ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/IO-AIO/AIO.pm

Comparing IO-AIO/AIO.pm (file contents):
Revision 1.2 by root, Sun Jul 10 18:16:49 2005 UTC vs.
Revision 1.203 by root, Thu Jul 7 22:36:18 2011 UTC

…		…
4		4
5	=head1 SYNOPSIS	5	=head1 SYNOPSIS
6		6
7	use IO::AIO;	7	use IO::AIO;
8		8
		9	aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub {
		10	my $fh = shift
		11	or die "/etc/passwd: $!";
		12	...
		13	};
		14
		15	aio_unlink "/tmp/file", sub { };
		16
		17	aio_read $fh, 30000, 1024, $buffer, 0, sub {
		18	$_[0] > 0 or die "read error: $!";
		19	};
		20
		21	# version 2+ has request and group objects
		22	use IO::AIO 2;
		23
		24	aioreq_pri 4; # give next request a very high priority
		25	my $req = aio_unlink "/tmp/file", sub { };
		26	$req->cancel; # cancel request if still in queue
		27
		28	my $grp = aio_group sub { print "all stats done\n" };
		29	add $grp aio_stat "..." for ...;
		30
9	=head1 DESCRIPTION	31	=head1 DESCRIPTION
10		32
11	This module implements asynchronous I/O using whatever means your	33	This module implements asynchronous I/O using whatever means your
12	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>).
13		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
14	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
15	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
16	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
17	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
18	aio functions available on many operating systems. However, they are often	57	functions available on many operating systems. However, they are often
19	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
20	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
21	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.
22		62
23	Although the module will work with in the presence of other threads, it is	63	Although the module will work in the presence of other (Perl-) threads,
24	currently not reentrant, so use appropriate locking yourself.	64	it is currently not reentrant in any way, so use appropriate locking
		65	yourself, always call C<poll_cb> from within the same thread, or never
		66	call C<poll_cb> (or other C<aio_> functions) recursively.
25		67
26	=head2 API NOTES	68	=head2 EXAMPLE
		69
		70	This is a simple example that uses the EV module and loads
		71	F</etc/passwd> asynchronously:
		72
		73	use Fcntl;
		74	use EV;
		75	use IO::AIO;
		76
		77	# register the IO::AIO callback with EV
		78	my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
		79
		80	# queue the request to open /etc/passwd
		81	aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub {
		82	my $fh = shift
		83	or die "error while opening: $!";
		84
		85	# stat'ing filehandles is generally non-blocking
		86	my $size = -s $fh;
		87
		88	# queue a request to read the file
		89	my $contents;
		90	aio_read $fh, 0, $size, $contents, 0, sub {
		91	$_[0] == $size
		92	or die "short read: $!";
		93
		94	close $fh;
		95
		96	# file contents now in $contents
		97	print $contents;
		98
		99	# exit event loop and program
		100	EV::unloop;
		101	};
		102	};
		103
		104	# possibly queue up other requests, or open GUI windows,
		105	# check for sockets etc. etc.
		106
		107	# process events as long as there are some:
		108	EV::loop;
		109
		110	=head1 REQUEST ANATOMY AND LIFETIME
		111
		112	Every C<aio_*> function creates a request. which is a C data structure not
		113	directly visible to Perl.
		114
		115	If called in non-void context, every request function returns a Perl
		116	object representing the request. In void context, nothing is returned,
		117	which saves a bit of memory.
		118
		119	The perl object is a fairly standard ref-to-hash object. The hash contents
		120	are not used by IO::AIO so you are free to store anything you like in it.
		121
		122	During their existance, aio requests travel through the following states,
		123	in order:
		124
		125	=over 4
		126
		127	=item ready
		128
		129	Immediately after a request is created it is put into the ready state,
		130	waiting for a thread to execute it.
		131
		132	=item execute
		133
		134	A thread has accepted the request for processing and is currently
		135	executing it (e.g. blocking in read).
		136
		137	=item pending
		138
		139	The request has been executed and is waiting for result processing.
		140
		141	While request submission and execution is fully asynchronous, result
		142	processing is not and relies on the perl interpreter calling C<poll_cb>
		143	(or another function with the same effect).
		144
		145	=item result
		146
		147	The request results are processed synchronously by C<poll_cb>.
		148
		149	The C<poll_cb> function will process all outstanding aio requests by
		150	calling their callbacks, freeing memory associated with them and managing
		151	any groups they are contained in.
		152
		153	=item done
		154
		155	Request has reached the end of its lifetime and holds no resources anymore
		156	(except possibly for the Perl object, but its connection to the actual
		157	aio request is severed and calling its methods will either do nothing or
		158	result in a runtime error).
		159
		160	=back
		161
		162	=cut
		163
		164	package IO::AIO;
		165
		166	use Carp ();
		167
		168	use common::sense;
		169
		170	use base 'Exporter';
		171
		172	BEGIN {
		173	our $VERSION = '3.93';
		174
		175	our @AIO_REQ = qw(aio_sendfile aio_read aio_write aio_open aio_close
		176	aio_stat aio_lstat aio_unlink aio_rmdir aio_readdir aio_readdirx
		177	aio_scandir aio_symlink aio_readlink aio_realpath aio_sync aio_fsync
		178	aio_fdatasync aio_sync_file_range aio_fallocate
		179	aio_pathsync aio_readahead
		180	aio_rename aio_link aio_move aio_copy aio_group
		181	aio_nop aio_mknod aio_load aio_rmtree aio_mkdir aio_chown
		182	aio_chmod aio_utime aio_truncate
		183	aio_msync aio_mtouch aio_mlock aio_mlockall
		184	aio_statvfs);
		185
		186	our @EXPORT = (@AIO_REQ, qw(aioreq_pri aioreq_nice));
		187	our @EXPORT_OK = qw(poll_fileno poll_cb poll_wait flush
		188	min_parallel max_parallel max_idle idle_timeout
		189	nreqs nready npending nthreads
		190	max_poll_time max_poll_reqs
		191	sendfile fadvise madvise
		192	mmap munmap munlock munlockall);
		193
		194	push @AIO_REQ, qw(aio_busy); # not exported
		195
		196	@IO::AIO::GRP::ISA = 'IO::AIO::REQ';
		197
		198	require XSLoader;
		199	XSLoader::load ("IO::AIO", $VERSION);
		200	}
		201
		202	=head1 FUNCTIONS
		203
		204	=head2 QUICK OVERVIEW
		205
		206	This section simply lists the prototypes of the most important functions
		207	for quick reference. See the following sections for function-by-function
		208	documentation.
		209
		210	aio_open $pathname, $flags, $mode, $callback->($fh)
		211	aio_close $fh, $callback->($status)
		212	aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
		213	aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
		214	aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
		215	aio_readahead $fh,$offset,$length, $callback->($retval)
		216	aio_stat $fh_or_path, $callback->($status)
		217	aio_lstat $fh, $callback->($status)
		218	aio_statvfs $fh_or_path, $callback->($statvfs)
		219	aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
		220	aio_chown $fh_or_path, $uid, $gid, $callback->($status)
		221	aio_truncate $fh_or_path, $offset, $callback->($status)
		222	aio_chmod $fh_or_path, $mode, $callback->($status)
		223	aio_unlink $pathname, $callback->($status)
		224	aio_mknod $path, $mode, $dev, $callback->($status)
		225	aio_link $srcpath, $dstpath, $callback->($status)
		226	aio_symlink $srcpath, $dstpath, $callback->($status)
		227	aio_readlink $path, $callback->($link)
		228	aio_realpath $path, $callback->($link)
		229	aio_rename $srcpath, $dstpath, $callback->($status)
		230	aio_mkdir $pathname, $mode, $callback->($status)
		231	aio_rmdir $pathname, $callback->($status)
		232	aio_readdir $pathname, $callback->($entries)
		233	aio_readdirx $pathname, $flags, $callback->($entries, $flags)
		234	IO::AIO::READDIR_DENTS IO::AIO::READDIR_DIRS_FIRST
		235	IO::AIO::READDIR_STAT_ORDER IO::AIO::READDIR_FOUND_UNKNOWN
		236	aio_load $path, $data, $callback->($status)
		237	aio_copy $srcpath, $dstpath, $callback->($status)
		238	aio_move $srcpath, $dstpath, $callback->($status)
		239	aio_scandir $path, $maxreq, $callback->($dirs, $nondirs)
		240	aio_rmtree $path, $callback->($status)
		241	aio_sync $callback->($status)
		242	aio_fsync $fh, $callback->($status)
		243	aio_fdatasync $fh, $callback->($status)
		244	aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
		245	aio_pathsync $path, $callback->($status)
		246	aio_msync $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		247	aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		248	aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status)
		249	aio_mlockall $flags, $callback->($status)
		250	aio_group $callback->(...)
		251	aio_nop $callback->()
		252
		253	$prev_pri = aioreq_pri [$pri]
		254	aioreq_nice $pri_adjust
		255
		256	IO::AIO::poll_wait
		257	IO::AIO::poll_cb
		258	IO::AIO::poll
		259	IO::AIO::flush
		260	IO::AIO::max_poll_reqs $nreqs
		261	IO::AIO::max_poll_time $seconds
		262	IO::AIO::min_parallel $nthreads
		263	IO::AIO::max_parallel $nthreads
		264	IO::AIO::max_idle $nthreads
		265	IO::AIO::idle_timeout $seconds
		266	IO::AIO::max_outstanding $maxreqs
		267	IO::AIO::nreqs
		268	IO::AIO::nready
		269	IO::AIO::npending
		270
		271	IO::AIO::sendfile $ofh, $ifh, $offset, $count
		272	IO::AIO::fadvise $fh, $offset, $len, $advice
		273	IO::AIO::madvise $scalar, $offset, $length, $advice
		274	IO::AIO::mprotect $scalar, $offset, $length, $protect
		275	IO::AIO::munlock $scalar, $offset = 0, $length = undef
		276	IO::AIO::munlockall
		277
		278	=head2 AIO REQUEST FUNCTIONS
27		279
28	All the C<aio_*> calls are more or less thin wrappers around the syscall	280	All the C<aio_*> calls are more or less thin wrappers around the syscall
29	with the same name (sans C<aio_>). The arguments are similar or identical,	281	with the same name (sans C<aio_>). The arguments are similar or identical,
30	and they all accept an additional C<$callback> argument which must be	282	and they all accept an additional (and optional) C<$callback> argument
31	a code reference. This code reference will get called with the syscall	283	which must be a code reference. This code reference will get called with
32	return code (e.g. most syscalls return C<-1> on error, unlike perl, which	284	the syscall return code (e.g. most syscalls return C<-1> on error, unlike
33	usually delivers "false") as it's sole argument when the given syscall has	285	perl, which usually delivers "false") as its sole argument after the given
34	been executed asynchronously.	286	syscall has been executed asynchronously.
