ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/IO-AIO/AIO.pm

Comparing IO-AIO/AIO.pm (file contents):
Revision 1.12 by root, Mon Jul 11 01:03:17 2005 UTC vs.
Revision 1.146 by root, Tue Apr 21 20:06:05 2009 UTC

…		…
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", O_RDONLY, 0, sub {
10	my ($fh) = @_;	10	my $fh = shift
		11	or die "/etc/passwd: $!";
11	...	12	...
12	};	13	};
13		14
14	aio_unlink "/tmp/file", sub { };	15	aio_unlink "/tmp/file", sub { };
15		16
16	aio_read $fh, 30000, 1024, $buffer, 0, sub {	17	aio_read $fh, 30000, 1024, $buffer, 0, sub {
17	$_[0] > 0 or die "read error: $!";	18	$_[0] > 0 or die "read error: $!";
18	};	19	};
19		20
20	# Event	21	# version 2+ has request and group objects
		22	use IO::AIO 2;
		23
		24	aioreq_pri 4; # give next request a very high priority
		25	my $req = aio_unlink "/tmp/file", sub { };
		26	$req->cancel; # cancel request if still in queue
		27
		28	my $grp = aio_group sub { print "all stats done\n" };
		29	add $grp aio_stat "..." for ...;
		30
		31	# AnyEvent integration (EV, Event, Glib, Tk, POE, urxvt, pureperl...)
		32	use AnyEvent::AIO;
		33
		34	# EV integration
		35	my $w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
		36
		37	# Event integration
21	Event->io (fd => IO::AIO::poll_fileno,	38	Event->io (fd => IO::AIO::poll_fileno,
22	poll => 'r',	39	poll => 'r',
23	cb => \&IO::AIO::poll_cb);	40	cb => \&IO::AIO::poll_cb);
24		41
25	# Glib/Gtk2	42	# Glib/Gtk2 integration
26	add_watch Glib::IO IO::AIO::poll_fileno,	43	add_watch Glib::IO IO::AIO::poll_fileno,
27	\&IO::AIO::poll_cb;	44	in => sub { IO::AIO::poll_cb; 1 };
28		45
29	# Tk	46	# Tk integration
30	Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",	47	Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",
31	readable => \&IO::AIO::poll_cb);	48	readable => \&IO::AIO::poll_cb);
32		49
33	# Danga::Socket	50	# Danga::Socket integration
34	Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>	51	Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
35	\&IO::AIO::poll_cb);	52	\&IO::AIO::poll_cb);
36		53
37
38	=head1 DESCRIPTION	54	=head1 DESCRIPTION
39		55
40	This module implements asynchronous I/O using whatever means your	56	This module implements asynchronous I/O using whatever means your
41	operating system supports.	57	operating system supports.
42		58
		59	Asynchronous means that operations that can normally block your program
		60	(e.g. reading from disk) will be done asynchronously: the operation
		61	will still block, but you can do something else in the meantime. This
		62	is extremely useful for programs that need to stay interactive even
		63	when doing heavy I/O (GUI programs, high performance network servers
		64	etc.), but can also be used to easily do operations in parallel that are
		65	normally done sequentially, e.g. stat'ing many files, which is much faster
		66	on a RAID volume or over NFS when you do a number of stat operations
		67	concurrently.
		68
		69	While most of this works on all types of file descriptors (for
		70	example sockets), using these functions on file descriptors that
		71	support nonblocking operation (again, sockets, pipes etc.) is very
		72	inefficient. Use an event loop for that (such as the L<Event|Event>
		73	module): IO::AIO will naturally fit into such an event loop itself.
		74
43	Currently, a number of threads are started that execute your read/writes	75	In this version, a number of threads are started that execute your
44	and signal their completion. You don't need thread support in your libc or	76	requests and signal their completion. You don't need thread support
45	perl, and the threads created by this module will not be visible to the	77	in perl, and the threads created by this module will not be visible
46	pthreads library. In the future, this module might make use of the native	78	to perl. In the future, this module might make use of the native aio
47	aio functions available on many operating systems. However, they are often	79	functions available on many operating systems. However, they are often
48	not well-supported (Linux doesn't allow them on normal files currently,	80	not well-supported or restricted (GNU/Linux doesn't allow them on normal
49	for example), and they would only support aio_read and aio_write, so the	81	files currently, for example), and they would only support aio_read and
50	remaining functionality would have to be implemented using threads anyway.	82	aio_write, so the remaining functionality would have to be implemented
		83	using threads anyway.
