	container for other aio requests, which is useful if you want to bundle
		1466	many requests into a single, composite, request with a definite callback
		1467	and the ability to cancel the whole request with its subrequests.
		1468
		1469	Returns an object of class L<IO::AIO::GRP>. See its documentation below
		1470	for more info.
		1471
		1472	Example:
		1473
		1474	my $grp = aio_group sub {
		1475	print "all stats done\n";
		1476	};
		1477
		1478	add $grp
		1479	(aio_stat ...),
		1480	(aio_stat ...),
		1481	...;
		1482
		1483	=item aio_nop $callback->()
		1484
		1485	This is a special request - it does nothing in itself and is only used for
		1486	side effects, such as when you want to add a dummy request to a group so
		1487	that finishing the requests in the group depends on executing the given
		1488	code.
		1489
		1490	While this request does nothing, it still goes through the execution
		1491	phase and still requires a worker thread. Thus, the callback will not
		1492	be executed immediately but only after other requests in the queue have
		1493	entered their execution phase. This can be used to measure request
		1494	latency.
		1495
		1496	=item IO::AIO::aio_busy $fractional_seconds, $callback->() *NOT EXPORTED*
		1497
		1498	Mainly used for debugging and benchmarking, this aio request puts one of
		1499	the request workers to sleep for the given time.
		1500
		1501	While it is theoretically handy to have simple I/O scheduling requests
		1502	like sleep and file handle readable/writable, the overhead this creates is
		1503	immense (it blocks a thread for a long time) so do not use this function
		1504	except to put your application under artificial I/O pressure.
		1505
189	=back	1506	=back
190		1507
		1508
		1509	=head2 IO::AIO::WD - multiple working directories
		1510
		1511	Your process only has one current working directory, which is used by all
		1512	threads. This makes it hard to use relative paths (some other component
		1513	could call C<chdir> at any time, and it is hard to control when the path
		1514	will be used by IO::AIO).
		1515
		1516	One solution for this is to always use absolute paths. This usually works,
		1517	but can be quite slow (the kernel has to walk the whole path on every
		1518	access), and can also be a hassle to implement.
		1519
		1520	Newer POSIX systems have a number of functions (openat, fdopendir,
		1521	futimensat and so on) that make it possible to specify working directories
		1522	per operation.
		1523
		1524	For portability, and because the clowns who "designed", or shall I write,
		1525	perpetrated this new interface were obviously half-drunk, this abstraction
		1526	cannot be perfect, though.
		1527
		1528	IO::AIO allows you to convert directory paths into a so-called IO::AIO::WD
		1529	object. This object stores the canonicalised, absolute version of the
		1530	path, and on systems that allow it, also a directory file descriptor.
		1531
		1532	Everywhere where a pathname is accepted by IO::AIO (e.g. in C<aio_stat>
		1533	or C<aio_unlink>), one can specify an array reference with an IO::AIO::WD
		1534	object and a pathname instead (or the IO::AIO::WD object alone, which
		1535	gets interpreted as C<[$wd, "."]>). If the pathname is absolute, the
		1536	IO::AIO::WD object is ignored, otherwise the pathname is resolved relative
		1537	to that IO::AIO::WD object.
		1538
		1539	For example, to get a wd object for F</etc> and then stat F<passwd>
		1540	inside, you would write:
		1541
		1542	aio_wd "/etc", sub {
		1543	my $etcdir = shift;
		1544
		1545	# although $etcdir can be undef on error, there is generally no reason
		1546	# to check for errors here, as aio_stat will fail with ENOENT
		1547	# when $etcdir is undef.
		1548
		1549	aio_stat [$etcdir, "passwd"], sub {
		1550	# yay
		1551	};
		1552	};
		1553
		1554	The fact that C<aio_wd> is a request and not a normal function shows that
		1555	creating an IO::AIO::WD object is itself a potentially blocking operation,
		1556	which is why it is done asynchronously.
		1557
		1558	To stat the directory obtained with C<aio_wd> above, one could write
		1559	either of the following three request calls:
		1560
		1561	aio_lstat "/etc" , sub { ... # pathname as normal string
		1562	aio_lstat [$wd, "."], sub { ... # "." relative to $wd (i.e. $wd itself)
		1563	aio_lstat $wd , sub { ... # shorthand for the previous
		1564
		1565	As with normal pathnames, IO::AIO keeps a copy of the working directory
		1566	object and the pathname string, so you could write the following without
		1567	causing any issues due to C<$path> getting reused:
		1568
		1569	my $path = [$wd, undef];
		1570
		1571	for my $name (qw(abc def ghi)) {
		1572	$path->[1] = $name;
		1573	aio_stat $path, sub {
		1574	# ...
