ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/IO-AIO/AIO.pm

Comparing IO-AIO/AIO.pm (file contents):
Revision 1.43 by root, Mon Dec 26 18:18:32 2005 UTC vs.
Revision 1.225 by root, Tue Apr 10 05:01:33 2012 UTC

…		…
4		4
5	=head1 SYNOPSIS	5	=head1 SYNOPSIS
6		6
7	use IO::AIO;	7	use IO::AIO;
8		8
9	aio_open "/etc/passwd", O_RDONLY, 0, sub {	9	aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub {
10	my ($fh) = @_;	10	my $fh = shift
		11	or die "/etc/passwd: $!";
11	...	12	...
12	};	13	};
13		14
14	aio_unlink "/tmp/file", sub { };	15	aio_unlink "/tmp/file", sub { };
15		16
16	aio_read $fh, 30000, 1024, $buffer, 0, sub {	17	aio_read $fh, 30000, 1024, $buffer, 0, sub {
17	$_[0] > 0 or die "read error: $!";	18	$_[0] > 0 or die "read error: $!";
18	};	19	};
19		20
20	# AnyEvent	21	# version 2+ has request and group objects
21	open my $fh, "<&=" . IO::AIO::poll_fileno or die "$!";	22	use IO::AIO 2;
22	my $w = AnyEvent->io (fh => $fh, poll => 'r', cb => sub { IO::AIO::poll_cb });
23		23
24	# Event	24	aioreq_pri 4; # give next request a very high priority
25	Event->io (fd => IO::AIO::poll_fileno,	25	my $req = aio_unlink "/tmp/file", sub { };
26	poll => 'r',	26	$req->cancel; # cancel request if still in queue
27	cb => \&IO::AIO::poll_cb);
28		27
29	# Glib/Gtk2	28	my $grp = aio_group sub { print "all stats done\n" };
30	add_watch Glib::IO IO::AIO::poll_fileno,	29	add $grp aio_stat "..." for ...;
31	in => sub { IO::AIO::poll_cb; 1 };
32
33	# Tk
34	Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",
35	readable => \&IO::AIO::poll_cb);
36
37	# Danga::Socket
38	Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
39	\&IO::AIO::poll_cb);
40
41		30
42	=head1 DESCRIPTION	31	=head1 DESCRIPTION
43		32
44	This module implements asynchronous I/O using whatever means your	33	This module implements asynchronous I/O using whatever means your
45	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>).
46		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
47	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
48	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
49	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
50	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
51	aio functions available on many operating systems. However, they are often	57	functions available on many operating systems. However, they are often
52	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
53	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
54	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.
55		62
56	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,
57	currently not reentrant, so use appropriate locking yourself, always call	64	it is currently not reentrant in any way, so use appropriate locking
58	C<poll_cb> from within the same thread, or never call C<poll_cb> (or other	65	yourself, always call C<poll_cb> from within the same thread, or never
59	C<aio_> functions) recursively.	66	call C<poll_cb> (or other C<aio_> functions) recursively.
		67
		68	=head2 EXAMPLE
		69
		70	This is a simple example that uses the EV module and loads
		71	F</etc/passwd> asynchronously:
		72
		73	use 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
60		161
61	=cut	162	=cut
62		163
63	package IO::AIO;	164	package IO::AIO;
64		165
65	no warnings;	166	use Carp ();
		167
		168	use common::sense;
66		169
67	use base 'Exporter';	170	use base 'Exporter';
68		171
69	use Fcntl ();
70
71	BEGIN {	172	BEGIN {
72	$VERSION = '1.61';	173	our $VERSION = '4.15';
73		174
74	@EXPORT = qw(aio_sendfile aio_read aio_write aio_open aio_close aio_stat	175	our @AIO_REQ = qw(aio_sendfile aio_seek aio_read aio_write aio_open aio_close
75	aio_lstat aio_unlink aio_rmdir aio_readdir aio_scandir aio_symlink	176	aio_stat aio_lstat aio_unlink aio_rmdir aio_readdir aio_readdirx
76	aio_fsync aio_fdatasync aio_readahead);	177	aio_scandir aio_symlink aio_readlink aio_realpath aio_sync
77	@EXPORT_OK = qw(poll_fileno poll_cb min_parallel max_parallel	178	aio_fsync aio_syncfs aio_fdatasync aio_sync_file_range aio_fallocate
78	max_outstanding nreqs);	179	aio_pathsync aio_readahead aio_fiemap
		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	aio_wd);
		186
		187	our @EXPORT = (@AIO_REQ, qw(aioreq_pri aioreq_nice));
		188	our @EXPORT_OK = qw(poll_fileno poll_cb poll_wait flush
		189	min_parallel max_parallel max_idle idle_timeout
		190	nreqs nready npending nthreads
		191	max_poll_time max_poll_reqs
		192	sendfile fadvise madvise
		193	mmap munmap munlock munlockall);
		194
		195	push @AIO_REQ, qw(aio_busy); # not exported
		196
		197	@IO::AIO::GRP::ISA = 'IO::AIO::REQ';
79		198
80	require XSLoader;	199	require XSLoader;
81	XSLoader::load IO::AIO, $VERSION;	200	XSLoader::load ("IO::AIO", $VERSION);
82	}	201	}
83		202
84	=head1 FUNCTIONS	203	=head1 FUNCTIONS
85		204
86	=head2 AIO FUNCTIONS	205	=head2 QUICK OVERVIEW
		206
		207	This section simply lists the prototypes of the most important functions
		208	for quick reference. See the following sections for function-by-function
		209	documentation.
		210
		211	aio_wd $pathname, $callback->($wd)
		212	aio_open $pathname, $flags, $mode, $callback->($fh)
		213	aio_close $fh, $callback->($status)
		214	aio_seek $fh,$offset,$whence, $callback->($offs)
		215	aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
		216	aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
		217	aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
		218	aio_readahead $fh,$offset,$length, $callback->($retval)
		219	aio_stat $fh_or_path, $callback->($status)
		220	aio_lstat $fh, $callback->($status)
		221	aio_statvfs $fh_or_path, $callback->($statvfs)
		222	aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
		223	aio_chown $fh_or_path, $uid, $gid, $callback->($status)
		224	aio_chmod $fh_or_path, $mode, $callback->($status)
		225	aio_truncate $fh_or_path, $offset, $callback->($status)
		226	aio_unlink $pathname, $callback->($status)
		227	aio_mknod $pathname, $mode, $dev, $callback->($status)
		228	aio_link $srcpath, $dstpath, $callback->($status)
		229	aio_symlink $srcpath, $dstpath, $callback->($status)
		230	aio_readlink $pathname, $callback->($link)
		231	aio_realpath $pathname, $callback->($link)
		232	aio_rename $srcpath, $dstpath, $callback->($status)
		233	aio_mkdir $pathname, $mode, $callback->($status)
		234	aio_rmdir $pathname, $callback->($status)
		235	aio_readdir $pathname, $callback->($entries)
		236	aio_readdirx $pathname, $flags, $callback->($entries, $flags)
		237	IO::AIO::READDIR_DENTS IO::AIO::READDIR_DIRS_FIRST
		238	IO::AIO::READDIR_STAT_ORDER IO::AIO::READDIR_FOUND_UNKNOWN
		239	aio_scandir $pathname, $maxreq, $callback->($dirs, $nondirs)
		240	aio_load $pathname, $data, $callback->($status)
		241	aio_copy $srcpath, $dstpath, $callback->($status)
		242	aio_move $srcpath, $dstpath, $callback->($status)
		243	aio_rmtree $pathname, $callback->($status)
		244	aio_sync $callback->($status)
		245	aio_syncfs $fh, $callback->($status)
		246	aio_fsync $fh, $callback->($status)
		247	aio_fdatasync $fh, $callback->($status)
		248	aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
		249	aio_pathsync $pathname, $callback->($status)
		250	aio_msync $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		251	aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		252	aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status)
		253	aio_mlockall $flags, $callback->($status)
		254	aio_group $callback->(...)
		255	aio_nop $callback->()
		256
		257	$prev_pri = aioreq_pri [$pri]
		258	aioreq_nice $pri_adjust
		259
		260	IO::AIO::poll_wait
		261	IO::AIO::poll_cb
		262	IO::AIO::poll
		263	IO::AIO::flush
		264	IO::AIO::max_poll_reqs $nreqs
		265	IO::AIO::max_poll_time $seconds
		266	IO::AIO::min_parallel $nthreads
		267	IO::AIO::max_parallel $nthreads
		268	IO::AIO::max_idle $nthreads
		269	IO::AIO::idle_timeout $seconds
		270	IO::AIO::max_outstanding $maxreqs
		271	IO::AIO::nreqs
		272	IO::AIO::nready
		273	IO::AIO::npending
		274
		275	IO::AIO::sendfile $ofh, $ifh, $offset, $count
		276	IO::AIO::fadvise $fh, $offset, $len, $advice
		277	IO::AIO::madvise $scalar, $offset, $length, $advice
		278	IO::AIO::mprotect $scalar, $offset, $length, $protect
		279	IO::AIO::munlock $scalar, $offset = 0, $length = undef
		280	IO::AIO::munlockall
		281
		282	=head2 API NOTES
87		283
88	All the C<aio_*> calls are more or less thin wrappers around the syscall	284	All the C<aio_*> calls are more or less thin wrappers around the syscall
89	with the same name (sans C<aio_>). The arguments are similar or identical,	285	with the same name (sans C<aio_>). The arguments are similar or identical,
90	and they all accept an additional (and optional) C<$callback> argument	286	and they all accept an additional (and optional) C<$callback> argument
91	which must be a code reference. This code reference will get called with	287	which must be a code reference. This code reference will be called after
92	the syscall return code (e.g. most syscalls return C<-1> on error, unlike	288	the syscall has been executed in an asynchronous fashion. The results
93	perl, which usually delivers "false") as it's sole argument when the given	289	of the request will be passed as arguments to the callback (and, if an
94	syscall has been executed asynchronously.	290	error occured, in C<$!>) - for most requests the syscall return code (e.g.
		291	most syscalls return C<-1> on error, unlike perl, which usually delivers
		292	"false").
		293
		294	Some requests (such as C<aio_readdir>) pass the actual results and
		295	communicate failures by passing C<undef>.
95		296
96	All functions expecting a filehandle keep a copy of the filehandle	297	All functions expecting a filehandle keep a copy of the filehandle
97	internally until the request has finished.	298	internally until the request has finished.
98		299
		300	All functions return request objects of type L<IO::AIO::REQ> that allow
		301	further manipulation of those requests while they are in-flight.
		302
99	The pathnames you pass to these routines I<must> be absolute and	303	The pathnames you pass to these routines I<should> be absolute. The
100	encoded in byte form. The reason for the former is that at the time the	304	reason for this is that at the time the request is being executed, the
101	request is being executed, the current working directory could have	305	current working directory could have changed. Alternatively, you can
102	changed. Alternatively, you can make sure that you never change the	306	make sure that you never change the current working directory anywhere
103	current working directory.	307	in the program and then use relative paths. You can also take advantage
		308	of IO::AIOs working directory abstraction, that lets you specify paths
		309	relative to some previously-opened "working directory object" - see the
		310	description of the C<IO::AIO::WD> class later in this document.
104		311
105	To encode pathnames to byte form, either make sure you either: a)	312	To encode pathnames as octets, either make sure you either: a) always pass
106	always pass in filenames you got from outside (command line, readdir	313	in filenames you got from outside (command line, readdir etc.) without
107	etc.), b) are ASCII or ISO 8859-1, c) use the Encode module and encode	314	tinkering, b) are in your native filesystem encoding, c) use the Encode
108	your pathnames to the locale (or other) encoding in effect in the user	315	module and encode your pathnames to the locale (or other) encoding in
109	environment, d) use Glib::filename_from_unicode on unicode filenames or e)	316	effect in the user environment, d) use Glib::filename_from_unicode on
110	use something else.	317	unicode filenames or e) use something else to ensure your scalar has the
		318	correct contents.