51		84
52	Although the module will work with in the presence of other threads, it is	85	Although the module will work in the presence of other (Perl-) threads,
53	currently not reentrant, so use appropriate locking yourself.	86	it is currently not reentrant in any way, so use appropriate locking
		87	yourself, always call C<poll_cb> from within the same thread, or never
		88	call C<poll_cb> (or other C<aio_> functions) recursively.
		89
		90	=head2 EXAMPLE
		91
		92	This is a simple example that uses the Event module and loads
		93	F</etc/passwd> asynchronously:
		94
		95	use Fcntl;
		96	use Event;
		97	use IO::AIO;
		98
		99	# register the IO::AIO callback with Event
		100	Event->io (fd => IO::AIO::poll_fileno,
		101	poll => 'r',
		102	cb => \&IO::AIO::poll_cb);
		103
		104	# queue the request to open /etc/passwd
		105	aio_open "/etc/passwd", O_RDONLY, 0, sub {
		106	my $fh = shift
		107	or die "error while opening: $!";
		108
		109	# stat'ing filehandles is generally non-blocking
		110	my $size = -s $fh;
		111
		112	# queue a request to read the file
		113	my $contents;
		114	aio_read $fh, 0, $size, $contents, 0, sub {
		115	$_[0] == $size
		116	or die "short read: $!";
		117
		118	close $fh;
		119
		120	# file contents now in $contents
		121	print $contents;
		122
		123	# exit event loop and program
		124	Event::unloop;
		125	};
		126	};
		127
		128	# possibly queue up other requests, or open GUI windows,
		129	# check for sockets etc. etc.
		130
		131	# process events as long as there are some:
		132	Event::loop;
		133
		134	=head1 REQUEST ANATOMY AND LIFETIME
		135
		136	Every C<aio_*> function creates a request. which is a C data structure not
		137	directly visible to Perl.
		138
		139	If called in non-void context, every request function returns a Perl
		140	object representing the request. In void context, nothing is returned,
		141	which saves a bit of memory.
		142
		143	The perl object is a fairly standard ref-to-hash object. The hash contents
		144	are not used by IO::AIO so you are free to store anything you like in it.
		145
		146	During their existance, aio requests travel through the following states,
		147	in order:
		148
		149	=over 4
		150
		151	=item ready
		152
		153	Immediately after a request is created it is put into the ready state,
		154	waiting for a thread to execute it.
		155
		156	=item execute
		157
		158	A thread has accepted the request for processing and is currently
		159	executing it (e.g. blocking in read).
		160
		161	=item pending
		162
		163	The request has been executed and is waiting for result processing.
		164
		165	While request submission and execution is fully asynchronous, result
		166	processing is not and relies on the perl interpreter calling C<poll_cb>
		167	(or another function with the same effect).
		168
		169	=item result
		170
		171	The request results are processed synchronously by C<poll_cb>.
		172
		173	The C<poll_cb> function will process all outstanding aio requests by
		174	calling their callbacks, freeing memory associated with them and managing
		175	any groups they are contained in.
		176
		177	=item done
		178
		179	Request has reached the end of its lifetime and holds no resources anymore
		180	(except possibly for the Perl object, but its connection to the actual
		181	aio request is severed and calling its methods will either do nothing or
		182	result in a runtime error).