		1575	};
		1576	}
		1577
		1578	There are some caveats: when directories get renamed (or deleted), the
		1579	pathname string doesn't change, so will point to the new directory (or
		1580	nowhere at all), while the directory fd, if available on the system,
		1581	will still point to the original directory. Most functions accepting a
		1582	pathname will use the directory fd on newer systems, and the string on
		1583	older systems. Some functions (such as realpath) will always rely on the
		1584	string form of the pathname.
		1585
		1586	So this functionality is mainly useful to get some protection against
		1587	C<chdir>, to easily get an absolute path out of a relative path for future
		1588	reference, and to speed up doing many operations in the same directory
		1589	(e.g. when stat'ing all files in a directory).
		1590
		1591	The following functions implement this working directory abstraction:
		1592
		1593	=over 4
		1594
		1595	=item aio_wd $pathname, $callback->($wd)
		1596
		1597	Asynchonously canonicalise the given pathname and convert it to an
		1598	IO::AIO::WD object representing it. If possible and supported on the
		1599	system, also open a directory fd to speed up pathname resolution relative
		1600	to this working directory.
		1601
		1602	If something goes wrong, then C<undef> is passwd to the callback instead
		1603	of a working directory object and C<$!> is set appropriately. Since
		1604	passing C<undef> as working directory component of a pathname fails the
		1605	request with C<ENOENT>, there is often no need for error checking in the
		1606	C<aio_wd> callback, as future requests using the value will fail in the
		1607	expected way.
		1608
		1609	=item IO::AIO::CWD
		1610
		1611	This is a compiletime constant (object) that represents the process
		1612	current working directory.
		1613
		1614	Specifying this object as working directory object for a pathname is as if
		1615	the pathname would be specified directly, without a directory object. For
		1616	example, these calls are functionally identical:
		1617
		1618	aio_stat "somefile", sub { ... };
		1619	aio_stat [IO::AIO::CWD, "somefile"], sub { ... };
		1620
		1621	=back
		1622
		1623	To recover the path associated with an IO::AIO::WD object, you can use
		1624	C<aio_realpath>:
		1625
		1626	aio_realpath $wd, sub {
		1627	warn "path is $_[0]\n";
		1628	};
		1629
		1630	Currently, C<aio_statvfs> always, and C<aio_rename> and C<aio_rmdir>
		1631	sometimes, fall back to using an absolue path.
		1632
		1633	=head2 IO::AIO::REQ CLASS
		1634
		1635	All non-aggregate C<aio_*> functions return an object of this class when
		1636	called in non-void context.
		1637
		1638	=over 4
		1639
		1640	=item cancel $req
		1641
		1642	Cancels the request, if possible. Has the effect of skipping execution
		1643	when entering the B<execute> state and skipping calling the callback when
		1644	entering the the B<result> state, but will leave the request otherwise
		1645	untouched (with the exception of readdir). That means that requests that
		1646	currently execute will not be stopped and resources held by the request
		1647	will not be freed prematurely.
		1648
		1649	=item cb $req $callback->(...)
		1650
		1651	Replace (or simply set) the callback registered to the request.
		1652
		1653	=back
		1654
		1655	=head2 IO::AIO::GRP CLASS
		1656
		1657	This class is a subclass of L<IO::AIO::REQ>, so all its methods apply to
		1658	objects of this class, too.
		1659
		1660	A IO::AIO::GRP object is a special request that can contain multiple other
		1661	aio requests.
		1662
		1663	You create one by calling the C<aio_group> constructing function with a
		1664	callback that will be called when all contained requests have entered the
		1665	C<done> state:
		1666
		1667	my $grp = aio_group sub {
		1668	print "all requests are done\n";
		1669	};
		1670
		1671	You add requests by calling the C<add> method with one or more
		1672	C<IO::AIO::REQ> objects:
		1673
		1674	$grp->add (aio_unlink "...");
		1675
		1676	add $grp aio_stat "...", sub {
		1677	$_[0] or return $grp->result ("error");
		1678
		1679	# add another request dynamically, if first succeeded
		1680	add $grp aio_open "...", sub {
		1681	$grp->result ("ok");
		1682	};
		1683	};
		1684
		1685	This makes it very easy to create composite requests (see the source of
		1686	C<aio_move> for an application) that work and feel like simple requests.
		1687
		1688	=over 4
		1689
		1690	=item * The IO::AIO::GRP objects will be cleaned up during calls to
		1691	C<IO::AIO::poll_cb>, just like any other request.
		1692
		1693	=item * They can be canceled like any other request. Canceling will cancel not
		1694	only the request itself, but also all requests it contains.
		1695
		1696	=item * They can also can also be added to other IO::AIO::GRP objects.
		1697
		1698	=item * You must not add requests to a group from within the group callback (or
		1699	any later time).