		319
		320	This works, btw. independent of the internal UTF-8 bit, which IO::AIO
		321	handles correctly whether it is set or not.
		322
		323	=head2 AIO REQUEST FUNCTIONS
111		324
112	=over 4	325	=over 4
		326
		327	=item $prev_pri = aioreq_pri [$pri]
		328
		329	Returns the priority value that would be used for the next request and, if
		330	C<$pri> is given, sets the priority for the next aio request.
		331
		332	The default priority is C<0>, the minimum and maximum priorities are C<-4>
		333	and C<4>, respectively. Requests with higher priority will be serviced
		334	first.
		335
		336	The priority will be reset to C<0> after each call to one of the C<aio_*>
		337	functions.
		338
		339	Example: open a file with low priority, then read something from it with
		340	higher priority so the read request is serviced before other low priority
		341	open requests (potentially spamming the cache):
		342
		343	aioreq_pri -3;
		344	aio_open ..., sub {
		345	return unless $_[0];
		346
		347	aioreq_pri -2;
		348	aio_read $_[0], ..., sub {
		349	...
		350	};
		351	};
		352
		353
		354	=item aioreq_nice $pri_adjust
		355
		356	Similar to C<aioreq_pri>, but subtracts the given value from the current
		357	priority, so the effect is cumulative.
		358
113		359
114	=item aio_open $pathname, $flags, $mode, $callback->($fh)	360	=item aio_open $pathname, $flags, $mode, $callback->($fh)
115		361
116	Asynchronously open or create a file and call the callback with a newly	362	Asynchronously open or create a file and call the callback with a newly
117	created filehandle for the file.	363	created filehandle for the file.
…		…
123	list. They are the same as used by C<sysopen>.	369	list. They are the same as used by C<sysopen>.
124		370
125	Likewise, C<$mode> specifies the mode of the newly created file, if it	371	Likewise, C<$mode> specifies the mode of the newly created file, if it
126	didn't exist and C<O_CREAT> has been given, just like perl's C<sysopen>,	372	didn't exist and C<O_CREAT> has been given, just like perl's C<sysopen>,
127	except that it is mandatory (i.e. use C<0> if you don't create new files,	373	except that it is mandatory (i.e. use C<0> if you don't create new files,
128	and C<0666> or C<0777> if you do).	374	and C<0666> or C<0777> if you do). Note that the C<$mode> will be modified
		375	by the umask in effect then the request is being executed, so better never
		376	change the umask.
129		377
130	Example:	378	Example:
131		379
132	aio_open "/etc/passwd", O_RDONLY, 0, sub {	380	aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub {
133	if ($_[0]) {	381	if ($_[0]) {
134	print "open successful, fh is $_[0]\n";	382	print "open successful, fh is $_[0]\n";
135	...	383	...
136	} else {	384	} else {
137	die "open failed: $!\n";	385	die "open failed: $!\n";
138	}	386	}
139	};	387	};
140		388
		389	In addition to all the common open modes/flags (C<O_RDONLY>, C<O_WRONLY>,
		390	C<O_RDWR>, C<O_CREAT>, C<O_TRUNC>, C<O_EXCL> and C<O_APPEND>), the
		391	following POSIX and non-POSIX constants are available (missing ones on
		392	your system are, as usual, C<0>):
		393
		394	C<O_ASYNC>, C<O_DIRECT>, C<O_NOATIME>, C<O_CLOEXEC>, C<O_NOCTTY>, C<O_NOFOLLOW>,
		395	C<O_NONBLOCK>, C<O_EXEC>, C<O_SEARCH>, C<O_DIRECTORY>, C<O_DSYNC>,
		396	C<O_RSYNC>, C<O_SYNC> and C<O_TTY_INIT>.
		397
		398
141	=item aio_close $fh, $callback->($status)	399	=item aio_close $fh, $callback->($status)
142		400
143	Asynchronously close a file and call the callback with the result	401	Asynchronously close a file and call the callback with the result
144	code. I<WARNING:> although accepted, you should not pass in a perl	402	code.
145	filehandle here, as perl will likely close the file descriptor another
146	time when the filehandle is destroyed. Normally, you can safely call perls
147	C<close> or just let filehandles go out of scope.
148		403
149	This is supposed to be a bug in the API, so that might change. It's	404	Unfortunately, you can't do this to perl. Perl I<insists> very strongly on
150	therefore best to avoid this function.	405	closing the file descriptor associated with the filehandle itself.
		406
		407	Therefore, C<aio_close> will not close the filehandle - instead it will
		408	use dup2 to overwrite the file descriptor with the write-end of a pipe
		409	(the pipe fd will be created on demand and will be cached).
		410
		411	Or in other words: the file descriptor will be closed, but it will not be
		412	free for reuse until the perl filehandle is closed.
		413
		414	=cut
		415
		416	=item aio_seek $fh, $offset, $whence, $callback->($offs)
		417
		418	Seeks the filehandle to the new C<$offset>, similarly to perl's
		419	C<sysseek>. The C<$whence> can use the traditional values (C<0> for
		420	C<IO::AIO::SEEK_SET>, C<1> for C<IO::AIO::SEEK_CUR> or C<2> for
		421	C<IO::AIO::SEEK_END>).
		422
		423	The resulting absolute offset will be passed to the callback, or C<-1> in
		424	case of an error.
		425
		426	In theory, the C<$whence> constants could be different than the
		427	corresponding values from L<Fcntl>, but perl guarantees they are the same,
		428	so don't panic.
		429
		430	As a GNU/Linux (and maybe Solaris) extension, also the constants
		431	C<IO::AIO::SEEK_DATA> and C<IO::AIO::SEEK_HOLE> are available, if they
		432	could be found. No guarantees about suitability for use in C<aio_seek> or
		433	Perl's C<sysseek> can be made though, although I would naively assume they
		434	"just work".
151		435
152	=item aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)	436	=item aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
153		437
154	=item aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)	438	=item aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
155		439
156	Reads or writes C<length> bytes from the specified C<fh> and C<offset>	440	Reads or writes C<$length> bytes from or to the specified C<$fh> and
157	into the scalar given by C<data> and offset C<dataoffset> and calls the	441	C<$offset> into the scalar given by C<$data> and offset C<$dataoffset>
158	callback without the actual number of bytes read (or -1 on error, just	442	and calls the callback without the actual number of bytes read (or -1 on
159	like the syscall).	443	error, just like the syscall).
		444
		445	C<aio_read> will, like C<sysread>, shrink or grow the C<$data> scalar to
		446	offset plus the actual number of bytes read.
		447
		448	If C<$offset> is undefined, then the current file descriptor offset will
		449	be used (and updated), otherwise the file descriptor offset will not be
		450	changed by these calls.
		451
		452	If C<$length> is undefined in C<aio_write>, use the remaining length of
		453	C<$data>.
		454
		455	If C<$dataoffset> is less than zero, it will be counted from the end of
		456	C<$data>.
160		457
161	The C<$data> scalar I<MUST NOT> be modified in any way while the request	458	The C<$data> scalar I<MUST NOT> be modified in any way while the request
162	is outstanding. Modifying it can result in segfaults or WW3 (if the	459	is outstanding. Modifying it can result in segfaults or World War III (if
163	necessary/optional hardware is installed).	460	the necessary/optional hardware is installed).
164		461
165	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at	462	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at
166	offset C<0> within the scalar:	463	offset C<0> within the scalar:
167		464
168	aio_read $fh, 7, 15, $buffer, 0, sub {	465	aio_read $fh, 7, 15, $buffer, 0, sub {
169	$_[0] > 0 or die "read error: $!";	466	$_[0] > 0 or die "read error: $!";
170	print "read $_[0] bytes: <$buffer>\n";	467	print "read $_[0] bytes: <$buffer>\n";
171	};	468	};
172		469
		470
173	=item aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)	471	=item aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
174		472
175	Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts	473	Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts
176	reading at byte offset C<$in_offset>, and starts writing at the current	474	reading at byte offset C<$in_offset>, and starts writing at the current
177	file offset of C<$out_fh>. Because of that, it is not safe to issue more	475	file offset of C<$out_fh>. Because of that, it is not safe to issue more
178	than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each	476	than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each
179	other.	477	other. The same C<$in_fh> works fine though, as this function does not
		478	move or use the file offset of C<$in_fh>.
180		479
		480	Please note that C<aio_sendfile> can read more bytes from C<$in_fh> than
		481	are written, and there is no way to find out how many more bytes have been
		482	read from C<aio_sendfile> alone, as C<aio_sendfile> only provides the
		483	number of bytes written to C<$out_fh>. Only if the result value equals
		484	C<$length> one can assume that C<$length> bytes have been read.
		485
		486	Unlike with other C<aio_> functions, it makes a lot of sense to use
		487	C<aio_sendfile> on non-blocking sockets, as long as one end (typically
		488	the C<$in_fh>) is a file - the file I/O will then be asynchronous, while
		489	the socket I/O will be non-blocking. Note, however, that you can run
		490	into a trap where C<aio_sendfile> reads some data with readahead, then
		491	fails to write all data, and when the socket is ready the next time, the
		492	data in the cache is already lost, forcing C<aio_sendfile> to again hit
		493	the disk. Explicit C<aio_read> + C<aio_write> let's you better control
		494	resource usage.
		495
181	This call tries to make use of a native C<sendfile> syscall to provide	496	This call tries to make use of a native C<sendfile>-like syscall to
182	zero-copy operation. For this to work, C<$out_fh> should refer to a	497	provide zero-copy operation. For this to work, C<$out_fh> should refer to
183	socket, and C<$in_fh> should refer to mmap'able file.	498	a socket, and C<$in_fh> should refer to an mmap'able file.
184		499
185	If the native sendfile call fails or is not implemented, it will be	500	If a native sendfile cannot be found or it fails with C<ENOSYS>,
186	emulated, so you can call C<aio_sendfile> on any type of filehandle	501	C<EINVAL>, C<ENOTSUP>, C<EOPNOTSUPP>, C<EAFNOSUPPORT>, C<EPROTOTYPE> or
		502	C<ENOTSOCK>, it will be emulated, so you can call C<aio_sendfile> on any
187	regardless of the limitations of the operating system.	503	type of filehandle regardless of the limitations of the operating system.
188		504
189	Please note, however, that C<aio_sendfile> can read more bytes from	505	As native sendfile syscalls (as practically any non-POSIX interface hacked
190	C<$in_fh> than are written, and there is no way to find out how many	506	together in a hurry to improve benchmark numbers) tend to be rather buggy
191	bytes have been read from C<aio_sendfile> alone, as C<aio_sendfile> only	507	on many systems, this implementation tries to work around some known bugs
192	provides the number of bytes written to C<$out_fh>. Only if the result	508	in Linux and FreeBSD kernels (probably others, too), but that might fail,
193	value equals C<$length> one can assume that C<$length> bytes have been	509	so you really really should check the return value of C<aio_sendfile> -
194	read.	510	fewre bytes than expected might have been transferred.
		511
195		512
196	=item aio_readahead $fh,$offset,$length, $callback->($retval)	513	=item aio_readahead $fh,$offset,$length, $callback->($retval)
197		514
198	C<aio_readahead> populates the page cache with data from a file so that	515	C<aio_readahead> populates the page cache with data from a file so that
199	subsequent reads from that file will not block on disk I/O. The C<$offset>	516	subsequent reads from that file will not block on disk I/O. The C<$offset>
…		…
205	file. The current file offset of the file is left unchanged.	522	file. The current file offset of the file is left unchanged.
206		523
207	If that syscall doesn't exist (likely if your OS isn't Linux) it will be	524	If that syscall doesn't exist (likely if your OS isn't Linux) it will be
208	emulated by simply reading the data, which would have a similar effect.	525	emulated by simply reading the data, which would have a similar effect.
209		526
		527
210	=item aio_stat $fh_or_path, $callback->($status)	528	=item aio_stat $fh_or_path, $callback->($status)
211		529
212	=item aio_lstat $fh, $callback->($status)	530	=item aio_lstat $fh, $callback->($status)
213		531
214	Works like perl's C<stat> or C<lstat> in void context. The callback will	532	Works like perl's C<stat> or C<lstat> in void context. The callback will
…		…
219	for an explanation.	537	for an explanation.
