ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/IO-AIO/AIO.pm

Comparing IO-AIO/AIO.pm (file contents):
Revision 1.59 by root, Sun Oct 22 10:33:26 2006 UTC vs.
Revision 1.275 by root, Fri Sep 22 05:20:39 2017 UTC

1	=head1 NAME	1	=head1 NAME
2		2
3	IO::AIO - Asynchronous Input/Output	3	IO::AIO - Asynchronous/Advanced Input/Output
4		4
5	=head1 SYNOPSIS	5	=head1 SYNOPSIS
6		6
7	use IO::AIO;	7	use IO::AIO;
8		8
9	aio_open "/etc/passwd", O_RDONLY, 0, sub {	9	aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub {
10	my ($fh) = @_;	10	my $fh = shift
		11	or die "/etc/passwd: $!";
11	...	12	...
12	};	13	};
13		14
14	aio_unlink "/tmp/file", sub { };	15	aio_unlink "/tmp/file", sub { };
15		16
…		…
18	};	19	};
19		20
20	# version 2+ has request and group objects	21	# version 2+ has request and group objects
21	use IO::AIO 2;	22	use IO::AIO 2;
22		23
		24	aioreq_pri 4; # give next request a very high priority
23	my $req = aio_unlink "/tmp/file", sub { };	25	my $req = aio_unlink "/tmp/file", sub { };
24	$req->cancel; # cancel request if still in queue	26	$req->cancel; # cancel request if still in queue
25		27
26	my $grp = aio_group sub { print "all stats done\n" };	28	my $grp = aio_group sub { print "all stats done\n" };
27	add $grp aio_stat "..." for ...;	29	add $grp aio_stat "..." for ...;
28		30
29	# AnyEvent integration
30	open my $fh, "<&=" . IO::AIO::poll_fileno or die "$!";
31	my $w = AnyEvent->io (fh => $fh, poll => 'r', cb => sub { IO::AIO::poll_cb });
32
33	# Event integration
34	Event->io (fd => IO::AIO::poll_fileno,
35	poll => 'r',
36	cb => \&IO::AIO::poll_cb);
37
38	# Glib/Gtk2 integration
39	add_watch Glib::IO IO::AIO::poll_fileno,
40	in => sub { IO::AIO::poll_cb; 1 };
41
42	# Tk integration
43	Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",
44	readable => \&IO::AIO::poll_cb);
45
46	# Danga::Socket integration
47	Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
48	\&IO::AIO::poll_cb);
49
50	=head1 DESCRIPTION	31	=head1 DESCRIPTION
51		32
52	This module implements asynchronous I/O using whatever means your	33	This module implements asynchronous I/O using whatever means your
53	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>).
54		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
55	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
56	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
57	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
58	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
59	aio functions available on many operating systems. However, they are often	57	functions available on many operating systems. However, they are often
60	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
61	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
62	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.
63		62
		63	In addition to asynchronous I/O, this module also exports some rather
		64	arcane interfaces, such as C<madvise> or linux's C<splice> system call,
		65	which is why the C<A> in C<AIO> can also mean I<advanced>.
		66
64	Although the module will work with in the presence of other threads, it is	67	Although the module will work in the presence of other (Perl-) threads,
65	currently not reentrant, so use appropriate locking yourself, always call	68	it is currently not reentrant in any way, so use appropriate locking
66	C<poll_cb> from within the same thread, or never call C<poll_cb> (or other	69	yourself, always call C<poll_cb> from within the same thread, or never
67	C<aio_> functions) recursively.	70	call C<poll_cb> (or other C<aio_> functions) recursively.
		71
		72	=head2 EXAMPLE
		73
		74	This is a simple example that uses the EV module and loads
		75	F</etc/passwd> asynchronously:
		76
		77	use EV;
		78	use IO::AIO;
		79
		80	# register the IO::AIO callback with EV
		81	my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
		82
		83	# queue the request to open /etc/passwd
		84	aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub {
		85	my $fh = shift
		86	or die "error while opening: $!";
		87
		88	# stat'ing filehandles is generally non-blocking
		89	my $size = -s $fh;
		90
		91	# queue a request to read the file
		92	my $contents;
		93	aio_read $fh, 0, $size, $contents, 0, sub {
		94	$_[0] == $size
		95	or die "short read: $!";
		96
		97	close $fh;
		98
		99	# file contents now in $contents
		100	print $contents;
		101
		102	# exit event loop and program
		103	EV::break;
		104	};
		105	};
		106
		107	# possibly queue up other requests, or open GUI windows,
		108	# check for sockets etc. etc.
		109
		110	# process events as long as there are some:
		111	EV::run;
		112
		113	=head1 REQUEST ANATOMY AND LIFETIME
		114
		115	Every C<aio_*> function creates a request. which is a C data structure not
		116	directly visible to Perl.
		117
		118	If called in non-void context, every request function returns a Perl
		119	object representing the request. In void context, nothing is returned,
		120	which saves a bit of memory.
		121
		122	The perl object is a fairly standard ref-to-hash object. The hash contents
		123	are not used by IO::AIO so you are free to store anything you like in it.
		124
		125	During their existance, aio requests travel through the following states,
		126	in order:
		127
		128	=over 4
		129
		130	=item ready
		131
		132	Immediately after a request is created it is put into the ready state,
		133	waiting for a thread to execute it.
		134
		135	=item execute
		136
		137	A thread has accepted the request for processing and is currently
		138	executing it (e.g. blocking in read).
		139
		140	=item pending
		141
		142	The request has been executed and is waiting for result processing.
		143
		144	While request submission and execution is fully asynchronous, result
		145	processing is not and relies on the perl interpreter calling C<poll_cb>
		146	(or another function with the same effect).
		147
		148	=item result
		149
		150	The request results are processed synchronously by C<poll_cb>.
		151
		152	The C<poll_cb> function will process all outstanding aio requests by
		153	calling their callbacks, freeing memory associated with them and managing
		154	any groups they are contained in.
		155
		156	=item done
		157
		158	Request has reached the end of its lifetime and holds no resources anymore
		159	(except possibly for the Perl object, but its connection to the actual
		160	aio request is severed and calling its methods will either do nothing or
		161	result in a runtime error).
		162
		163	=back
68		164
69	=cut	165	=cut
70		166
71	package IO::AIO;	167	package IO::AIO;
72		168
73	no warnings;	169	use Carp ();
74	use strict 'vars';	170
		171	use common::sense;
75		172
76	use base 'Exporter';	173	use base 'Exporter';
77		174
78	BEGIN {	175	BEGIN {
79	our $VERSION = '2.0';	176	our $VERSION = 4.35;
80		177
81	our @EXPORT = qw(aio_sendfile aio_read aio_write aio_open aio_close aio_stat	178	our @AIO_REQ = qw(aio_sendfile aio_seek aio_read aio_write aio_open aio_close
82	aio_lstat aio_unlink aio_rmdir aio_readdir aio_scandir aio_symlink	179	aio_stat aio_lstat aio_unlink aio_rmdir aio_readdir aio_readdirx
83	aio_fsync aio_fdatasync aio_readahead aio_rename aio_link aio_move	180	aio_scandir aio_symlink aio_readlink aio_realpath aio_fcntl aio_ioctl
		181	aio_sync aio_fsync aio_syncfs aio_fdatasync aio_sync_file_range
		182	aio_pathsync aio_readahead aio_fiemap aio_allocate
		183	aio_rename aio_rename2 aio_link aio_move aio_copy aio_group
		184	aio_nop aio_mknod aio_load aio_rmtree aio_mkdir aio_chown
		185	aio_chmod aio_utime aio_truncate
		186	aio_msync aio_mtouch aio_mlock aio_mlockall
		187	aio_statvfs
84	aio_group);	188	aio_wd);
85	our @EXPORT_OK = qw(poll_fileno poll_cb min_parallel max_parallel max_outstanding nreqs);	189
		190	our @EXPORT = (@AIO_REQ, qw(aioreq_pri aioreq_nice));
		191	our @EXPORT_OK = qw(poll_fileno poll_cb poll_wait flush
		192	min_parallel max_parallel max_idle idle_timeout
		193	nreqs nready npending nthreads
		194	max_poll_time max_poll_reqs
		195	sendfile fadvise madvise
		196	mmap munmap munlock munlockall);
		197
		198	push @AIO_REQ, qw(aio_busy); # not exported
86		199
87	@IO::AIO::GRP::ISA = 'IO::AIO::REQ';	200	@IO::AIO::GRP::ISA = 'IO::AIO::REQ';
88		201
89	require XSLoader;	202	require XSLoader;
90	XSLoader::load ("IO::AIO", $VERSION);	203	XSLoader::load ("IO::AIO", $VERSION);
91	}	204	}
92		205
93	=head1 FUNCTIONS	206	=head1 FUNCTIONS
94		207
95	=head2 AIO FUNCTIONS	208	=head2 QUICK OVERVIEW
		209
		210	This section simply lists the prototypes most of the functions for
		211	quick reference. See the following sections for function-by-function
		212	documentation.
		213
		214	aio_wd $pathname, $callback->($wd)
		215	aio_open $pathname, $flags, $mode, $callback->($fh)
		216	aio_close $fh, $callback->($status)
		217	aio_seek $fh,$offset,$whence, $callback->($offs)
		218	aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
		219	aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
		220	aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
		221	aio_readahead $fh,$offset,$length, $callback->($retval)
		222	aio_stat $fh_or_path, $callback->($status)
		223	aio_lstat $fh, $callback->($status)
		224	aio_statvfs $fh_or_path, $callback->($statvfs)
		225	aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
		226	aio_chown $fh_or_path, $uid, $gid, $callback->($status)
		227	aio_chmod $fh_or_path, $mode, $callback->($status)
		228	aio_truncate $fh_or_path, $offset, $callback->($status)
		229	aio_allocate $fh, $mode, $offset, $len, $callback->($status)
		230	aio_fiemap $fh, $start, $length, $flags, $count, $cb->(\@extents)
		231	aio_unlink $pathname, $callback->($status)
		232	aio_mknod $pathname, $mode, $dev, $callback->($status)
		233	aio_link $srcpath, $dstpath, $callback->($status)
		234	aio_symlink $srcpath, $dstpath, $callback->($status)
		235	aio_readlink $pathname, $callback->($link)
		236	aio_realpath $pathname, $callback->($path)
		237	aio_rename $srcpath, $dstpath, $callback->($status)
		238	aio_rename2 $srcpath, $dstpath, $flags, $callback->($status)
		239	aio_mkdir $pathname, $mode, $callback->($status)
		240	aio_rmdir $pathname, $callback->($status)
		241	aio_readdir $pathname, $callback->($entries)
		242	aio_readdirx $pathname, $flags, $callback->($entries, $flags)
		243	IO::AIO::READDIR_DENTS IO::AIO::READDIR_DIRS_FIRST
		244	IO::AIO::READDIR_STAT_ORDER IO::AIO::READDIR_FOUND_UNKNOWN
		245	aio_scandir $pathname, $maxreq, $callback->($dirs, $nondirs)
		246	aio_load $pathname, $data, $callback->($status)
		247	aio_copy $srcpath, $dstpath, $callback->($status)
		248	aio_move $srcpath, $dstpath, $callback->($status)
		249	aio_rmtree $pathname, $callback->($status)
		250	aio_fcntl $fh, $cmd, $arg, $callback->($status)
		251	aio_ioctl $fh, $request, $buf, $callback->($status)
		252	aio_sync $callback->($status)
		253	aio_syncfs $fh, $callback->($status)
		254	aio_fsync $fh, $callback->($status)
		255	aio_fdatasync $fh, $callback->($status)
		256	aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
		257	aio_pathsync $pathname, $callback->($status)
		258	aio_msync $scalar, $offset = 0, $length = undef, flags = MS_SYNC, $callback->($status)
		259	aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		260	aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status)
		261	aio_mlockall $flags, $callback->($status)
		262	aio_group $callback->(...)
		263	aio_nop $callback->()
		264
		265	$prev_pri = aioreq_pri [$pri]
		266	aioreq_nice $pri_adjust
		267
		268	IO::AIO::poll_wait
		269	IO::AIO::poll_cb
		270	IO::AIO::poll
		271	IO::AIO::flush
		272	IO::AIO::max_poll_reqs $nreqs
		273	IO::AIO::max_poll_time $seconds
		274	IO::AIO::min_parallel $nthreads
		275	IO::AIO::max_parallel $nthreads
		276	IO::AIO::max_idle $nthreads
		277	IO::AIO::idle_timeout $seconds
		278	IO::AIO::max_outstanding $maxreqs
		279	IO::AIO::nreqs
		280	IO::AIO::nready
		281	IO::AIO::npending
		282	IO::AIO::min_fdlimit $nfd;
		283
		284	IO::AIO::sendfile $ofh, $ifh, $offset, $count
		285	IO::AIO::fadvise $fh, $offset, $len, $advice
		286	IO::AIO::mmap $scalar, $length, $prot, $flags[, $fh[, $offset]]
		287	IO::AIO::munmap $scalar
		288	IO::AIO::madvise $scalar, $offset, $length, $advice
		289	IO::AIO::mprotect $scalar, $offset, $length, $protect
		290	IO::AIO::munlock $scalar, $offset = 0, $length = undef
		291	IO::AIO::munlockall
		292
		293	=head2 API NOTES
96		294
97	All the C<aio_*> calls are more or less thin wrappers around the syscall	295	All the C<aio_*> calls are more or less thin wrappers around the syscall
98	with the same name (sans C<aio_>). The arguments are similar or identical,	296	with the same name (sans C<aio_>). The arguments are similar or identical,
99	and they all accept an additional (and optional) C<$callback> argument	297	and they all accept an additional (and optional) C<$callback> argument
100	which must be a code reference. This code reference will get called with	298	which must be a code reference. This code reference will be called after
101	the syscall return code (e.g. most syscalls return C<-1> on error, unlike	299	the syscall has been executed in an asynchronous fashion. The results
102	perl, which usually delivers "false") as it's sole argument when the given	300	of the request will be passed as arguments to the callback (and, if an
103	syscall has been executed asynchronously.	301	error occured, in C<$!>) - for most requests the syscall return code (e.g.
		302	most syscalls return C<-1> on error, unlike perl, which usually delivers
		303	"false").
		304
		305	Some requests (such as C<aio_readdir>) pass the actual results and
		306	communicate failures by passing C<undef>.
104		307
105	All functions expecting a filehandle keep a copy of the filehandle	308	All functions expecting a filehandle keep a copy of the filehandle
106	internally until the request has finished.	309	internally until the request has finished.
107		310
108	All requests return objects of type L<IO::AIO::REQ> that allow further	311	All functions return request objects of type L<IO::AIO::REQ> that allow
109	manipulation of those requests while they are in-flight.	312	further manipulation of those requests while they are in-flight.