35		287
36	All functions that expect a filehandle will also accept a file descriptor.	288	All functions expecting a filehandle keep a copy of the filehandle
		289	internally until the request has finished.
37		290
		291	All functions return request objects of type L<IO::AIO::REQ> that allow
		292	further manipulation of those requests while they are in-flight.
		293
38	The filenames you pass to these routines I<must> be absolute. The reason	294	The pathnames you pass to these routines I<must> be absolute and
39	is that at the time the request is being executed, the current working	295	encoded as octets. The reason for the former is that at the time the
40	directory could have changed. Alternatively, you can make sure that you	296	request is being executed, the current working directory could have
41	never change the current working directory.	297	changed. Alternatively, you can make sure that you never change the
		298	current working directory anywhere in the program and then use relative
		299	paths.
		300
		301	To encode pathnames as octets, either make sure you either: a) always pass
		302	in filenames you got from outside (command line, readdir etc.) without
		303	tinkering, b) are ASCII or ISO 8859-1, c) use the Encode module and encode
		304	your pathnames to the locale (or other) encoding in effect in the user
		305	environment, d) use Glib::filename_from_unicode on unicode filenames or e)
		306	use something else to ensure your scalar has the correct contents.
		307
		308	This works, btw. independent of the internal UTF-8 bit, which IO::AIO
		309	handles correctly whether it is set or not.
42		310
43	=over 4	311	=over 4
44		312
45	=cut	313	=item $prev_pri = aioreq_pri [$pri]
46		314
47	package IO::AIO;	315	Returns the priority value that would be used for the next request and, if
		316	C<$pri> is given, sets the priority for the next aio request.
48		317
49	use base 'Exporter';	318	The default priority is C<0>, the minimum and maximum priorities are C<-4>
		319	and C<4>, respectively. Requests with higher priority will be serviced
		320	first.
50		321
51	use Fcntl ();	322	The priority will be reset to C<0> after each call to one of the C<aio_*>
		323	functions.
52		324
53	BEGIN {	325	Example: open a file with low priority, then read something from it with
54	$VERSION = 0.1;	326	higher priority so the read request is serviced before other low priority
		327	open requests (potentially spamming the cache):
55		328
56	@EXPORT = qw(aio_read aio_write aio_open aio_close aio_stat aio_lstat aio_unlink	329	aioreq_pri -3;
57	aio_fsync aio_fdatasync aio_readahead);	330	aio_open ..., sub {
58	@EXPORT_OK = qw(poll_fileno poll_cb min_parallel max_parallel nreqs);	331	return unless $_[0];
59		332
60	require XSLoader;	333	aioreq_pri -2;
61	XSLoader::load IO::AIO, $VERSION;	334	aio_read $_[0], ..., sub {
62	}	335	...
		336	};
		337	};
63		338
64	=item IO::AIO::min_parallel $nthreads
65		339
66	Set the minimum number of AIO threads to C<$nthreads>. The default is	340	=item aioreq_nice $pri_adjust
67	C<1>, which means a single asynchronous operation can be done at one time
68	(the number of outstanding operations, however, is unlimited).
69		341
70	It is recommended to keep the number of threads low, as some linux	342	Similar to C<aioreq_pri>, but subtracts the given value from the current
71	kernel versions will scale negatively with the number of threads (higher	343	priority, so the effect is cumulative.
72	parallelity => MUCH higher latency).
73		344
74	Under normal circumstances you don't need to call this function, as this
75	module automatically starts a single async thread.
76		345
77	=item IO::AIO::max_parallel $nthreads
78
79	Sets the maximum number of AIO threads to C<$nthreads>. If more than
80	the specified number of threads are currently running, kill them. This
81	function blocks until the limit is reached.
82
83	This module automatically runs C<max_parallel 0> at program end, to ensure
84	that all threads are killed and that there are no outstanding requests.
85
86	Under normal circumstances you don't need to call this function.
87
88	=item $fileno = IO::AIO::poll_fileno
89
90	Return the I<request result pipe filehandle>. This filehandle must be
91	polled for reading by some mechanism outside this module (e.g. Event
92	or select, see below). If the pipe becomes readable you have to call
93	C<poll_cb> to check the results.
94
95	See C<poll_cb> for an example.
96
97	=item IO::AIO::poll_cb
98
99	Process all outstanding events on the result pipe. You have to call this
100	regularly. Returns the number of events processed. Returns immediately
101	when no events are outstanding.
102
103	You can use Event to multiplex, e.g.:
104
105	Event->io (fd => IO::AIO::poll_fileno,
106	poll => 'r', async => 1,
107	cb => \&IO::AIO::poll_cb);
108
109	=item IO::AIO::poll_wait
110
111	Wait till the result filehandle becomes ready for reading (simply does a
112	select on the filehandle. This is useful if you want to synchronously wait
113	for some requests to finish).
114
115	See C<nreqs> for an example.
116
117	=item IO::AIO::nreqs
118
119	Returns the number of requests currently outstanding.
120
121	Example: wait till there are no outstanding requests anymore:
122
123	IO::AIO::poll_wait, IO::AIO::poll_cb
124	while IO::AIO::nreqs;
125
126	=item aio_open $pathname, $flags, $mode, $callback	346	=item aio_open $pathname, $flags, $mode, $callback->($fh)
127		347
128	Asynchronously open or create a file and call the callback with a newly	348	Asynchronously open or create a file and call the callback with a newly
129	created filehandle for the file.	349	created filehandle for the file.
130		350
131	The pathname passed to C<aio_open> must be absolute. See API NOTES, above,	351	The pathname passed to C<aio_open> must be absolute. See API NOTES, above,
132	for an explanation.	352	for an explanation.
133		353
134	The C<$mode> argument is a bitmask. See the C<Fcntl> module for a	354	The C<$flags> argument is a bitmask. See the C<Fcntl> module for a
135	list. They are the same as used in C<sysopen>.	355	list. They are the same as used by C<sysopen>.
		356
		357	Likewise, C<$mode> specifies the mode of the newly created file, if it
		358	didn't exist and C<O_CREAT> has been given, just like perl's C<sysopen>,
		359	except that it is mandatory (i.e. use C<0> if you don't create new files,
		360	and C<0666> or C<0777> if you do). Note that the C<$mode> will be modified
		361	by the umask in effect then the request is being executed, so better never
		362	change the umask.
136		363
137	Example:	364	Example:
138		365
139	aio_open "/etc/passwd", O_RDONLY, 0, sub {	366	aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub {
140	if ($_[0]) {	367	if ($_[0]) {
141	print "open successful, fh is $_[0]\n";	368	print "open successful, fh is $_[0]\n";
142	...	369	...
143	} else {	370	} else {
144	die "open failed: $!\n";	371	die "open failed: $!\n";
145	}	372	}
146	};	373	};
147		374
		375	In addition to all the common open modes/flags (C<O_RDONLY>, C<O_WRONLY>,
		376	C<O_RDWR>, C<O_CREAT>, C<O_TRUNC>, C<O_EXCL> and C<O_APPEND>), the
		377	following POSIX and non-POSIX constants are available (missing ones on
		378	your system are, as usual, C<0>):
		379
		380	C<O_ASYNC>, C<O_DIRECT>, C<O_NOATIME>, C<O_CLOEXEC>, C<O_NOCTTY>, C<O_NOFOLLOW>,
		381	C<O_NONBLOCK>, C<O_EXEC>, C<O_SEARCH>, C<O_DIRECTORY>, C<O_DSYNC>,
		382	C<O_RSYNC>, C<O_SYNC> and C<O_TTY_INIT>.
		383
		384
148	=item aio_close $fh, $callback	385	=item aio_close $fh, $callback->($status)
149		386
150	Asynchronously close a file and call the callback with the result	387	Asynchronously close a file and call the callback with the result
151	code. I<WARNING:> although accepted, you should not pass in a perl	388	code.
152	filehandle here, as perl will likely close the file descriptor itself when
153	the filehandle is destroyed. Normally, you can safely call perls C<close>
154	or just let filehandles go out of scope.
155		389
		390	Unfortunately, you can't do this to perl. Perl I<insists> very strongly on
		391	closing the file descriptor associated with the filehandle itself.
		392
		393	Therefore, C<aio_close> will not close the filehandle - instead it will
		394	use dup2 to overwrite the file descriptor with the write-end of a pipe
		395	(the pipe fd will be created on demand and will be cached).
		396
		397	Or in other words: the file descriptor will be closed, but it will not be
		398	free for reuse until the perl filehandle is closed.
		399
		400	=cut
		401
156	=item aio_read $fh,$offset,$length, $data,$dataoffset,$callback	402	=item aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
157		403
158	=item aio_write $fh,$offset,$length, $data,$dataoffset,$callback	404	=item aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
159		405
160	Reads or writes C<length> bytes from the specified C<fh> and C<offset>	406	Reads or writes C<$length> bytes from or to the specified C<$fh> and
161	into the scalar given by C<data> and offset C<dataoffset> and calls the	407	C<$offset> into the scalar given by C<$data> and offset C<$dataoffset>
162	callback without the actual number of bytes read (or -1 on error, just	408	and calls the callback without the actual number of bytes read (or -1 on
163	like the syscall).	409	error, just like the syscall).