		1700
		1701	=back
		1702
		1703	Their lifetime, simplified, looks like this: when they are empty, they
		1704	will finish very quickly. If they contain only requests that are in the
		1705	C<done> state, they will also finish. Otherwise they will continue to
		1706	exist.
		1707
		1708	That means after creating a group you have some time to add requests
		1709	(precisely before the callback has been invoked, which is only done within
		1710	the C<poll_cb>). And in the callbacks of those requests, you can add
		1711	further requests to the group. And only when all those requests have
		1712	finished will the the group itself finish.
		1713
		1714	=over 4
		1715
		1716	=item add $grp ...
		1717
		1718	=item $grp->add (...)
		1719
		1720	Add one or more requests to the group. Any type of L<IO::AIO::REQ> can
		1721	be added, including other groups, as long as you do not create circular
		1722	dependencies.
		1723
		1724	Returns all its arguments.
		1725
		1726	=item $grp->cancel_subs
		1727
		1728	Cancel all subrequests and clears any feeder, but not the group request
		1729	itself. Useful when you queued a lot of events but got a result early.
		1730
		1731	The group request will finish normally (you cannot add requests to the
		1732	group).
		1733
		1734	=item $grp->result (...)
		1735
		1736	Set the result value(s) that will be passed to the group callback when all
		1737	subrequests have finished and set the groups errno to the current value
		1738	of errno (just like calling C<errno> without an error number). By default,
		1739	no argument will be passed and errno is zero.
		1740
		1741	=item $grp->errno ([$errno])
		1742
		1743	Sets the group errno value to C<$errno>, or the current value of errno
		1744	when the argument is missing.
		1745
		1746	Every aio request has an associated errno value that is restored when
		1747	the callback is invoked. This method lets you change this value from its
		1748	default (0).
		1749
		1750	Calling C<result> will also set errno, so make sure you either set C<$!>
		1751	before the call to C<result>, or call c<errno> after it.
		1752
		1753	=item feed $grp $callback->($grp)
		1754
		1755	Sets a feeder/generator on this group: every group can have an attached
		1756	generator that generates requests if idle. The idea behind this is that,
		1757	although you could just queue as many requests as you want in a group,
		1758	this might starve other requests for a potentially long time. For example,
		1759	C<aio_scandir> might generate hundreds of thousands of C<aio_stat>
		1760	requests, delaying any later requests for a long time.
		1761
		1762	To avoid this, and allow incremental generation of requests, you can
		1763	instead a group and set a feeder on it that generates those requests. The
		1764	feed callback will be called whenever there are few enough (see C<limit>,
		1765	below) requests active in the group itself and is expected to queue more
		1766	requests.
		1767
		1768	The feed callback can queue as many requests as it likes (i.e. C<add> does
		1769	not impose any limits).
		1770
		1771	If the feed does not queue more requests when called, it will be
		1772	automatically removed from the group.
		1773
		1774	If the feed limit is C<0> when this method is called, it will be set to
		1775	C<2> automatically.
		1776
		1777	Example:
		1778
		1779	# stat all files in @files, but only ever use four aio requests concurrently:
		1780
		1781	my $grp = aio_group sub { print "finished\n" };
		1782	limit $grp 4;
		1783	feed $grp sub {
		1784	my $file = pop @files
		1785	or return;
		1786
		1787	add $grp aio_stat $file, sub { ... };
		1788	};
		1789
		1790	=item limit $grp $num
		1791
		1792	Sets the feeder limit for the group: The feeder will be called whenever
		1793	the group contains less than this many requests.
		1794
		1795	Setting the limit to C<0> will pause the feeding process.
		1796
		1797	The default value for the limit is C<0>, but note that setting a feeder
		1798	automatically bumps it up to C<2>.
		1799
		1800	=back
		1801
191	=head2 SUPPORT FUNCTIONS	1802	=head2 SUPPORT FUNCTIONS
192		1803
		1804	=head3 EVENT PROCESSING AND EVENT LOOP INTEGRATION
		1805
193	=over 4	1806	=over 4
194		1807
195	=item $fileno = IO::AIO::poll_fileno	1808	=item $fileno = IO::AIO::poll_fileno
196		1809
197	Return the I<request result pipe filehandle>. This filehandle must be	1810	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	1811	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	1812	select and so on, see below or the SYNOPSIS). If the pipe becomes readable
200	C<poll_cb> to check the results.	1813	you have to call C<poll_cb> to check the results.
201		1814
202	See C<poll_cb> for an example.	1815	See C<poll_cb> for an example.
203		1816
204	=item IO::AIO::poll_cb	1817	=item IO::AIO::poll_cb
205		1818
206	Process all outstanding events on the result pipe. You have to call this	1819	Process some requests that have reached the result phase (i.e. they have
207	regularly. Returns the number of events processed. Returns immediately	1820	been executed but the results are not yet reported). You have to call
208	when no events are outstanding.	1821	this "regularly" to finish outstanding requests.