		183
		184	=back
54		185
55	=cut	186	=cut
56		187
57	package IO::AIO;	188	package IO::AIO;
58		189
		190	use Carp ();
		191
		192	no warnings;
		193	use strict 'vars';
		194
59	use base 'Exporter';	195	use base 'Exporter';
60		196
61	use Fcntl ();
62
63	BEGIN {	197	BEGIN {
64	$VERSION = 0.3;	198	our $VERSION = '3.19';
65		199
66	@EXPORT = qw(aio_read aio_write aio_open aio_close aio_stat aio_lstat aio_unlink	200	our @AIO_REQ = qw(aio_sendfile aio_read aio_write aio_open aio_close
67	aio_fsync aio_fdatasync aio_readahead);	201	aio_stat aio_lstat aio_unlink aio_rmdir aio_readdir
68	@EXPORT_OK = qw(poll_fileno poll_cb min_parallel max_parallel max_outstanding nreqs);	202	aio_scandir aio_symlink aio_readlink aio_sync aio_fsync
		203	aio_fdatasync aio_sync_file_range aio_pathsync aio_readahead
		204	aio_rename aio_link aio_move aio_copy aio_group
		205	aio_nop aio_mknod aio_load aio_rmtree aio_mkdir aio_chown
		206	aio_chmod aio_utime aio_truncate);
		207
		208	our @EXPORT = (@AIO_REQ, qw(aioreq_pri aioreq_nice));
		209	our @EXPORT_OK = qw(poll_fileno poll_cb poll_wait flush
		210	min_parallel max_parallel max_idle
		211	nreqs nready npending nthreads
		212	max_poll_time max_poll_reqs);
		213
		214	push @AIO_REQ, qw(aio_busy); # not exported
		215
		216	@IO::AIO::GRP::ISA = 'IO::AIO::REQ';
69		217
70	require XSLoader;	218	require XSLoader;
71	XSLoader::load IO::AIO, $VERSION;	219	XSLoader::load ("IO::AIO", $VERSION);
72	}	220	}
73		221
74	=head1 FUNCTIONS	222	=head1 FUNCTIONS
75		223
76	=head2 AIO FUNCTIONS	224	=head2 AIO REQUEST FUNCTIONS
77		225
78	All the C<aio_*> calls are more or less thin wrappers around the syscall	226	All the C<aio_*> calls are more or less thin wrappers around the syscall
79	with the same name (sans C<aio_>). The arguments are similar or identical,	227	with the same name (sans C<aio_>). The arguments are similar or identical,
80	and they all accept an additional C<$callback> argument which must be	228	and they all accept an additional (and optional) C<$callback> argument
81	a code reference. This code reference will get called with the syscall	229	which must be a code reference. This code reference will get called with
82	return code (e.g. most syscalls return C<-1> on error, unlike perl, which	230	the syscall return code (e.g. most syscalls return C<-1> on error, unlike
83	usually delivers "false") as it's sole argument when the given syscall has	231	perl, which usually delivers "false") as its sole argument after the given
84	been executed asynchronously.	232	syscall has been executed asynchronously.
85		233
86	All functions that expect a filehandle will also accept a file descriptor.	234	All functions expecting a filehandle keep a copy of the filehandle
		235	internally until the request has finished.
87		236
		237	All functions return request objects of type L<IO::AIO::REQ> that allow
		238	further manipulation of those requests while they are in-flight.
		239
88	The filenames you pass to these routines I<must> be absolute. The reason	240	The pathnames you pass to these routines I<must> be absolute and
89	is that at the time the request is being executed, the current working	241	encoded as octets. The reason for the former is that at the time the
90	directory could have changed. Alternatively, you can make sure that you	242	request is being executed, the current working directory could have
91	never change the current working directory.	243	changed. Alternatively, you can make sure that you never change the
		244	current working directory anywhere in the program and then use relative
		245	paths.
		246
		247	To encode pathnames as octets, either make sure you either: a) always pass
		248	in filenames you got from outside (command line, readdir etc.) without
		249	tinkering, b) are ASCII or ISO 8859-1, c) use the Encode module and encode
		250	your pathnames to the locale (or other) encoding in effect in the user
		251	environment, d) use Glib::filename_from_unicode on unicode filenames or e)
		252	use something else to ensure your scalar has the correct contents.
		253
		254	This works, btw. independent of the internal UTF-8 bit, which IO::AIO
		255	handles correctly whether it is set or not.
92		256
93	=over 4	257	=over 4
94		258
		259	=item $prev_pri = aioreq_pri [$pri]
		260
		261	Returns the priority value that would be used for the next request and, if
		262	C<$pri> is given, sets the priority for the next aio request.
		263
		264	The default priority is C<0>, the minimum and maximum priorities are C<-4>
		265	and C<4>, respectively. Requests with higher priority will be serviced
		266	first.