164		410
		411	C<aio_read> will, like C<sysread>, shrink or grow the C<$data> scalar to
		412	offset plus the actual number of bytes read.
		413
		414	If C<$offset> is undefined, then the current file descriptor offset will
		415	be used (and updated), otherwise the file descriptor offset will not be
		416	changed by these calls.
		417
		418	If C<$length> is undefined in C<aio_write>, use the remaining length of
		419	C<$data>.
		420
		421	If C<$dataoffset> is less than zero, it will be counted from the end of
		422	C<$data>.
		423
		424	The C<$data> scalar I<MUST NOT> be modified in any way while the request
		425	is outstanding. Modifying it can result in segfaults or World War III (if
		426	the necessary/optional hardware is installed).
		427
165	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, strating at	428	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at
166	offset C<0> within the scalar:	429	offset C<0> within the scalar:
167		430
168	aio_read $fh, 7, 15, $buffer, 0, sub {	431	aio_read $fh, 7, 15, $buffer, 0, sub {
169	$_[0] >= 0 or die "read error: $!";	432	$_[0] > 0 or die "read error: $!";
170	print "read <$buffer>\n";	433	print "read $_[0] bytes: <$buffer>\n";
171	};	434	};
172		435
		436
		437	=item aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
		438
		439	Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts
		440	reading at byte offset C<$in_offset>, and starts writing at the current
		441	file offset of C<$out_fh>. Because of that, it is not safe to issue more
		442	than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each
		443	other. The same C<$in_fh> works fine though, as this function does not
		444	move or use the file offset of C<$in_fh>.
		445
		446	Please note that C<aio_sendfile> can read more bytes from C<$in_fh> than
		447	are written, and there is no way to find out how many more bytes have been
		448	read from C<aio_sendfile> alone, as C<aio_sendfile> only provides the
		449	number of bytes written to C<$out_fh>. Only if the result value equals
		450	C<$length> one can assume that C<$length> bytes have been read.
		451
		452	Unlike with other C<aio_> functions, it makes a lot of sense to use
		453	C<aio_sendfile> on non-blocking sockets, as long as one end (typically
		454	the C<$in_fh>) is a file - the file I/O will then be asynchronous, while
		455	the socket I/O will be non-blocking. Note, however, that you can run
		456	into a trap where C<aio_sendfile> reads some data with readahead, then
		457	fails to write all data, and when the socket is ready the next time, the
		458	data in the cache is already lost, forcing C<aio_sendfile> to again hit
		459	the disk. Explicit C<aio_read> + C<aio_write> let's you better control
		460	resource usage.
		461
		462	This call tries to make use of a native C<sendfile>-like syscall to
		463	provide zero-copy operation. For this to work, C<$out_fh> should refer to
		464	a socket, and C<$in_fh> should refer to an mmap'able file.
		465
		466	If a native sendfile cannot be found or it fails with C<ENOSYS>,
		467	C<EINVAL>, C<ENOTSUP>, C<EOPNOTSUPP>, C<EAFNOSUPPORT>, C<EPROTOTYPE> or
		468	C<ENOTSOCK>, it will be emulated, so you can call C<aio_sendfile> on any
		469	type of filehandle regardless of the limitations of the operating system.
		470
		471	As native sendfile syscalls (as practically any non-POSIX interface hacked
		472	together in a hurry to improve benchmark numbers) tend to be rather buggy
		473	on many systems, this implementation tries to work around some known bugs
		474	in Linux and FreeBSD kernels (probably others, too), but that might fail,
		475	so you really really should check the return value of C<aio_sendfile> -
		476	fewre bytes than expected might have been transferred.
		477
		478
173	=item aio_readahead $fh,$offset,$length, $callback	479	=item aio_readahead $fh,$offset,$length, $callback->($retval)
174		480
175	Asynchronously reads the specified byte range into the page cache, using
176	the C<readahead> syscall. If that syscall doesn't exist the status will be
177	C<-1> and C<$!> is set to ENOSYS.
178
179	readahead() populates the page cache with data from a file so that	481	C<aio_readahead> populates the page cache with data from a file so that
180	subsequent reads from that file will not block on disk I/O. The C<$offset>	482	subsequent reads from that file will not block on disk I/O. The C<$offset>
181	argument specifies the starting point from which data is to be read and	483	argument specifies the starting point from which data is to be read and
182	C<$length> specifies the number of bytes to be read. I/O is performed in	484	C<$length> specifies the number of bytes to be read. I/O is performed in
183	whole pages, so that offset is effectively rounded down to a page boundary	485	whole pages, so that offset is effectively rounded down to a page boundary
184	and bytes are read up to the next page boundary greater than or equal to	486	and bytes are read up to the next page boundary greater than or equal to
185	(off-set+length). aio_readahead() does not read beyond the end of the	487	(off-set+length). C<aio_readahead> does not read beyond the end of the
186	file. The current file offset of the file is left unchanged.	488	file. The current file offset of the file is left unchanged.
187		489
		490	If that syscall doesn't exist (likely if your OS isn't Linux) it will be
		491	emulated by simply reading the data, which would have a similar effect.
		492
		493
188	=item aio_stat $fh_or_path, $callback	494	=item aio_stat $fh_or_path, $callback->($status)
189		495
190	=item aio_lstat $fh, $callback	496	=item aio_lstat $fh, $callback->($status)
191		497
192	Works like perl's C<stat> or C<lstat> in void context. The callback will	498	Works like perl's C<stat> or C<lstat> in void context. The callback will
193	be called after the stat and the results will be available using C<stat _>	499	be called after the stat and the results will be available using C<stat _>
194	or C<-s _> etc...	500	or C<-s _> etc...
195		501
…		…
197	for an explanation.	503	for an explanation.
198		504
199	Currently, the stats are always 64-bit-stats, i.e. instead of returning an	505	Currently, the stats are always 64-bit-stats, i.e. instead of returning an
200	error when stat'ing a large file, the results will be silently truncated	506	error when stat'ing a large file, the results will be silently truncated
201	unless perl itself is compiled with large file support.	507	unless perl itself is compiled with large file support.
		508
		509	To help interpret the mode and dev/rdev stat values, IO::AIO offers the
		510	following constants and functions (if not implemented, the constants will
		511	be C<0> and the functions will either C<croak> or fall back on traditional
		512	behaviour).
		513
		514	C<S_IFMT>, C<S_IFIFO>, C<S_IFCHR>, C<S_IFBLK>, C<S_IFLNK>, C<S_IFREG>,
		515	C<S_IFDIR>, C<S_IFWHT>, C<S_IFSOCK>, C<IO::AIO::major $dev_t>,
		516	C<IO::AIO::minor $dev_t>, C<IO::AIO::makedev $major, $minor>.
202		517
203	Example: Print the length of F</etc/passwd>:	518	Example: Print the length of F</etc/passwd>:
204		519
205	aio_stat "/etc/passwd", sub {	520	aio_stat "/etc/passwd", sub {
206	$_[0] and die "stat failed: $!";	521	$_[0] and die "stat failed: $!";
207	print "size is ", -s _, "\n";	522	print "size is ", -s _, "\n";
208	};	523	};
209		524
		525
		526	=item aio_statvfs $fh_or_path, $callback->($statvfs)
		527
		528	Works like the POSIX C<statvfs> or C<fstatvfs> syscalls, depending on
		529	whether a file handle or path was passed.
		530
		531	On success, the callback is passed a hash reference with the following
		532	members: C<bsize>, C<frsize>, C<blocks>, C<bfree>, C<bavail>, C<files>,
		533	C<ffree>, C<favail>, C<fsid>, C<flag> and C<namemax>. On failure, C<undef>
		534	is passed.
		535
		536	The following POSIX IO::AIO::ST_* constants are defined: C<ST_RDONLY> and
		537	C<ST_NOSUID>.
		538
		539	The following non-POSIX IO::AIO::ST_* flag masks are defined to
		540	their correct value when available, or to C<0> on systems that do
		541	not support them: C<ST_NODEV>, C<ST_NOEXEC>, C<ST_SYNCHRONOUS>,
		542	C<ST_MANDLOCK>, C<ST_WRITE>, C<ST_APPEND>, C<ST_IMMUTABLE>, C<ST_NOATIME>,
		543	C<ST_NODIRATIME> and C<ST_RELATIME>.
		544
		545	Example: stat C</wd> and dump out the data if successful.
		546
		547	aio_statvfs "/wd", sub {
		548	my $f = $_[0]
		549	or die "statvfs: $!";
		550
		551	use Data::Dumper;
		552	say Dumper $f;
		553	};
		554
		555	# result:
		556	{
		557	bsize => 1024,
		558	bfree => 4333064312,
		559	blocks => 10253828096,
		560	files => 2050765568,
		561	flag => 4096,
		562	favail => 2042092649,
		563	bavail => 4333064312,
		564	ffree => 2042092649,
		565	namemax => 255,
		566	frsize => 1024,
		567	fsid => 1810
		568	}
		569
		570
		571	=item aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
		572
		573	Works like perl's C<utime> function (including the special case of $atime
		574	and $mtime being undef). Fractional times are supported if the underlying
		575	syscalls support them.
		576
		577	When called with a pathname, uses utimes(2) if available, otherwise
		578	utime(2). If called on a file descriptor, uses futimes(2) if available,
		579	otherwise returns ENOSYS, so this is not portable.
		580
		581	Examples:
		582
		583	# set atime and mtime to current time (basically touch(1)):
		584	aio_utime "path", undef, undef;
		585	# set atime to current time and mtime to beginning of the epoch:
		586	aio_utime "path", time, undef; # undef==0
		587
		588
		589	=item aio_chown $fh_or_path, $uid, $gid, $callback->($status)
		590
		591	Works like perl's C<chown> function, except that C<undef> for either $uid
		592	or $gid is being interpreted as "do not change" (but -1 can also be used).