220		538
221	Currently, the stats are always 64-bit-stats, i.e. instead of returning an	539	Currently, the stats are always 64-bit-stats, i.e. instead of returning an
222	error when stat'ing a large file, the results will be silently truncated	540	error when stat'ing a large file, the results will be silently truncated
223	unless perl itself is compiled with large file support.	541	unless perl itself is compiled with large file support.
		542
		543	To help interpret the mode and dev/rdev stat values, IO::AIO offers the
		544	following constants and functions (if not implemented, the constants will
		545	be C<0> and the functions will either C<croak> or fall back on traditional
		546	behaviour).
		547
		548	C<S_IFMT>, C<S_IFIFO>, C<S_IFCHR>, C<S_IFBLK>, C<S_IFLNK>, C<S_IFREG>,
		549	C<S_IFDIR>, C<S_IFWHT>, C<S_IFSOCK>, C<IO::AIO::major $dev_t>,
		550	C<IO::AIO::minor $dev_t>, C<IO::AIO::makedev $major, $minor>.
224		551
225	Example: Print the length of F</etc/passwd>:	552	Example: Print the length of F</etc/passwd>:
226		553
227	aio_stat "/etc/passwd", sub {	554	aio_stat "/etc/passwd", sub {
228	$_[0] and die "stat failed: $!";	555	$_[0] and die "stat failed: $!";
229	print "size is ", -s _, "\n";	556	print "size is ", -s _, "\n";
230	};	557	};
231		558
		559
		560	=item aio_statvfs $fh_or_path, $callback->($statvfs)
		561
		562	Works like the POSIX C<statvfs> or C<fstatvfs> syscalls, depending on
		563	whether a file handle or path was passed.
		564
		565	On success, the callback is passed a hash reference with the following
		566	members: C<bsize>, C<frsize>, C<blocks>, C<bfree>, C<bavail>, C<files>,
		567	C<ffree>, C<favail>, C<fsid>, C<flag> and C<namemax>. On failure, C<undef>
		568	is passed.
		569
		570	The following POSIX IO::AIO::ST_* constants are defined: C<ST_RDONLY> and
		571	C<ST_NOSUID>.
		572
		573	The following non-POSIX IO::AIO::ST_* flag masks are defined to
		574	their correct value when available, or to C<0> on systems that do
		575	not support them: C<ST_NODEV>, C<ST_NOEXEC>, C<ST_SYNCHRONOUS>,
		576	C<ST_MANDLOCK>, C<ST_WRITE>, C<ST_APPEND>, C<ST_IMMUTABLE>, C<ST_NOATIME>,
		577	C<ST_NODIRATIME> and C<ST_RELATIME>.
		578
		579	Example: stat C</wd> and dump out the data if successful.
		580
		581	aio_statvfs "/wd", sub {
		582	my $f = $_[0]
		583	or die "statvfs: $!";
		584
		585	use Data::Dumper;
		586	say Dumper $f;
		587	};
		588
		589	# result:
		590	{
		591	bsize => 1024,
		592	bfree => 4333064312,
		593	blocks => 10253828096,
		594	files => 2050765568,
		595	flag => 4096,
		596	favail => 2042092649,
		597	bavail => 4333064312,
		598	ffree => 2042092649,
		599	namemax => 255,
		600	frsize => 1024,
		601	fsid => 1810
		602	}
		603
		604
		605	=item aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
		606
		607	Works like perl's C<utime> function (including the special case of $atime
		608	and $mtime being undef). Fractional times are supported if the underlying
		609	syscalls support them.
		610
		611	When called with a pathname, uses utimes(2) if available, otherwise
		612	utime(2). If called on a file descriptor, uses futimes(2) if available,
		613	otherwise returns ENOSYS, so this is not portable.
		614
		615	Examples:
		616
		617	# set atime and mtime to current time (basically touch(1)):
		618	aio_utime "path", undef, undef;
		619	# set atime to current time and mtime to beginning of the epoch:
		620	aio_utime "path", time, undef; # undef==0
		621
		622
		623	=item aio_chown $fh_or_path, $uid, $gid, $callback->($status)
		624
		625	Works like perl's C<chown> function, except that C<undef> for either $uid
		626	or $gid is being interpreted as "do not change" (but -1 can also be used).
		627
		628	Examples:
		629
		630	# same as "chown root path" in the shell:
		631	aio_chown "path", 0, -1;
		632	# same as above:
		633	aio_chown "path", 0, undef;
		634
		635
		636	=item aio_truncate $fh_or_path, $offset, $callback->($status)
		637
		638	Works like truncate(2) or ftruncate(2).
		639
		640
		641	=item aio_chmod $fh_or_path, $mode, $callback->($status)
		642
		643	Works like perl's C<chmod> function.
		644
		645
232	=item aio_unlink $pathname, $callback->($status)	646	=item aio_unlink $pathname, $callback->($status)
233		647
234	Asynchronously unlink (delete) a file and call the callback with the	648	Asynchronously unlink (delete) a file and call the callback with the
235	result code.	649	result code.
236		650
		651
		652	=item aio_mknod $pathname, $mode, $dev, $callback->($status)
		653
		654	[EXPERIMENTAL]
		655
		656	Asynchronously create a device node (or fifo). See mknod(2).
		657
		658	The only (POSIX-) portable way of calling this function is:
		659
		660	aio_mknod $pathname, IO::AIO::S_IFIFO | $mode, 0, sub { ...
		661
		662	See C<aio_stat> for info about some potentially helpful extra constants
		663	and functions.
		664
		665	=item aio_link $srcpath, $dstpath, $callback->($status)
		666
		667	Asynchronously create a new link to the existing object at C<$srcpath> at
		668	the path C<$dstpath> and call the callback with the result code.
		669
		670
		671	=item aio_symlink $srcpath, $dstpath, $callback->($status)
		672
		673	Asynchronously create a new symbolic link to the existing object at C<$srcpath> at
		674	the path C<$dstpath> and call the callback with the result code.
		675
		676
		677	=item aio_readlink $pathname, $callback->($link)
		678
		679	Asynchronously read the symlink specified by C<$path> and pass it to
		680	the callback. If an error occurs, nothing or undef gets passed to the
		681	callback.
		682
		683
		684	=item aio_realpath $pathname, $callback->($path)
		685
		686	Asynchronously make the path absolute and resolve any symlinks in
		687	C<$path>. The resulting path only consists of directories (Same as
		688	L<Cwd::realpath>).
		689
		690	This request can be used to get the absolute path of the current working
		691	directory by passing it a path of F<.> (a single dot).
		692
		693
		694	=item aio_rename $srcpath, $dstpath, $callback->($status)
		695
		696	Asynchronously rename the object at C<$srcpath> to C<$dstpath>, just as
		697	rename(2) and call the callback with the result code.
		698
		699
		700	=item aio_mkdir $pathname, $mode, $callback->($status)
		701
		702	Asynchronously mkdir (create) a directory and call the callback with
		703	the result code. C<$mode> will be modified by the umask at the time the
		704	request is executed, so do not change your umask.
		705
		706
237	=item aio_rmdir $pathname, $callback->($status)	707	=item aio_rmdir $pathname, $callback->($status)
238		708
239	Asynchronously rmdir (delete) a directory and call the callback with the	709	Asynchronously rmdir (delete) a directory and call the callback with the
240	result code.	710	result code.
241		711
		712
242	=item aio_readdir $pathname $callback->($entries)	713	=item aio_readdir $pathname, $callback->($entries)
243		714
244	Unlike the POSIX call of the same name, C<aio_readdir> reads an entire	715	Unlike the POSIX call of the same name, C<aio_readdir> reads an entire
245	directory (i.e. opendir + readdir + closedir). The entries will not be	716	directory (i.e. opendir + readdir + closedir). The entries will not be
246	sorted, and will B<NOT> include the C<.> and C<..> entries.	717	sorted, and will B<NOT> include the C<.> and C<..> entries.
247		718
248	The callback a single argument which is either C<undef> or an array-ref	719	The callback is passed a single argument which is either C<undef> or an
249	with the filenames.	720	array-ref with the filenames.
250		721
		722
		723	=item aio_readdirx $pathname, $flags, $callback->($entries, $flags)
		724
		725	Quite similar to C<aio_readdir>, but the C<$flags> argument allows one to
		726	tune behaviour and output format. In case of an error, C<$entries> will be
		727	C<undef>.
		728
		729	The flags are a combination of the following constants, ORed together (the
		730	flags will also be passed to the callback, possibly modified):
		731
		732	=over 4
		733
		734	=item IO::AIO::READDIR_DENTS
		735
		736	When this flag is off, then the callback gets an arrayref consisting of
		737	names only (as with C<aio_readdir>), otherwise it gets an arrayref with
		738	C<[$name, $type, $inode]> arrayrefs, each describing a single directory
		739	entry in more detail.
		740
		741	C<$name> is the name of the entry.
		742
		743	C<$type> is one of the C<IO::AIO::DT_xxx> constants:
		744
		745	C<IO::AIO::DT_UNKNOWN>, C<IO::AIO::DT_FIFO>, C<IO::AIO::DT_CHR>, C<IO::AIO::DT_DIR>,
		746	C<IO::AIO::DT_BLK>, C<IO::AIO::DT_REG>, C<IO::AIO::DT_LNK>, C<IO::AIO::DT_SOCK>,
		747	C<IO::AIO::DT_WHT>.
		748
		749	C<IO::AIO::DT_UNKNOWN> means just that: readdir does not know. If you need to
		750	know, you have to run stat yourself. Also, for speed reasons, the C<$type>
		751	scalars are read-only: you can not modify them.
		752
		753	C<$inode> is the inode number (which might not be exact on systems with 64
		754	bit inode numbers and 32 bit perls). This field has unspecified content on
		755	systems that do not deliver the inode information.
		756
		757	=item IO::AIO::READDIR_DIRS_FIRST
		758
		759	When this flag is set, then the names will be returned in an order where
		760	likely directories come first, in optimal stat order. This is useful when
		761	you need to quickly find directories, or you want to find all directories
		762	while avoiding to stat() each entry.
		763
		764	If the system returns type information in readdir, then this is used
		765	to find directories directly. Otherwise, likely directories are names
		766	beginning with ".", or otherwise names with no dots, of which names with
		767	short names are tried first.
		768
		769	=item IO::AIO::READDIR_STAT_ORDER
		770
		771	When this flag is set, then the names will be returned in an order
		772	suitable for stat()'ing each one. That is, when you plan to stat()
		773	all files in the given directory, then the returned order will likely
		774	be fastest.
		775
		776	If both this flag and C<IO::AIO::READDIR_DIRS_FIRST> are specified, then
		777	the likely dirs come first, resulting in a less optimal stat order.
		778
		779	=item IO::AIO::READDIR_FOUND_UNKNOWN
		780
		781	This flag should not be set when calling C<aio_readdirx>. Instead, it
		782	is being set by C<aio_readdirx>, when any of the C<$type>'s found were
		783	C<IO::AIO::DT_UNKNOWN>. The absence of this flag therefore indicates that all
		784	C<$type>'s are known, which can be used to speed up some algorithms.
		785
		786	=back
		787
		788
		789	=item aio_load $pathname, $data, $callback->($status)
		790
		791	This is a composite request that tries to fully load the given file into
		792	memory. Status is the same as with aio_read.
		793
		794	=cut
		795
		796	sub aio_load($$;$) {
		797	my ($path, undef, $cb) = @_;
		798	my $data = \$_[1];
		799
		800	my $pri = aioreq_pri;
		801	my $grp = aio_group $cb;
		802
		803	aioreq_pri $pri;
		804	add $grp aio_open $path, O_RDONLY, 0, sub {
		805	my $fh = shift
		806	or return $grp->result (-1);
		807
		808	aioreq_pri $pri;
		809	add $grp aio_read $fh, 0, (-s $fh), $$data, 0, sub {
		810	$grp->result ($_[0]);
		811	};
		812	};
		813
		814	$grp
		815	}
		816
		817	=item aio_copy $srcpath, $dstpath, $callback->($status)
		818
		819	Try to copy the I<file> (directories not supported as either source or
		820	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		821	a status of C<0> (ok) or C<-1> (error, see C<$!>).