110		313
111	The pathnames you pass to these routines I<must> be absolute and	314	The pathnames you pass to these routines I<should> be absolute. The
112	encoded in byte form. The reason for the former is that at the time the	315	reason for this is that at the time the request is being executed, the
113	request is being executed, the current working directory could have	316	current working directory could have changed. Alternatively, you can
114	changed. Alternatively, you can make sure that you never change the	317	make sure that you never change the current working directory anywhere
115	current working directory.	318	in the program and then use relative paths. You can also take advantage
		319	of IO::AIOs working directory abstraction, that lets you specify paths
		320	relative to some previously-opened "working directory object" - see the
		321	description of the C<IO::AIO::WD> class later in this document.
116		322
117	To encode pathnames to byte form, either make sure you either: a)	323	To encode pathnames as octets, either make sure you either: a) always pass
118	always pass in filenames you got from outside (command line, readdir	324	in filenames you got from outside (command line, readdir etc.) without
119	etc.), b) are ASCII or ISO 8859-1, c) use the Encode module and encode	325	tinkering, b) are in your native filesystem encoding, c) use the Encode
120	your pathnames to the locale (or other) encoding in effect in the user	326	module and encode your pathnames to the locale (or other) encoding in
121	environment, d) use Glib::filename_from_unicode on unicode filenames or e)	327	effect in the user environment, d) use Glib::filename_from_unicode on
122	use something else.	328	unicode filenames or e) use something else to ensure your scalar has the
		329	correct contents.
		330
		331	This works, btw. independent of the internal UTF-8 bit, which IO::AIO
		332	handles correctly whether it is set or not.
		333
		334	=head2 AIO REQUEST FUNCTIONS
123		335
124	=over 4	336	=over 4
125		337
		338	=item $prev_pri = aioreq_pri [$pri]
		339
		340	Returns the priority value that would be used for the next request and, if
		341	C<$pri> is given, sets the priority for the next aio request.
		342
		343	The default priority is C<0>, the minimum and maximum priorities are C<-4>
		344	and C<4>, respectively. Requests with higher priority will be serviced
		345	first.
		346
		347	The priority will be reset to C<0> after each call to one of the C<aio_*>
		348	functions.
		349
		350	Example: open a file with low priority, then read something from it with
		351	higher priority so the read request is serviced before other low priority
		352	open requests (potentially spamming the cache):
		353
		354	aioreq_pri -3;
		355	aio_open ..., sub {
		356	return unless $_[0];
		357
		358	aioreq_pri -2;
		359	aio_read $_[0], ..., sub {
		360	...
		361	};
		362	};
		363
		364
		365	=item aioreq_nice $pri_adjust
		366
		367	Similar to C<aioreq_pri>, but subtracts the given value from the current
		368	priority, so the effect is cumulative.
		369
		370
126	=item aio_open $pathname, $flags, $mode, $callback->($fh)	371	=item aio_open $pathname, $flags, $mode, $callback->($fh)
127		372
128	Asynchronously open or create a file and call the callback with a newly	373	Asynchronously open or create a file and call the callback with a newly
129	created filehandle for the file.	374	created filehandle for the file (or C<undef> in case of an error).
130		375
131	The pathname passed to C<aio_open> must be absolute. See API NOTES, above,	376	The pathname passed to C<aio_open> must be absolute. See API NOTES, above,
132	for an explanation.	377	for an explanation.
133		378
134	The C<$flags> argument is a bitmask. See the C<Fcntl> module for a	379	The C<$flags> argument is a bitmask. See the C<Fcntl> module for a
135	list. They are the same as used by C<sysopen>.	380	list. They are the same as used by C<sysopen>.
136		381
137	Likewise, C<$mode> specifies the mode of the newly created file, if it	382	Likewise, C<$mode> specifies the mode of the newly created file, if it
138	didn't exist and C<O_CREAT> has been given, just like perl's C<sysopen>,	383	didn't exist and C<O_CREAT> has been given, just like perl's C<sysopen>,
139	except that it is mandatory (i.e. use C<0> if you don't create new files,	384	except that it is mandatory (i.e. use C<0> if you don't create new files,
140	and C<0666> or C<0777> if you do).	385	and C<0666> or C<0777> if you do). Note that the C<$mode> will be modified
		386	by the umask in effect then the request is being executed, so better never
		387	change the umask.
141		388
142	Example:	389	Example:
143		390
144	aio_open "/etc/passwd", O_RDONLY, 0, sub {	391	aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub {
145	if ($_[0]) {	392	if ($_[0]) {
146	print "open successful, fh is $_[0]\n";	393	print "open successful, fh is $_[0]\n";
147	...	394	...
148	} else {	395	} else {
149	die "open failed: $!\n";	396	die "open failed: $!\n";
150	}	397	}
151	};	398	};
152		399
		400	In addition to all the common open modes/flags (C<O_RDONLY>, C<O_WRONLY>,
		401	C<O_RDWR>, C<O_CREAT>, C<O_TRUNC>, C<O_EXCL> and C<O_APPEND>), the
		402	following POSIX and non-POSIX constants are available (missing ones on
		403	your system are, as usual, C<0>):
		404
		405	C<O_ASYNC>, C<O_DIRECT>, C<O_NOATIME>, C<O_CLOEXEC>, C<O_NOCTTY>, C<O_NOFOLLOW>,
		406	C<O_NONBLOCK>, C<O_EXEC>, C<O_SEARCH>, C<O_DIRECTORY>, C<O_DSYNC>,
		407	C<O_RSYNC>, C<O_SYNC>, C<O_PATH>, C<O_TMPFILE>, and C<O_TTY_INIT>.
		408
		409
153	=item aio_close $fh, $callback->($status)	410	=item aio_close $fh, $callback->($status)
154		411
155	Asynchronously close a file and call the callback with the result	412	Asynchronously close a file and call the callback with the result
156	code. I<WARNING:> although accepted, you should not pass in a perl	413	code.
157	filehandle here, as perl will likely close the file descriptor another
158	time when the filehandle is destroyed. Normally, you can safely call perls
159	C<close> or just let filehandles go out of scope.
160		414
161	This is supposed to be a bug in the API, so that might change. It's	415	Unfortunately, you can't do this to perl. Perl I<insists> very strongly on
162	therefore best to avoid this function.	416	closing the file descriptor associated with the filehandle itself.
		417
		418	Therefore, C<aio_close> will not close the filehandle - instead it will
		419	use dup2 to overwrite the file descriptor with the write-end of a pipe
		420	(the pipe fd will be created on demand and will be cached).
		421
		422	Or in other words: the file descriptor will be closed, but it will not be
		423	free for reuse until the perl filehandle is closed.
		424
		425	=cut
		426
		427	=item aio_seek $fh, $offset, $whence, $callback->($offs)
		428
		429	Seeks the filehandle to the new C<$offset>, similarly to perl's
		430	C<sysseek>. The C<$whence> can use the traditional values (C<0> for
		431	C<IO::AIO::SEEK_SET>, C<1> for C<IO::AIO::SEEK_CUR> or C<2> for
		432	C<IO::AIO::SEEK_END>).
		433
		434	The resulting absolute offset will be passed to the callback, or C<-1> in
		435	case of an error.
		436
		437	In theory, the C<$whence> constants could be different than the
		438	corresponding values from L<Fcntl>, but perl guarantees they are the same,
		439	so don't panic.
		440
		441	As a GNU/Linux (and maybe Solaris) extension, also the constants
		442	C<IO::AIO::SEEK_DATA> and C<IO::AIO::SEEK_HOLE> are available, if they
		443	could be found. No guarantees about suitability for use in C<aio_seek> or
		444	Perl's C<sysseek> can be made though, although I would naively assume they
		445	"just work".
163		446
164	=item aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)	447	=item aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
165		448
166	=item aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)	449	=item aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
167		450
168	Reads or writes C<length> bytes from the specified C<fh> and C<offset>	451	Reads or writes C<$length> bytes from or to the specified C<$fh> and
169	into the scalar given by C<data> and offset C<dataoffset> and calls the	452	C<$offset> into the scalar given by C<$data> and offset C<$dataoffset> and
170	callback without the actual number of bytes read (or -1 on error, just	453	calls the callback with the actual number of bytes transferred (or -1 on
171	like the syscall).	454	error, just like the syscall).
		455
		456	C<aio_read> will, like C<sysread>, shrink or grow the C<$data> scalar to
		457	offset plus the actual number of bytes read.
		458
		459	If C<$offset> is undefined, then the current file descriptor offset will
		460	be used (and updated), otherwise the file descriptor offset will not be
		461	changed by these calls.
		462
		463	If C<$length> is undefined in C<aio_write>, use the remaining length of
		464	C<$data>.
		465
		466	If C<$dataoffset> is less than zero, it will be counted from the end of
		467	C<$data>.
172		468
173	The C<$data> scalar I<MUST NOT> be modified in any way while the request	469	The C<$data> scalar I<MUST NOT> be modified in any way while the request
174	is outstanding. Modifying it can result in segfaults or WW3 (if the	470	is outstanding. Modifying it can result in segfaults or World War III (if
175	necessary/optional hardware is installed).	471	the necessary/optional hardware is installed).
176		472
177	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at	473	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at
178	offset C<0> within the scalar:	474	offset C<0> within the scalar:
179		475
180	aio_read $fh, 7, 15, $buffer, 0, sub {	476	aio_read $fh, 7, 15, $buffer, 0, sub {
181	$_[0] > 0 or die "read error: $!";	477	$_[0] > 0 or die "read error: $!";
182	print "read $_[0] bytes: <$buffer>\n";	478	print "read $_[0] bytes: <$buffer>\n";
183	};	479	};
184		480
185	=item aio_move $srcpath, $dstpath, $callback->($status)
186
187	[EXPERIMENTAL due to internal aio_group use]
188
189	Try to move the I<file> (directories not supported as either source or
190	destination) from C<$srcpath> to C<$dstpath> and call the callback with
191	the C<0> (error) or C<-1> ok.
192
193	This is a composite request that tries to rename(2) the file first. If
194	rename files with C<EXDEV>, it creates the destination file with mode 0200
195	and copies the contents of the source file into it using C<aio_sendfile>,
196	followed by restoring atime, mtime, access mode and uid/gid, in that
197	order, and unlinking the C<$srcpath>.
198
199	If an error occurs, the partial destination file will be unlinked, if
200	possible, except when setting atime, mtime, access mode and uid/gid, where
201	errors are being ignored.
202
203	=cut
204
205	sub aio_move($$$) {
206	my ($src, $dst, $cb) = @_;
207
208	my $grp = aio_group $cb;
209
210	add $grp aio_rename $src, $dst, sub {
211	if ($_[0] && $! == EXDEV) {
212	add $grp aio_open $src, O_RDONLY, 0, sub {
213	if (my $src_fh = $_[0]) {
214	my @stat = stat $src_fh;
215
216	add $grp aio_open $dst, O_WRONLY, 0200, sub {
217	if (my $dst_fh = $_[0]) {
218	add $grp aio_sendfile $dst_fh, $src_fh, 0, $stat[7], sub {
219	close $src_fh;
220
221	if ($_[0] == $stat[7]) {
222	utime $stat[8], $stat[9], $dst;
223	chmod $stat[2] & 07777, $dst_fh;
224	chown $stat[4], $stat[5], $dst_fh;
225	close $dst_fh;
226
227	add $grp aio_unlink $src, sub {
228	$grp->result ($_[0]);
229	};
230	} else {
231	my $errno = $!;
232	add $grp aio_unlink $dst, sub {
233	$! = $errno;
234	$grp->result (-1);
235	};
236	}
237	};
238	} else {
239	$grp->result (-1);
240	}
241	},
242
243	} else {
244	$grp->result (-1);
245	}
246	};
247	} else {
248	$grp->result ($_[0]);
249	}
250	};
251
252	$grp
253	}
254		481
255	=item aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)	482	=item aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
256		483
257	Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts	484	Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts
258	reading at byte offset C<$in_offset>, and starts writing at the current	485	reading at byte offset C<$in_offset>, and starts writing at the current
259	file offset of C<$out_fh>. Because of that, it is not safe to issue more	486	file offset of C<$out_fh>. Because of that, it is not safe to issue more
260	than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each	487	than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each
261	other.	488	other. The same C<$in_fh> works fine though, as this function does not
		489	move or use the file offset of C<$in_fh>.
262		490
		491	Please note that C<aio_sendfile> can read more bytes from C<$in_fh> than
		492	are written, and there is no way to find out how many more bytes have been
		493	read from C<aio_sendfile> alone, as C<aio_sendfile> only provides the
		494	number of bytes written to C<$out_fh>. Only if the result value equals
		495	C<$length> one can assume that C<$length> bytes have been read.
		496
		497	Unlike with other C<aio_> functions, it makes a lot of sense to use
		498	C<aio_sendfile> on non-blocking sockets, as long as one end (typically
		499	the C<$in_fh>) is a file - the file I/O will then be asynchronous, while
		500	the socket I/O will be non-blocking. Note, however, that you can run
		501	into a trap where C<aio_sendfile> reads some data with readahead, then
		502	fails to write all data, and when the socket is ready the next time, the
		503	data in the cache is already lost, forcing C<aio_sendfile> to again hit
		504	the disk. Explicit C<aio_read> + C<aio_write> let's you better control
		505	resource usage.
		506
263	This call tries to make use of a native C<sendfile> syscall to provide	507	This call tries to make use of a native C<sendfile>-like syscall to
264	zero-copy operation. For this to work, C<$out_fh> should refer to a	508	provide zero-copy operation. For this to work, C<$out_fh> should refer to
265	socket, and C<$in_fh> should refer to mmap'able file.	509	a socket, and C<$in_fh> should refer to an mmap'able file.
266		510
267	If the native sendfile call fails or is not implemented, it will be	511	If a native sendfile cannot be found or it fails with C<ENOSYS>,
268	emulated, so you can call C<aio_sendfile> on any type of filehandle	512	C<EINVAL>, C<ENOTSUP>, C<EOPNOTSUPP>, C<EAFNOSUPPORT>, C<EPROTOTYPE> or
		513	C<ENOTSOCK>, it will be emulated, so you can call C<aio_sendfile> on any
269	regardless of the limitations of the operating system.	514	type of filehandle regardless of the limitations of the operating system.
270		515
271	Please note, however, that C<aio_sendfile> can read more bytes from	516	As native sendfile syscalls (as practically any non-POSIX interface hacked
272	C<$in_fh> than are written, and there is no way to find out how many	517	together in a hurry to improve benchmark numbers) tend to be rather buggy
273	bytes have been read from C<aio_sendfile> alone, as C<aio_sendfile> only	518	on many systems, this implementation tries to work around some known bugs
274	provides the number of bytes written to C<$out_fh>. Only if the result	519	in Linux and FreeBSD kernels (probably others, too), but that might fail,
275	value equals C<$length> one can assume that C<$length> bytes have been	520	so you really really should check the return value of C<aio_sendfile> -
276	read.	521	fewer bytes than expected might have been transferred.