		593
		594	Examples:
		595
		596	# same as "chown root path" in the shell:
		597	aio_chown "path", 0, -1;
		598	# same as above:
		599	aio_chown "path", 0, undef;
		600
		601
		602	=item aio_truncate $fh_or_path, $offset, $callback->($status)
		603
		604	Works like truncate(2) or ftruncate(2).
		605
		606
		607	=item aio_chmod $fh_or_path, $mode, $callback->($status)
		608
		609	Works like perl's C<chmod> function.
		610
		611
210	=item aio_unlink $pathname, $callback	612	=item aio_unlink $pathname, $callback->($status)
211		613
212	Asynchronously unlink (delete) a file and call the callback with the	614	Asynchronously unlink (delete) a file and call the callback with the
213	result code.	615	result code.
214		616
		617
		618	=item aio_mknod $path, $mode, $dev, $callback->($status)
		619
		620	[EXPERIMENTAL]
		621
		622	Asynchronously create a device node (or fifo). See mknod(2).
		623
		624	The only (POSIX-) portable way of calling this function is:
		625
		626	aio_mknod $path, IO::AIO::S_IFIFO | $mode, 0, sub { ...
		627
		628	See C<aio_stat> for info about some potentially helpful extra constants
		629	and functions.
		630
		631	=item aio_link $srcpath, $dstpath, $callback->($status)
		632
		633	Asynchronously create a new link to the existing object at C<$srcpath> at
		634	the path C<$dstpath> and call the callback with the result code.
		635
		636
		637	=item aio_symlink $srcpath, $dstpath, $callback->($status)
		638
		639	Asynchronously create a new symbolic link to the existing object at C<$srcpath> at
		640	the path C<$dstpath> and call the callback with the result code.
		641
		642
		643	=item aio_readlink $path, $callback->($link)
		644
		645	Asynchronously read the symlink specified by C<$path> and pass it to
		646	the callback. If an error occurs, nothing or undef gets passed to the
		647	callback.
		648
		649
		650	=item aio_realpath $path, $callback->($path)
		651
		652	Asynchronously make the path absolute and resolve any symlinks in
		653	C<$path>. The resulting path only consists of directories (Same as
		654	L<Cwd::realpath>).
		655
		656	This request can be used to get the absolute path of the current working
		657	directory by passing it a path of F<.> (a single dot).
		658
		659
		660	=item aio_rename $srcpath, $dstpath, $callback->($status)
		661
		662	Asynchronously rename the object at C<$srcpath> to C<$dstpath>, just as
		663	rename(2) and call the callback with the result code.
		664
		665
		666	=item aio_mkdir $pathname, $mode, $callback->($status)
		667
		668	Asynchronously mkdir (create) a directory and call the callback with
		669	the result code. C<$mode> will be modified by the umask at the time the
		670	request is executed, so do not change your umask.
		671
		672
		673	=item aio_rmdir $pathname, $callback->($status)
		674
		675	Asynchronously rmdir (delete) a directory and call the callback with the
		676	result code.
		677
		678
		679	=item aio_readdir $pathname, $callback->($entries)
		680
		681	Unlike the POSIX call of the same name, C<aio_readdir> reads an entire
		682	directory (i.e. opendir + readdir + closedir). The entries will not be
		683	sorted, and will B<NOT> include the C<.> and C<..> entries.
		684
		685	The callback is passed a single argument which is either C<undef> or an
		686	array-ref with the filenames.
		687
		688
		689	=item aio_readdirx $pathname, $flags, $callback->($entries, $flags)
		690
		691	Quite similar to C<aio_readdir>, but the C<$flags> argument allows to tune
		692	behaviour and output format. In case of an error, C<$entries> will be
		693	C<undef>.
		694
		695	The flags are a combination of the following constants, ORed together (the
		696	flags will also be passed to the callback, possibly modified):
		697
		698	=over 4
		699
		700	=item IO::AIO::READDIR_DENTS
		701
		702	When this flag is off, then the callback gets an arrayref consisting of
		703	names only (as with C<aio_readdir>), otherwise it gets an arrayref with
		704	C<[$name, $type, $inode]> arrayrefs, each describing a single directory
		705	entry in more detail.
		706
		707	C<$name> is the name of the entry.
		708
		709	C<$type> is one of the C<IO::AIO::DT_xxx> constants:
		710
		711	C<IO::AIO::DT_UNKNOWN>, C<IO::AIO::DT_FIFO>, C<IO::AIO::DT_CHR>, C<IO::AIO::DT_DIR>,
		712	C<IO::AIO::DT_BLK>, C<IO::AIO::DT_REG>, C<IO::AIO::DT_LNK>, C<IO::AIO::DT_SOCK>,
		713	C<IO::AIO::DT_WHT>.
		714
		715	C<IO::AIO::DT_UNKNOWN> means just that: readdir does not know. If you need to
		716	know, you have to run stat yourself. Also, for speed reasons, the C<$type>
		717	scalars are read-only: you can not modify them.
		718
		719	C<$inode> is the inode number (which might not be exact on systems with 64
		720	bit inode numbers and 32 bit perls). This field has unspecified content on
		721	systems that do not deliver the inode information.
		722
		723	=item IO::AIO::READDIR_DIRS_FIRST
		724
		725	When this flag is set, then the names will be returned in an order where
		726	likely directories come first, in optimal stat order. This is useful when
		727	you need to quickly find directories, or you want to find all directories
		728	while avoiding to stat() each entry.
		729
		730	If the system returns type information in readdir, then this is used
		731	to find directories directly. Otherwise, likely directories are names
		732	beginning with ".", or otherwise names with no dots, of which names with
		733	short names are tried first.
		734
		735	=item IO::AIO::READDIR_STAT_ORDER
		736
		737	When this flag is set, then the names will be returned in an order
		738	suitable for stat()'ing each one. That is, when you plan to stat()
		739	all files in the given directory, then the returned order will likely
		740	be fastest.
		741
		742	If both this flag and C<IO::AIO::READDIR_DIRS_FIRST> are specified, then
		743	the likely dirs come first, resulting in a less optimal stat order.
		744
		745	=item IO::AIO::READDIR_FOUND_UNKNOWN
		746
		747	This flag should not be set when calling C<aio_readdirx>. Instead, it
		748	is being set by C<aio_readdirx>, when any of the C<$type>'s found were
		749	C<IO::AIO::DT_UNKNOWN>. The absense of this flag therefore indicates that all
		750	C<$type>'s are known, which can be used to speed up some algorithms.
		751
		752	=back
		753
		754
		755	=item aio_load $path, $data, $callback->($status)
		756
		757	This is a composite request that tries to fully load the given file into
		758	memory. Status is the same as with aio_read.
		759
		760	=cut
		761
		762	sub aio_load($$;$) {
		763	my ($path, undef, $cb) = @_;
		764	my $data = \$_[1];
		765
		766	my $pri = aioreq_pri;
		767	my $grp = aio_group $cb;
		768
		769	aioreq_pri $pri;
		770	add $grp aio_open $path, O_RDONLY, 0, sub {
		771	my $fh = shift
		772	or return $grp->result (-1);
		773
		774	aioreq_pri $pri;
		775	add $grp aio_read $fh, 0, (-s $fh), $$data, 0, sub {
		776	$grp->result ($_[0]);
		777	};
		778	};
		779
		780	$grp
		781	}
		782
		783	=item aio_copy $srcpath, $dstpath, $callback->($status)
		784
		785	Try to copy the I<file> (directories not supported as either source or
		786	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		787	a status of C<0> (ok) or C<-1> (error, see C<$!>).
		788
		789	This is a composite request that creates the destination file with
		790	mode 0200 and copies the contents of the source file into it using
		791	C<aio_sendfile>, followed by restoring atime, mtime, access mode and
		792	uid/gid, in that order.
		793
		794	If an error occurs, the partial destination file will be unlinked, if
		795	possible, except when setting atime, mtime, access mode and uid/gid, where
		796	errors are being ignored.
		797
		798	=cut
		799
		800	sub aio_copy($$;$) {
		801	my ($src, $dst, $cb) = @_;
		802
		803	my $pri = aioreq_pri;
		804	my $grp = aio_group $cb;
		805
		806	aioreq_pri $pri;
		807	add $grp aio_open $src, O_RDONLY, 0, sub {
		808	if (my $src_fh = $_[0]) {
		809	my @stat = stat $src_fh; # hmm, might block over nfs?
		810
		811	aioreq_pri $pri;
		812	add $grp aio_open $dst, O_CREAT | O_WRONLY | O_TRUNC, 0200, sub {
		813	if (my $dst_fh = $_[0]) {
		814	aioreq_pri $pri;
		815	add $grp aio_sendfile $dst_fh, $src_fh, 0, $stat[7], sub {
		816	if ($_[0] == $stat[7]) {
		817	$grp->result (0);
		818	close $src_fh;
		819
		820	my $ch = sub {
		821	aioreq_pri $pri;
		822	add $grp aio_chmod $dst_fh, $stat[2] & 07777, sub {
		823	aioreq_pri $pri;
		824	add $grp aio_chown $dst_fh, $stat[4], $stat[5], sub {
		825	aioreq_pri $pri;
		826	add $grp aio_close $dst_fh;
		827	}
		828	};
		829	};
		830
		831	aioreq_pri $pri;
		832	add $grp aio_utime $dst_fh, $stat[8], $stat[9], sub {
		833	if ($_[0] < 0 && $! == ENOSYS) {
		834	aioreq_pri $pri;
		835	add $grp aio_utime $dst, $stat[8], $stat[9], $ch;
		836	} else {
		837	$ch->();
		838	}
		839	};
		840	} else {
		841	$grp->result (-1);
		842	close $src_fh;
		843	close $dst_fh;
		844
		845	aioreq $pri;
		846	add $grp aio_unlink $dst;
		847	}
		848	};
		849	} else {
		850	$grp->result (-1);
		851	}
		852	},
		853
		854	} else {
		855	$grp->result (-1);
		856	}
		857	};
		858
		859	$grp
		860	}
		861
		862	=item aio_move $srcpath, $dstpath, $callback->($status)
		863
		864	Try to move the I<file> (directories not supported as either source or
		865	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		866	a status of C<0> (ok) or C<-1> (error, see C<$!>).