209		1822
210	You can use Event to multiplex, e.g.:	1823	Returns C<0> if all events could be processed (or there were no
		1824	events to process), or C<-1> if it returned earlier for whatever
		1825	reason. Returns immediately when no events are outstanding. The amount
		1826	of events processed depends on the settings of C<IO::AIO::max_poll_req>,
		1827	C<IO::AIO::max_poll_time> and C<IO::AIO::max_outstanding>.
		1828
		1829	If not all requests were processed for whatever reason, the poll file
		1830	descriptor will still be ready when C<poll_cb> returns, so normally you
		1831	don't have to do anything special to have it called later.
		1832
		1833	Apart from calling C<IO::AIO::poll_cb> when the event filehandle becomes
		1834	ready, it can be beneficial to call this function from loops which submit
		1835	a lot of requests, to make sure the results get processed when they become
		1836	available and not just when the loop is finished and the event loop takes
		1837	over again. This function returns very fast when there are no outstanding
		1838	requests.
		1839
		1840	Example: Install an Event watcher that automatically calls
		1841	IO::AIO::poll_cb with high priority (more examples can be found in the
		1842	SYNOPSIS section, at the top of this document):
211		1843
212	Event->io (fd => IO::AIO::poll_fileno,	1844	Event->io (fd => IO::AIO::poll_fileno,
213	poll => 'r', async => 1,	1845	poll => 'r', async => 1,
214	cb => \&IO::AIO::poll_cb);	1846	cb => \&IO::AIO::poll_cb);
215		1847
216	=item IO::AIO::poll_wait	1848	=item IO::AIO::poll_wait
217		1849
218	Wait till the result filehandle becomes ready for reading (simply does a	1850	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	1851	requests are outstanding anymore.
220	for some requests to finish).	1852
		1853	This is useful if you want to synchronously wait for some requests to
		1854	become ready, without actually handling them.
221		1855
222	See C<nreqs> for an example.	1856	See C<nreqs> for an example.
223		1857
		1858	=item IO::AIO::poll
		1859
		1860	Waits until some requests have been handled.
		1861
		1862	Returns the number of requests processed, but is otherwise strictly
		1863	equivalent to:
		1864
		1865	IO::AIO::poll_wait, IO::AIO::poll_cb
		1866
224	=item IO::AIO::nreqs	1867	=item IO::AIO::flush
225		1868
226	Returns the number of requests currently outstanding.	1869	Wait till all outstanding AIO requests have been handled.
227		1870
228	Example: wait till there are no outstanding requests anymore:	1871	Strictly equivalent to:
229		1872
230	IO::AIO::poll_wait, IO::AIO::poll_cb	1873	IO::AIO::poll_wait, IO::AIO::poll_cb
231	while IO::AIO::nreqs;	1874	while IO::AIO::nreqs;
232		1875
		1876	=item IO::AIO::max_poll_reqs $nreqs
		1877
		1878	=item IO::AIO::max_poll_time $seconds
		1879
		1880	These set the maximum number of requests (default C<0>, meaning infinity)
		1881	that are being processed by C<IO::AIO::poll_cb> in one call, respectively
		1882	the maximum amount of time (default C<0>, meaning infinity) spent in
		1883	C<IO::AIO::poll_cb> to process requests (more correctly the mininum amount
		1884	of time C<poll_cb> is allowed to use).
		1885
		1886	Setting C<max_poll_time> to a non-zero value creates an overhead of one
		1887	syscall per request processed, which is not normally a problem unless your
		1888	callbacks are really really fast or your OS is really really slow (I am
		1889	not mentioning Solaris here). Using C<max_poll_reqs> incurs no overhead.
		1890
		1891	Setting these is useful if you want to ensure some level of
		1892	interactiveness when perl is not fast enough to process all requests in
		1893	time.
		1894
		1895	For interactive programs, values such as C<0.01> to C<0.1> should be fine.
		1896
		1897	Example: Install an Event watcher that automatically calls
		1898	IO::AIO::poll_cb with low priority, to ensure that other parts of the
		1899	program get the CPU sometimes even under high AIO load.
		1900
		1901	# try not to spend much more than 0.1s in poll_cb
		1902	IO::AIO::max_poll_time 0.1;
		1903
		1904	# use a low priority so other tasks have priority
		1905	Event->io (fd => IO::AIO::poll_fileno,
		1906	poll => 'r', nice => 1,
		1907	cb => &IO::AIO::poll_cb);
		1908
		1909	=back
		1910
		1911	=head3 CONTROLLING THE NUMBER OF THREADS
		1912
		1913	=over
		1914
233	=item IO::AIO::min_parallel $nthreads	1915	=item IO::AIO::min_parallel $nthreads
234		1916
235	Set the minimum number of AIO threads to C<$nthreads>. The default is	1917	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	1918	default is C<8>, which means eight asynchronous operations can execute
237	(the number of outstanding operations, however, is unlimited).	1919	concurrently at any one time (the number of outstanding requests,
		1920	however, is unlimited).