		822
		823	This is a composite request that creates the destination file with
		824	mode 0200 and copies the contents of the source file into it using
		825	C<aio_sendfile>, followed by restoring atime, mtime, access mode and
		826	uid/gid, in that order.
		827
		828	If an error occurs, the partial destination file will be unlinked, if
		829	possible, except when setting atime, mtime, access mode and uid/gid, where
		830	errors are being ignored.
		831
		832	=cut
		833
		834	sub aio_copy($$;$) {
		835	my ($src, $dst, $cb) = @_;
		836
		837	my $pri = aioreq_pri;
		838	my $grp = aio_group $cb;
		839
		840	aioreq_pri $pri;
		841	add $grp aio_open $src, O_RDONLY, 0, sub {
		842	if (my $src_fh = $_[0]) {
		843	my @stat = stat $src_fh; # hmm, might block over nfs?
		844
		845	aioreq_pri $pri;
		846	add $grp aio_open $dst, O_CREAT | O_WRONLY | O_TRUNC, 0200, sub {
		847	if (my $dst_fh = $_[0]) {
		848	aioreq_pri $pri;
		849	add $grp aio_sendfile $dst_fh, $src_fh, 0, $stat[7], sub {
		850	if ($_[0] == $stat[7]) {
		851	$grp->result (0);
		852	close $src_fh;
		853
		854	my $ch = sub {
		855	aioreq_pri $pri;
		856	add $grp aio_chmod $dst_fh, $stat[2] & 07777, sub {
		857	aioreq_pri $pri;
		858	add $grp aio_chown $dst_fh, $stat[4], $stat[5], sub {
		859	aioreq_pri $pri;
		860	add $grp aio_close $dst_fh;
		861	}
		862	};
		863	};
		864
		865	aioreq_pri $pri;
		866	add $grp aio_utime $dst_fh, $stat[8], $stat[9], sub {
		867	if ($_[0] < 0 && $! == ENOSYS) {
		868	aioreq_pri $pri;
		869	add $grp aio_utime $dst, $stat[8], $stat[9], $ch;
		870	} else {
		871	$ch->();
		872	}
		873	};
		874	} else {
		875	$grp->result (-1);
		876	close $src_fh;
		877	close $dst_fh;
		878
		879	aioreq $pri;
		880	add $grp aio_unlink $dst;
		881	}
		882	};
		883	} else {
		884	$grp->result (-1);
		885	}
		886	},
		887
		888	} else {
		889	$grp->result (-1);
		890	}
		891	};
		892
		893	$grp
		894	}
		895
		896	=item aio_move $srcpath, $dstpath, $callback->($status)
		897
		898	Try to move the I<file> (directories not supported as either source or
		899	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		900	a status of C<0> (ok) or C<-1> (error, see C<$!>).
		901
		902	This is a composite request that tries to rename(2) the file first; if
		903	rename fails with C<EXDEV>, it copies the file with C<aio_copy> and, if
		904	that is successful, unlinks the C<$srcpath>.
		905
		906	=cut
		907
		908	sub aio_move($$;$) {
		909	my ($src, $dst, $cb) = @_;
		910
		911	my $pri = aioreq_pri;
		912	my $grp = aio_group $cb;
		913
		914	aioreq_pri $pri;
		915	add $grp aio_rename $src, $dst, sub {
		916	if ($_[0] && $! == EXDEV) {
		917	aioreq_pri $pri;
		918	add $grp aio_copy $src, $dst, sub {
		919	$grp->result ($_[0]);
		920
		921	unless ($_[0]) {
		922	aioreq_pri $pri;
		923	add $grp aio_unlink $src;
		924	}
		925	};
		926	} else {
		927	$grp->result ($_[0]);
		928	}
		929	};
		930
		931	$grp
		932	}
		933
251	=item aio_scandir $path, $maxreq, $callback->($dirs, $nondirs)	934	=item aio_scandir $pathname, $maxreq, $callback->($dirs, $nondirs)
252		935
253	Scans a directory (similar to C<aio_readdir>) and tries to separate the	936	Scans a directory (similar to C<aio_readdir>) but additionally tries to
254	entries of directory C<$path> into two sets of names, ones you can recurse	937	efficiently separate the entries of directory C<$path> into two sets of
255	into (directories), and ones you cannot recurse into (everything else).	938	names, directories you can recurse into (directories), and ones you cannot
		939	recurse into (everything else, including symlinks to directories).
256		940
257	C<aio_scandir> is a composite request that consists of many	941	C<aio_scandir> is a composite request that creates of many sub requests_
258	aio-primitives. C<$maxreq> specifies the maximum number of outstanding	942	C<$maxreq> specifies the maximum number of outstanding aio requests that
259	aio requests that this function generates. If it is C<< <= 0 >>, then a	943	this function generates. If it is C<< <= 0 >>, then a suitable default
260	suitable default will be chosen (currently 8).	944	will be chosen (currently 4).
261		945
262	On error, the callback is called without arguments, otherwise it receives	946	On error, the callback is called without arguments, otherwise it receives
263	two array-refs with path-relative entry names.	947	two array-refs with path-relative entry names.
264		948
265	Example:	949	Example:
…		…
272		956
273	Implementation notes.	957	Implementation notes.
274		958
275	The C<aio_readdir> cannot be avoided, but C<stat()>'ing every entry can.	959	The C<aio_readdir> cannot be avoided, but C<stat()>'ing every entry can.
276		960
		961	If readdir returns file type information, then this is used directly to
		962	find directories.
		963
277	After reading the directory, the modification time, size etc. of the	964	Otherwise, after reading the directory, the modification time, size etc.
278	directory before and after the readdir is checked, and if they match, the	965	of the directory before and after the readdir is checked, and if they
279	link count will be used to decide how many entries are directories (if	966	match (and isn't the current time), the link count will be used to decide
280	>= 2). Otherwise, no knowledge of the number of subdirectories will be	967	how many entries are directories (if >= 2). Otherwise, no knowledge of the
281	assumed.	968	number of subdirectories will be assumed.
282		969
283	Then entires will be sorted into likely directories (everything without a	970	Then entries will be sorted into likely directories a non-initial dot
284	non-initial dot) and likely non-directories (everything else). Then every	971	currently) and likely non-directories (see C<aio_readdirx>). Then every
285	entry + C</.> will be C<stat>'ed, likely directories first. This is often	972	entry plus an appended C</.> will be C<stat>'ed, likely directories first,
		973	in order of their inode numbers. If that succeeds, it assumes that the
		974	entry is a directory or a symlink to directory (which will be checked
		975	separately). This is often faster than stat'ing the entry itself because
286	faster because filesystems might detect the type of the entry without	976	filesystems might detect the type of the entry without reading the inode
287	reading the inode data (e.g. ext2s filetype feature). If that succeeds,	977	data (e.g. ext2fs filetype feature), even on systems that cannot return
288	it assumes that the entry is a directory or a symlink to directory (which	978	the filetype information on readdir.
289	will be checked seperately).
290		979
291	If the known number of directories has been reached, the rest of the	980	If the known number of directories (link count - 2) has been reached, the
292	entries is assumed to be non-directories.	981	rest of the entries is assumed to be non-directories.
		982
		983	This only works with certainty on POSIX (= UNIX) filesystems, which
		984	fortunately are the vast majority of filesystems around.
		985
		986	It will also likely work on non-POSIX filesystems with reduced efficiency
		987	as those tend to return 0 or 1 as link counts, which disables the
		988	directory counting heuristic.
293		989
294	=cut	990	=cut
295		991
296	sub aio_scandir($$$) {	992	sub aio_scandir($$;$) {
297	my ($path, $maxreq, $cb) = @_;	993	my ($path, $maxreq, $cb) = @_;
298		994
		995	my $pri = aioreq_pri;
		996
		997	my $grp = aio_group $cb;
		998
299	$maxreq = 8 if $maxreq <= 0;	999	$maxreq = 4 if $maxreq <= 0;
300		1000
		1001	# get a wd object
		1002	aioreq_pri $pri;
		1003	add $grp aio_wd $path, sub {
		1004	$_[0]
		1005	or return $grp->result ();
		1006
		1007	my $wd = [shift, "."];
		1008
301	# stat once	1009	# stat once
302	aio_stat $path, sub {	1010	aioreq_pri $pri;
303	$cb->() if $_[0];	1011	add $grp aio_stat $wd, sub {
		1012	return $grp->result () if $_[0];
		1013	my $now = time;
304	my $hash1 = join ":", (stat _)[0,1,3,7,9];	1014	my $hash1 = join ":", (stat _)[0,1,3,7,9];
305		1015
306	# read the directory entries	1016	# read the directory entries
307	aio_readdir $path, sub {	1017	aioreq_pri $pri;
		1018	add $grp aio_readdirx $wd, READDIR_DIRS_FIRST, sub {
308	my $entries = shift	1019	my $entries = shift
309	or return $cb->();	1020	or return $grp->result ();
310		1021
311	# stat the dir another time	1022	# stat the dir another time
312	aio_stat $path, sub {	1023	aioreq_pri $pri;
		1024	add $grp aio_stat $wd, sub {
313	my $hash2 = join ":", (stat _)[0,1,3,7,9];	1025	my $hash2 = join ":", (stat _)[0,1,3,7,9];
314		1026
315	my $ndirs;	1027	my $ndirs;
316		1028
317	# take the slow route if anything looks fishy	1029	# take the slow route if anything looks fishy
318	if ($hash1 ne $hash2) {	1030	if ($hash1 ne $hash2 or (stat _)[9] == $now) {
319	$ndirs = -1;	1031	$ndirs = -1;
320	} else {	1032	} else {
321	# if nlink == 2, we are finished	1033	# if nlink == 2, we are finished
322	# on non-posix-fs's, we rely on nlink < 2	1034	# for non-posix-fs's, we rely on nlink < 2
323	$ndirs = (stat _)[3] - 2	1035	$ndirs = (stat _)[3] - 2
324	or return $cb->([], $entries);	1036	or return $grp->result ([], $entries);
325	}
326
327	# sort into likely dirs and likely nondirs
328	# dirs == files without ".", short entries first
329	$entries = [map $_->[0],
330	sort { $b->[1] cmp $a->[1] }
331	map [$_, sprintf "%s%04d", (/.\./ ? "1" : "0"), length],
332	@$entries];
333
334	my (@dirs, @nondirs);
335
336	my ($statcb, $schedcb);
337	my $nreq = 0;
338
339	$schedcb = sub {
340	if (@$entries) {
341	if ($nreq < $maxreq) {
342	my $ent = pop @$entries;
343	$nreq++;
344	aio_stat "$path/$ent/.", sub { $statcb->($_[0], $ent) };
345	}
346	} elsif (!$nreq) {
347	# finished
348	undef $statcb;
349	undef $schedcb;
350	$cb->(\@dirs, \@nondirs);
351	undef $cb;
352	}	1037	}
		1038
		1039	my (@dirs, @nondirs);
		1040
		1041	my $statgrp = add $grp aio_group sub {
		1042	$grp->result (\@dirs, \@nondirs);
		1043	};
		1044
		1045	limit $statgrp $maxreq;
		1046	feed $statgrp sub {
		1047	return unless @$entries;
		1048	my $entry = shift @$entries;
		1049
		1050	aioreq_pri $pri;
		1051	$wd->[1] = "$entry/.";
		1052	add $statgrp aio_stat $wd, sub {
		1053	if ($_[0] < 0) {
		1054	push @nondirs, $entry;
		1055	} else {
		1056	# need to check for real directory
		1057	aioreq_pri $pri;
		1058	$wd->[1] = $entry;
		1059	add $statgrp aio_lstat $wd, sub {
		1060	if (-d _) {
		1061	push @dirs, $entry;
		1062
		1063	unless (--$ndirs) {
		1064	push @nondirs, @$entries;
		1065	feed $statgrp;
		1066	}
		1067	} else {
		1068	push @nondirs, $entry;
		1069	}
		1070	}
		1071	}
		1072	};
		1073	};
353	};	1074	};
354	$statcb = sub {
355	my ($status, $entry) = @_;
356
357	if ($status < 0) {
358	$nreq--;
359	push @nondirs, $entry;
360	&$schedcb;
361	} else {
362	# need to check for real directory
363	aio_lstat "$path/$entry", sub {
364	$nreq--;
365
366	if (-d _) {
367	push @dirs, $entry;
368
369	if (!--$ndirs) {
370	push @nondirs, @$entries;
371	$entries = [];
372	}
373	} else {
374	push @nondirs, $entry;
375	}
376
377	&$schedcb;
378	}
379	}
380	};
381
382	&$schedcb while @$entries && $nreq < $maxreq;
383	};	1075	};
384	};	1076	};
385	};	1077	};
		1078
		1079	$grp
386	}	1080	}
		1081
		1082	=item aio_rmtree $pathname, $callback->($status)
		1083
		1084	Delete a directory tree starting (and including) C<$path>, return the
		1085	status of the final C<rmdir> only. This is a composite request that
		1086	uses C<aio_scandir> to recurse into and rmdir directories, and unlink
		1087	everything else.