		522
277		523
278	=item aio_readahead $fh,$offset,$length, $callback->($retval)	524	=item aio_readahead $fh,$offset,$length, $callback->($retval)
279		525
280	C<aio_readahead> populates the page cache with data from a file so that	526	C<aio_readahead> populates the page cache with data from a file so that
281	subsequent reads from that file will not block on disk I/O. The C<$offset>	527	subsequent reads from that file will not block on disk I/O. The C<$offset>
…		…
284	whole pages, so that offset is effectively rounded down to a page boundary	530	whole pages, so that offset is effectively rounded down to a page boundary
285	and bytes are read up to the next page boundary greater than or equal to	531	and bytes are read up to the next page boundary greater than or equal to
286	(off-set+length). C<aio_readahead> does not read beyond the end of the	532	(off-set+length). C<aio_readahead> does not read beyond the end of the
287	file. The current file offset of the file is left unchanged.	533	file. The current file offset of the file is left unchanged.
288		534
289	If that syscall doesn't exist (likely if your OS isn't Linux) it will be	535	If that syscall doesn't exist (likely if your kernel isn't Linux) it will
290	emulated by simply reading the data, which would have a similar effect.	536	be emulated by simply reading the data, which would have a similar effect.
		537
291		538
292	=item aio_stat $fh_or_path, $callback->($status)	539	=item aio_stat $fh_or_path, $callback->($status)
293		540
294	=item aio_lstat $fh, $callback->($status)	541	=item aio_lstat $fh, $callback->($status)
295		542
…		…
301	for an explanation.	548	for an explanation.
302		549
303	Currently, the stats are always 64-bit-stats, i.e. instead of returning an	550	Currently, the stats are always 64-bit-stats, i.e. instead of returning an
304	error when stat'ing a large file, the results will be silently truncated	551	error when stat'ing a large file, the results will be silently truncated
305	unless perl itself is compiled with large file support.	552	unless perl itself is compiled with large file support.
		553
		554	To help interpret the mode and dev/rdev stat values, IO::AIO offers the
		555	following constants and functions (if not implemented, the constants will
		556	be C<0> and the functions will either C<croak> or fall back on traditional
		557	behaviour).
		558
		559	C<S_IFMT>, C<S_IFIFO>, C<S_IFCHR>, C<S_IFBLK>, C<S_IFLNK>, C<S_IFREG>,
		560	C<S_IFDIR>, C<S_IFWHT>, C<S_IFSOCK>, C<IO::AIO::major $dev_t>,
		561	C<IO::AIO::minor $dev_t>, C<IO::AIO::makedev $major, $minor>.
306		562
307	Example: Print the length of F</etc/passwd>:	563	Example: Print the length of F</etc/passwd>:
308		564
309	aio_stat "/etc/passwd", sub {	565	aio_stat "/etc/passwd", sub {
310	$_[0] and die "stat failed: $!";	566	$_[0] and die "stat failed: $!";
311	print "size is ", -s _, "\n";	567	print "size is ", -s _, "\n";
312	};	568	};
313		569
		570
		571	=item aio_statvfs $fh_or_path, $callback->($statvfs)
		572
		573	Works like the POSIX C<statvfs> or C<fstatvfs> syscalls, depending on
		574	whether a file handle or path was passed.
		575
		576	On success, the callback is passed a hash reference with the following
		577	members: C<bsize>, C<frsize>, C<blocks>, C<bfree>, C<bavail>, C<files>,
		578	C<ffree>, C<favail>, C<fsid>, C<flag> and C<namemax>. On failure, C<undef>
		579	is passed.
		580
		581	The following POSIX IO::AIO::ST_* constants are defined: C<ST_RDONLY> and
		582	C<ST_NOSUID>.
		583
		584	The following non-POSIX IO::AIO::ST_* flag masks are defined to
		585	their correct value when available, or to C<0> on systems that do
		586	not support them: C<ST_NODEV>, C<ST_NOEXEC>, C<ST_SYNCHRONOUS>,
		587	C<ST_MANDLOCK>, C<ST_WRITE>, C<ST_APPEND>, C<ST_IMMUTABLE>, C<ST_NOATIME>,
		588	C<ST_NODIRATIME> and C<ST_RELATIME>.
		589
		590	Example: stat C</wd> and dump out the data if successful.
		591
		592	aio_statvfs "/wd", sub {
		593	my $f = $_[0]
		594	or die "statvfs: $!";
		595
		596	use Data::Dumper;
		597	say Dumper $f;
		598	};
		599
		600	# result:
		601	{
		602	bsize => 1024,
		603	bfree => 4333064312,
		604	blocks => 10253828096,
		605	files => 2050765568,
		606	flag => 4096,
		607	favail => 2042092649,
		608	bavail => 4333064312,
		609	ffree => 2042092649,
		610	namemax => 255,
		611	frsize => 1024,
		612	fsid => 1810
		613	}
		614
		615	Here is a (likely partial - send me updates!) list of fsid values used by
		616	Linux - it is safe to hardcode these when C<$^O> is C<linux>:
		617
		618	0x0000adf5 adfs
		619	0x0000adff affs
		620	0x5346414f afs
		621	0x09041934 anon-inode filesystem
		622	0x00000187 autofs
		623	0x42465331 befs
		624	0x1badface bfs
		625	0x42494e4d binfmt_misc
		626	0x9123683e btrfs
		627	0x0027e0eb cgroupfs
		628	0xff534d42 cifs
		629	0x73757245 coda
		630	0x012ff7b7 coh
		631	0x28cd3d45 cramfs
		632	0x453dcd28 cramfs-wend (wrong endianness)
		633	0x64626720 debugfs
		634	0x00001373 devfs
		635	0x00001cd1 devpts
		636	0x0000f15f ecryptfs
		637	0x00414a53 efs
		638	0x0000137d ext
		639	0x0000ef53 ext2/ext3/ext4
		640	0x0000ef51 ext2
		641	0xf2f52010 f2fs
		642	0x00004006 fat
		643	0x65735546 fuseblk
		644	0x65735543 fusectl
		645	0x0bad1dea futexfs
		646	0x01161970 gfs2
		647	0x47504653 gpfs
		648	0x00004244 hfs
		649	0xf995e849 hpfs
		650	0x00c0ffee hostfs
		651	0x958458f6 hugetlbfs
		652	0x2bad1dea inotifyfs
		653	0x00009660 isofs
		654	0x000072b6 jffs2
		655	0x3153464a jfs
		656	0x6b414653 k-afs
		657	0x0bd00bd0 lustre
		658	0x0000137f minix
		659	0x0000138f minix 30 char names
		660	0x00002468 minix v2
		661	0x00002478 minix v2 30 char names
		662	0x00004d5a minix v3
		663	0x19800202 mqueue
		664	0x00004d44 msdos
		665	0x0000564c novell
		666	0x00006969 nfs
		667	0x6e667364 nfsd
		668	0x00003434 nilfs
		669	0x5346544e ntfs
		670	0x00009fa1 openprom
		671	0x7461636F ocfs2
		672	0x00009fa0 proc
		673	0x6165676c pstorefs
		674	0x0000002f qnx4
		675	0x68191122 qnx6
		676	0x858458f6 ramfs
		677	0x52654973 reiserfs
		678	0x00007275 romfs
		679	0x67596969 rpc_pipefs
		680	0x73636673 securityfs
		681	0xf97cff8c selinux
		682	0x0000517b smb
		683	0x534f434b sockfs
		684	0x73717368 squashfs
		685	0x62656572 sysfs
		686	0x012ff7b6 sysv2
		687	0x012ff7b5 sysv4
		688	0x01021994 tmpfs
		689	0x15013346 udf
		690	0x00011954 ufs
		691	0x54190100 ufs byteswapped
		692	0x00009fa2 usbdevfs
		693	0x01021997 v9fs
		694	0xa501fcf5 vxfs
		695	0xabba1974 xenfs
		696	0x012ff7b4 xenix
		697	0x58465342 xfs
		698	0x012fd16d xia
		699
		700	=item aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
		701
		702	Works like perl's C<utime> function (including the special case of $atime
		703	and $mtime being undef). Fractional times are supported if the underlying
		704	syscalls support them.
		705
		706	When called with a pathname, uses utimes(2) if available, otherwise
		707	utime(2). If called on a file descriptor, uses futimes(2) if available,
		708	otherwise returns ENOSYS, so this is not portable.
		709
		710	Examples:
		711
		712	# set atime and mtime to current time (basically touch(1)):
		713	aio_utime "path", undef, undef;
		714	# set atime to current time and mtime to beginning of the epoch:
		715	aio_utime "path", time, undef; # undef==0
		716
		717
		718	=item aio_chown $fh_or_path, $uid, $gid, $callback->($status)
		719
		720	Works like perl's C<chown> function, except that C<undef> for either $uid
		721	or $gid is being interpreted as "do not change" (but -1 can also be used).
		722
		723	Examples:
		724
		725	# same as "chown root path" in the shell:
		726	aio_chown "path", 0, -1;
		727	# same as above:
		728	aio_chown "path", 0, undef;
		729
		730
		731	=item aio_truncate $fh_or_path, $offset, $callback->($status)
		732
		733	Works like truncate(2) or ftruncate(2).
		734
		735
		736	=item aio_allocate $fh, $mode, $offset, $len, $callback->($status)
		737
		738	Allocates or frees disk space according to the C<$mode> argument. See the
		739	linux C<fallocate> documentation for details.
		740
		741	C<$mode> is usually C<0> or C<IO::AIO::FALLOC_FL_KEEP_SIZE> to allocate
		742	space, or C<IO::AIO::FALLOC_FL_PUNCH_HOLE | IO::AIO::FALLOC_FL_KEEP_SIZE>,
		743	to deallocate a file range.
		744
		745	IO::AIO also supports C<FALLOC_FL_COLLAPSE_RANGE>, to remove a range
		746	(without leaving a hole), C<FALLOC_FL_ZERO_RANGE>, to zero a range,
		747	C<FALLOC_FL_INSERT_RANGE> to insert a range and C<FALLOC_FL_UNSHARE_RANGE>
		748	to unshare shared blocks (see your L<fallocate(2)> manpage).
		749
		750	The file system block size used by C<fallocate> is presumably the
		751	C<f_bsize> returned by C<statvfs>, but different filesystems and filetypes
		752	can dictate other limitations.
		753
		754	If C<fallocate> isn't available or cannot be emulated (currently no
		755	emulation will be attempted), passes C<-1> and sets C<$!> to C<ENOSYS>.
		756
		757
		758	=item aio_chmod $fh_or_path, $mode, $callback->($status)
		759
		760	Works like perl's C<chmod> function.
		761
		762
314	=item aio_unlink $pathname, $callback->($status)	763	=item aio_unlink $pathname, $callback->($status)
315		764
316	Asynchronously unlink (delete) a file and call the callback with the	765	Asynchronously unlink (delete) a file and call the callback with the
317	result code.	766	result code.
318		767
		768
		769	=item aio_mknod $pathname, $mode, $dev, $callback->($status)
		770
		771	[EXPERIMENTAL]
		772
		773	Asynchronously create a device node (or fifo). See mknod(2).
		774
		775	The only (POSIX-) portable way of calling this function is:
		776
		777	aio_mknod $pathname, IO::AIO::S_IFIFO | $mode, 0, sub { ...
		778
		779	See C<aio_stat> for info about some potentially helpful extra constants
		780	and functions.
		781
319	=item aio_link $srcpath, $dstpath, $callback->($status)	782	=item aio_link $srcpath, $dstpath, $callback->($status)
320		783
321	Asynchronously create a new link to the existing object at C<$srcpath> at	784	Asynchronously create a new link to the existing object at C<$srcpath> at
322	the path C<$dstpath> and call the callback with the result code.	785	the path C<$dstpath> and call the callback with the result code.
323		786
		787
324	=item aio_symlink $srcpath, $dstpath, $callback->($status)	788	=item aio_symlink $srcpath, $dstpath, $callback->($status)
325		789
326	Asynchronously create a new symbolic link to the existing object at C<$srcpath> at	790	Asynchronously create a new symbolic link to the existing object at C<$srcpath> at
327	the path C<$dstpath> and call the callback with the result code.	791	the path C<$dstpath> and call the callback with the result code.
328		792
		793
		794	=item aio_readlink $pathname, $callback->($link)
		795
		796	Asynchronously read the symlink specified by C<$path> and pass it to
		797	the callback. If an error occurs, nothing or undef gets passed to the
		798	callback.
		799
		800
		801	=item aio_realpath $pathname, $callback->($path)
		802
		803	Asynchronously make the path absolute and resolve any symlinks in
		804	C<$path>. The resulting path only consists of directories (same as
		805	L<Cwd::realpath>).
		806
		807	This request can be used to get the absolute path of the current working
		808	directory by passing it a path of F<.> (a single dot).
		809
		810
329	=item aio_rename $srcpath, $dstpath, $callback->($status)	811	=item aio_rename $srcpath, $dstpath, $callback->($status)
330		812
331	Asynchronously rename the object at C<$srcpath> to C<$dstpath>, just as	813	Asynchronously rename the object at C<$srcpath> to C<$dstpath>, just as
332	rename(2) and call the callback with the result code.	814	rename(2) and call the callback with the result code.
333		815
		816	On systems that support the AIO::WD working directory abstraction
		817	natively, the case C<[$wd, "."]> as C<$srcpath> is specialcased - instead
		818	of failing, C<rename> is called on the absolute path of C<$wd>.
		819
		820
		821	=item aio_rename2 $srcpath, $dstpath, $flags, $callback->($status)
		822
		823	Basically a version of C<aio_rename> with an additional C<$flags>
		824	argument. Calling this with C<$flags=0> is the same as calling
		825	C<aio_rename>.
		826
		827	Non-zero flags are currently only supported on GNU/Linux systems that
		828	support renameat2. Other systems fail with C<ENOSYS> in this case.
		829
		830	The following constants are available (missing ones are, as usual C<0>),
		831	see renameat2(2) for details:
		832
		833	C<IO::AIO::RENAME_NOREPLACE>, C<IO::AIO::RENAME_EXCHANGE>
		834	and C<IO::AIO::RENAME_WHITEOUT>.
		835
		836
		837	=item aio_mkdir $pathname, $mode, $callback->($status)
		838
		839	Asynchronously mkdir (create) a directory and call the callback with
		840	the result code. C<$mode> will be modified by the umask at the time the
		841	request is executed, so do not change your umask.
		842
		843
334	=item aio_rmdir $pathname, $callback->($status)	844	=item aio_rmdir $pathname, $callback->($status)
335		845
336	Asynchronously rmdir (delete) a directory and call the callback with the	846	Asynchronously rmdir (delete) a directory and call the callback with the
337	result code.	847	result code.
		848
		849	On systems that support the AIO::WD working directory abstraction
		850	natively, the case C<[$wd, "."]> is specialcased - instead of failing,
		851	C<rmdir> is called on the absolute path of C<$wd>.
		852
338		853
339	=item aio_readdir $pathname, $callback->($entries)	854	=item aio_readdir $pathname, $callback->($entries)
340		855
341	Unlike the POSIX call of the same name, C<aio_readdir> reads an entire	856	Unlike the POSIX call of the same name, C<aio_readdir> reads an entire
342	directory (i.e. opendir + readdir + closedir). The entries will not be	857	directory (i.e. opendir + readdir + closedir). The entries will not be
343	sorted, and will B<NOT> include the C<.> and C<..> entries.	858	sorted, and will B<NOT> include the C<.> and C<..> entries.