238		1921
		1922	IO::AIO starts threads only on demand, when an AIO request is queued and
		1923	no free thread exists. Please note that queueing up a hundred requests can
		1924	create demand for a hundred threads, even if it turns out that everything
		1925	is in the cache and could have been processed faster by a single thread.
		1926
239	It is recommended to keep the number of threads low, as some Linux	1927	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	1928	Linux kernel versions will scale negatively with the number of threads
241	parallelity => MUCH higher latency). With current Linux 2.6 versions, 4-32	1929	(higher parallelity => MUCH higher latency). With current Linux 2.6
242	threads should be fine.	1930	versions, 4-32 threads should be fine.
243		1931
244	Under normal circumstances you don't need to call this function, as this	1932	Under most circumstances you don't need to call this function, as the
245	module automatically starts some threads (the exact number might change,	1933	module selects a default that is suitable for low to moderate load.
246	and is currently 4).
247		1934
248	=item IO::AIO::max_parallel $nthreads	1935	=item IO::AIO::max_parallel $nthreads
249		1936
250	Sets the maximum number of AIO threads to C<$nthreads>. If more than	1937	Sets the maximum number of AIO threads to C<$nthreads>. If more than the
251	the specified number of threads are currently running, kill them. This	1938	specified number of threads are currently running, this function kills
252	function blocks until the limit is reached.	1939	them. This function blocks until the limit is reached.
		1940
		1941	While C<$nthreads> are zero, aio requests get queued but not executed
		1942	until the number of threads has been increased again.
253		1943
254	This module automatically runs C<max_parallel 0> at program end, to ensure	1944	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.	1945	that all threads are killed and that there are no outstanding requests.
256		1946
257	Under normal circumstances you don't need to call this function.	1947	Under normal circumstances you don't need to call this function.
258		1948
		1949	=item IO::AIO::max_idle $nthreads
		1950
		1951	Limit the number of threads (default: 4) that are allowed to idle
		1952	(i.e., threads that did not get a request to process within the idle
		1953	timeout (default: 10 seconds). That means if a thread becomes idle while
		1954	C<$nthreads> other threads are also idle, it will free its resources and
		1955	exit.
		1956
		1957	This is useful when you allow a large number of threads (e.g. 100 or 1000)
		1958	to allow for extremely high load situations, but want to free resources
		1959	under normal circumstances (1000 threads can easily consume 30MB of RAM).
		1960
		1961	The default is probably ok in most situations, especially if thread
		1962	creation is fast. If thread creation is very slow on your system you might
		1963	want to use larger values.
		1964
		1965	=item IO::AIO::idle_timeout $seconds
		1966
		1967	Sets the minimum idle timeout (default 10) after which worker threads are
		1968	allowed to exit. SEe C<IO::AIO::max_idle>.
		1969
259	=item $oldnreqs = IO::AIO::max_outstanding $nreqs	1970	=item IO::AIO::max_outstanding $maxreqs
260		1971
261	Sets the maximum number of outstanding requests to C<$nreqs>. If you	1972	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	1973	you do queue up more than this number of requests, the next call to
263	some requests have been handled.	1974	C<IO::AIO::poll_cb> (and other functions calling C<poll_cb>, such as
		1975	C<IO::AIO::flush> or C<IO::AIO::poll>) will block until the limit is no
		1976	longer exceeded.
264		1977
265	The default is very large, so normally there is no practical limit. If you	1978	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	1979	used to make poll functions block if the limit is exceeded.
267	this to a relatively low number, such as C<100>.
268		1980
269	Under normal circumstances you don't need to call this function.	1981	This is a very bad function to use in interactive programs because it
		1982	blocks, and a bad way to reduce concurrency because it is inexact: Better
		1983	use an C<aio_group> together with a feed callback.
		1984
		1985	Its main use is in scripts without an event loop - when you want to stat
		1986	a lot of files, you can write somehting like this:
		1987
		1988	IO::AIO::max_outstanding 32;
		1989
		1990	for my $path (...) {
		1991	aio_stat $path , ...;
		1992	IO::AIO::poll_cb;
		1993	}
		1994
		1995	IO::AIO::flush;
		1996
		1997	The call to C<poll_cb> inside the loop will normally return instantly, but
		1998	as soon as more thna C<32> reqeusts are in-flight, it will block until
		1999	some requests have been handled. This keeps the loop from pushing a large
		2000	number of C<aio_stat> requests onto the queue.
		2001
		2002	The default value for C<max_outstanding> is very large, so there is no
		2003	practical limit on the number of outstanding requests.