		1088
		1089	=cut
		1090
		1091	sub aio_rmtree;
		1092	sub aio_rmtree($;$) {
		1093	my ($path, $cb) = @_;
		1094
		1095	my $pri = aioreq_pri;
		1096	my $grp = aio_group $cb;
		1097
		1098	aioreq_pri $pri;
		1099	add $grp aio_scandir $path, 0, sub {
		1100	my ($dirs, $nondirs) = @_;
		1101
		1102	my $dirgrp = aio_group sub {
		1103	add $grp aio_rmdir $path, sub {
		1104	$grp->result ($_[0]);
		1105	};
		1106	};
		1107
		1108	(aioreq_pri $pri), add $dirgrp aio_rmtree "$path/$_" for @$dirs;
		1109	(aioreq_pri $pri), add $dirgrp aio_unlink "$path/$_" for @$nondirs;
		1110
		1111	add $grp $dirgrp;
		1112	};
		1113
		1114	$grp
		1115	}
		1116
		1117	=item aio_sync $callback->($status)
		1118
		1119	Asynchronously call sync and call the callback when finished.
387		1120
388	=item aio_fsync $fh, $callback->($status)	1121	=item aio_fsync $fh, $callback->($status)
389		1122
390	Asynchronously call fsync on the given filehandle and call the callback	1123	Asynchronously call fsync on the given filehandle and call the callback
391	with the fsync result code.	1124	with the fsync result code.
…		…
396	callback with the fdatasync result code.	1129	callback with the fdatasync result code.
397		1130
398	If this call isn't available because your OS lacks it or it couldn't be	1131	If this call isn't available because your OS lacks it or it couldn't be
399	detected, it will be emulated by calling C<fsync> instead.	1132	detected, it will be emulated by calling C<fsync> instead.
400		1133
		1134	=item aio_syncfs $fh, $callback->($status)
		1135
		1136	Asynchronously call the syncfs syscall to sync the filesystem associated
		1137	to the given filehandle and call the callback with the syncfs result
		1138	code. If syncfs is not available, calls sync(), but returns C<-1> and sets
		1139	errno to C<ENOSYS> nevertheless.
		1140
		1141	=item aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
		1142
		1143	Sync the data portion of the file specified by C<$offset> and C<$length>
		1144	to disk (but NOT the metadata), by calling the Linux-specific
		1145	sync_file_range call. If sync_file_range is not available or it returns
		1146	ENOSYS, then fdatasync or fsync is being substituted.
		1147
		1148	C<$flags> can be a combination of C<IO::AIO::SYNC_FILE_RANGE_WAIT_BEFORE>,
		1149	C<IO::AIO::SYNC_FILE_RANGE_WRITE> and
		1150	C<IO::AIO::SYNC_FILE_RANGE_WAIT_AFTER>: refer to the sync_file_range
		1151	manpage for details.
		1152
		1153	=item aio_pathsync $pathname, $callback->($status)
		1154
		1155	This request tries to open, fsync and close the given path. This is a
		1156	composite request intended to sync directories after directory operations
		1157	(E.g. rename). This might not work on all operating systems or have any
		1158	specific effect, but usually it makes sure that directory changes get
		1159	written to disc. It works for anything that can be opened for read-only,
		1160	not just directories.
		1161
		1162	Future versions of this function might fall back to other methods when
		1163	C<fsync> on the directory fails (such as calling C<sync>).
		1164
		1165	Passes C<0> when everything went ok, and C<-1> on error.
		1166
		1167	=cut
		1168
		1169	sub aio_pathsync($;$) {
		1170	my ($path, $cb) = @_;
		1171
		1172	my $pri = aioreq_pri;
		1173	my $grp = aio_group $cb;
		1174
		1175	aioreq_pri $pri;
		1176	add $grp aio_open $path, O_RDONLY, 0, sub {
		1177	my ($fh) = @_;
		1178	if ($fh) {
		1179	aioreq_pri $pri;
		1180	add $grp aio_fsync $fh, sub {
		1181	$grp->result ($_[0]);
		1182
		1183	aioreq_pri $pri;
		1184	add $grp aio_close $fh;
		1185	};
		1186	} else {
		1187	$grp->result (-1);
		1188	}
		1189	};
		1190
		1191	$grp
		1192	}
		1193
		1194	=item aio_msync $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		1195
		1196	This is a rather advanced IO::AIO call, which only works on mmap(2)ed
		1197	scalars (see the C<IO::AIO::mmap> function, although it also works on data
		1198	scalars managed by the L<Sys::Mmap> or L<Mmap> modules, note that the
		1199	scalar must only be modified in-place while an aio operation is pending on
		1200	it).
		1201
		1202	It calls the C<msync> function of your OS, if available, with the memory
		1203	area starting at C<$offset> in the string and ending C<$length> bytes
		1204	later. If C<$length> is negative, counts from the end, and if C<$length>
		1205	is C<undef>, then it goes till the end of the string. The flags can be
		1206	a combination of C<IO::AIO::MS_ASYNC>, C<IO::AIO::MS_INVALIDATE> and
		1207	C<IO::AIO::MS_SYNC>.
		1208
		1209	=item aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		1210
		1211	This is a rather advanced IO::AIO call, which works best on mmap(2)ed
		1212	scalars.
		1213
		1214	It touches (reads or writes) all memory pages in the specified
		1215	range inside the scalar. All caveats and parameters are the same
		1216	as for C<aio_msync>, above, except for flags, which must be either
		1217	C<0> (which reads all pages and ensures they are instantiated) or
		1218	C<IO::AIO::MT_MODIFY>, which modifies the memory page s(by reading and
		1219	writing an octet from it, which dirties the page).
		1220
		1221	=item aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status)
		1222
		1223	This is a rather advanced IO::AIO call, which works best on mmap(2)ed
		1224	scalars.
		1225
		1226	It reads in all the pages of the underlying storage into memory (if any)
		1227	and locks them, so they are not getting swapped/paged out or removed.
		1228
		1229	If C<$length> is undefined, then the scalar will be locked till the end.
		1230
		1231	On systems that do not implement C<mlock>, this function returns C<-1>
		1232	and sets errno to C<ENOSYS>.
		1233
		1234	Note that the corresponding C<munlock> is synchronous and is
		1235	documented under L<MISCELLANEOUS FUNCTIONS>.
		1236
		1237	Example: open a file, mmap and mlock it - both will be undone when
		1238	C<$data> gets destroyed.
		1239
		1240	open my $fh, "<", $path or die "$path: $!";
		1241	my $data;
		1242	IO::AIO::mmap $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh;
		1243	aio_mlock $data; # mlock in background
		1244
		1245	=item aio_mlockall $flags, $callback->($status)
		1246
		1247	Calls the C<mlockall> function with the given C<$flags> (a combination of
		1248	C<IO::AIO::MCL_CURRENT> and C<IO::AIO::MCL_FUTURE>).
		1249
		1250	On systems that do not implement C<mlockall>, this function returns C<-1>
		1251	and sets errno to C<ENOSYS>.
		1252
		1253	Note that the corresponding C<munlockall> is synchronous and is
		1254	documented under L<MISCELLANEOUS FUNCTIONS>.
		1255
		1256	Example: asynchronously lock all current and future pages into memory.
		1257
		1258	aio_mlockall IO::AIO::MCL_FUTURE;
		1259
		1260	=item aio_fiemap $fh, $start, $length, $flags, $count, $cb->(\@extents)
		1261
		1262	Queries the extents of the given file (by calling the Linux FIEMAP ioctl,
		1263	see L<http://cvs.schmorp.de/IO-AIO/doc/fiemap.txt> for details). If the
		1264	C<ioctl> is not available on your OS, then this rquiest will fail with
		1265	C<ENOSYS>.
		1266
		1267	C<$start> is the starting offset to query extents for, C<$length> is the
		1268	size of the range to query - if it is C<undef>, then the whole file will
		1269	be queried.
		1270
		1271	C<$flags> is a combination of flags (C<IO::AIO::FIEMAP_FLAG_SYNC> or
		1272	C<IO::AIO::FIEMAP_FLAG_XATTR> - C<IO::AIO::FIEMAP_FLAGS_COMPAT> is also
		1273	exported), and is normally C<0> or C<IO::AIO::FIEMAP_FLAG_SYNC> to query
		1274	the data portion.
		1275
		1276	C<$count> is the maximum number of extent records to return. If it is
		1277	C<undef>, then IO::AIO queries all extents of the file. As a very special
		1278	case, if it is C<0>, then the callback receives the number of extents
		1279	instead of the extents themselves.
		1280
		1281	If an error occurs, the callback receives no arguments. The special
		1282	C<errno> value C<IO::AIO::EBADR> is available to test for flag errors.
		1283
		1284	Otherwise, the callback receives an array reference with extent
		1285	structures. Each extent structure is an array reference itself, with the
		1286	following members:
		1287
		1288	[$logical, $physical, $length, $flags]
		1289
		1290	Flags is any combination of the following flag values (typically either C<0>
		1291	or C<IO::AIO::FIEMAP_EXTENT_LAST>):
		1292
		1293	C<IO::AIO::FIEMAP_EXTENT_LAST>, C<IO::AIO::FIEMAP_EXTENT_UNKNOWN>,
		1294	C<IO::AIO::FIEMAP_EXTENT_DELALLOC>, C<IO::AIO::FIEMAP_EXTENT_ENCODED>,
		1295	C<IO::AIO::FIEMAP_EXTENT_DATA_ENCRYPTED>, C<IO::AIO::FIEMAP_EXTENT_NOT_ALIGNED>,
		1296	C<IO::AIO::FIEMAP_EXTENT_DATA_INLINE>, C<IO::AIO::FIEMAP_EXTENT_DATA_TAIL>,
		1297	C<IO::AIO::FIEMAP_EXTENT_UNWRITTEN>, C<IO::AIO::FIEMAP_EXTENT_MERGED> or
		1298	C<IO::AIO::FIEMAP_EXTENT_SHARED>.
		1299
		1300	=item aio_group $callback->(...)
		1301
		1302	This is a very special aio request: Instead of doing something, it is a
		1303	container for other aio requests, which is useful if you want to bundle
		1304	many requests into a single, composite, request with a definite callback
		1305	and the ability to cancel the whole request with its subrequests.
		1306
		1307	Returns an object of class L<IO::AIO::GRP>. See its documentation below
		1308	for more info.
		1309
		1310	Example:
		1311
		1312	my $grp = aio_group sub {
		1313	print "all stats done\n";
		1314	};
		1315
		1316	add $grp
		1317	(aio_stat ...),
		1318	(aio_stat ...),
		1319	...;
		1320
		1321	=item aio_nop $callback->()
		1322
		1323	This is a special request - it does nothing in itself and is only used for
		1324	side effects, such as when you want to add a dummy request to a group so
		1325	that finishing the requests in the group depends on executing the given
		1326	code.