344		859
345	The callback a single argument which is either C<undef> or an array-ref	860	The callback is passed a single argument which is either C<undef> or an
346	with the filenames.	861	array-ref with the filenames.
347		862
		863
		864	=item aio_readdirx $pathname, $flags, $callback->($entries, $flags)
		865
		866	Quite similar to C<aio_readdir>, but the C<$flags> argument allows one to
		867	tune behaviour and output format. In case of an error, C<$entries> will be
		868	C<undef>.
		869
		870	The flags are a combination of the following constants, ORed together (the
		871	flags will also be passed to the callback, possibly modified):
		872
		873	=over 4
		874
		875	=item IO::AIO::READDIR_DENTS
		876
		877	When this flag is off, then the callback gets an arrayref consisting of
		878	names only (as with C<aio_readdir>), otherwise it gets an arrayref with
		879	C<[$name, $type, $inode]> arrayrefs, each describing a single directory
		880	entry in more detail.
		881
		882	C<$name> is the name of the entry.
		883
		884	C<$type> is one of the C<IO::AIO::DT_xxx> constants:
		885
		886	C<IO::AIO::DT_UNKNOWN>, C<IO::AIO::DT_FIFO>, C<IO::AIO::DT_CHR>, C<IO::AIO::DT_DIR>,
		887	C<IO::AIO::DT_BLK>, C<IO::AIO::DT_REG>, C<IO::AIO::DT_LNK>, C<IO::AIO::DT_SOCK>,
		888	C<IO::AIO::DT_WHT>.
		889
		890	C<IO::AIO::DT_UNKNOWN> means just that: readdir does not know. If you need to
		891	know, you have to run stat yourself. Also, for speed reasons, the C<$type>
		892	scalars are read-only: you can not modify them.
		893
		894	C<$inode> is the inode number (which might not be exact on systems with 64
		895	bit inode numbers and 32 bit perls). This field has unspecified content on
		896	systems that do not deliver the inode information.
		897
		898	=item IO::AIO::READDIR_DIRS_FIRST
		899
		900	When this flag is set, then the names will be returned in an order where
		901	likely directories come first, in optimal stat order. This is useful when
		902	you need to quickly find directories, or you want to find all directories
		903	while avoiding to stat() each entry.
		904
		905	If the system returns type information in readdir, then this is used
		906	to find directories directly. Otherwise, likely directories are names
		907	beginning with ".", or otherwise names with no dots, of which names with
		908	short names are tried first.
		909
		910	=item IO::AIO::READDIR_STAT_ORDER
		911
		912	When this flag is set, then the names will be returned in an order
		913	suitable for stat()'ing each one. That is, when you plan to stat()
		914	all files in the given directory, then the returned order will likely
		915	be fastest.
		916
		917	If both this flag and C<IO::AIO::READDIR_DIRS_FIRST> are specified, then
		918	the likely dirs come first, resulting in a less optimal stat order.
		919
		920	=item IO::AIO::READDIR_FOUND_UNKNOWN
		921
		922	This flag should not be set when calling C<aio_readdirx>. Instead, it
		923	is being set by C<aio_readdirx>, when any of the C<$type>'s found were
		924	C<IO::AIO::DT_UNKNOWN>. The absence of this flag therefore indicates that all
		925	C<$type>'s are known, which can be used to speed up some algorithms.
		926
		927	=back
		928
		929
		930	=item aio_load $pathname, $data, $callback->($status)
		931
		932	This is a composite request that tries to fully load the given file into
		933	memory. Status is the same as with aio_read.
		934
		935	=cut
		936
		937	sub aio_load($$;$) {
		938	my ($path, undef, $cb) = @_;
		939	my $data = \$_[1];
		940
		941	my $pri = aioreq_pri;
		942	my $grp = aio_group $cb;
		943
		944	aioreq_pri $pri;
		945	add $grp aio_open $path, O_RDONLY, 0, sub {
		946	my $fh = shift
		947	or return $grp->result (-1);
		948
		949	aioreq_pri $pri;
		950	add $grp aio_read $fh, 0, (-s $fh), $$data, 0, sub {
		951	$grp->result ($_[0]);
		952	};
		953	};
		954
		955	$grp
		956	}
		957
		958	=item aio_copy $srcpath, $dstpath, $callback->($status)
		959
		960	Try to copy the I<file> (directories not supported as either source or
		961	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		962	a status of C<0> (ok) or C<-1> (error, see C<$!>).
		963
		964	Existing destination files will be truncated.
		965
		966	This is a composite request that creates the destination file with
		967	mode 0200 and copies the contents of the source file into it using
		968	C<aio_sendfile>, followed by restoring atime, mtime, access mode and
		969	uid/gid, in that order.
		970
		971	If an error occurs, the partial destination file will be unlinked, if
		972	possible, except when setting atime, mtime, access mode and uid/gid, where
		973	errors are being ignored.
		974
		975	=cut
		976
		977	sub aio_copy($$;$) {
		978	my ($src, $dst, $cb) = @_;
		979
		980	my $pri = aioreq_pri;
		981	my $grp = aio_group $cb;
		982
		983	aioreq_pri $pri;
		984	add $grp aio_open $src, O_RDONLY, 0, sub {
		985	if (my $src_fh = $_[0]) {
		986	my @stat = stat $src_fh; # hmm, might block over nfs?
		987
		988	aioreq_pri $pri;
		989	add $grp aio_open $dst, O_CREAT | O_WRONLY | O_TRUNC, 0200, sub {
		990	if (my $dst_fh = $_[0]) {
		991	aioreq_pri $pri;
		992	add $grp aio_sendfile $dst_fh, $src_fh, 0, $stat[7], sub {
		993	if ($_[0] == $stat[7]) {
		994	$grp->result (0);
		995	close $src_fh;
		996
		997	my $ch = sub {
		998	aioreq_pri $pri;
		999	add $grp aio_chmod $dst_fh, $stat[2] & 07777, sub {
		1000	aioreq_pri $pri;
		1001	add $grp aio_chown $dst_fh, $stat[4], $stat[5], sub {
		1002	aioreq_pri $pri;
		1003	add $grp aio_close $dst_fh;
		1004	}
		1005	};
		1006	};
		1007
		1008	aioreq_pri $pri;
		1009	add $grp aio_utime $dst_fh, $stat[8], $stat[9], sub {
		1010	if ($_[0] < 0 && $! == ENOSYS) {
		1011	aioreq_pri $pri;
		1012	add $grp aio_utime $dst, $stat[8], $stat[9], $ch;
		1013	} else {
		1014	$ch->();
		1015	}
		1016	};
		1017	} else {
		1018	$grp->result (-1);
		1019	close $src_fh;
		1020	close $dst_fh;
		1021
		1022	aioreq $pri;
		1023	add $grp aio_unlink $dst;
		1024	}
		1025	};
		1026	} else {
		1027	$grp->result (-1);
		1028	}
		1029	},
		1030
		1031	} else {
		1032	$grp->result (-1);
		1033	}
		1034	};
		1035
		1036	$grp
		1037	}
		1038
		1039	=item aio_move $srcpath, $dstpath, $callback->($status)
		1040
		1041	Try to move the I<file> (directories not supported as either source or
		1042	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		1043	a status of C<0> (ok) or C<-1> (error, see C<$!>).
		1044
		1045	This is a composite request that tries to rename(2) the file first; if
		1046	rename fails with C<EXDEV>, it copies the file with C<aio_copy> and, if
		1047	that is successful, unlinks the C<$srcpath>.
		1048
		1049	=cut
		1050
		1051	sub aio_move($$;$) {
		1052	my ($src, $dst, $cb) = @_;
		1053
		1054	my $pri = aioreq_pri;
		1055	my $grp = aio_group $cb;
		1056
		1057	aioreq_pri $pri;
		1058	add $grp aio_rename $src, $dst, sub {
		1059	if ($_[0] && $! == EXDEV) {
		1060	aioreq_pri $pri;
		1061	add $grp aio_copy $src, $dst, sub {
		1062	$grp->result ($_[0]);
		1063
		1064	unless ($_[0]) {
		1065	aioreq_pri $pri;
		1066	add $grp aio_unlink $src;
		1067	}
		1068	};
		1069	} else {
		1070	$grp->result ($_[0]);
		1071	}
		1072	};
		1073
		1074	$grp
		1075	}
		1076
348	=item aio_scandir $path, $maxreq, $callback->($dirs, $nondirs)	1077	=item aio_scandir $pathname, $maxreq, $callback->($dirs, $nondirs)
349
350	[EXPERIMENTAL due to internal aio_group use]
351		1078
352	Scans a directory (similar to C<aio_readdir>) but additionally tries to	1079	Scans a directory (similar to C<aio_readdir>) but additionally tries to
353	separate the entries of directory C<$path> into two sets of names, ones	1080	efficiently separate the entries of directory C<$path> into two sets of
354	you can recurse into (directories or links to them), and ones you cannot	1081	names, directories you can recurse into (directories), and ones you cannot
355	recurse into (everything else).	1082	recurse into (everything else, including symlinks to directories).
356		1083
357	C<aio_scandir> is a composite request that consists of many sub	1084	C<aio_scandir> is a composite request that creates of many sub requests_
358	requests. C<$maxreq> specifies the maximum number of outstanding aio	1085	C<$maxreq> specifies the maximum number of outstanding aio requests that
359	requests that this function generates. If it is C<< <= 0 >>, then a	1086	this function generates. If it is C<< <= 0 >>, then a suitable default
360	suitable default will be chosen (currently 8).	1087	will be chosen (currently 4).
361		1088
362	On error, the callback is called without arguments, otherwise it receives	1089	On error, the callback is called without arguments, otherwise it receives
363	two array-refs with path-relative entry names.	1090	two array-refs with path-relative entry names.
364		1091
365	Example:	1092	Example:
…		…
372		1099
373	Implementation notes.	1100	Implementation notes.
374		1101
375	The C<aio_readdir> cannot be avoided, but C<stat()>'ing every entry can.	1102	The C<aio_readdir> cannot be avoided, but C<stat()>'ing every entry can.
376		1103
		1104	If readdir returns file type information, then this is used directly to
		1105	find directories.
		1106
377	After reading the directory, the modification time, size etc. of the	1107	Otherwise, after reading the directory, the modification time, size etc.
378	directory before and after the readdir is checked, and if they match (and	1108	of the directory before and after the readdir is checked, and if they
379	isn't the current time), the link count will be used to decide how many	1109	match (and isn't the current time), the link count will be used to decide
380	entries are directories (if >= 2). Otherwise, no knowledge of the number	1110	how many entries are directories (if >= 2). Otherwise, no knowledge of the
381	of subdirectories will be assumed.	1111	number of subdirectories will be assumed.
382		1112
383	Then entries will be sorted into likely directories (everything without	1113	Then entries will be sorted into likely directories a non-initial dot
384	a non-initial dot currently) and likely non-directories (everything	1114	currently) and likely non-directories (see C<aio_readdirx>). Then every
385	else). Then every entry plus an appended C</.> will be C<stat>'ed,	1115	entry plus an appended C</.> will be C<stat>'ed, likely directories first,
386	likely directories first. If that succeeds, it assumes that the entry	1116	in order of their inode numbers. If that succeeds, it assumes that the
387	is a directory or a symlink to directory (which will be checked	1117	entry is a directory or a symlink to directory (which will be checked
388	seperately). This is often faster than stat'ing the entry itself because	1118	separately). This is often faster than stat'ing the entry itself because
389	filesystems might detect the type of the entry without reading the inode	1119	filesystems might detect the type of the entry without reading the inode
390	data (e.g. ext2fs filetype feature).	1120	data (e.g. ext2fs filetype feature), even on systems that cannot return
		1121	the filetype information on readdir.
391		1122
392	If the known number of directories (link count - 2) has been reached, the	1123	If the known number of directories (link count - 2) has been reached, the
393	rest of the entries is assumed to be non-directories.	1124	rest of the entries is assumed to be non-directories.
394		1125
395	This only works with certainty on POSIX (= UNIX) filesystems, which	1126	This only works with certainty on POSIX (= UNIX) filesystems, which
…		…
399	as those tend to return 0 or 1 as link counts, which disables the	1130	as those tend to return 0 or 1 as link counts, which disables the
400	directory counting heuristic.	1131	directory counting heuristic.