		867
		868	This is a composite request that tries to rename(2) the file first; if
		869	rename fails with C<EXDEV>, it copies the file with C<aio_copy> and, if
		870	that is successful, unlinks the C<$srcpath>.
		871
		872	=cut
		873
		874	sub aio_move($$;$) {
		875	my ($src, $dst, $cb) = @_;
		876
		877	my $pri = aioreq_pri;
		878	my $grp = aio_group $cb;
		879
		880	aioreq_pri $pri;
		881	add $grp aio_rename $src, $dst, sub {
		882	if ($_[0] && $! == EXDEV) {
		883	aioreq_pri $pri;
		884	add $grp aio_copy $src, $dst, sub {
		885	$grp->result ($_[0]);
		886
		887	unless ($_[0]) {
		888	aioreq_pri $pri;
		889	add $grp aio_unlink $src;
		890	}
		891	};
		892	} else {
		893	$grp->result ($_[0]);
		894	}
		895	};
		896
		897	$grp
		898	}
		899
		900	=item aio_scandir $path, $maxreq, $callback->($dirs, $nondirs)
		901
		902	Scans a directory (similar to C<aio_readdir>) but additionally tries to
		903	efficiently separate the entries of directory C<$path> into two sets of
		904	names, directories you can recurse into (directories), and ones you cannot
		905	recurse into (everything else, including symlinks to directories).
		906
		907	C<aio_scandir> is a composite request that creates of many sub requests_
		908	C<$maxreq> specifies the maximum number of outstanding aio requests that
		909	this function generates. If it is C<< <= 0 >>, then a suitable default
		910	will be chosen (currently 4).
		911
		912	On error, the callback is called without arguments, otherwise it receives
		913	two array-refs with path-relative entry names.
		914
		915	Example:
		916
		917	aio_scandir $dir, 0, sub {
		918	my ($dirs, $nondirs) = @_;
		919	print "real directories: @$dirs\n";
		920	print "everything else: @$nondirs\n";
		921	};
		922
		923	Implementation notes.
		924
		925	The C<aio_readdir> cannot be avoided, but C<stat()>'ing every entry can.
		926
		927	If readdir returns file type information, then this is used directly to
		928	find directories.
		929
		930	Otherwise, after reading the directory, the modification time, size etc.
		931	of the directory before and after the readdir is checked, and if they
		932	match (and isn't the current time), the link count will be used to decide
		933	how many entries are directories (if >= 2). Otherwise, no knowledge of the
		934	number of subdirectories will be assumed.
		935
		936	Then entries will be sorted into likely directories a non-initial dot
		937	currently) and likely non-directories (see C<aio_readdirx>). Then every
		938	entry plus an appended C</.> will be C<stat>'ed, likely directories first,
		939	in order of their inode numbers. If that succeeds, it assumes that the
		940	entry is a directory or a symlink to directory (which will be checked
		941	seperately). This is often faster than stat'ing the entry itself because
		942	filesystems might detect the type of the entry without reading the inode
		943	data (e.g. ext2fs filetype feature), even on systems that cannot return
		944	the filetype information on readdir.
		945
		946	If the known number of directories (link count - 2) has been reached, the
		947	rest of the entries is assumed to be non-directories.
		948
		949	This only works with certainty on POSIX (= UNIX) filesystems, which
		950	fortunately are the vast majority of filesystems around.
		951
		952	It will also likely work on non-POSIX filesystems with reduced efficiency
		953	as those tend to return 0 or 1 as link counts, which disables the
		954	directory counting heuristic.
		955
		956	=cut
		957
		958	sub aio_scandir($$;$) {
		959	my ($path, $maxreq, $cb) = @_;
		960
		961	my $pri = aioreq_pri;
		962
		963	my $grp = aio_group $cb;
		964
		965	$maxreq = 4 if $maxreq <= 0;
		966
		967	# stat once
		968	aioreq_pri $pri;
		969	add $grp aio_stat $path, sub {
		970	return $grp->result () if $_[0];
		971	my $now = time;
		972	my $hash1 = join ":", (stat _)[0,1,3,7,9];
		973
		974	# read the directory entries
		975	aioreq_pri $pri;
		976	add $grp aio_readdirx $path, READDIR_DIRS_FIRST, sub {
		977	my $entries = shift
		978	or return $grp->result ();
		979
		980	# stat the dir another time
		981	aioreq_pri $pri;
		982	add $grp aio_stat $path, sub {
		983	my $hash2 = join ":", (stat _)[0,1,3,7,9];
		984
		985	my $ndirs;
		986
		987	# take the slow route if anything looks fishy
		988	if ($hash1 ne $hash2 or (stat _)[9] == $now) {
		989	$ndirs = -1;
		990	} else {
		991	# if nlink == 2, we are finished
		992	# for non-posix-fs's, we rely on nlink < 2
		993	$ndirs = (stat _)[3] - 2
		994	or return $grp->result ([], $entries);
		995	}
		996
		997	my (@dirs, @nondirs);
		998
		999	my $statgrp = add $grp aio_group sub {
		1000	$grp->result (\@dirs, \@nondirs);
		1001	};
		1002
		1003	limit $statgrp $maxreq;
		1004	feed $statgrp sub {
		1005	return unless @$entries;
		1006	my $entry = shift @$entries;
		1007
		1008	aioreq_pri $pri;
		1009	add $statgrp aio_stat "$path/$entry/.", sub {
		1010	if ($_[0] < 0) {
		1011	push @nondirs, $entry;
		1012	} else {
		1013	# need to check for real directory
		1014	aioreq_pri $pri;
		1015	add $statgrp aio_lstat "$path/$entry", sub {
		1016	if (-d _) {
		1017	push @dirs, $entry;
		1018
		1019	unless (--$ndirs) {
		1020	push @nondirs, @$entries;
		1021	feed $statgrp;
		1022	}
		1023	} else {
		1024	push @nondirs, $entry;
		1025	}
		1026	}
		1027	}
		1028	};
		1029	};
		1030	};
		1031	};
		1032	};
		1033
		1034	$grp
		1035	}
		1036
		1037	=item aio_rmtree $path, $callback->($status)
		1038
		1039	Delete a directory tree starting (and including) C<$path>, return the
		1040	status of the final C<rmdir> only. This is a composite request that
		1041	uses C<aio_scandir> to recurse into and rmdir directories, and unlink
		1042	everything else.
		1043
		1044	=cut
		1045
		1046	sub aio_rmtree;
		1047	sub aio_rmtree($;$) {
		1048	my ($path, $cb) = @_;
		1049
		1050	my $pri = aioreq_pri;
		1051	my $grp = aio_group $cb;
		1052
		1053	aioreq_pri $pri;
		1054	add $grp aio_scandir $path, 0, sub {
		1055	my ($dirs, $nondirs) = @_;
		1056
		1057	my $dirgrp = aio_group sub {
		1058	add $grp aio_rmdir $path, sub {
		1059	$grp->result ($_[0]);
		1060	};
		1061	};
		1062
		1063	(aioreq_pri $pri), add $dirgrp aio_rmtree "$path/$_" for @$dirs;
		1064	(aioreq_pri $pri), add $dirgrp aio_unlink "$path/$_" for @$nondirs;
		1065
		1066	add $grp $dirgrp;
		1067	};
		1068
		1069	$grp
		1070	}
		1071
		1072	=item aio_sync $callback->($status)
		1073
		1074	Asynchronously call sync and call the callback when finished.
		1075
215	=item aio_fsync $fh, $callback	1076	=item aio_fsync $fh, $callback->($status)
216		1077
217	Asynchronously call fsync on the given filehandle and call the callback	1078	Asynchronously call fsync on the given filehandle and call the callback
218	with the fsync result code.	1079	with the fsync result code.
219		1080
220	=item aio_fdatasync $fh, $callback	1081	=item aio_fdatasync $fh, $callback->($status)
221		1082
222	Asynchronously call fdatasync on the given filehandle and call the	1083	Asynchronously call fdatasync on the given filehandle and call the
223	callback with the fdatasync result code.	1084	callback with the fdatasync result code.
224		1085
		1086	If this call isn't available because your OS lacks it or it couldn't be
		1087	detected, it will be emulated by calling C<fsync> instead.
		1088
		1089	=item aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
		1090
		1091	Sync the data portion of the file specified by C<$offset> and C<$length>
		1092	to disk (but NOT the metadata), by calling the Linux-specific
		1093	sync_file_range call. If sync_file_range is not available or it returns
		1094	ENOSYS, then fdatasync or fsync is being substituted.
		1095
		1096	C<$flags> can be a combination of C<IO::AIO::SYNC_FILE_RANGE_WAIT_BEFORE>,
		1097	C<IO::AIO::SYNC_FILE_RANGE_WRITE> and
		1098	C<IO::AIO::SYNC_FILE_RANGE_WAIT_AFTER>: refer to the sync_file_range
		1099	manpage for details.
		1100
		1101	=item aio_pathsync $path, $callback->($status)
		1102
		1103	This request tries to open, fsync and close the given path. This is a
		1104	composite request intended to sync directories after directory operations
		1105	(E.g. rename). This might not work on all operating systems or have any
		1106	specific effect, but usually it makes sure that directory changes get
		1107	written to disc. It works for anything that can be opened for read-only,
		1108	not just directories.
		1109
		1110	Future versions of this function might fall back to other methods when
		1111	C<fsync> on the directory fails (such as calling C<sync>).
		1112
		1113	Passes C<0> when everything went ok, and C<-1> on error.
		1114
225	=cut	1115	=cut
226		1116
227	# support function to convert a fd into a perl filehandle	1117	sub aio_pathsync($;$) {
228	sub _fd2fh {	1118	my ($path, $cb) = @_;
229	return undef if $_[0] < 0;
230		1119
231	# try to be perl5.6-compatible	1120	my $pri = aioreq_pri;
232	local *AIO_FH;	1121	my $grp = aio_group $cb;
233	open AIO_FH, "+<&=$_[0]"
234	or return undef;
235		1122
236	*AIO_FH	1123	aioreq_pri $pri;
		1124	add $grp aio_open $path, O_RDONLY, 0, sub {
		1125	my ($fh) = @_;
		1126	if ($fh) {
		1127	aioreq_pri $pri;
		1128	add $grp aio_fsync $fh, sub {
		1129	$grp->result ($_[0]);
		1130
		1131	aioreq_pri $pri;
		1132	add $grp aio_close $fh;
		1133	};
		1134	} else {
		1135	$grp->result (-1);
		1136	}
		1137	};
		1138
		1139	$grp
237	}	1140	}
238		1141
		1142	=item aio_msync $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		1143
		1144	This is a rather advanced IO::AIO call, which only works on mmap(2)ed
		1145	scalars (see the C<IO::AIO::mmap> function, although it also works on data
		1146	scalars managed by the L<Sys::Mmap> or L<Mmap> modules, note that the
		1147	scalar must only be modified in-place while an aio operation is pending on
		1148	it).