		1327
		1328	While this request does nothing, it still goes through the execution
		1329	phase and still requires a worker thread. Thus, the callback will not
		1330	be executed immediately but only after other requests in the queue have
		1331	entered their execution phase. This can be used to measure request
		1332	latency.
		1333
		1334	=item IO::AIO::aio_busy $fractional_seconds, $callback->() *NOT EXPORTED*
		1335
		1336	Mainly used for debugging and benchmarking, this aio request puts one of
		1337	the request workers to sleep for the given time.
		1338
		1339	While it is theoretically handy to have simple I/O scheduling requests
		1340	like sleep and file handle readable/writable, the overhead this creates is
		1341	immense (it blocks a thread for a long time) so do not use this function
		1342	except to put your application under artificial I/O pressure.
		1343
401	=back	1344	=back
402		1345
		1346
		1347	=head2 IO::AIO::WD - multiple working directories
		1348
		1349	Your process only has one current working directory, which is used by all
		1350	threads. This makes it hard to use relative paths (some other component
		1351	could call C<chdir> at any time, and it is hard to control when the path
		1352	will be used by IO::AIO).
		1353
		1354	One solution for this is to always use absolute paths. This usually works,
		1355	but can be quite slow (the kernel has to walk the whole path on every
		1356	access), and can also be a hassle to implement.
		1357
		1358	Newer POSIX systems have a number of functions (openat, fdopendir,
		1359	futimensat and so on) that make it possible to specify working directories
		1360	per operation.
		1361
		1362	For portability, and because the clowns who "designed", or shall I write,
		1363	perpetrated this new interface were obviously half-drunk, this abstraction
		1364	cannot be perfect, though.
		1365
		1366	IO::AIO allows you to convert directory paths into a so-called IO::AIO::WD
		1367	object. This object stores the canonicalised, absolute version of the
		1368	path, and on systems that allow it, also a directory file descriptor.
		1369
		1370	Everywhere where a pathname is accepted by IO::AIO (e.g. in C<aio_stat>
		1371	or C<aio_unlink>), one can specify an array reference with an IO::AIO::WD
		1372	object and a pathname instead (or the IO::AIO::WD object alone, which
		1373	gets interpreted as C<[$wd, "."]>). If the pathname is absolute, the
		1374	IO::AIO::WD object is ignored, otherwise the pathname is resolved relative
		1375	to that IO::AIO::WD object.
		1376
		1377	For example, to get a wd object for F</etc> and then stat F<passwd>
		1378	inside, you would write:
		1379
		1380	aio_wd "/etc", sub {
		1381	my $etcdir = shift;
		1382
		1383	# although $etcdir can be undef on error, there is generally no reason
		1384	# to check for errors here, as aio_stat will fail with ENOENT
		1385	# when $etcdir is undef.
		1386
		1387	aio_stat [$etcdir, "passwd"], sub {
		1388	# yay
		1389	};
		1390	};
		1391
		1392	That C<aio_wd> is a request and not a normal function shows that creating
		1393	an IO::AIO::WD object is itself a potentially blocking operation, which is
		1394	why it is done asynchronously.
		1395
		1396	To stat the directory obtained with C<aio_wd> above, one could write
		1397	either of the following three request calls:
		1398
		1399	aio_lstat "/etc" , sub { ... # pathname as normal string
		1400	aio_lstat [$wd, "."], sub { ... # "." relative to $wd (i.e. $wd itself)
		1401	aio_lstat $wd , sub { ... # shorthand for the previous
		1402
		1403	As with normal pathnames, IO::AIO keeps a copy of the working directory
		1404	object and the pathname string, so you could write the following without
		1405	causing any issues due to C<$path> getting reused:
		1406
		1407	my $path = [$wd, undef];
		1408
		1409	for my $name (qw(abc def ghi)) {
		1410	$path->[1] = $name;
		1411	aio_stat $path, sub {
		1412	# ...
		1413	};
		1414	}
		1415
		1416	There are some caveats: when directories get renamed (or deleted), the
		1417	pathname string doesn't change, so will point to the new directory (or
		1418	nowhere at all), while the directory fd, if available on the system,
		1419	will still point to the original directory. Most functions accepting a
		1420	pathname will use the directory fd on newer systems, and the string on
		1421	older systems. Some functions (such as realpath) will always rely on the
		1422	string form of the pathname.
		1423
		1424	So this fucntionality is mainly useful to get some protection against
		1425	C<chdir>, to easily get an absolute path out of a relative path for future
		1426	reference, and to speed up doing many operations in the same directory
		1427	(e.g. when stat'ing all files in a directory).
		1428
		1429	The following functions implement this working directory abstraction:
		1430
		1431	=over 4
		1432
		1433	=item aio_wd $pathname, $callback->($wd)
		1434
		1435	Asynchonously canonicalise the given pathname and convert it to an
		1436	IO::AIO::WD object representing it. If possible and supported on the
		1437	system, also open a directory fd to speed up pathname resolution relative
		1438	to this working directory.
		1439
		1440	If something goes wrong, then C<undef> is passwd to the callback instead
		1441	of a working directory object and C<$!> is set appropriately. Since
		1442	passing C<undef> as working directory component of a pathname fails the
		1443	request with C<ENOENT>, there is often no need for error checking in the
		1444	C<aio_wd> callback, as future requests using the value will fail in the
		1445	expected way.
		1446
		1447	If this call isn't available because your OS lacks it or it couldn't be
		1448	detected, it will be emulated by calling C<fsync> instead.
		1449
		1450	=item IO::AIO::CWD
		1451
		1452	This is a compiletime constant (object) that represents the process
		1453	current working directory.
		1454
		1455	Specifying this object as working directory object for a pathname is as
		1456	if the pathname would be specified directly, without a directory object,
		1457	e.g., these calls are functionally identical:
		1458
		1459	aio_stat "somefile", sub { ... };
		1460	aio_stat [IO::AIO::CWD, "somefile"], sub { ... };
		1461
		1462	=back
		1463
		1464
		1465	=head2 IO::AIO::REQ CLASS
		1466
		1467	All non-aggregate C<aio_*> functions return an object of this class when
		1468	called in non-void context.
		1469
		1470	=over 4
		1471
		1472	=item cancel $req
		1473
		1474	Cancels the request, if possible. Has the effect of skipping execution
		1475	when entering the B<execute> state and skipping calling the callback when
		1476	entering the the B<result> state, but will leave the request otherwise
		1477	untouched (with the exception of readdir). That means that requests that
		1478	currently execute will not be stopped and resources held by the request
		1479	will not be freed prematurely.
		1480
		1481	=item cb $req $callback->(...)
		1482
		1483	Replace (or simply set) the callback registered to the request.
		1484
		1485	=back
		1486
		1487	=head2 IO::AIO::GRP CLASS
		1488
		1489	This class is a subclass of L<IO::AIO::REQ>, so all its methods apply to
		1490	objects of this class, too.
		1491
		1492	A IO::AIO::GRP object is a special request that can contain multiple other
		1493	aio requests.
		1494
		1495	You create one by calling the C<aio_group> constructing function with a
		1496	callback that will be called when all contained requests have entered the
		1497	C<done> state:
		1498
		1499	my $grp = aio_group sub {
		1500	print "all requests are done\n";
		1501	};
		1502
		1503	You add requests by calling the C<add> method with one or more
		1504	C<IO::AIO::REQ> objects:
		1505
		1506	$grp->add (aio_unlink "...");
		1507
		1508	add $grp aio_stat "...", sub {
		1509	$_[0] or return $grp->result ("error");
		1510
		1511	# add another request dynamically, if first succeeded
		1512	add $grp aio_open "...", sub {
		1513	$grp->result ("ok");
		1514	};
		1515	};
		1516
		1517	This makes it very easy to create composite requests (see the source of
		1518	C<aio_move> for an application) that work and feel like simple requests.
		1519
		1520	=over 4
		1521
		1522	=item * The IO::AIO::GRP objects will be cleaned up during calls to
		1523	C<IO::AIO::poll_cb>, just like any other request.
		1524
		1525	=item * They can be canceled like any other request. Canceling will cancel not
		1526	only the request itself, but also all requests it contains.
		1527
		1528	=item * They can also can also be added to other IO::AIO::GRP objects.
		1529
		1530	=item * You must not add requests to a group from within the group callback (or
		1531	any later time).
		1532
		1533	=back
		1534
		1535	Their lifetime, simplified, looks like this: when they are empty, they
		1536	will finish very quickly. If they contain only requests that are in the
		1537	C<done> state, they will also finish. Otherwise they will continue to
		1538	exist.
		1539
		1540	That means after creating a group you have some time to add requests
		1541	(precisely before the callback has been invoked, which is only done within
		1542	the C<poll_cb>). And in the callbacks of those requests, you can add
		1543	further requests to the group. And only when all those requests have
		1544	finished will the the group itself finish.
		1545
		1546	=over 4
		1547
		1548	=item add $grp ...
		1549
		1550	=item $grp->add (...)
		1551
		1552	Add one or more requests to the group. Any type of L<IO::AIO::REQ> can
		1553	be added, including other groups, as long as you do not create circular
		1554	dependencies.
		1555
		1556	Returns all its arguments.
		1557
		1558	=item $grp->cancel_subs
		1559
		1560	Cancel all subrequests and clears any feeder, but not the group request
		1561	itself. Useful when you queued a lot of events but got a result early.
		1562
		1563	The group request will finish normally (you cannot add requests to the
		1564	group).
		1565
		1566	=item $grp->result (...)
		1567
		1568	Set the result value(s) that will be passed to the group callback when all
		1569	subrequests have finished and set the groups errno to the current value
		1570	of errno (just like calling C<errno> without an error number). By default,
		1571	no argument will be passed and errno is zero.
		1572
		1573	=item $grp->errno ([$errno])
		1574
		1575	Sets the group errno value to C<$errno>, or the current value of errno
		1576	when the argument is missing.
		1577
		1578	Every aio request has an associated errno value that is restored when
		1579	the callback is invoked. This method lets you change this value from its
		1580	default (0).
		1581
		1582	Calling C<result> will also set errno, so make sure you either set C<$!>
		1583	before the call to C<result>, or call c<errno> after it.
		1584
		1585	=item feed $grp $callback->($grp)
		1586
		1587	Sets a feeder/generator on this group: every group can have an attached
		1588	generator that generates requests if idle. The idea behind this is that,
		1589	although you could just queue as many requests as you want in a group,
		1590	this might starve other requests for a potentially long time. For example,
		1591	C<aio_scandir> might generate hundreds of thousands of C<aio_stat>
		1592	requests, delaying any later requests for a long time.
		1593
		1594	To avoid this, and allow incremental generation of requests, you can
		1595	instead a group and set a feeder on it that generates those requests. The
		1596	feed callback will be called whenever there are few enough (see C<limit>,
		1597	below) requests active in the group itself and is expected to queue more
		1598	requests.
		1599
		1600	The feed callback can queue as many requests as it likes (i.e. C<add> does
		1601	not impose any limits).
		1602
		1603	If the feed does not queue more requests when called, it will be
		1604	automatically removed from the group.
		1605
		1606	If the feed limit is C<0> when this method is called, it will be set to
		1607	C<2> automatically.
		1608
		1609	Example:
		1610
		1611	# stat all files in @files, but only ever use four aio requests concurrently:
		1612
		1613	my $grp = aio_group sub { print "finished\n" };
		1614	limit $grp 4;
		1615	feed $grp sub {
		1616	my $file = pop @files
		1617	or return;
		1618
		1619	add $grp aio_stat $file, sub { ... };
		1620	};
		1621
		1622	=item limit $grp $num
		1623
		1624	Sets the feeder limit for the group: The feeder will be called whenever
		1625	the group contains less than this many requests.
		1626
		1627	Setting the limit to C<0> will pause the feeding process.
		1628
		1629	The default value for the limit is C<0>, but note that setting a feeder
		1630	automatically bumps it up to C<2>.