401		1132
402	=cut	1133	=cut
403		1134
404	sub aio_scandir($$$) {	1135	sub aio_scandir($$;$) {
405	my ($path, $maxreq, $cb) = @_;	1136	my ($path, $maxreq, $cb) = @_;
406		1137
		1138	my $pri = aioreq_pri;
		1139
407	my $grp = aio_group $cb;	1140	my $grp = aio_group $cb;
408		1141
409	$maxreq = 8 if $maxreq <= 0;	1142	$maxreq = 4 if $maxreq <= 0;
410		1143
411	# stat once	1144	# get a wd object
		1145	aioreq_pri $pri;
412	add $grp aio_stat $path, sub {	1146	add $grp aio_wd $path, sub {
		1147	$_[0]
		1148	or return $grp->result ();
		1149
		1150	my $wd = [shift, "."];
		1151
		1152	# stat once
		1153	aioreq_pri $pri;
		1154	add $grp aio_stat $wd, sub {
413	return $grp->result () if $_[0];	1155	return $grp->result () if $_[0];
414	my $now = time;	1156	my $now = time;
415	my $hash1 = join ":", (stat _)[0,1,3,7,9];	1157	my $hash1 = join ":", (stat _)[0,1,3,7,9];
416		1158
417	# read the directory entries	1159	# read the directory entries
418	add $grp aio_readdir $path, sub {	1160	aioreq_pri $pri;
		1161	add $grp aio_readdirx $wd, READDIR_DIRS_FIRST, sub {
419	my $entries = shift	1162	my $entries = shift
420	or return $grp->result ();	1163	or return $grp->result ();
421		1164
422	# stat the dir another time	1165	# stat the dir another time
		1166	aioreq_pri $pri;
423	add $grp aio_stat $path, sub {	1167	add $grp aio_stat $wd, sub {
424	my $hash2 = join ":", (stat _)[0,1,3,7,9];	1168	my $hash2 = join ":", (stat _)[0,1,3,7,9];
425		1169
426	my $ndirs;	1170	my $ndirs;
427		1171
428	# take the slow route if anything looks fishy	1172	# take the slow route if anything looks fishy
429	if ($hash1 ne $hash2 or (stat _)[9] == $now) {	1173	if ($hash1 ne $hash2 or (stat _)[9] == $now) {
430	$ndirs = -1;	1174	$ndirs = -1;
431	} else {	1175	} else {
432	# if nlink == 2, we are finished	1176	# if nlink == 2, we are finished
433	# on non-posix-fs's, we rely on nlink < 2	1177	# for non-posix-fs's, we rely on nlink < 2
434	$ndirs = (stat _)[3] - 2	1178	$ndirs = (stat _)[3] - 2
435	or return $grp->result ([], $entries);	1179	or return $grp->result ([], $entries);
436	}
437
438	# sort into likely dirs and likely nondirs
439	# dirs == files without ".", short entries first
440	$entries = [map $_->[0],
441	sort { $b->[1] cmp $a->[1] }
442	map [$_, sprintf "%s%04d", (/.\./ ? "1" : "0"), length],
443	@$entries];
444
445	my (@dirs, @nondirs);
446
447	my ($statcb, $schedcb);
448	my $nreq = 0;
449
450	$schedcb = sub {
451	if (@$entries) {
452	if ($nreq < $maxreq) {
453	my $ent = pop @$entries;
454	$nreq++;
455	add $grp aio_stat "$path/$ent/.", sub { $statcb->($_[0], $ent) };
456	}
457	} elsif (!$nreq) {
458	# finished
459	undef $statcb;
460	undef $schedcb;
461	$grp->result (\@dirs, \@nondirs) if $cb;
462	undef $cb;
463	}	1180	}
		1181
		1182	my (@dirs, @nondirs);
		1183
		1184	my $statgrp = add $grp aio_group sub {
		1185	$grp->result (\@dirs, \@nondirs);
		1186	};
		1187
		1188	limit $statgrp $maxreq;
		1189	feed $statgrp sub {
		1190	return unless @$entries;
		1191	my $entry = shift @$entries;
		1192
		1193	aioreq_pri $pri;
		1194	$wd->[1] = "$entry/.";
		1195	add $statgrp aio_stat $wd, sub {
		1196	if ($_[0] < 0) {
		1197	push @nondirs, $entry;
		1198	} else {
		1199	# need to check for real directory
		1200	aioreq_pri $pri;
		1201	$wd->[1] = $entry;
		1202	add $statgrp aio_lstat $wd, sub {
		1203	if (-d _) {
		1204	push @dirs, $entry;
		1205
		1206	unless (--$ndirs) {
		1207	push @nondirs, @$entries;
		1208	feed $statgrp;
		1209	}
		1210	} else {
		1211	push @nondirs, $entry;
		1212	}
		1213	}
		1214	}
		1215	};
		1216	};
464	};	1217	};
465	$statcb = sub {
466	my ($status, $entry) = @_;
467
468	if ($status < 0) {
469	$nreq--;
470	push @nondirs, $entry;
471	&$schedcb;
472	} else {
473	# need to check for real directory
474	add $grp aio_lstat "$path/$entry", sub {
475	$nreq--;
476
477	if (-d _) {
478	push @dirs, $entry;
479
480	if (!--$ndirs) {
481	push @nondirs, @$entries;
482	$entries = [];
483	}
484	} else {
485	push @nondirs, $entry;
486	}
487
488	&$schedcb;
489	}
490	}
491	};
492
493	&$schedcb while @$entries && $nreq < $maxreq;
494	};	1218	};
495	};	1219	};
496	};	1220	};
497		1221
498	$grp	1222	$grp
499	}	1223	}
500		1224
		1225	=item aio_rmtree $pathname, $callback->($status)
		1226
		1227	Delete a directory tree starting (and including) C<$path>, return the
		1228	status of the final C<rmdir> only. This is a composite request that
		1229	uses C<aio_scandir> to recurse into and rmdir directories, and unlink
		1230	everything else.
		1231
		1232	=cut
		1233
		1234	sub aio_rmtree;
		1235	sub aio_rmtree($;$) {
		1236	my ($path, $cb) = @_;
		1237
		1238	my $pri = aioreq_pri;
		1239	my $grp = aio_group $cb;
		1240
		1241	aioreq_pri $pri;
		1242	add $grp aio_scandir $path, 0, sub {
		1243	my ($dirs, $nondirs) = @_;
		1244
		1245	my $dirgrp = aio_group sub {
		1246	add $grp aio_rmdir $path, sub {
		1247	$grp->result ($_[0]);
		1248	};
		1249	};
		1250
		1251	(aioreq_pri $pri), add $dirgrp aio_rmtree "$path/$_" for @$dirs;
		1252	(aioreq_pri $pri), add $dirgrp aio_unlink "$path/$_" for @$nondirs;
		1253
		1254	add $grp $dirgrp;
		1255	};
		1256
		1257	$grp
		1258	}
		1259
		1260	=item aio_fcntl $fh, $cmd, $arg, $callback->($status)
		1261
		1262	=item aio_ioctl $fh, $request, $buf, $callback->($status)
		1263
		1264	These work just like the C<fcntl> and C<ioctl> built-in functions, except
		1265	they execute asynchronously and pass the return value to the callback.
		1266
		1267	Both calls can be used for a lot of things, some of which make more sense
		1268	to run asynchronously in their own thread, while some others make less
		1269	sense. For example, calls that block waiting for external events, such
		1270	as locking, will also lock down an I/O thread while it is waiting, which
		1271	can deadlock the whole I/O system. At the same time, there might be no
		1272	alternative to using a thread to wait.
		1273
		1274	So in general, you should only use these calls for things that do
		1275	(filesystem) I/O, not for things that wait for other events (network,
		1276	other processes), although if you are careful and know what you are doing,
		1277	you still can.
		1278
		1279	The following constants are available (missing ones are, as usual C<0>):
		1280
		1281	C<F_DUPFD_CLOEXEC>,
		1282
		1283	C<F_OFD_GETLK>, C<F_OFD_SETLK>, C<F_OFD_GETLKW>,
		1284
		1285	C<FIFREEZE>, C<FITHAW>, C<FITRIM>, C<FICLONE>, C<FICLONERANGE>, C<FIDEDUPERANGE>.
		1286
		1287	C<FS_IOC_GETFLAGS>, C<FS_IOC_SETFLAGS>, C<FS_IOC_GETVERSION>, C<FS_IOC_SETVERSION>,
		1288	C<FS_IOC_FIEMAP>.
		1289
		1290	C<FS_IOC_FSGETXATTR>, C<FS_IOC_FSSETXATTR>, C<FS_IOC_SET_ENCRYPTION_POLICY>,
		1291	C<FS_IOC_GET_ENCRYPTION_PWSALT>, C<FS_IOC_GET_ENCRYPTION_POLICY>, C<FS_KEY_DESCRIPTOR_SIZE>.
		1292
		1293	C<FS_SECRM_FL>, C<FS_UNRM_FL>, C<FS_COMPR_FL>, C<FS_SYNC_FL>, C<FS_IMMUTABLE_FL>,
		1294	C<FS_APPEND_FL>, C<FS_NODUMP_FL>, C<FS_NOATIME_FL>, C<FS_DIRTY_FL>,
		1295	C<FS_COMPRBLK_FL>, C<FS_NOCOMP_FL>, C<FS_ENCRYPT_FL>, C<FS_BTREE_FL>,
		1296	C<FS_INDEX_FL>, C<FS_JOURNAL_DATA_FL>, C<FS_NOTAIL_FL>, C<FS_DIRSYNC_FL>, C<FS_TOPDIR_FL>,
		1297	C<FS_FL_USER_MODIFIABLE>.
		1298
		1299	C<FS_XFLAG_REALTIME>, C<FS_XFLAG_PREALLOC>, C<FS_XFLAG_IMMUTABLE>, C<FS_XFLAG_APPEND>,
		1300	C<FS_XFLAG_SYNC>, C<FS_XFLAG_NOATIME>, C<FS_XFLAG_NODUMP>, C<FS_XFLAG_RTINHERIT>,
		1301	C<FS_XFLAG_PROJINHERIT>, C<FS_XFLAG_NOSYMLINKS>, C<FS_XFLAG_EXTSIZE>, C<FS_XFLAG_EXTSZINHERIT>,
		1302	C<FS_XFLAG_NODEFRAG>, C<FS_XFLAG_FILESTREAM>, C<FS_XFLAG_DAX>, C<FS_XFLAG_HASATTR>,
		1303
		1304	=item aio_sync $callback->($status)
		1305
		1306	Asynchronously call sync and call the callback when finished.
		1307
501	=item aio_fsync $fh, $callback->($status)	1308	=item aio_fsync $fh, $callback->($status)
502		1309
503	Asynchronously call fsync on the given filehandle and call the callback	1310	Asynchronously call fsync on the given filehandle and call the callback
504	with the fsync result code.	1311	with the fsync result code.
505		1312
…		…
509	callback with the fdatasync result code.	1316	callback with the fdatasync result code.
510		1317
511	If this call isn't available because your OS lacks it or it couldn't be	1318	If this call isn't available because your OS lacks it or it couldn't be
512	detected, it will be emulated by calling C<fsync> instead.	1319	detected, it will be emulated by calling C<fsync> instead.
513		1320
		1321	=item aio_syncfs $fh, $callback->($status)
		1322
		1323	Asynchronously call the syncfs syscall to sync the filesystem associated
		1324	to the given filehandle and call the callback with the syncfs result
		1325	code. If syncfs is not available, calls sync(), but returns C<-1> and sets
		1326	errno to C<ENOSYS> nevertheless.
		1327
		1328	=item aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
		1329
		1330	Sync the data portion of the file specified by C<$offset> and C<$length>
		1331	to disk (but NOT the metadata), by calling the Linux-specific
		1332	sync_file_range call. If sync_file_range is not available or it returns
		1333	ENOSYS, then fdatasync or fsync is being substituted.
		1334
		1335	C<$flags> can be a combination of C<IO::AIO::SYNC_FILE_RANGE_WAIT_BEFORE>,
		1336	C<IO::AIO::SYNC_FILE_RANGE_WRITE> and
		1337	C<IO::AIO::SYNC_FILE_RANGE_WAIT_AFTER>: refer to the sync_file_range
		1338	manpage for details.
		1339
		1340	=item aio_pathsync $pathname, $callback->($status)
		1341
		1342	This request tries to open, fsync and close the given path. This is a
		1343	composite request intended to sync directories after directory operations
		1344	(E.g. rename). This might not work on all operating systems or have any
		1345	specific effect, but usually it makes sure that directory changes get
		1346	written to disc. It works for anything that can be opened for read-only,
		1347	not just directories.
		1348
		1349	Future versions of this function might fall back to other methods when
		1350	C<fsync> on the directory fails (such as calling C<sync>).
		1351
		1352	Passes C<0> when everything went ok, and C<-1> on error.
		1353
		1354	=cut
		1355
		1356	sub aio_pathsync($;$) {
		1357	my ($path, $cb) = @_;
		1358
		1359	my $pri = aioreq_pri;
		1360	my $grp = aio_group $cb;
		1361
		1362	aioreq_pri $pri;
		1363	add $grp aio_open $path, O_RDONLY, 0, sub {
		1364	my ($fh) = @_;
		1365	if ($fh) {
		1366	aioreq_pri $pri;
		1367	add $grp aio_fsync $fh, sub {
		1368	$grp->result ($_[0]);
		1369
		1370	aioreq_pri $pri;
		1371	add $grp aio_close $fh;
		1372	};
		1373	} else {
		1374	$grp->result (-1);
		1375	}
		1376	};
		1377
		1378	$grp
		1379	}
		1380
		1381	=item aio_msync $scalar, $offset = 0, $length = undef, flags = MS_SYNC, $callback->($status)
		1382
		1383	This is a rather advanced IO::AIO call, which only works on mmap(2)ed
		1384	scalars (see the C<IO::AIO::mmap> function, although it also works on data
		1385	scalars managed by the L<Sys::Mmap> or L<Mmap> modules, note that the
		1386	scalar must only be modified in-place while an aio operation is pending on
		1387	it).
		1388
		1389	It calls the C<msync> function of your OS, if available, with the memory
		1390	area starting at C<$offset> in the string and ending C<$length> bytes
		1391	later. If C<$length> is negative, counts from the end, and if C<$length>
		1392	is C<undef>, then it goes till the end of the string. The flags can be
		1393	either C<IO::AIO::MS_ASYNC> or C<IO::AIO::MS_SYNC>, plus an optional
		1394	C<IO::AIO::MS_INVALIDATE>.
		1395
		1396	=item aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		1397
		1398	This is a rather advanced IO::AIO call, which works best on mmap(2)ed
		1399	scalars.
		1400
		1401	It touches (reads or writes) all memory pages in the specified
		1402	range inside the scalar. All caveats and parameters are the same
		1403	as for C<aio_msync>, above, except for flags, which must be either
		1404	C<0> (which reads all pages and ensures they are instantiated) or
		1405	C<IO::AIO::MT_MODIFY>, which modifies the memory pages (by reading and
		1406	writing an octet from it, which dirties the page).
		1407
		1408	=item aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status)
		1409
		1410	This is a rather advanced IO::AIO call, which works best on mmap(2)ed
		1411	scalars.
		1412
		1413	It reads in all the pages of the underlying storage into memory (if any)
		1414	and locks them, so they are not getting swapped/paged out or removed.
		1415
		1416	If C<$length> is undefined, then the scalar will be locked till the end.
		1417
		1418	On systems that do not implement C<mlock>, this function returns C<-1>
		1419	and sets errno to C<ENOSYS>.
		1420
		1421	Note that the corresponding C<munlock> is synchronous and is
		1422	documented under L<MISCELLANEOUS FUNCTIONS>.
		1423
		1424	Example: open a file, mmap and mlock it - both will be undone when
		1425	C<$data> gets destroyed.
		1426
		1427	open my $fh, "<", $path or die "$path: $!";
		1428	my $data;
		1429	IO::AIO::mmap $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh;
		1430	aio_mlock $data; # mlock in background
		1431
		1432	=item aio_mlockall $flags, $callback->($status)
		1433
		1434	Calls the C<mlockall> function with the given C<$flags> (a combination of
		1435	C<IO::AIO::MCL_CURRENT> and C<IO::AIO::MCL_FUTURE>).
		1436
		1437	On systems that do not implement C<mlockall>, this function returns C<-1>
		1438	and sets errno to C<ENOSYS>.
		1439
		1440	Note that the corresponding C<munlockall> is synchronous and is
		1441	documented under L<MISCELLANEOUS FUNCTIONS>.
		1442
		1443	Example: asynchronously lock all current and future pages into memory.
		1444
		1445	aio_mlockall IO::AIO::MCL_FUTURE;
		1446
		1447	=item aio_fiemap $fh, $start, $length, $flags, $count, $cb->(\@extents)
		1448
		1449	Queries the extents of the given file (by calling the Linux C<FIEMAP>
		1450	ioctl, see L<http://cvs.schmorp.de/IO-AIO/doc/fiemap.txt> for details). If
		1451	the ioctl is not available on your OS, then this request will fail with
		1452	C<ENOSYS>.
		1453
		1454	C<$start> is the starting offset to query extents for, C<$length> is the
		1455	size of the range to query - if it is C<undef>, then the whole file will
		1456	be queried.
		1457
		1458	C<$flags> is a combination of flags (C<IO::AIO::FIEMAP_FLAG_SYNC> or
		1459	C<IO::AIO::FIEMAP_FLAG_XATTR> - C<IO::AIO::FIEMAP_FLAGS_COMPAT> is also
		1460	exported), and is normally C<0> or C<IO::AIO::FIEMAP_FLAG_SYNC> to query
		1461	the data portion.