270		2004
271	=back	2005	=back
272		2006
		2007	=head3 STATISTICAL INFORMATION
		2008
		2009	=over
		2010
		2011	=item IO::AIO::nreqs
		2012
		2013	Returns the number of requests currently in the ready, execute or pending
		2014	states (i.e. for which their callback has not been invoked yet).
		2015
		2016	Example: wait till there are no outstanding requests anymore:
		2017
		2018	IO::AIO::poll_wait, IO::AIO::poll_cb
		2019	while IO::AIO::nreqs;
		2020
		2021	=item IO::AIO::nready
		2022
		2023	Returns the number of requests currently in the ready state (not yet
		2024	executed).
		2025
		2026	=item IO::AIO::npending
		2027
		2028	Returns the number of requests currently in the pending state (executed,
		2029	but not yet processed by poll_cb).
		2030
		2031	=back
		2032
		2033	=head3 MISCELLANEOUS FUNCTIONS
		2034
		2035	IO::AIO implements some functions that are useful when you want to use
		2036	some "Advanced I/O" function not available to in Perl, without going the
		2037	"Asynchronous I/O" route. Many of these have an asynchronous C<aio_*>
		2038	counterpart.
		2039
		2040	=over 4
		2041
		2042	=item IO::AIO::sendfile $ofh, $ifh, $offset, $count
		2043
		2044	Calls the C<eio_sendfile_sync> function, which is like C<aio_sendfile>,
		2045	but is blocking (this makes most sense if you know the input data is
		2046	likely cached already and the output filehandle is set to non-blocking
		2047	operations).
		2048
		2049	Returns the number of bytes copied, or C<-1> on error.
		2050
		2051	=item IO::AIO::fadvise $fh, $offset, $len, $advice
		2052
		2053	Simply calls the C<posix_fadvise> function (see its
		2054	manpage for details). The following advice constants are
		2055	available: C<IO::AIO::FADV_NORMAL>, C<IO::AIO::FADV_SEQUENTIAL>,
		2056	C<IO::AIO::FADV_RANDOM>, C<IO::AIO::FADV_NOREUSE>,
		2057	C<IO::AIO::FADV_WILLNEED>, C<IO::AIO::FADV_DONTNEED>.
		2058
		2059	On systems that do not implement C<posix_fadvise>, this function returns
		2060	ENOSYS, otherwise the return value of C<posix_fadvise>.
		2061
		2062	=item IO::AIO::madvise $scalar, $offset, $len, $advice
		2063
		2064	Simply calls the C<posix_madvise> function (see its
		2065	manpage for details). The following advice constants are
		2066	available: C<IO::AIO::MADV_NORMAL>, C<IO::AIO::MADV_SEQUENTIAL>,
		2067	C<IO::AIO::MADV_RANDOM>, C<IO::AIO::MADV_WILLNEED>, C<IO::AIO::MADV_DONTNEED>,
		2068	C<IO::AIO::MADV_FREE>.
		2069
		2070	On systems that do not implement C<posix_madvise>, this function returns
		2071	ENOSYS, otherwise the return value of C<posix_madvise>.
		2072
		2073	=item IO::AIO::mprotect $scalar, $offset, $len, $protect
		2074
		2075	Simply calls the C<mprotect> function on the preferably AIO::mmap'ed
		2076	$scalar (see its manpage for details). The following protect
		2077	constants are available: C<IO::AIO::PROT_NONE>, C<IO::AIO::PROT_READ>,
		2078	C<IO::AIO::PROT_WRITE>, C<IO::AIO::PROT_EXEC>.
		2079
		2080	On systems that do not implement C<mprotect>, this function returns
		2081	ENOSYS, otherwise the return value of C<mprotect>.
		2082
		2083	=item IO::AIO::mmap $scalar, $length, $prot, $flags, $fh[, $offset]
		2084
		2085	Memory-maps a file (or anonymous memory range) and attaches it to the
		2086	given C<$scalar>, which will act like a string scalar. Returns true on
		2087	success, and false otherwise.
		2088
		2089	The only operations allowed on the scalar are C<substr>/C<vec> that don't
		2090	change the string length, and most read-only operations such as copying it
		2091	or searching it with regexes and so on.
		2092
		2093	Anything else is unsafe and will, at best, result in memory leaks.
		2094
		2095	The memory map associated with the C<$scalar> is automatically removed
		2096	when the C<$scalar> is destroyed, or when the C<IO::AIO::mmap> or
		2097	C<IO::AIO::munmap> functions are called.
		2098
		2099	This calls the C<mmap>(2) function internally. See your system's manual
		2100	page for details on the C<$length>, C<$prot> and C<$flags> parameters.
		2101
		2102	The C<$length> must be larger than zero and smaller than the actual
		2103	filesize.