		1631
		1632	=back
		1633
403	=head2 SUPPORT FUNCTIONS	1634	=head2 SUPPORT FUNCTIONS
404		1635
		1636	=head3 EVENT PROCESSING AND EVENT LOOP INTEGRATION
		1637
405	=over 4	1638	=over 4
406		1639
407	=item $fileno = IO::AIO::poll_fileno	1640	=item $fileno = IO::AIO::poll_fileno
408		1641
409	Return the I<request result pipe file descriptor>. This filehandle must be	1642	Return the I<request result pipe file descriptor>. This filehandle must be
410	polled for reading by some mechanism outside this module (e.g. Event or	1643	polled for reading by some mechanism outside this module (e.g. EV, Glib,
411	select, see below or the SYNOPSIS). If the pipe becomes readable you have	1644	select and so on, see below or the SYNOPSIS). If the pipe becomes readable
412	to call C<poll_cb> to check the results.	1645	you have to call C<poll_cb> to check the results.
413		1646
414	See C<poll_cb> for an example.	1647	See C<poll_cb> for an example.
415		1648
416	=item IO::AIO::poll_cb	1649	=item IO::AIO::poll_cb
417		1650
418	Process all outstanding events on the result pipe. You have to call this	1651	Process some outstanding events on the result pipe. You have to call
419	regularly. Returns the number of events processed. Returns immediately	1652	this regularly. Returns C<0> if all events could be processed (or there
420	when no events are outstanding.	1653	were no events to process), or C<-1> if it returned earlier for whatever
		1654	reason. Returns immediately when no events are outstanding. The amount of
		1655	events processed depends on the settings of C<IO::AIO::max_poll_req> and
		1656	C<IO::AIO::max_poll_time>.
		1657
		1658	If not all requests were processed for whatever reason, the filehandle
		1659	will still be ready when C<poll_cb> returns, so normally you don't have to
		1660	do anything special to have it called later.
		1661
		1662	Apart from calling C<IO::AIO::poll_cb> when the event filehandle becomes
		1663	ready, it can be beneficial to call this function from loops which submit
		1664	a lot of requests, to make sure the results get processed when they become
		1665	available and not just when the loop is finished and the event loop takes
		1666	over again. This function returns very fast when there are no outstanding
		1667	requests.
421		1668
422	Example: Install an Event watcher that automatically calls	1669	Example: Install an Event watcher that automatically calls
423	IO::AIO::poll_cb with high priority:	1670	IO::AIO::poll_cb with high priority (more examples can be found in the
		1671	SYNOPSIS section, at the top of this document):
424		1672
425	Event->io (fd => IO::AIO::poll_fileno,	1673	Event->io (fd => IO::AIO::poll_fileno,
426	poll => 'r', async => 1,	1674	poll => 'r', async => 1,
427	cb => \&IO::AIO::poll_cb);	1675	cb => \&IO::AIO::poll_cb);
428		1676
429	=item IO::AIO::poll_wait	1677	=item IO::AIO::poll_wait
430		1678
		1679	If there are any outstanding requests and none of them in the result
431	Wait till the result filehandle becomes ready for reading (simply does a	1680	phase, wait till the result filehandle becomes ready for reading (simply
432	C<select> on the filehandle. This is useful if you want to synchronously wait	1681	does a C<select> on the filehandle. This is useful if you want to
433	for some requests to finish).	1682	synchronously wait for some requests to finish).
434		1683
435	See C<nreqs> for an example.	1684	See C<nreqs> for an example.
436		1685
		1686	=item IO::AIO::poll
		1687
		1688	Waits until some requests have been handled.
		1689
		1690	Returns the number of requests processed, but is otherwise strictly
		1691	equivalent to:
		1692
		1693	IO::AIO::poll_wait, IO::AIO::poll_cb
		1694
437	=item IO::AIO::nreqs	1695	=item IO::AIO::flush
438		1696
439	Returns the number of requests currently outstanding (i.e. for which their	1697	Wait till all outstanding AIO requests have been handled.
440	callback has not been invoked yet).
441		1698
442	Example: wait till there are no outstanding requests anymore:	1699	Strictly equivalent to:
443		1700
444	IO::AIO::poll_wait, IO::AIO::poll_cb	1701	IO::AIO::poll_wait, IO::AIO::poll_cb
445	while IO::AIO::nreqs;	1702	while IO::AIO::nreqs;
446		1703
447	=item IO::AIO::flush	1704	=item IO::AIO::max_poll_reqs $nreqs
448		1705
449	Wait till all outstanding AIO requests have been handled.	1706	=item IO::AIO::max_poll_time $seconds
450		1707
451	Strictly equivalent to:	1708	These set the maximum number of requests (default C<0>, meaning infinity)
		1709	that are being processed by C<IO::AIO::poll_cb> in one call, respectively
		1710	the maximum amount of time (default C<0>, meaning infinity) spent in
		1711	C<IO::AIO::poll_cb> to process requests (more correctly the mininum amount
		1712	of time C<poll_cb> is allowed to use).
452		1713
453	IO::AIO::poll_wait, IO::AIO::poll_cb	1714	Setting C<max_poll_time> to a non-zero value creates an overhead of one
454	while IO::AIO::nreqs;	1715	syscall per request processed, which is not normally a problem unless your
		1716	callbacks are really really fast or your OS is really really slow (I am
		1717	not mentioning Solaris here). Using C<max_poll_reqs> incurs no overhead.
455		1718
456	=item IO::AIO::poll	1719	Setting these is useful if you want to ensure some level of
		1720	interactiveness when perl is not fast enough to process all requests in
		1721	time.
457		1722
458	Waits until some requests have been handled.	1723	For interactive programs, values such as C<0.01> to C<0.1> should be fine.
459		1724
460	Strictly equivalent to:	1725	Example: Install an Event watcher that automatically calls
		1726	IO::AIO::poll_cb with low priority, to ensure that other parts of the
		1727	program get the CPU sometimes even under high AIO load.
461		1728
462	IO::AIO::poll_wait, IO::AIO::poll_cb	1729	# try not to spend much more than 0.1s in poll_cb
463	if IO::AIO::nreqs;	1730	IO::AIO::max_poll_time 0.1;
		1731
		1732	# use a low priority so other tasks have priority
		1733	Event->io (fd => IO::AIO::poll_fileno,
		1734	poll => 'r', nice => 1,
		1735	cb => &IO::AIO::poll_cb);
		1736
		1737	=back
		1738
		1739	=head3 CONTROLLING THE NUMBER OF THREADS
		1740
		1741	=over
464		1742
465	=item IO::AIO::min_parallel $nthreads	1743	=item IO::AIO::min_parallel $nthreads
466		1744
467	Set the minimum number of AIO threads to C<$nthreads>. The current default	1745	Set the minimum number of AIO threads to C<$nthreads>. The current
468	is C<4>, which means four asynchronous operations can be done at one time	1746	default is C<8>, which means eight asynchronous operations can execute
469	(the number of outstanding operations, however, is unlimited).	1747	concurrently at any one time (the number of outstanding requests,
		1748	however, is unlimited).
470		1749
471	IO::AIO starts threads only on demand, when an AIO request is queued and	1750	IO::AIO starts threads only on demand, when an AIO request is queued and
472	no free thread exists.	1751	no free thread exists. Please note that queueing up a hundred requests can
		1752	create demand for a hundred threads, even if it turns out that everything
		1753	is in the cache and could have been processed faster by a single thread.
473		1754
474	It is recommended to keep the number of threads low, as some Linux	1755	It is recommended to keep the number of threads relatively low, as some
475	kernel versions will scale negatively with the number of threads (higher	1756	Linux kernel versions will scale negatively with the number of threads
476	parallelity => MUCH higher latency). With current Linux 2.6 versions, 4-32	1757	(higher parallelity => MUCH higher latency). With current Linux 2.6
477	threads should be fine.	1758	versions, 4-32 threads should be fine.
478		1759
479	Under most circumstances you don't need to call this function, as the	1760	Under most circumstances you don't need to call this function, as the
480	module selects a default that is suitable for low to moderate load.	1761	module selects a default that is suitable for low to moderate load.
481		1762
482	=item IO::AIO::max_parallel $nthreads	1763	=item IO::AIO::max_parallel $nthreads
…		…
491	This module automatically runs C<max_parallel 0> at program end, to ensure	1772	This module automatically runs C<max_parallel 0> at program end, to ensure
492	that all threads are killed and that there are no outstanding requests.	1773	that all threads are killed and that there are no outstanding requests.
493		1774
494	Under normal circumstances you don't need to call this function.	1775	Under normal circumstances you don't need to call this function.
495		1776
		1777	=item IO::AIO::max_idle $nthreads
		1778
		1779	Limit the number of threads (default: 4) that are allowed to idle
		1780	(i.e., threads that did not get a request to process within the idle
		1781	timeout (default: 10 seconds). That means if a thread becomes idle while
		1782	C<$nthreads> other threads are also idle, it will free its resources and
		1783	exit.
		1784
		1785	This is useful when you allow a large number of threads (e.g. 100 or 1000)
		1786	to allow for extremely high load situations, but want to free resources
		1787	under normal circumstances (1000 threads can easily consume 30MB of RAM).
		1788
		1789	The default is probably ok in most situations, especially if thread
		1790	creation is fast. If thread creation is very slow on your system you might
		1791	want to use larger values.
		1792
		1793	=item IO::AIO::idle_timeout $seconds
		1794
		1795	Sets the minimum idle timeout (default 10) after which worker threads are
		1796	allowed to exit. SEe C<IO::AIO::max_idle>.
		1797
496	=item $oldnreqs = IO::AIO::max_outstanding $nreqs	1798	=item IO::AIO::max_outstanding $maxreqs
497		1799
498	Sets the maximum number of outstanding requests to C<$nreqs>. If you	1800	Sets the maximum number of outstanding requests to C<$nreqs>. If
499	try to queue up more than this number of requests, the caller will block until	1801	you do queue up more than this number of requests, the next call to
500	some requests have been handled.	1802	C<IO::AIO::poll_cb> (and other functions calling C<poll_cb>, such as
		1803	C<IO::AIO::flush> or C<IO::AIO::poll>) will block until the limit is no
		1804	longer exceeded.
501		1805
502	The default is very large, so normally there is no practical limit. If you	1806	In other words, this setting does not enforce a queue limit, but can be
503	queue up many requests in a loop it often improves speed if you set	1807	used to make poll functions block if the limit is exceeded.
504	this to a relatively low number, such as C<100>.
505		1808
506	Under normal circumstances you don't need to call this function.	1809	This is a very bad function to use in interactive programs because it
		1810	blocks, and a bad way to reduce concurrency because it is inexact: Better
		1811	use an C<aio_group> together with a feed callback.
		1812
		1813	It's main use is in scripts without an event loop - when you want to stat
		1814	a lot of files, you can write somehting like this:
		1815
		1816	IO::AIO::max_outstanding 32;
		1817
		1818	for my $path (...) {
		1819	aio_stat $path , ...;
		1820	IO::AIO::poll_cb;
		1821	}
		1822
		1823	IO::AIO::flush;
		1824
		1825	The call to C<poll_cb> inside the loop will normally return instantly, but
		1826	as soon as more thna C<32> reqeusts are in-flight, it will block until
		1827	some requests have been handled. This keeps the loop from pushing a large
		1828	number of C<aio_stat> requests onto the queue.
		1829
		1830	The default value for C<max_outstanding> is very large, so there is no
		1831	practical limit on the number of outstanding requests.
507		1832
508	=back	1833	=back
509		1834
		1835	=head3 STATISTICAL INFORMATION
		1836
		1837	=over
		1838
		1839	=item IO::AIO::nreqs
		1840
		1841	Returns the number of requests currently in the ready, execute or pending
		1842	states (i.e. for which their callback has not been invoked yet).
		1843
		1844	Example: wait till there are no outstanding requests anymore:
		1845
		1846	IO::AIO::poll_wait, IO::AIO::poll_cb
		1847	while IO::AIO::nreqs;
		1848
		1849	=item IO::AIO::nready
		1850
		1851	Returns the number of requests currently in the ready state (not yet
		1852	executed).