		1149
		1150	It calls the C<msync> function of your OS, if available, with the memory
		1151	area starting at C<$offset> in the string and ending C<$length> bytes
		1152	later. If C<$length> is negative, counts from the end, and if C<$length>
		1153	is C<undef>, then it goes till the end of the string. The flags can be
		1154	a combination of C<IO::AIO::MS_ASYNC>, C<IO::AIO::MS_INVALIDATE> and
		1155	C<IO::AIO::MS_SYNC>.
		1156
		1157	=item aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		1158
		1159	This is a rather advanced IO::AIO call, which works best on mmap(2)ed
		1160	scalars.
		1161
		1162	It touches (reads or writes) all memory pages in the specified
		1163	range inside the scalar. All caveats and parameters are the same
		1164	as for C<aio_msync>, above, except for flags, which must be either
		1165	C<0> (which reads all pages and ensures they are instantiated) or
		1166	C<IO::AIO::MT_MODIFY>, which modifies the memory page s(by reading and
		1167	writing an octet from it, which dirties the page).
		1168
		1169	=item aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status)
		1170
		1171	This is a rather advanced IO::AIO call, which works best on mmap(2)ed
		1172	scalars.
		1173
		1174	It reads in all the pages of the underlying storage into memory (if any)
		1175	and locks them, so they are not getting swapped/paged out or removed.
		1176
		1177	If C<$length> is undefined, then the scalar will be locked till the end.
		1178
		1179	On systems that do not implement C<mlock>, this function returns C<-1>
		1180	and sets errno to C<ENOSYS>.
		1181
		1182	Note that the corresponding C<munlock> is synchronous and is
		1183	documented under L<MISCELLANEOUS FUNCTIONS>.
		1184
		1185	Example: open a file, mmap and mlock it - both will be undone when
		1186	C<$data> gets destroyed.
		1187
		1188	open my $fh, "<", $path or die "$path: $!";
		1189	my $data;
		1190	IO::AIO::mmap $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh;
		1191	aio_mlock $data; # mlock in background
		1192
		1193	=item aio_mlockall $flags, $callback->($status)
		1194
		1195	Calls the C<mlockall> function with the given C<$flags> (a combination of
		1196	C<IO::AIO::MCL_CURRENT> and C<IO::AIO::MCL_FUTURE>).
		1197
		1198	On systems that do not implement C<mlockall>, this function returns C<-1>
		1199	and sets errno to C<ENOSYS>.
		1200
		1201	Note that the corresponding C<munlockall> is synchronous and is
		1202	documented under L<MISCELLANEOUS FUNCTIONS>.
		1203
		1204	Example: asynchronously lock all current and future pages into memory.
		1205
		1206	aio_mlockall IO::AIO::MCL_FUTURE;
		1207
		1208	=item aio_group $callback->(...)
		1209
		1210	This is a very special aio request: Instead of doing something, it is a
		1211	container for other aio requests, which is useful if you want to bundle
		1212	many requests into a single, composite, request with a definite callback
		1213	and the ability to cancel the whole request with its subrequests.
		1214
		1215	Returns an object of class L<IO::AIO::GRP>. See its documentation below
		1216	for more info.
		1217
		1218	Example:
		1219
		1220	my $grp = aio_group sub {
		1221	print "all stats done\n";
		1222	};
		1223
		1224	add $grp
		1225	(aio_stat ...),
		1226	(aio_stat ...),
		1227	...;
		1228
		1229	=item aio_nop $callback->()
		1230
		1231	This is a special request - it does nothing in itself and is only used for
		1232	side effects, such as when you want to add a dummy request to a group so
		1233	that finishing the requests in the group depends on executing the given
		1234	code.
		1235
		1236	While this request does nothing, it still goes through the execution
		1237	phase and still requires a worker thread. Thus, the callback will not
		1238	be executed immediately but only after other requests in the queue have
		1239	entered their execution phase. This can be used to measure request
		1240	latency.
		1241
		1242	=item IO::AIO::aio_busy $fractional_seconds, $callback->() *NOT EXPORTED*
		1243
		1244	Mainly used for debugging and benchmarking, this aio request puts one of
		1245	the request workers to sleep for the given time.
		1246
		1247	While it is theoretically handy to have simple I/O scheduling requests
		1248	like sleep and file handle readable/writable, the overhead this creates is
		1249	immense (it blocks a thread for a long time) so do not use this function
		1250	except to put your application under artificial I/O pressure.
		1251
		1252	=back
		1253
		1254	=head2 IO::AIO::REQ CLASS
		1255
		1256	All non-aggregate C<aio_*> functions return an object of this class when
		1257	called in non-void context.
		1258
		1259	=over 4
		1260
		1261	=item cancel $req
		1262
		1263	Cancels the request, if possible. Has the effect of skipping execution
		1264	when entering the B<execute> state and skipping calling the callback when
		1265	entering the the B<result> state, but will leave the request otherwise
		1266	untouched (with the exception of readdir). That means that requests that
		1267	currently execute will not be stopped and resources held by the request
		1268	will not be freed prematurely.
		1269
		1270	=item cb $req $callback->(...)
		1271
		1272	Replace (or simply set) the callback registered to the request.
		1273
		1274	=back
		1275
		1276	=head2 IO::AIO::GRP CLASS
		1277
		1278	This class is a subclass of L<IO::AIO::REQ>, so all its methods apply to
		1279	objects of this class, too.
		1280
		1281	A IO::AIO::GRP object is a special request that can contain multiple other
		1282	aio requests.
		1283
		1284	You create one by calling the C<aio_group> constructing function with a
		1285	callback that will be called when all contained requests have entered the
		1286	C<done> state:
		1287
		1288	my $grp = aio_group sub {
		1289	print "all requests are done\n";
		1290	};
		1291
		1292	You add requests by calling the C<add> method with one or more
		1293	C<IO::AIO::REQ> objects:
		1294
		1295	$grp->add (aio_unlink "...");
		1296
		1297	add $grp aio_stat "...", sub {
		1298	$_[0] or return $grp->result ("error");
		1299
		1300	# add another request dynamically, if first succeeded
		1301	add $grp aio_open "...", sub {
		1302	$grp->result ("ok");
		1303	};
		1304	};
		1305
		1306	This makes it very easy to create composite requests (see the source of
		1307	C<aio_move> for an application) that work and feel like simple requests.
		1308
		1309	=over 4
		1310
		1311	=item * The IO::AIO::GRP objects will be cleaned up during calls to
		1312	C<IO::AIO::poll_cb>, just like any other request.
		1313
		1314	=item * They can be canceled like any other request. Canceling will cancel not
		1315	only the request itself, but also all requests it contains.
		1316
		1317	=item * They can also can also be added to other IO::AIO::GRP objects.
		1318
		1319	=item * You must not add requests to a group from within the group callback (or
		1320	any later time).
		1321
		1322	=back
		1323
		1324	Their lifetime, simplified, looks like this: when they are empty, they
		1325	will finish very quickly. If they contain only requests that are in the
		1326	C<done> state, they will also finish. Otherwise they will continue to
		1327	exist.
		1328
		1329	That means after creating a group you have some time to add requests
		1330	(precisely before the callback has been invoked, which is only done within
		1331	the C<poll_cb>). And in the callbacks of those requests, you can add
		1332	further requests to the group. And only when all those requests have
		1333	finished will the the group itself finish.
		1334
		1335	=over 4
		1336
		1337	=item add $grp ...
		1338
		1339	=item $grp->add (...)
		1340
		1341	Add one or more requests to the group. Any type of L<IO::AIO::REQ> can
		1342	be added, including other groups, as long as you do not create circular
		1343	dependencies.
		1344
		1345	Returns all its arguments.
		1346
		1347	=item $grp->cancel_subs
		1348
		1349	Cancel all subrequests and clears any feeder, but not the group request
		1350	itself. Useful when you queued a lot of events but got a result early.
		1351
		1352	The group request will finish normally (you cannot add requests to the
		1353	group).
		1354
		1355	=item $grp->result (...)
		1356
		1357	Set the result value(s) that will be passed to the group callback when all
		1358	subrequests have finished and set the groups errno to the current value
		1359	of errno (just like calling C<errno> without an error number). By default,
		1360	no argument will be passed and errno is zero.
		1361
		1362	=item $grp->errno ([$errno])
		1363
		1364	Sets the group errno value to C<$errno>, or the current value of errno
		1365	when the argument is missing.
		1366
		1367	Every aio request has an associated errno value that is restored when
		1368	the callback is invoked. This method lets you change this value from its
		1369	default (0).
		1370
		1371	Calling C<result> will also set errno, so make sure you either set C<$!>
		1372	before the call to C<result>, or call c<errno> after it.
		1373
		1374	=item feed $grp $callback->($grp)
		1375
		1376	Sets a feeder/generator on this group: every group can have an attached
		1377	generator that generates requests if idle. The idea behind this is that,
		1378	although you could just queue as many requests as you want in a group,
		1379	this might starve other requests for a potentially long time. For example,
		1380	C<aio_scandir> might generate hundreds of thousands C<aio_stat> requests,
		1381	delaying any later requests for a long time.
		1382
		1383	To avoid this, and allow incremental generation of requests, you can
		1384	instead a group and set a feeder on it that generates those requests. The
		1385	feed callback will be called whenever there are few enough (see C<limit>,
		1386	below) requests active in the group itself and is expected to queue more
		1387	requests.