		1462
		1463	C<$count> is the maximum number of extent records to return. If it is
		1464	C<undef>, then IO::AIO queries all extents of the range. As a very special
		1465	case, if it is C<0>, then the callback receives the number of extents
		1466	instead of the extents themselves (which is unreliable, see below).
		1467
		1468	If an error occurs, the callback receives no arguments. The special
		1469	C<errno> value C<IO::AIO::EBADR> is available to test for flag errors.
		1470
		1471	Otherwise, the callback receives an array reference with extent
		1472	structures. Each extent structure is an array reference itself, with the
		1473	following members:
		1474
		1475	[$logical, $physical, $length, $flags]
		1476
		1477	Flags is any combination of the following flag values (typically either C<0>
		1478	or C<IO::AIO::FIEMAP_EXTENT_LAST> (1)):
		1479
		1480	C<IO::AIO::FIEMAP_EXTENT_LAST>, C<IO::AIO::FIEMAP_EXTENT_UNKNOWN>,
		1481	C<IO::AIO::FIEMAP_EXTENT_DELALLOC>, C<IO::AIO::FIEMAP_EXTENT_ENCODED>,
		1482	C<IO::AIO::FIEMAP_EXTENT_DATA_ENCRYPTED>, C<IO::AIO::FIEMAP_EXTENT_NOT_ALIGNED>,
		1483	C<IO::AIO::FIEMAP_EXTENT_DATA_INLINE>, C<IO::AIO::FIEMAP_EXTENT_DATA_TAIL>,
		1484	C<IO::AIO::FIEMAP_EXTENT_UNWRITTEN>, C<IO::AIO::FIEMAP_EXTENT_MERGED> or
		1485	C<IO::AIO::FIEMAP_EXTENT_SHARED>.
		1486
		1487	At the time of this writing (Linux 3.2), this requets is unreliable unless
		1488	C<$count> is C<undef>, as the kernel has all sorts of bugs preventing
		1489	it to return all extents of a range for files with large number of
		1490	extents. The code works around all these issues if C<$count> is undef.
		1491
514	=item aio_group $callback->(...)	1492	=item aio_group $callback->(...)
515
516	[EXPERIMENTAL]
517		1493
518	This is a very special aio request: Instead of doing something, it is a	1494	This is a very special aio request: Instead of doing something, it is a
519	container for other aio requests, which is useful if you want to bundle	1495	container for other aio requests, which is useful if you want to bundle
520	many requests into a single, composite, request.	1496	many requests into a single, composite, request with a definite callback
		1497	and the ability to cancel the whole request with its subrequests.
521		1498
522	Returns an object of class L<IO::AIO::GRP>. See its documentation below	1499	Returns an object of class L<IO::AIO::GRP>. See its documentation below
523	for more info.	1500	for more info.
524		1501
525	Example:	1502	Example:
…		…
531	add $grp	1508	add $grp
532	(aio_stat ...),	1509	(aio_stat ...),
533	(aio_stat ...),	1510	(aio_stat ...),
534	...;	1511	...;
535		1512
		1513	=item aio_nop $callback->()
		1514
		1515	This is a special request - it does nothing in itself and is only used for
		1516	side effects, such as when you want to add a dummy request to a group so
		1517	that finishing the requests in the group depends on executing the given
		1518	code.
		1519
		1520	While this request does nothing, it still goes through the execution
		1521	phase and still requires a worker thread. Thus, the callback will not
		1522	be executed immediately but only after other requests in the queue have
		1523	entered their execution phase. This can be used to measure request
		1524	latency.
		1525
536	=item IO::AIO::aio_sleep $fractional_seconds, $callback->() *NOT EXPORTED*	1526	=item IO::AIO::aio_busy $fractional_seconds, $callback->() *NOT EXPORTED*
537		1527
538	Mainly used for debugging and benchmarking, this aio request puts one of	1528	Mainly used for debugging and benchmarking, this aio request puts one of
539	the request workers to sleep for the given time.	1529	the request workers to sleep for the given time.
540		1530
541	While it is theoretically handy to have simple I/O scheduling requests	1531	While it is theoretically handy to have simple I/O scheduling requests
542	like sleep and file handle readable/writable, the overhead this creates	1532	like sleep and file handle readable/writable, the overhead this creates is
543	is immense, so do not use this function except to put your application	1533	immense (it blocks a thread for a long time) so do not use this function
544	under artificial I/O pressure.	1534	except to put your application under artificial I/O pressure.
545		1535
546	=back	1536	=back
		1537
		1538
		1539	=head2 IO::AIO::WD - multiple working directories
		1540
		1541	Your process only has one current working directory, which is used by all
		1542	threads. This makes it hard to use relative paths (some other component
		1543	could call C<chdir> at any time, and it is hard to control when the path
		1544	will be used by IO::AIO).
		1545
		1546	One solution for this is to always use absolute paths. This usually works,
		1547	but can be quite slow (the kernel has to walk the whole path on every
		1548	access), and can also be a hassle to implement.
		1549
		1550	Newer POSIX systems have a number of functions (openat, fdopendir,
		1551	futimensat and so on) that make it possible to specify working directories
		1552	per operation.
		1553
		1554	For portability, and because the clowns who "designed", or shall I write,
		1555	perpetrated this new interface were obviously half-drunk, this abstraction
		1556	cannot be perfect, though.
		1557
		1558	IO::AIO allows you to convert directory paths into a so-called IO::AIO::WD
		1559	object. This object stores the canonicalised, absolute version of the
		1560	path, and on systems that allow it, also a directory file descriptor.
		1561
		1562	Everywhere where a pathname is accepted by IO::AIO (e.g. in C<aio_stat>
		1563	or C<aio_unlink>), one can specify an array reference with an IO::AIO::WD
		1564	object and a pathname instead (or the IO::AIO::WD object alone, which
		1565	gets interpreted as C<[$wd, "."]>). If the pathname is absolute, the
		1566	IO::AIO::WD object is ignored, otherwise the pathname is resolved relative
		1567	to that IO::AIO::WD object.
		1568
		1569	For example, to get a wd object for F</etc> and then stat F<passwd>
		1570	inside, you would write:
		1571
		1572	aio_wd "/etc", sub {
		1573	my $etcdir = shift;
		1574
		1575	# although $etcdir can be undef on error, there is generally no reason
		1576	# to check for errors here, as aio_stat will fail with ENOENT
		1577	# when $etcdir is undef.
		1578
		1579	aio_stat [$etcdir, "passwd"], sub {
		1580	# yay
		1581	};
		1582	};
		1583
		1584	The fact that C<aio_wd> is a request and not a normal function shows that
		1585	creating an IO::AIO::WD object is itself a potentially blocking operation,
		1586	which is why it is done asynchronously.
		1587
		1588	To stat the directory obtained with C<aio_wd> above, one could write
		1589	either of the following three request calls:
		1590
		1591	aio_lstat "/etc" , sub { ... # pathname as normal string
		1592	aio_lstat [$wd, "."], sub { ... # "." relative to $wd (i.e. $wd itself)
		1593	aio_lstat $wd , sub { ... # shorthand for the previous
		1594
		1595	As with normal pathnames, IO::AIO keeps a copy of the working directory
		1596	object and the pathname string, so you could write the following without
		1597	causing any issues due to C<$path> getting reused:
		1598
		1599	my $path = [$wd, undef];
		1600
		1601	for my $name (qw(abc def ghi)) {
		1602	$path->[1] = $name;
		1603	aio_stat $path, sub {
		1604	# ...
		1605	};
		1606	}
		1607
		1608	There are some caveats: when directories get renamed (or deleted), the
		1609	pathname string doesn't change, so will point to the new directory (or
		1610	nowhere at all), while the directory fd, if available on the system,
		1611	will still point to the original directory. Most functions accepting a
		1612	pathname will use the directory fd on newer systems, and the string on
		1613	older systems. Some functions (such as realpath) will always rely on the
		1614	string form of the pathname.
		1615
		1616	So this functionality is mainly useful to get some protection against
		1617	C<chdir>, to easily get an absolute path out of a relative path for future
		1618	reference, and to speed up doing many operations in the same directory
		1619	(e.g. when stat'ing all files in a directory).
		1620
		1621	The following functions implement this working directory abstraction:
		1622
		1623	=over 4
		1624
		1625	=item aio_wd $pathname, $callback->($wd)
		1626
		1627	Asynchonously canonicalise the given pathname and convert it to an
		1628	IO::AIO::WD object representing it. If possible and supported on the
		1629	system, also open a directory fd to speed up pathname resolution relative
		1630	to this working directory.
		1631
		1632	If something goes wrong, then C<undef> is passwd to the callback instead
		1633	of a working directory object and C<$!> is set appropriately. Since
		1634	passing C<undef> as working directory component of a pathname fails the
		1635	request with C<ENOENT>, there is often no need for error checking in the
		1636	C<aio_wd> callback, as future requests using the value will fail in the
		1637	expected way.
		1638
		1639	=item IO::AIO::CWD
		1640
		1641	This is a compiletime constant (object) that represents the process
		1642	current working directory.
		1643
		1644	Specifying this object as working directory object for a pathname is as if
		1645	the pathname would be specified directly, without a directory object. For
		1646	example, these calls are functionally identical:
		1647
		1648	aio_stat "somefile", sub { ... };
		1649	aio_stat [IO::AIO::CWD, "somefile"], sub { ... };
		1650
		1651	=back
		1652
		1653	To recover the path associated with an IO::AIO::WD object, you can use
		1654	C<aio_realpath>:
		1655
		1656	aio_realpath $wd, sub {
		1657	warn "path is $_[0]\n";
		1658	};
		1659
		1660	Currently, C<aio_statvfs> always, and C<aio_rename> and C<aio_rmdir>
		1661	sometimes, fall back to using an absolue path.
547		1662
548	=head2 IO::AIO::REQ CLASS	1663	=head2 IO::AIO::REQ CLASS
549		1664
550	All non-aggregate C<aio_*> functions return an object of this class when	1665	All non-aggregate C<aio_*> functions return an object of this class when
551	called in non-void context.	1666	called in non-void context.
552		1667
553	A request always moves through the following five states in its lifetime,
554	in order: B<ready> (request has been created, but has not been executed
555	yet), B<execute> (request is currently being executed), B<pending>
556	(request has been executed but callback has not been called yet),
557	B<result> (results are being processed synchronously, includes calling the
558	callback) and B<done> (request has reached the end of its lifetime and
559	holds no resources anymore).
560
561	=over 4	1668	=over 4
562		1669
563	=item $req->cancel	1670	=item cancel $req
564		1671
565	Cancels the request, if possible. Has the effect of skipping execution	1672	Cancels the request, if possible. Has the effect of skipping execution
566	when entering the B<execute> state and skipping calling the callback when	1673	when entering the B<execute> state and skipping calling the callback when
567	entering the the B<result> state, but will leave the request otherwise	1674	entering the the B<result> state, but will leave the request otherwise
568	untouched. That means that requests that currently execute will not be	1675	untouched (with the exception of readdir). That means that requests that
569	stopped and resources held by the request will not be freed prematurely.	1676	currently execute will not be stopped and resources held by the request
		1677	will not be freed prematurely.
		1678
		1679	=item cb $req $callback->(...)
		1680
		1681	Replace (or simply set) the callback registered to the request.
570		1682
571	=back	1683	=back
572		1684
573	=head2 IO::AIO::GRP CLASS	1685	=head2 IO::AIO::GRP CLASS
574		1686
…		…
601	};	1713	};
602		1714
603	This makes it very easy to create composite requests (see the source of	1715	This makes it very easy to create composite requests (see the source of
604	C<aio_move> for an application) that work and feel like simple requests.	1716	C<aio_move> for an application) that work and feel like simple requests.
605		1717
		1718	=over 4
		1719
606	The IO::AIO::GRP objects will be cleaned up during calls to	1720	=item * The IO::AIO::GRP objects will be cleaned up during calls to
607	C<IO::AIO::poll_cb>, just like any other request.	1721	C<IO::AIO::poll_cb>, just like any other request.
608		1722
609	They can be canceled like any other request. Canceling will cancel not	1723	=item * They can be canceled like any other request. Canceling will cancel not
610	only the request itself, but also all requests it contains.	1724	only the request itself, but also all requests it contains.
611		1725
612	They can also can also be added to other IO::AIO::GRP objects.	1726	=item * They can also can also be added to other IO::AIO::GRP objects.
		1727
		1728	=item * You must not add requests to a group from within the group callback (or
		1729	any later time).
		1730
		1731	=back
613		1732
614	Their lifetime, simplified, looks like this: when they are empty, they	1733	Their lifetime, simplified, looks like this: when they are empty, they
615	will finish very quickly. If they contain only requests that are in the	1734	will finish very quickly. If they contain only requests that are in the
616	C<done> state, they will also finish. Otherwise they will continue to	1735	C<done> state, they will also finish. Otherwise they will continue to
617	exist.	1736	exist.
618		1737
619	That means after creating a group you have some time to add requests. And	1738	That means after creating a group you have some time to add requests
620	in the callbacks of those requests, you can add further requests to the	1739	(precisely before the callback has been invoked, which is only done within
621	group. And only when all those requests have finished will the the group	1740	the C<poll_cb>). And in the callbacks of those requests, you can add
622	itself finish.	1741	further requests to the group. And only when all those requests have
		1742	finished will the the group itself finish.
623		1743
624	=over 4	1744	=over 4
625		1745
		1746	=item add $grp ...
		1747
626	=item $grp->add (...)	1748	=item $grp->add (...)
627
628	=item add $grp ...
629		1749
630	Add one or more requests to the group. Any type of L<IO::AIO::REQ> can	1750	Add one or more requests to the group. Any type of L<IO::AIO::REQ> can
631	be added, including other groups, as long as you do not create circular	1751	be added, including other groups, as long as you do not create circular
632	dependencies.	1752	dependencies.
633		1753
634	Returns all its arguments.	1754	Returns all its arguments.
635		1755
		1756	=item $grp->cancel_subs
		1757
		1758	Cancel all subrequests and clears any feeder, but not the group request
		1759	itself. Useful when you queued a lot of events but got a result early.
		1760
		1761	The group request will finish normally (you cannot add requests to the
		1762	group).
		1763
636	=item $grp->result (...)	1764	=item $grp->result (...)
637		1765
638	Set the result value(s) that will be passed to the group callback when all	1766	Set the result value(s) that will be passed to the group callback when all
639	subrequests have finished. By default, no argument will be passed.	1767	subrequests have finished and set the groups errno to the current value
		1768	of errno (just like calling C<errno> without an error number). By default,
		1769	no argument will be passed and errno is zero.
		1770
		1771	=item $grp->errno ([$errno])
		1772
		1773	Sets the group errno value to C<$errno>, or the current value of errno
		1774	when the argument is missing.
		1775
		1776	Every aio request has an associated errno value that is restored when
		1777	the callback is invoked. This method lets you change this value from its
		1778	default (0).