		1853
		1854	=item IO::AIO::npending
		1855
		1856	Returns the number of requests currently in the pending state (executed,
		1857	but not yet processed by poll_cb).
		1858
		1859	=back
		1860
		1861	=head3 MISCELLANEOUS FUNCTIONS
		1862
		1863	IO::AIO implements some functions that might be useful, but are not
		1864	asynchronous.
		1865
		1866	=over 4
		1867
		1868	=item IO::AIO::sendfile $ofh, $ifh, $offset, $count
		1869
		1870	Calls the C<eio_sendfile_sync> function, which is like C<aio_sendfile>,
		1871	but is blocking (this makes most sense if you know the input data is
		1872	likely cached already and the output filehandle is set to non-blocking
		1873	operations).
		1874
		1875	Returns the number of bytes copied, or C<-1> on error.
		1876
		1877	=item IO::AIO::fadvise $fh, $offset, $len, $advice
		1878
		1879	Simply calls the C<posix_fadvise> function (see its
		1880	manpage for details). The following advice constants are
		1881	available: C<IO::AIO::FADV_NORMAL>, C<IO::AIO::FADV_SEQUENTIAL>,
		1882	C<IO::AIO::FADV_RANDOM>, C<IO::AIO::FADV_NOREUSE>,
		1883	C<IO::AIO::FADV_WILLNEED>, C<IO::AIO::FADV_DONTNEED>.
		1884
		1885	On systems that do not implement C<posix_fadvise>, this function returns
		1886	ENOSYS, otherwise the return value of C<posix_fadvise>.
		1887
		1888	=item IO::AIO::madvise $scalar, $offset, $len, $advice
		1889
		1890	Simply calls the C<posix_madvise> function (see its
		1891	manpage for details). The following advice constants are
		1892	available: C<IO::AIO::MADV_NORMAL>, C<IO::AIO::MADV_SEQUENTIAL>,
		1893	C<IO::AIO::MADV_RANDOM>, C<IO::AIO::MADV_WILLNEED>, C<IO::AIO::MADV_DONTNEED>.
		1894
		1895	On systems that do not implement C<posix_madvise>, this function returns
		1896	ENOSYS, otherwise the return value of C<posix_madvise>.
		1897
		1898	=item IO::AIO::mprotect $scalar, $offset, $len, $protect
		1899
		1900	Simply calls the C<mprotect> function on the preferably AIO::mmap'ed
		1901	$scalar (see its manpage for details). The following protect
		1902	constants are available: C<IO::AIO::PROT_NONE>, C<IO::AIO::PROT_READ>,
		1903	C<IO::AIO::PROT_WRITE>, C<IO::AIO::PROT_EXEC>.
		1904
		1905	On systems that do not implement C<mprotect>, this function returns
		1906	ENOSYS, otherwise the return value of C<mprotect>.
		1907
		1908	=item IO::AIO::mmap $scalar, $length, $prot, $flags, $fh[, $offset]
		1909
		1910	Memory-maps a file (or anonymous memory range) and attaches it to the
		1911	given C<$scalar>, which will act like a string scalar.
		1912
		1913	The only operations allowed on the scalar are C<substr>/C<vec> that don't
		1914	change the string length, and most read-only operations such as copying it
		1915	or searching it with regexes and so on.
		1916
		1917	Anything else is unsafe and will, at best, result in memory leaks.
		1918
		1919	The memory map associated with the C<$scalar> is automatically removed
		1920	when the C<$scalar> is destroyed, or when the C<IO::AIO::mmap> or
		1921	C<IO::AIO::munmap> functions are called.
		1922
		1923	This calls the C<mmap>(2) function internally. See your system's manual
		1924	page for details on the C<$length>, C<$prot> and C<$flags> parameters.
		1925
		1926	The C<$length> must be larger than zero and smaller than the actual
		1927	filesize.
		1928
		1929	C<$prot> is a combination of C<IO::AIO::PROT_NONE>, C<IO::AIO::PROT_EXEC>,
		1930	C<IO::AIO::PROT_READ> and/or C<IO::AIO::PROT_WRITE>,
		1931
		1932	C<$flags> can be a combination of C<IO::AIO::MAP_SHARED> or
		1933	C<IO::AIO::MAP_PRIVATE>, or a number of system-specific flags (when
		1934	not available, the are defined as 0): C<IO::AIO::MAP_ANONYMOUS>
		1935	(which is set to C<MAP_ANON> if your system only provides this
		1936	constant), C<IO::AIO::MAP_HUGETLB>, C<IO::AIO::MAP_LOCKED>,
		1937	C<IO::AIO::MAP_NORESERVE>, C<IO::AIO::MAP_POPULATE> or
		1938	C<IO::AIO::MAP_NONBLOCK>
		1939
		1940	If C<$fh> is C<undef>, then a file descriptor of C<-1> is passed.
		1941
		1942	C<$offset> is the offset from the start of the file - it generally must be
		1943	a multiple of C<IO::AIO::PAGESIZE> and defaults to C<0>.
		1944
		1945	Example:
		1946
		1947	use Digest::MD5;
		1948	use IO::AIO;
		1949
		1950	open my $fh, "<verybigfile"
		1951	or die "$!";
		1952
		1953	IO::AIO::mmap my $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh
		1954	or die "verybigfile: $!";
		1955
		1956	my $fast_md5 = md5 $data;
		1957
		1958	=item IO::AIO::munmap $scalar
		1959
		1960	Removes a previous mmap and undefines the C<$scalar>.
		1961
		1962	=item IO::AIO::munlock $scalar, $offset = 0, $length = undef
		1963
		1964	Calls the C<munlock> function, undoing the effects of a previous
		1965	C<aio_mlock> call (see its description for details).
		1966
		1967	=item IO::AIO::munlockall
		1968
		1969	Calls the C<munlockall> function.
		1970
		1971	On systems that do not implement C<munlockall>, this function returns
		1972	ENOSYS, otherwise the return value of C<munlockall>.
		1973
		1974	=item IO::AIO::splice $r_fh, $r_off, $w_fh, $w_off, $length, $flags
		1975
		1976	Calls the GNU/Linux C<splice(2)> syscall, if available. If C<$r_off> or
		1977	C<$w_off> are C<undef>, then C<NULL> is passed for these, otherwise they
		1978	should be the file offset.
		1979
		1980	The following symbol flag values are available: C<IO::AIO::SPLICE_F_MOVE>,
		1981	C<IO::AIO::SPLICE_F_NONBLOCK>, C<IO::AIO::SPLICE_F_MORE> and
		1982	C<IO::AIO::SPLICE_F_GIFT>.
		1983
		1984	See the C<splice(2)> manpage for details.
		1985
		1986	=item IO::AIO::tee $r_fh, $w_fh, $length, $flags
		1987
		1988	Calls the GNU/Linux C<tee(2)> syscall, see it's manpage and the
		1989	description for C<IO::AIO::splice> above for details.
		1990
		1991	=back
		1992
510	=cut	1993	=cut
511		1994
512	# support function to convert a fd into a perl filehandle
513	sub _fd2fh {
514	return undef if $_[0] < 0;
515
516	# try to generate nice filehandles
517	my $sym = "IO::AIO::fd#$_[0]";
518	local *$sym;
519
520	open *$sym, "+<&=$_[0]" # usually works under any unix
521	or open *$sym, "<&=$_[0]" # cygwin needs this
522	or open *$sym, ">&=$_[0]" # or this
523	or return undef;
524
525	*$sym
526	}
527
528	min_parallel 4;	1995	min_parallel 8;
529		1996
530	END {	1997	END { flush }
531	max_parallel 0;
532	}
533		1998
534	1;	1999	1;
535		2000
		2001	=head1 EVENT LOOP INTEGRATION
		2002
		2003	It is recommended to use L<AnyEvent::AIO> to integrate IO::AIO
		2004	automatically into many event loops:
		2005
		2006	# AnyEvent integration (EV, Event, Glib, Tk, POE, urxvt, pureperl...)
		2007	use AnyEvent::AIO;
		2008
		2009	You can also integrate IO::AIO manually into many event loops, here are
		2010	some examples of how to do this:
		2011
		2012	# EV integration
		2013	my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
		2014
		2015	# Event integration
		2016	Event->io (fd => IO::AIO::poll_fileno,
		2017	poll => 'r',
		2018	cb => \&IO::AIO::poll_cb);
		2019
		2020	# Glib/Gtk2 integration
		2021	add_watch Glib::IO IO::AIO::poll_fileno,
		2022	in => sub { IO::AIO::poll_cb; 1 };
		2023
		2024	# Tk integration
		2025	Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",
		2026	readable => \&IO::AIO::poll_cb);
		2027
		2028	# Danga::Socket integration
		2029	Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
		2030	\&IO::AIO::poll_cb);
		2031
536	=head2 FORK BEHAVIOUR	2032	=head2 FORK BEHAVIOUR
537		2033
538	Before the fork, IO::AIO enters a quiescent state where no requests	2034	Usage of pthreads in a program changes the semantics of fork
539	can be added in other threads and no results will be processed. After	2035	considerably. Specifically, only async-safe functions can be called after
540	the fork the parent simply leaves the quiescent state and continues	2036	fork. Perl doesn't know about this, so in general, you cannot call fork
541	request/result processing, while the child clears the request/result	2037	with defined behaviour in perl if pthreads are involved. IO::AIO uses
542	queue (so the requests started before the fork will only be handled in	2038	pthreads, so this applies, but many other extensions and (for inexplicable
543	the parent). Threats will be started on demand until the limit ste in the	2039	reasons) perl itself often is linked against pthreads, so this limitation
544	parent process has been reached again.	2040	applies to quite a lot of perls.
		2041
		2042	This module no longer tries to fight your OS, or POSIX. That means IO::AIO
		2043	only works in the process that loaded it. Forking is fully supported, but
		2044	using IO::AIO in the child is not.
		2045
		2046	You might get around by not I<using> IO::AIO before (or after)
		2047	forking. You could also try to call the L<IO::AIO::reinit> function in the
		2048	child:
		2049
		2050	=over 4
		2051
		2052	=item IO::AIO::reinit
		2053
		2054	Abandons all current requests and I/O threads and simply reinitialises all
		2055	data structures. This is not an operation supported by any standards, but
		2056	happens to work on GNU/Linux and some newer BSD systems.
		2057
		2058	The only reasonable use for this function is to call it after forking, if
		2059	C<IO::AIO> was used in the parent. Calling it while IO::AIO is active in
		2060	the process will result in undefined behaviour. Calling it at any time
		2061	will also result in any undefined (by POSIX) behaviour.
		2062
		2063	=back
		2064
		2065	=head2 MEMORY USAGE
		2066
		2067	Per-request usage:
		2068
		2069	Each aio request uses - depending on your architecture - around 100-200
		2070	bytes of memory. In addition, stat requests need a stat buffer (possibly
		2071	a few hundred bytes), readdir requires a result buffer and so on. Perl
		2072	scalars and other data passed into aio requests will also be locked and
		2073	will consume memory till the request has entered the done state.
		2074
		2075	This is not awfully much, so queuing lots of requests is not usually a
		2076	problem.
		2077
		2078	Per-thread usage:
		2079
		2080	In the execution phase, some aio requests require more memory for
		2081	temporary buffers, and each thread requires a stack and other data
		2082	structures (usually around 16k-128k, depending on the OS).
		2083
		2084	=head1 KNOWN BUGS
		2085
		2086	Known bugs will be fixed in the next release.
545		2087
546	=head1 SEE ALSO	2088	=head1 SEE ALSO
547		2089
548	L<Coro>, L<Linux::AIO>.	2090	L<AnyEvent::AIO> for easy integration into event loops, L<Coro::AIO> for a
		2091	more natural syntax.
549		2092
550	=head1 AUTHOR	2093	=head1 AUTHOR
551		2094
552	Marc Lehmann <schmorp@schmorp.de>	2095	Marc Lehmann <schmorp@schmorp.de>
553	http://home.schmorp.de/	2096	http://home.schmorp.de/

Diff Legend

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