		1388
		1389	The feed callback can queue as many requests as it likes (i.e. C<add> does
		1390	not impose any limits).
		1391
		1392	If the feed does not queue more requests when called, it will be
		1393	automatically removed from the group.
		1394
		1395	If the feed limit is C<0> when this method is called, it will be set to
		1396	C<2> automatically.
		1397
		1398	Example:
		1399
		1400	# stat all files in @files, but only ever use four aio requests concurrently:
		1401
		1402	my $grp = aio_group sub { print "finished\n" };
		1403	limit $grp 4;
		1404	feed $grp sub {
		1405	my $file = pop @files
		1406	or return;
		1407
		1408	add $grp aio_stat $file, sub { ... };
		1409	};
		1410
		1411	=item limit $grp $num
		1412
		1413	Sets the feeder limit for the group: The feeder will be called whenever
		1414	the group contains less than this many requests.
		1415
		1416	Setting the limit to C<0> will pause the feeding process.
		1417
		1418	The default value for the limit is C<0>, but note that setting a feeder
		1419	automatically bumps it up to C<2>.
		1420
		1421	=back
		1422
		1423	=head2 SUPPORT FUNCTIONS
		1424
		1425	=head3 EVENT PROCESSING AND EVENT LOOP INTEGRATION
		1426
		1427	=over 4
		1428
		1429	=item $fileno = IO::AIO::poll_fileno
		1430
		1431	Return the I<request result pipe file descriptor>. This filehandle must be
		1432	polled for reading by some mechanism outside this module (e.g. EV, Glib,
		1433	select and so on, see below or the SYNOPSIS). If the pipe becomes readable
		1434	you have to call C<poll_cb> to check the results.
		1435
		1436	See C<poll_cb> for an example.
		1437
		1438	=item IO::AIO::poll_cb
		1439
		1440	Process some outstanding events on the result pipe. You have to call
		1441	this regularly. Returns C<0> if all events could be processed (or there
		1442	were no events to process), or C<-1> if it returned earlier for whatever
		1443	reason. Returns immediately when no events are outstanding. The amount of
		1444	events processed depends on the settings of C<IO::AIO::max_poll_req> and
		1445	C<IO::AIO::max_poll_time>.
		1446
		1447	If not all requests were processed for whatever reason, the filehandle
		1448	will still be ready when C<poll_cb> returns, so normally you don't have to
		1449	do anything special to have it called later.
		1450
		1451	Apart from calling C<IO::AIO::poll_cb> when the event filehandle becomes
		1452	ready, it can be beneficial to call this function from loops which submit
		1453	a lot of requests, to make sure the results get processed when they become
		1454	available and not just when the loop is finished and the event loop takes
		1455	over again. This function returns very fast when there are no outstanding
		1456	requests.
		1457
		1458	Example: Install an Event watcher that automatically calls
		1459	IO::AIO::poll_cb with high priority (more examples can be found in the
		1460	SYNOPSIS section, at the top of this document):
		1461
		1462	Event->io (fd => IO::AIO::poll_fileno,
		1463	poll => 'r', async => 1,
		1464	cb => \&IO::AIO::poll_cb);
		1465
		1466	=item IO::AIO::poll_wait
		1467
		1468	If there are any outstanding requests and none of them in the result
		1469	phase, wait till the result filehandle becomes ready for reading (simply
		1470	does a C<select> on the filehandle. This is useful if you want to
		1471	synchronously wait for some requests to finish).
		1472
		1473	See C<nreqs> for an example.
		1474
		1475	=item IO::AIO::poll
		1476
		1477	Waits until some requests have been handled.
		1478
		1479	Returns the number of requests processed, but is otherwise strictly
		1480	equivalent to:
		1481
		1482	IO::AIO::poll_wait, IO::AIO::poll_cb
		1483
		1484	=item IO::AIO::flush
		1485
		1486	Wait till all outstanding AIO requests have been handled.
		1487
		1488	Strictly equivalent to:
		1489
		1490	IO::AIO::poll_wait, IO::AIO::poll_cb
		1491	while IO::AIO::nreqs;
		1492
		1493	=item IO::AIO::max_poll_reqs $nreqs
		1494
		1495	=item IO::AIO::max_poll_time $seconds
		1496
		1497	These set the maximum number of requests (default C<0>, meaning infinity)
		1498	that are being processed by C<IO::AIO::poll_cb> in one call, respectively
		1499	the maximum amount of time (default C<0>, meaning infinity) spent in
		1500	C<IO::AIO::poll_cb> to process requests (more correctly the mininum amount
		1501	of time C<poll_cb> is allowed to use).
		1502
		1503	Setting C<max_poll_time> to a non-zero value creates an overhead of one
		1504	syscall per request processed, which is not normally a problem unless your
		1505	callbacks are really really fast or your OS is really really slow (I am
		1506	not mentioning Solaris here). Using C<max_poll_reqs> incurs no overhead.
		1507
		1508	Setting these is useful if you want to ensure some level of
		1509	interactiveness when perl is not fast enough to process all requests in
		1510	time.
		1511
		1512	For interactive programs, values such as C<0.01> to C<0.1> should be fine.
		1513
		1514	Example: Install an Event watcher that automatically calls
		1515	IO::AIO::poll_cb with low priority, to ensure that other parts of the
		1516	program get the CPU sometimes even under high AIO load.
		1517
		1518	# try not to spend much more than 0.1s in poll_cb
		1519	IO::AIO::max_poll_time 0.1;
		1520
		1521	# use a low priority so other tasks have priority
		1522	Event->io (fd => IO::AIO::poll_fileno,
		1523	poll => 'r', nice => 1,
		1524	cb => &IO::AIO::poll_cb);
		1525
		1526	=back
		1527
		1528	=head3 CONTROLLING THE NUMBER OF THREADS
		1529
		1530	=over
		1531
		1532	=item IO::AIO::min_parallel $nthreads
		1533
		1534	Set the minimum number of AIO threads to C<$nthreads>. The current
		1535	default is C<8>, which means eight asynchronous operations can execute
		1536	concurrently at any one time (the number of outstanding requests,
		1537	however, is unlimited).
		1538
		1539	IO::AIO starts threads only on demand, when an AIO request is queued and
		1540	no free thread exists. Please note that queueing up a hundred requests can
		1541	create demand for a hundred threads, even if it turns out that everything
		1542	is in the cache and could have been processed faster by a single thread.
		1543
		1544	It is recommended to keep the number of threads relatively low, as some
		1545	Linux kernel versions will scale negatively with the number of threads
		1546	(higher parallelity => MUCH higher latency). With current Linux 2.6
		1547	versions, 4-32 threads should be fine.
		1548
		1549	Under most circumstances you don't need to call this function, as the
		1550	module selects a default that is suitable for low to moderate load.
		1551
		1552	=item IO::AIO::max_parallel $nthreads
		1553
		1554	Sets the maximum number of AIO threads to C<$nthreads>. If more than the
		1555	specified number of threads are currently running, this function kills
		1556	them. This function blocks until the limit is reached.
		1557
		1558	While C<$nthreads> are zero, aio requests get queued but not executed
		1559	until the number of threads has been increased again.
		1560
		1561	This module automatically runs C<max_parallel 0> at program end, to ensure
		1562	that all threads are killed and that there are no outstanding requests.
		1563
		1564	Under normal circumstances you don't need to call this function.
		1565
		1566	=item IO::AIO::max_idle $nthreads
		1567
		1568	Limit the number of threads (default: 4) that are allowed to idle
		1569	(i.e., threads that did not get a request to process within the idle
		1570	timeout (default: 10 seconds). That means if a thread becomes idle while
		1571	C<$nthreads> other threads are also idle, it will free its resources and
		1572	exit.
		1573
		1574	This is useful when you allow a large number of threads (e.g. 100 or 1000)
		1575	to allow for extremely high load situations, but want to free resources
		1576	under normal circumstances (1000 threads can easily consume 30MB of RAM).
		1577
		1578	The default is probably ok in most situations, especially if thread
		1579	creation is fast. If thread creation is very slow on your system you might
		1580	want to use larger values.
		1581
		1582	=item IO::AIO::idle_timeout $seconds
		1583
		1584	Sets the minimum idle timeout (default 10) after which worker threads are
		1585	allowed to exit. SEe C<IO::AIO::max_idle>.
		1586
		1587	=item IO::AIO::max_outstanding $maxreqs
		1588
		1589	Sets the maximum number of outstanding requests to C<$nreqs>. If
		1590	you do queue up more than this number of requests, the next call to
		1591	C<IO::AIO::poll_cb> (and other functions calling C<poll_cb>, such as
		1592	C<IO::AIO::flush> or C<IO::AIO::poll>) will block until the limit is no
		1593	longer exceeded.
		1594
		1595	In other words, this setting does not enforce a queue limit, but can be
		1596	used to make poll functions block if the limit is exceeded.
		1597
		1598	This is a very bad function to use in interactive programs because it
		1599	blocks, and a bad way to reduce concurrency because it is inexact: Better
		1600	use an C<aio_group> together with a feed callback.
		1601
		1602	It's main use is in scripts without an event loop - when you want to stat
		1603	a lot of files, you can write somehting like this:
		1604
		1605	IO::AIO::max_outstanding 32;
		1606
		1607	for my $path (...) {
		1608	aio_stat $path , ...;
		1609	IO::AIO::poll_cb;
		1610	}
		1611
		1612	IO::AIO::flush;
		1613
		1614	The call to C<poll_cb> inside the loop will normally return instantly, but
		1615	as soon as more thna C<32> reqeusts are in-flight, it will block until
		1616	some requests have been handled. This keeps the loop from pushing a large
		1617	number of C<aio_stat> requests onto the queue.
		1618
		1619	The default value for C<max_outstanding> is very large, so there is no
		1620	practical limit on the number of outstanding requests.