		1779
		1780	Calling C<result> will also set errno, so make sure you either set C<$!>
		1781	before the call to C<result>, or call c<errno> after it.
		1782
		1783	=item feed $grp $callback->($grp)
		1784
		1785	Sets a feeder/generator on this group: every group can have an attached
		1786	generator that generates requests if idle. The idea behind this is that,
		1787	although you could just queue as many requests as you want in a group,
		1788	this might starve other requests for a potentially long time. For example,
		1789	C<aio_scandir> might generate hundreds of thousands of C<aio_stat>
		1790	requests, delaying any later requests for a long time.
		1791
		1792	To avoid this, and allow incremental generation of requests, you can
		1793	instead a group and set a feeder on it that generates those requests. The
		1794	feed callback will be called whenever there are few enough (see C<limit>,
		1795	below) requests active in the group itself and is expected to queue more
		1796	requests.
		1797
		1798	The feed callback can queue as many requests as it likes (i.e. C<add> does
		1799	not impose any limits).
		1800
		1801	If the feed does not queue more requests when called, it will be
		1802	automatically removed from the group.
		1803
		1804	If the feed limit is C<0> when this method is called, it will be set to
		1805	C<2> automatically.
		1806
		1807	Example:
		1808
		1809	# stat all files in @files, but only ever use four aio requests concurrently:
		1810
		1811	my $grp = aio_group sub { print "finished\n" };
		1812	limit $grp 4;
		1813	feed $grp sub {
		1814	my $file = pop @files
		1815	or return;
		1816
		1817	add $grp aio_stat $file, sub { ... };
		1818	};
		1819
		1820	=item limit $grp $num
		1821
		1822	Sets the feeder limit for the group: The feeder will be called whenever
		1823	the group contains less than this many requests.
		1824
		1825	Setting the limit to C<0> will pause the feeding process.
		1826
		1827	The default value for the limit is C<0>, but note that setting a feeder
		1828	automatically bumps it up to C<2>.
640		1829
641	=back	1830	=back
642		1831
643	=head2 SUPPORT FUNCTIONS	1832	=head2 SUPPORT FUNCTIONS
644		1833
		1834	=head3 EVENT PROCESSING AND EVENT LOOP INTEGRATION
		1835
645	=over 4	1836	=over 4
646		1837
647	=item $fileno = IO::AIO::poll_fileno	1838	=item $fileno = IO::AIO::poll_fileno
648		1839
649	Return the I<request result pipe file descriptor>. This filehandle must be	1840	Return the I<request result pipe file descriptor>. This filehandle must be
650	polled for reading by some mechanism outside this module (e.g. Event or	1841	polled for reading by some mechanism outside this module (e.g. EV, Glib,
651	select, see below or the SYNOPSIS). If the pipe becomes readable you have	1842	select and so on, see below or the SYNOPSIS). If the pipe becomes readable
652	to call C<poll_cb> to check the results.	1843	you have to call C<poll_cb> to check the results.
653		1844
654	See C<poll_cb> for an example.	1845	See C<poll_cb> for an example.
655		1846
656	=item IO::AIO::poll_cb	1847	=item IO::AIO::poll_cb
657		1848
658	Process all outstanding events on the result pipe. You have to call this	1849	Process some requests that have reached the result phase (i.e. they have
659	regularly. Returns the number of events processed. Returns immediately	1850	been executed but the results are not yet reported). You have to call
660	when no events are outstanding.	1851	this "regularly" to finish outstanding requests.
		1852
		1853	Returns C<0> if all events could be processed (or there were no
		1854	events to process), or C<-1> if it returned earlier for whatever
		1855	reason. Returns immediately when no events are outstanding. The amount
		1856	of events processed depends on the settings of C<IO::AIO::max_poll_req>,
		1857	C<IO::AIO::max_poll_time> and C<IO::AIO::max_outstanding>.
		1858
		1859	If not all requests were processed for whatever reason, the poll file
		1860	descriptor will still be ready when C<poll_cb> returns, so normally you
		1861	don't have to do anything special to have it called later.
		1862
		1863	Apart from calling C<IO::AIO::poll_cb> when the event filehandle becomes
		1864	ready, it can be beneficial to call this function from loops which submit
		1865	a lot of requests, to make sure the results get processed when they become
		1866	available and not just when the loop is finished and the event loop takes
		1867	over again. This function returns very fast when there are no outstanding
		1868	requests.
661		1869
662	Example: Install an Event watcher that automatically calls	1870	Example: Install an Event watcher that automatically calls
663	IO::AIO::poll_cb with high priority:	1871	IO::AIO::poll_cb with high priority (more examples can be found in the
		1872	SYNOPSIS section, at the top of this document):
664		1873
665	Event->io (fd => IO::AIO::poll_fileno,	1874	Event->io (fd => IO::AIO::poll_fileno,
666	poll => 'r', async => 1,	1875	poll => 'r', async => 1,
667	cb => \&IO::AIO::poll_cb);	1876	cb => \&IO::AIO::poll_cb);
668		1877
669	=item IO::AIO::poll_wait	1878	=item IO::AIO::poll_wait
670		1879
671	Wait till the result filehandle becomes ready for reading (simply does a	1880	Wait until either at least one request is in the result phase or no
672	C<select> on the filehandle. This is useful if you want to synchronously wait	1881	requests are outstanding anymore.
673	for some requests to finish).	1882
		1883	This is useful if you want to synchronously wait for some requests to
		1884	become ready, without actually handling them.
674		1885
675	See C<nreqs> for an example.	1886	See C<nreqs> for an example.
676		1887
		1888	=item IO::AIO::poll
		1889
		1890	Waits until some requests have been handled.
		1891
		1892	Returns the number of requests processed, but is otherwise strictly
		1893	equivalent to:
		1894
		1895	IO::AIO::poll_wait, IO::AIO::poll_cb
		1896
677	=item IO::AIO::nreqs	1897	=item IO::AIO::flush
678		1898
679	Returns the number of requests currently outstanding (i.e. for which their	1899	Wait till all outstanding AIO requests have been handled.
680	callback has not been invoked yet).
681		1900
682	Example: wait till there are no outstanding requests anymore:	1901	Strictly equivalent to:
683		1902
684	IO::AIO::poll_wait, IO::AIO::poll_cb	1903	IO::AIO::poll_wait, IO::AIO::poll_cb
685	while IO::AIO::nreqs;	1904	while IO::AIO::nreqs;
686		1905
687	=item IO::AIO::flush	1906	=item IO::AIO::max_poll_reqs $nreqs
688		1907
689	Wait till all outstanding AIO requests have been handled.	1908	=item IO::AIO::max_poll_time $seconds
690		1909
691	Strictly equivalent to:	1910	These set the maximum number of requests (default C<0>, meaning infinity)
		1911	that are being processed by C<IO::AIO::poll_cb> in one call, respectively
		1912	the maximum amount of time (default C<0>, meaning infinity) spent in
		1913	C<IO::AIO::poll_cb> to process requests (more correctly the mininum amount
		1914	of time C<poll_cb> is allowed to use).
692		1915
693	IO::AIO::poll_wait, IO::AIO::poll_cb	1916	Setting C<max_poll_time> to a non-zero value creates an overhead of one
694	while IO::AIO::nreqs;	1917	syscall per request processed, which is not normally a problem unless your
		1918	callbacks are really really fast or your OS is really really slow (I am
		1919	not mentioning Solaris here). Using C<max_poll_reqs> incurs no overhead.
695		1920
696	=item IO::AIO::poll	1921	Setting these is useful if you want to ensure some level of
		1922	interactiveness when perl is not fast enough to process all requests in
		1923	time.
697		1924
698	Waits until some requests have been handled.	1925	For interactive programs, values such as C<0.01> to C<0.1> should be fine.
699		1926
700	Strictly equivalent to:	1927	Example: Install an Event watcher that automatically calls
		1928	IO::AIO::poll_cb with low priority, to ensure that other parts of the
		1929	program get the CPU sometimes even under high AIO load.
701		1930
702	IO::AIO::poll_wait, IO::AIO::poll_cb	1931	# try not to spend much more than 0.1s in poll_cb
703	if IO::AIO::nreqs;	1932	IO::AIO::max_poll_time 0.1;
		1933
		1934	# use a low priority so other tasks have priority
		1935	Event->io (fd => IO::AIO::poll_fileno,
		1936	poll => 'r', nice => 1,
		1937	cb => &IO::AIO::poll_cb);
		1938
		1939	=back
		1940
		1941	=head3 CONTROLLING THE NUMBER OF THREADS
		1942
		1943	=over
704		1944
705	=item IO::AIO::min_parallel $nthreads	1945	=item IO::AIO::min_parallel $nthreads
706		1946
707	Set the minimum number of AIO threads to C<$nthreads>. The current default	1947	Set the minimum number of AIO threads to C<$nthreads>. The current
708	is C<4>, which means four asynchronous operations can be done at one time	1948	default is C<8>, which means eight asynchronous operations can execute
709	(the number of outstanding operations, however, is unlimited).	1949	concurrently at any one time (the number of outstanding requests,
		1950	however, is unlimited).
710		1951
711	IO::AIO starts threads only on demand, when an AIO request is queued and	1952	IO::AIO starts threads only on demand, when an AIO request is queued and
712	no free thread exists.	1953	no free thread exists. Please note that queueing up a hundred requests can
		1954	create demand for a hundred threads, even if it turns out that everything
		1955	is in the cache and could have been processed faster by a single thread.
713		1956
714	It is recommended to keep the number of threads low, as some Linux	1957	It is recommended to keep the number of threads relatively low, as some
715	kernel versions will scale negatively with the number of threads (higher	1958	Linux kernel versions will scale negatively with the number of threads
716	parallelity => MUCH higher latency). With current Linux 2.6 versions, 4-32	1959	(higher parallelity => MUCH higher latency). With current Linux 2.6
717	threads should be fine.	1960	versions, 4-32 threads should be fine.
718		1961
719	Under most circumstances you don't need to call this function, as the	1962	Under most circumstances you don't need to call this function, as the
720	module selects a default that is suitable for low to moderate load.	1963	module selects a default that is suitable for low to moderate load.
721		1964
722	=item IO::AIO::max_parallel $nthreads	1965	=item IO::AIO::max_parallel $nthreads
…		…
731	This module automatically runs C<max_parallel 0> at program end, to ensure	1974	This module automatically runs C<max_parallel 0> at program end, to ensure
732	that all threads are killed and that there are no outstanding requests.	1975	that all threads are killed and that there are no outstanding requests.
733		1976
734	Under normal circumstances you don't need to call this function.	1977	Under normal circumstances you don't need to call this function.
735		1978
		1979	=item IO::AIO::max_idle $nthreads
		1980
		1981	Limit the number of threads (default: 4) that are allowed to idle
		1982	(i.e., threads that did not get a request to process within the idle
		1983	timeout (default: 10 seconds). That means if a thread becomes idle while
		1984	C<$nthreads> other threads are also idle, it will free its resources and
		1985	exit.
		1986
		1987	This is useful when you allow a large number of threads (e.g. 100 or 1000)
		1988	to allow for extremely high load situations, but want to free resources
		1989	under normal circumstances (1000 threads can easily consume 30MB of RAM).
		1990
		1991	The default is probably ok in most situations, especially if thread
		1992	creation is fast. If thread creation is very slow on your system you might
		1993	want to use larger values.
		1994
		1995	=item IO::AIO::idle_timeout $seconds
		1996
		1997	Sets the minimum idle timeout (default 10) after which worker threads are
		1998	allowed to exit. SEe C<IO::AIO::max_idle>.
		1999
736	=item $oldnreqs = IO::AIO::max_outstanding $nreqs	2000	=item IO::AIO::max_outstanding $maxreqs
737		2001
738	Sets the maximum number of outstanding requests to C<$nreqs>. If you	2002	Sets the maximum number of outstanding requests to C<$nreqs>. If
739	try to queue up more than this number of requests, the caller will block until	2003	you do queue up more than this number of requests, the next call to
740	some requests have been handled.	2004	C<IO::AIO::poll_cb> (and other functions calling C<poll_cb>, such as
		2005	C<IO::AIO::flush> or C<IO::AIO::poll>) will block until the limit is no
		2006	longer exceeded.
741		2007
742	The default is very large, so normally there is no practical limit. If you	2008	In other words, this setting does not enforce a queue limit, but can be
743	queue up many requests in a loop it often improves speed if you set	2009	used to make poll functions block if the limit is exceeded.
744	this to a relatively low number, such as C<100>.
745		2010
746	Under normal circumstances you don't need to call this function.	2011	This is a very bad function to use in interactive programs because it
		2012	blocks, and a bad way to reduce concurrency because it is inexact: Better
		2013	use an C<aio_group> together with a feed callback.
		2014
		2015	Its main use is in scripts without an event loop - when you want to stat
		2016	a lot of files, you can write something like this:
		2017
		2018	IO::AIO::max_outstanding 32;
		2019
		2020	for my $path (...) {
		2021	aio_stat $path , ...;
		2022	IO::AIO::poll_cb;
		2023	}
		2024
		2025	IO::AIO::flush;
		2026
		2027	The call to C<poll_cb> inside the loop will normally return instantly, but
		2028	as soon as more thna C<32> reqeusts are in-flight, it will block until
		2029	some requests have been handled. This keeps the loop from pushing a large
		2030	number of C<aio_stat> requests onto the queue.
		2031
		2032	The default value for C<max_outstanding> is very large, so there is no
		2033	practical limit on the number of outstanding requests.
747		2034
748	=back	2035	=back
749		2036
		2037	=head3 STATISTICAL INFORMATION
		2038
		2039	=over
		2040
		2041	=item IO::AIO::nreqs
		2042
		2043	Returns the number of requests currently in the ready, execute or pending
		2044	states (i.e. for which their callback has not been invoked yet).
		2045
		2046	Example: wait till there are no outstanding requests anymore:
		2047
		2048	IO::AIO::poll_wait, IO::AIO::poll_cb
		2049	while IO::AIO::nreqs;
		2050
		2051	=item IO::AIO::nready
		2052
		2053	Returns the number of requests currently in the ready state (not yet
		2054	executed).
		2055
		2056	=item IO::AIO::npending
		2057
		2058	Returns the number of requests currently in the pending state (executed,
		2059	but not yet processed by poll_cb).
		2060
		2061	=back
		2062
		2063	=head3 MISCELLANEOUS FUNCTIONS
		2064
		2065	IO::AIO implements some functions that are useful when you want to use
		2066	some "Advanced I/O" function not available to in Perl, without going the
		2067	"Asynchronous I/O" route. Many of these have an asynchronous C<aio_*>
		2068	counterpart.
		2069
		2070	=over 4
		2071
		2072	=item $numfd = IO::AIO::get_fdlimit
		2073
		2074	Tries to find the current file descriptor limit and returns it, or
		2075	C<undef> and sets C<$!> in case of an error. The limit is one larger than
		2076	the highest valid file descriptor number.