		2104
		2105	C<$prot> is a combination of C<IO::AIO::PROT_NONE>, C<IO::AIO::PROT_EXEC>,
		2106	C<IO::AIO::PROT_READ> and/or C<IO::AIO::PROT_WRITE>,
		2107
		2108	C<$flags> can be a combination of
		2109	C<IO::AIO::MAP_SHARED> or
		2110	C<IO::AIO::MAP_PRIVATE>,
		2111	or a number of system-specific flags (when not available, the are C<0>):
		2112	C<IO::AIO::MAP_ANONYMOUS> (which is set to C<MAP_ANON> if your system only provides this constant),
		2113	C<IO::AIO::MAP_LOCKED>,
		2114	C<IO::AIO::MAP_NORESERVE>,
		2115	C<IO::AIO::MAP_POPULATE>,
		2116	C<IO::AIO::MAP_NONBLOCK>,
		2117	C<IO::AIO::MAP_FIXED>,
		2118	C<IO::AIO::MAP_GROWSDOWN>,
		2119	C<IO::AIO::MAP_32BIT>,
		2120	C<IO::AIO::MAP_HUGETLB> or
		2121	C<IO::AIO::MAP_STACK>.
		2122
		2123	If C<$fh> is C<undef>, then a file descriptor of C<-1> is passed.
		2124
		2125	C<$offset> is the offset from the start of the file - it generally must be
		2126	a multiple of C<IO::AIO::PAGESIZE> and defaults to C<0>.
		2127
		2128	Example:
		2129
		2130	use Digest::MD5;
		2131	use IO::AIO;
		2132
		2133	open my $fh, "<verybigfile"
		2134	or die "$!";
		2135
		2136	IO::AIO::mmap my $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh
		2137	or die "verybigfile: $!";
		2138
		2139	my $fast_md5 = md5 $data;
		2140
		2141	=item IO::AIO::munmap $scalar
		2142
		2143	Removes a previous mmap and undefines the C<$scalar>.
		2144
		2145	=item IO::AIO::munlock $scalar, $offset = 0, $length = undef
		2146
		2147	Calls the C<munlock> function, undoing the effects of a previous
		2148	C<aio_mlock> call (see its description for details).
		2149
		2150	=item IO::AIO::munlockall
		2151
		2152	Calls the C<munlockall> function.
		2153
		2154	On systems that do not implement C<munlockall>, this function returns
		2155	ENOSYS, otherwise the return value of C<munlockall>.
		2156
		2157	=item IO::AIO::splice $r_fh, $r_off, $w_fh, $w_off, $length, $flags
		2158
		2159	Calls the GNU/Linux C<splice(2)> syscall, if available. If C<$r_off> or
		2160	C<$w_off> are C<undef>, then C<NULL> is passed for these, otherwise they
		2161	should be the file offset.
		2162
		2163	C<$r_fh> and C<$w_fh> should not refer to the same file, as splice might
		2164	silently corrupt the data in this case.
		2165
		2166	The following symbol flag values are available: C<IO::AIO::SPLICE_F_MOVE>,
		2167	C<IO::AIO::SPLICE_F_NONBLOCK>, C<IO::AIO::SPLICE_F_MORE> and
		2168	C<IO::AIO::SPLICE_F_GIFT>.
		2169
		2170	See the C<splice(2)> manpage for details.
		2171
		2172	=item IO::AIO::tee $r_fh, $w_fh, $length, $flags
		2173
		2174	Calls the GNU/Linux C<tee(2)> syscall, see its manpage and the
		2175	description for C<IO::AIO::splice> above for details.
		2176
		2177	=item $actual_size = IO::AIO::pipesize $r_fh[, $new_size]
		2178
		2179	Attempts to query or change the pipe buffer size. Obviously works only
		2180	on pipes, and currently works only on GNU/Linux systems, and fails with
		2181	C<-1>/C<ENOSYS> everywhere else. If anybody knows how to influence pipe buffer
		2182	size on other systems, drop me a note.
		2183
		2184	=item ($rfh, $wfh) = IO::AIO::pipe2 [$flags]
		2185
		2186	This is a direct interface to the Linux L<pipe2(2)> system call. If
		2187	C<$flags> is missing or C<0>, then this should be the same as a call to
		2188	perl's built-in C<pipe> function and create a new pipe, and works on
		2189	systems that lack the pipe2 syscall. On win32, this case invokes C<_pipe
		2190	(..., 4096, O_BINARY)>.
		2191
		2192	If C<$flags> is non-zero, it tries to invoke the pipe2 system call with
		2193	the given flags (Linux 2.6.27, glibc 2.9).