		1621
		1622	=back
		1623
		1624	=head3 STATISTICAL INFORMATION
		1625
		1626	=over
		1627
		1628	=item IO::AIO::nreqs
		1629
		1630	Returns the number of requests currently in the ready, execute or pending
		1631	states (i.e. for which their callback has not been invoked yet).
		1632
		1633	Example: wait till there are no outstanding requests anymore:
		1634
		1635	IO::AIO::poll_wait, IO::AIO::poll_cb
		1636	while IO::AIO::nreqs;
		1637
		1638	=item IO::AIO::nready
		1639
		1640	Returns the number of requests currently in the ready state (not yet
		1641	executed).
		1642
		1643	=item IO::AIO::npending
		1644
		1645	Returns the number of requests currently in the pending state (executed,
		1646	but not yet processed by poll_cb).
		1647
		1648	=back
		1649
		1650	=head3 MISCELLANEOUS FUNCTIONS
		1651
		1652	IO::AIO implements some functions that might be useful, but are not
		1653	asynchronous.
		1654
		1655	=over 4
		1656
		1657	=item IO::AIO::sendfile $ofh, $ifh, $offset, $count
		1658
		1659	Calls the C<eio_sendfile_sync> function, which is like C<aio_sendfile>,
		1660	but is blocking (this makes most sense if you know the input data is
		1661	likely cached already and the output filehandle is set to non-blocking
		1662	operations).
		1663
		1664	Returns the number of bytes copied, or C<-1> on error.
		1665
		1666	=item IO::AIO::fadvise $fh, $offset, $len, $advice
		1667
		1668	Simply calls the C<posix_fadvise> function (see its
		1669	manpage for details). The following advice constants are
		1670	avaiable: C<IO::AIO::FADV_NORMAL>, C<IO::AIO::FADV_SEQUENTIAL>,
		1671	C<IO::AIO::FADV_RANDOM>, C<IO::AIO::FADV_NOREUSE>,
		1672	C<IO::AIO::FADV_WILLNEED>, C<IO::AIO::FADV_DONTNEED>.
		1673
		1674	On systems that do not implement C<posix_fadvise>, this function returns
		1675	ENOSYS, otherwise the return value of C<posix_fadvise>.
		1676
		1677	=item IO::AIO::madvise $scalar, $offset, $len, $advice
		1678
		1679	Simply calls the C<posix_madvise> function (see its
		1680	manpage for details). The following advice constants are
		1681	avaiable: C<IO::AIO::MADV_NORMAL>, C<IO::AIO::MADV_SEQUENTIAL>,
		1682	C<IO::AIO::MADV_RANDOM>, C<IO::AIO::MADV_WILLNEED>, C<IO::AIO::MADV_DONTNEED>.
		1683
		1684	On systems that do not implement C<posix_madvise>, this function returns
		1685	ENOSYS, otherwise the return value of C<posix_madvise>.
		1686
		1687	=item IO::AIO::mprotect $scalar, $offset, $len, $protect
		1688
		1689	Simply calls the C<mprotect> function on the preferably AIO::mmap'ed
		1690	$scalar (see its manpage for details). The following protect
		1691	constants are avaiable: C<IO::AIO::PROT_NONE>, C<IO::AIO::PROT_READ>,
		1692	C<IO::AIO::PROT_WRITE>, C<IO::AIO::PROT_EXEC>.
		1693
		1694	On systems that do not implement C<mprotect>, this function returns
		1695	ENOSYS, otherwise the return value of C<mprotect>.
		1696
		1697	=item IO::AIO::mmap $scalar, $length, $prot, $flags, $fh[, $offset]
		1698
		1699	Memory-maps a file (or anonymous memory range) and attaches it to the
		1700	given C<$scalar>, which will act like a string scalar.
		1701
		1702	The only operations allowed on the scalar are C<substr>/C<vec> that don't
		1703	change the string length, and most read-only operations such as copying it
		1704	or searching it with regexes and so on.
		1705
		1706	Anything else is unsafe and will, at best, result in memory leaks.
		1707
		1708	The memory map associated with the C<$scalar> is automatically removed
		1709	when the C<$scalar> is destroyed, or when the C<IO::AIO::mmap> or
		1710	C<IO::AIO::munmap> functions are called.
		1711
		1712	This calls the C<mmap>(2) function internally. See your system's manual
		1713	page for details on the C<$length>, C<$prot> and C<$flags> parameters.
		1714
		1715	The C<$length> must be larger than zero and smaller than the actual
		1716	filesize.
		1717
		1718	C<$prot> is a combination of C<IO::AIO::PROT_NONE>, C<IO::AIO::PROT_EXEC>,
		1719	C<IO::AIO::PROT_READ> and/or C<IO::AIO::PROT_WRITE>,
		1720
		1721	C<$flags> can be a combination of C<IO::AIO::MAP_SHARED> or
		1722	C<IO::AIO::MAP_PRIVATE>, or a number of system-specific flags (when
		1723	not available, the are defined as 0): C<IO::AIO::MAP_ANONYMOUS>
		1724	(which is set to C<MAP_ANON> if your system only provides this
		1725	constant), C<IO::AIO::MAP_HUGETLB>, C<IO::AIO::MAP_LOCKED>,
		1726	C<IO::AIO::MAP_NORESERVE>, C<IO::AIO::MAP_POPULATE> or
		1727	C<IO::AIO::MAP_NONBLOCK>
		1728
		1729	If C<$fh> is C<undef>, then a file descriptor of C<-1> is passed.
		1730
		1731	C<$offset> is the offset from the start of the file - it generally must be
		1732	a multiple of C<IO::AIO::PAGESIZE> and defaults to C<0>.
		1733
		1734	Example:
		1735
		1736	use Digest::MD5;
		1737	use IO::AIO;
		1738
		1739	open my $fh, "<verybigfile"
		1740	or die "$!";
		1741
		1742	IO::AIO::mmap my $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh
		1743	or die "verybigfile: $!";
		1744
		1745	my $fast_md5 = md5 $data;
		1746
		1747	=item IO::AIO::munmap $scalar
		1748
		1749	Removes a previous mmap and undefines the C<$scalar>.
		1750
		1751	=item IO::AIO::munlock $scalar, $offset = 0, $length = undef
		1752
		1753	Calls the C<munlock> function, undoing the effects of a previous
		1754	C<aio_mlock> call (see its description for details).
		1755
		1756	=item IO::AIO::munlockall
		1757
		1758	Calls the C<munlockall> function.
		1759
		1760	On systems that do not implement C<munlockall>, this function returns
		1761	ENOSYS, otherwise the return value of C<munlockall>.
		1762
		1763	=back
		1764
		1765	=cut
		1766
239	min_parallel 4;	1767	min_parallel 8;
240		1768
241	END {	1769	END { flush }
242	max_parallel 0;
243	}
244		1770
245	1;	1771	1;
246		1772
247	=back	1773	=head1 EVENT LOOP INTEGRATION
248		1774
		1775	It is recommended to use L<AnyEvent::AIO> to integrate IO::AIO
		1776	automatically into many event loops:
		1777
		1778	# AnyEvent integration (EV, Event, Glib, Tk, POE, urxvt, pureperl...)
		1779	use AnyEvent::AIO;
		1780
		1781	You can also integrate IO::AIO manually into many event loops, here are
		1782	some examples of how to do this:
		1783
		1784	# EV integration
		1785	my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
		1786
		1787	# Event integration
		1788	Event->io (fd => IO::AIO::poll_fileno,
		1789	poll => 'r',
		1790	cb => \&IO::AIO::poll_cb);
		1791
		1792	# Glib/Gtk2 integration
		1793	add_watch Glib::IO IO::AIO::poll_fileno,
		1794	in => sub { IO::AIO::poll_cb; 1 };
		1795
		1796	# Tk integration
		1797	Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",
		1798	readable => \&IO::AIO::poll_cb);
		1799
		1800	# Danga::Socket integration
		1801	Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
		1802	\&IO::AIO::poll_cb);
		1803
		1804	=head2 FORK BEHAVIOUR
		1805
		1806	Usage of pthreads in a program changes the semantics of fork
		1807	considerably. Specifically, only async-safe functions can be called after
		1808	fork. Perl doesn't know about this, so in general, you cannot call fork
		1809	with defined behaviour in perl. IO::AIO uses pthreads, so this applies,
		1810	but many other extensions and (for inexplicable reasons) perl itself often
		1811	is linked against pthreads, so this limitation applies.
		1812
		1813	Some operating systems have extensions that allow safe use of fork, and
		1814	this module should do "the right thing" on those, and tries on others. At
		1815	the time of this writing (2011) only GNU/Linux supports these extensions
		1816	to POSIX.
		1817
		1818	=head2 MEMORY USAGE
		1819
		1820	Per-request usage:
		1821
		1822	Each aio request uses - depending on your architecture - around 100-200
		1823	bytes of memory. In addition, stat requests need a stat buffer (possibly
		1824	a few hundred bytes), readdir requires a result buffer and so on. Perl
		1825	scalars and other data passed into aio requests will also be locked and
		1826	will consume memory till the request has entered the done state.
		1827
		1828	This is not awfully much, so queuing lots of requests is not usually a
		1829	problem.
		1830
		1831	Per-thread usage:
		1832
		1833	In the execution phase, some aio requests require more memory for
		1834	temporary buffers, and each thread requires a stack and other data
		1835	structures (usually around 16k-128k, depending on the OS).
		1836
249	=head1 BUGS	1837	=head1 KNOWN BUGS
250		1838
251	- could be optimized to use more semaphores instead of filehandles.	1839	Known bugs will be fixed in the next release.
252		1840
253	=head1 SEE ALSO	1841	=head1 SEE ALSO
254		1842
255	L<Coro>, L<Linux::AIO>.	1843	L<AnyEvent::AIO> for easy integration into event loops, L<Coro::AIO> for a
		1844	more natural syntax.
256		1845
257	=head1 AUTHOR	1846	=head1 AUTHOR
258		1847
259	Marc Lehmann <schmorp@schmorp.de>	1848	Marc Lehmann <schmorp@schmorp.de>
260	http://home.schmorp.de/	1849	http://home.schmorp.de/

Diff Legend

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