		2077
		2078	=item IO::AIO::min_fdlimit [$numfd]
		2079
		2080	Try to increase the current file descriptor limit(s) to at least C<$numfd>
		2081	by changing the soft or hard file descriptor resource limit. If C<$numfd>
		2082	is missing, it will try to set a very high limit, although this is not
		2083	recommended when you know the actual minimum that you require.
		2084
		2085	If the limit cannot be raised enough, the function makes a best-effort
		2086	attempt to increase the limit as much as possible, using various
		2087	tricks, while still failing. You can query the resulting limit using
		2088	C<IO::AIO::get_fdlimit>.
		2089
		2090	If an error occurs, returns C<undef> and sets C<$!>.
		2091
		2092	=item IO::AIO::sendfile $ofh, $ifh, $offset, $count
		2093
		2094	Calls the C<eio_sendfile_sync> function, which is like C<aio_sendfile>,
		2095	but is blocking (this makes most sense if you know the input data is
		2096	likely cached already and the output filehandle is set to non-blocking
		2097	operations).
		2098
		2099	Returns the number of bytes copied, or C<-1> on error.
		2100
		2101	=item IO::AIO::fadvise $fh, $offset, $len, $advice
		2102
		2103	Simply calls the C<posix_fadvise> function (see its
		2104	manpage for details). The following advice constants are
		2105	available: C<IO::AIO::FADV_NORMAL>, C<IO::AIO::FADV_SEQUENTIAL>,
		2106	C<IO::AIO::FADV_RANDOM>, C<IO::AIO::FADV_NOREUSE>,
		2107	C<IO::AIO::FADV_WILLNEED>, C<IO::AIO::FADV_DONTNEED>.
		2108
		2109	On systems that do not implement C<posix_fadvise>, this function returns
		2110	ENOSYS, otherwise the return value of C<posix_fadvise>.
		2111
		2112	=item IO::AIO::madvise $scalar, $offset, $len, $advice
		2113
		2114	Simply calls the C<posix_madvise> function (see its
		2115	manpage for details). The following advice constants are
		2116	available: C<IO::AIO::MADV_NORMAL>, C<IO::AIO::MADV_SEQUENTIAL>,
		2117	C<IO::AIO::MADV_RANDOM>, C<IO::AIO::MADV_WILLNEED>,
		2118	C<IO::AIO::MADV_DONTNEED>.
		2119
		2120	If C<$offset> is negative, counts from the end. If C<$length> is negative,
		2121	the remaining length of the C<$scalar> is used. If possible, C<$length>
		2122	will be reduced to fit into the C<$scalar>.
		2123
		2124	On systems that do not implement C<posix_madvise>, this function returns
		2125	ENOSYS, otherwise the return value of C<posix_madvise>.
		2126
		2127	=item IO::AIO::mprotect $scalar, $offset, $len, $protect
		2128
		2129	Simply calls the C<mprotect> function on the preferably AIO::mmap'ed
		2130	$scalar (see its manpage for details). The following protect
		2131	constants are available: C<IO::AIO::PROT_NONE>, C<IO::AIO::PROT_READ>,
		2132	C<IO::AIO::PROT_WRITE>, C<IO::AIO::PROT_EXEC>.
		2133
		2134	If C<$offset> is negative, counts from the end. If C<$length> is negative,
		2135	the remaining length of the C<$scalar> is used. If possible, C<$length>
		2136	will be reduced to fit into the C<$scalar>.
		2137
		2138	On systems that do not implement C<mprotect>, this function returns
		2139	ENOSYS, otherwise the return value of C<mprotect>.
		2140
		2141	=item IO::AIO::mmap $scalar, $length, $prot, $flags, $fh[, $offset]
		2142
		2143	Memory-maps a file (or anonymous memory range) and attaches it to the
		2144	given C<$scalar>, which will act like a string scalar. Returns true on
		2145	success, and false otherwise.
		2146
		2147	The scalar must exist, but its contents do not matter - this means you
		2148	cannot use a nonexistant array or hash element. When in doubt, C<undef>
		2149	the scalar first.
		2150
		2151	The only operations allowed on the mmapped scalar are C<substr>/C<vec>,
		2152	which don't change the string length, and most read-only operations such
		2153	as copying it or searching it with regexes and so on.
		2154
		2155	Anything else is unsafe and will, at best, result in memory leaks.
		2156
		2157	The memory map associated with the C<$scalar> is automatically removed
		2158	when the C<$scalar> is undef'd or destroyed, or when the C<IO::AIO::mmap>
		2159	or C<IO::AIO::munmap> functions are called on it.
		2160
		2161	This calls the C<mmap>(2) function internally. See your system's manual
		2162	page for details on the C<$length>, C<$prot> and C<$flags> parameters.
		2163
		2164	The C<$length> must be larger than zero and smaller than the actual
		2165	filesize.
		2166
		2167	C<$prot> is a combination of C<IO::AIO::PROT_NONE>, C<IO::AIO::PROT_EXEC>,
		2168	C<IO::AIO::PROT_READ> and/or C<IO::AIO::PROT_WRITE>,
		2169
		2170	C<$flags> can be a combination of
		2171	C<IO::AIO::MAP_SHARED> or
		2172	C<IO::AIO::MAP_PRIVATE>,
		2173	or a number of system-specific flags (when not available, the are C<0>):
		2174	C<IO::AIO::MAP_ANONYMOUS> (which is set to C<MAP_ANON> if your system only provides this constant),
		2175	C<IO::AIO::MAP_LOCKED>,
		2176	C<IO::AIO::MAP_NORESERVE>,
		2177	C<IO::AIO::MAP_POPULATE>,
		2178	C<IO::AIO::MAP_NONBLOCK>,
		2179	C<IO::AIO::MAP_FIXED>,
		2180	C<IO::AIO::MAP_GROWSDOWN>,
		2181	C<IO::AIO::MAP_32BIT>,
		2182	C<IO::AIO::MAP_HUGETLB> or
		2183	C<IO::AIO::MAP_STACK>.
		2184
		2185	If C<$fh> is C<undef>, then a file descriptor of C<-1> is passed.
		2186
		2187	C<$offset> is the offset from the start of the file - it generally must be
		2188	a multiple of C<IO::AIO::PAGESIZE> and defaults to C<0>.
		2189
		2190	Example:
		2191
		2192	use Digest::MD5;
		2193	use IO::AIO;
		2194
		2195	open my $fh, "<verybigfile"
		2196	or die "$!";
		2197
		2198	IO::AIO::mmap my $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh
		2199	or die "verybigfile: $!";
		2200
		2201	my $fast_md5 = md5 $data;
		2202
		2203	=item IO::AIO::munmap $scalar
		2204
		2205	Removes a previous mmap and undefines the C<$scalar>.
		2206
		2207	=item IO::AIO::munlock $scalar, $offset = 0, $length = undef
		2208
		2209	Calls the C<munlock> function, undoing the effects of a previous
		2210	C<aio_mlock> call (see its description for details).
		2211
		2212	=item IO::AIO::munlockall
		2213
		2214	Calls the C<munlockall> function.
		2215
		2216	On systems that do not implement C<munlockall>, this function returns
		2217	ENOSYS, otherwise the return value of C<munlockall>.
		2218
		2219	=item IO::AIO::splice $r_fh, $r_off, $w_fh, $w_off, $length, $flags
		2220
		2221	Calls the GNU/Linux C<splice(2)> syscall, if available. If C<$r_off> or
		2222	C<$w_off> are C<undef>, then C<NULL> is passed for these, otherwise they
		2223	should be the file offset.
		2224
		2225	C<$r_fh> and C<$w_fh> should not refer to the same file, as splice might
		2226	silently corrupt the data in this case.
		2227
		2228	The following symbol flag values are available: C<IO::AIO::SPLICE_F_MOVE>,
		2229	C<IO::AIO::SPLICE_F_NONBLOCK>, C<IO::AIO::SPLICE_F_MORE> and
		2230	C<IO::AIO::SPLICE_F_GIFT>.
		2231
		2232	See the C<splice(2)> manpage for details.
		2233
		2234	=item IO::AIO::tee $r_fh, $w_fh, $length, $flags
		2235
		2236	Calls the GNU/Linux C<tee(2)> syscall, see its manpage and the
		2237	description for C<IO::AIO::splice> above for details.
		2238
		2239	=item $actual_size = IO::AIO::pipesize $r_fh[, $new_size]
		2240
		2241	Attempts to query or change the pipe buffer size. Obviously works only
		2242	on pipes, and currently works only on GNU/Linux systems, and fails with
		2243	C<-1>/C<ENOSYS> everywhere else. If anybody knows how to influence pipe buffer
		2244	size on other systems, drop me a note.
		2245
		2246	=item ($rfh, $wfh) = IO::AIO::pipe2 [$flags]
		2247
		2248	This is a direct interface to the Linux L<pipe2(2)> system call. If
		2249	C<$flags> is missing or C<0>, then this should be the same as a call to
		2250	perl's built-in C<pipe> function and create a new pipe, and works on
		2251	systems that lack the pipe2 syscall. On win32, this case invokes C<_pipe
		2252	(..., 4096, O_BINARY)>.
		2253
		2254	If C<$flags> is non-zero, it tries to invoke the pipe2 system call with
		2255	the given flags (Linux 2.6.27, glibc 2.9).
		2256
		2257	On success, the read and write file handles are returned.
		2258
		2259	On error, nothing will be returned. If the pipe2 syscall is missing and
		2260	C<$flags> is non-zero, fails with C<ENOSYS>.
		2261
		2262	Please refer to L<pipe2(2)> for more info on the C<$flags>, but at the
		2263	time of this writing, C<IO::AIO::O_CLOEXEC>, C<IO::AIO::O_NONBLOCK> and
		2264	C<IO::AIO::O_DIRECT> (Linux 3.4, for packet-based pipes) were supported.
		2265
		2266	=back
		2267
750	=cut	2268	=cut
751		2269
752	# support function to convert a fd into a perl filehandle
753	sub _fd2fh {
754	return undef if $_[0] < 0;
755
756	# try to generate nice filehandles
757	my $sym = "IO::AIO::fd#$_[0]";
758	local *$sym;
759
760	open *$sym, "+<&=$_[0]" # usually works under any unix
761	or open *$sym, "<&=$_[0]" # cygwin needs this
762	or open *$sym, ">&=$_[0]" # or this
763	or return undef;
764
765	*$sym
766	}
767
768	min_parallel 4;	2270	min_parallel 8;
769		2271
770	END {	2272	END { flush }
771	max_parallel 0;
772	}
773		2273
774	1;	2274	1;
775		2275
		2276	=head1 EVENT LOOP INTEGRATION
		2277
		2278	It is recommended to use L<AnyEvent::AIO> to integrate IO::AIO
		2279	automatically into many event loops:
		2280
		2281	# AnyEvent integration (EV, Event, Glib, Tk, POE, urxvt, pureperl...)
		2282	use AnyEvent::AIO;
		2283
		2284	You can also integrate IO::AIO manually into many event loops, here are
		2285	some examples of how to do this:
		2286
		2287	# EV integration
		2288	my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
		2289
		2290	# Event integration
		2291	Event->io (fd => IO::AIO::poll_fileno,
		2292	poll => 'r',
		2293	cb => \&IO::AIO::poll_cb);
		2294
		2295	# Glib/Gtk2 integration
		2296	add_watch Glib::IO IO::AIO::poll_fileno,
		2297	in => sub { IO::AIO::poll_cb; 1 };
		2298
		2299	# Tk integration
		2300	Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",
		2301	readable => \&IO::AIO::poll_cb);
		2302
		2303	# Danga::Socket integration
		2304	Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
		2305	\&IO::AIO::poll_cb);
		2306
776	=head2 FORK BEHAVIOUR	2307	=head2 FORK BEHAVIOUR
777		2308
778	This module should do "the right thing" when the process using it forks:	2309	Usage of pthreads in a program changes the semantics of fork
		2310	considerably. Specifically, only async-safe functions can be called after
		2311	fork. Perl doesn't know about this, so in general, you cannot call fork
		2312	with defined behaviour in perl if pthreads are involved. IO::AIO uses
		2313	pthreads, so this applies, but many other extensions and (for inexplicable
		2314	reasons) perl itself often is linked against pthreads, so this limitation
		2315	applies to quite a lot of perls.
779		2316
780	Before the fork, IO::AIO enters a quiescent state where no requests	2317	This module no longer tries to fight your OS, or POSIX. That means IO::AIO
781	can be added in other threads and no results will be processed. After	2318	only works in the process that loaded it. Forking is fully supported, but
782	the fork the parent simply leaves the quiescent state and continues	2319	using IO::AIO in the child is not.
783	request/result processing, while the child clears the request/result
784	queue (so the requests started before the fork will only be handled in
785	the parent). Threads will be started on demand until the limit ste in the
786	parent process has been reached again.
787		2320
788	In short: the parent will, after a short pause, continue as if fork had	2321	You might get around by not I<using> IO::AIO before (or after)
789	not been called, while the child will act as if IO::AIO has not been used	2322	forking. You could also try to call the L<IO::AIO::reinit> function in the
790	yet.	2323	child:
		2324
		2325	=over 4
		2326
		2327	=item IO::AIO::reinit
		2328
		2329	Abandons all current requests and I/O threads and simply reinitialises all
		2330	data structures. This is not an operation supported by any standards, but
		2331	happens to work on GNU/Linux and some newer BSD systems.
		2332
		2333	The only reasonable use for this function is to call it after forking, if
		2334	C<IO::AIO> was used in the parent. Calling it while IO::AIO is active in
		2335	the process will result in undefined behaviour. Calling it at any time
		2336	will also result in any undefined (by POSIX) behaviour.
		2337
		2338	=back
		2339
		2340	=head2 MEMORY USAGE
		2341
		2342	Per-request usage:
		2343
		2344	Each aio request uses - depending on your architecture - around 100-200
		2345	bytes of memory. In addition, stat requests need a stat buffer (possibly
		2346	a few hundred bytes), readdir requires a result buffer and so on. Perl
		2347	scalars and other data passed into aio requests will also be locked and
		2348	will consume memory till the request has entered the done state.
		2349
		2350	This is not awfully much, so queuing lots of requests is not usually a
		2351	problem.
		2352
		2353	Per-thread usage:
		2354
		2355	In the execution phase, some aio requests require more memory for
		2356	temporary buffers, and each thread requires a stack and other data
		2357	structures (usually around 16k-128k, depending on the OS).
		2358
		2359	=head1 KNOWN BUGS
		2360
		2361	Known bugs will be fixed in the next release.
791		2362
792	=head1 SEE ALSO	2363	=head1 SEE ALSO
793		2364
794	L<Coro>, L<Linux::AIO> (obsolete).	2365	L<AnyEvent::AIO> for easy integration into event loops, L<Coro::AIO> for a
		2366	more natural syntax.
795		2367
796	=head1 AUTHOR	2368	=head1 AUTHOR
797		2369
798	Marc Lehmann <schmorp@schmorp.de>	2370	Marc Lehmann <schmorp@schmorp.de>
799	http://home.schmorp.de/	2371	http://home.schmorp.de/

Diff Legend

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