		2194
		2195	On success, the read and write file handles are returned.
		2196
		2197	On error, nothing will be returned. If the pipe2 syscall is missing and
		2198	C<$flags> is non-zero, fails with C<ENOSYS>.
		2199
		2200	Please refer to L<pipe2(2)> for more info on the C<$flags>, but at the
		2201	time of this writing, C<IO::AIO::O_CLOEXEC>, C<IO::AIO::O_NONBLOCK> and
		2202	C<IO::AIO::O_DIRECT> (Linux 3.4, for packet-based pipes) were supported.
		2203
		2204	=back
		2205
273	=cut	2206	=cut
274		2207
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;	2208	min_parallel 8;
288		2209
289	END {	2210	END { flush }
290	max_parallel 0;
291	}
292		2211
293	1;	2212	1;
294		2213
		2214	=head1 EVENT LOOP INTEGRATION
		2215
		2216	It is recommended to use L<AnyEvent::AIO> to integrate IO::AIO
		2217	automatically into many event loops:
		2218
		2219	# AnyEvent integration (EV, Event, Glib, Tk, POE, urxvt, pureperl...)
		2220	use AnyEvent::AIO;
		2221
		2222	You can also integrate IO::AIO manually into many event loops, here are
		2223	some examples of how to do this:
		2224
		2225	# EV integration
		2226	my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
		2227
		2228	# Event integration
		2229	Event->io (fd => IO::AIO::poll_fileno,
		2230	poll => 'r',
		2231	cb => \&IO::AIO::poll_cb);
		2232
		2233	# Glib/Gtk2 integration
		2234	add_watch Glib::IO IO::AIO::poll_fileno,
		2235	in => sub { IO::AIO::poll_cb; 1 };
		2236
		2237	# Tk integration
		2238	Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",
		2239	readable => \&IO::AIO::poll_cb);
		2240
		2241	# Danga::Socket integration
		2242	Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
		2243	\&IO::AIO::poll_cb);
		2244
		2245	=head2 FORK BEHAVIOUR
		2246
		2247	Usage of pthreads in a program changes the semantics of fork
		2248	considerably. Specifically, only async-safe functions can be called after
		2249	fork. Perl doesn't know about this, so in general, you cannot call fork
		2250	with defined behaviour in perl if pthreads are involved. IO::AIO uses
		2251	pthreads, so this applies, but many other extensions and (for inexplicable
		2252	reasons) perl itself often is linked against pthreads, so this limitation
		2253	applies to quite a lot of perls.
		2254
		2255	This module no longer tries to fight your OS, or POSIX. That means IO::AIO
		2256	only works in the process that loaded it. Forking is fully supported, but
		2257	using IO::AIO in the child is not.
		2258
		2259	You might get around by not I<using> IO::AIO before (or after)
		2260	forking. You could also try to call the L<IO::AIO::reinit> function in the
		2261	child:
		2262
		2263	=over 4
		2264
		2265	=item IO::AIO::reinit
		2266
		2267	Abandons all current requests and I/O threads and simply reinitialises all
		2268	data structures. This is not an operation supported by any standards, but
		2269	happens to work on GNU/Linux and some newer BSD systems.
		2270
		2271	The only reasonable use for this function is to call it after forking, if
		2272	C<IO::AIO> was used in the parent. Calling it while IO::AIO is active in
		2273	the process will result in undefined behaviour. Calling it at any time
		2274	will also result in any undefined (by POSIX) behaviour.
		2275
		2276	=back
		2277
		2278	=head2 MEMORY USAGE
		2279
		2280	Per-request usage:
		2281
		2282	Each aio request uses - depending on your architecture - around 100-200
		2283	bytes of memory. In addition, stat requests need a stat buffer (possibly
		2284	a few hundred bytes), readdir requires a result buffer and so on. Perl
		2285	scalars and other data passed into aio requests will also be locked and
		2286	will consume memory till the request has entered the done state.
		2287
		2288	This is not awfully much, so queuing lots of requests is not usually a
		2289	problem.
		2290
		2291	Per-thread usage:
		2292
		2293	In the execution phase, some aio requests require more memory for
		2294	temporary buffers, and each thread requires a stack and other data
		2295	structures (usually around 16k-128k, depending on the OS).
		2296
		2297	=head1 KNOWN BUGS
		2298
		2299	Known bugs will be fixed in the next release.
		2300
295	=head1 SEE ALSO	2301	=head1 SEE ALSO
296		2302
297	L<Coro>, L<Linux::AIO>.	2303	L<AnyEvent::AIO> for easy integration into event loops, L<Coro::AIO> for a
		2304	more natural syntax.
298		2305
299	=head1 AUTHOR	2306	=head1 AUTHOR
300		2307
301	Marc Lehmann <schmorp@schmorp.de>	2308	Marc Lehmann <schmorp@schmorp.de>
302	http://home.schmorp.de/	2309	http://home.schmorp.de/

Diff Legend

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