		267
		268	The priority will be reset to C<0> after each call to one of the C<aio_*>
		269	functions.
		270
		271	Example: open a file with low priority, then read something from it with
		272	higher priority so the read request is serviced before other low priority
		273	open requests (potentially spamming the cache):
		274
		275	aioreq_pri -3;
		276	aio_open ..., sub {
		277	return unless $_[0];
		278
		279	aioreq_pri -2;
		280	aio_read $_[0], ..., sub {
		281	...
		282	};
		283	};
		284
		285
		286	=item aioreq_nice $pri_adjust
		287
		288	Similar to C<aioreq_pri>, but subtracts the given value from the current
		289	priority, so the effect is cumulative.
		290
		291
95	=item aio_open $pathname, $flags, $mode, $callback	292	=item aio_open $pathname, $flags, $mode, $callback->($fh)
96		293
97	Asynchronously open or create a file and call the callback with a newly	294	Asynchronously open or create a file and call the callback with a newly
98	created filehandle for the file.	295	created filehandle for the file.
99		296
100	The pathname passed to C<aio_open> must be absolute. See API NOTES, above,	297	The pathname passed to C<aio_open> must be absolute. See API NOTES, above,
101	for an explanation.	298	for an explanation.
102		299
103	The C<$mode> argument is a bitmask. See the C<Fcntl> module for a	300	The C<$flags> argument is a bitmask. See the C<Fcntl> module for a
104	list. They are the same as used in C<sysopen>.	301	list. They are the same as used by C<sysopen>.
		302
		303	Likewise, C<$mode> specifies the mode of the newly created file, if it
		304	didn't exist and C<O_CREAT> has been given, just like perl's C<sysopen>,
		305	except that it is mandatory (i.e. use C<0> if you don't create new files,
		306	and C<0666> or C<0777> if you do). Note that the C<$mode> will be modified
		307	by the umask in effect then the request is being executed, so better never
		308	change the umask.
105		309
106	Example:	310	Example:
107		311
108	aio_open "/etc/passwd", O_RDONLY, 0, sub {	312	aio_open "/etc/passwd", O_RDONLY, 0, sub {
109	if ($_[0]) {	313	if ($_[0]) {
…		…
112	} else {	316	} else {
113	die "open failed: $!\n";	317	die "open failed: $!\n";
114	}	318	}
115	};	319	};
116		320
		321
117	=item aio_close $fh, $callback	322	=item aio_close $fh, $callback->($status)
118		323
119	Asynchronously close a file and call the callback with the result	324	Asynchronously close a file and call the callback with the result
120	code. I<WARNING:> although accepted, you should not pass in a perl	325	code.
121	filehandle here, as perl will likely close the file descriptor itself when
122	the filehandle is destroyed. Normally, you can safely call perls C<close>
123	or just let filehandles go out of scope.
124		326
		327	Unfortunately, you can't do this to perl. Perl I<insists> very strongly on
		328	closing the file descriptor associated with the filehandle itself.
		329
		330	Therefore, C<aio_close> will not close the filehandle - instead it will
		331	use dup2 to overwrite the file descriptor with the write-end of a pipe
		332	(the pipe fd will be created on demand and will be cached).
		333
		334	Or in other words: the file descriptor will be closed, but it will not be
		335	free for reuse until the perl filehandle is closed.
		336
		337	=cut
		338
125	=item aio_read $fh,$offset,$length, $data,$dataoffset,$callback	339	=item aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
126		340
127	=item aio_write $fh,$offset,$length, $data,$dataoffset,$callback	341	=item aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
128		342
129	Reads or writes C<length> bytes from the specified C<fh> and C<offset>	343	Reads or writes C<$length> bytes from or to the specified C<$fh> and
130	into the scalar given by C<data> and offset C<dataoffset> and calls the	344	C<$offset> into the scalar given by C<$data> and offset C<$dataoffset>
131	callback without the actual number of bytes read (or -1 on error, just	345	and calls the callback without the actual number of bytes read (or -1 on
132	like the syscall).	346	error, just like the syscall).
133		347
		348	C<aio_read> will, like C<sysread>, shrink or grow the C<$data> scalar to
		349	offset plus the actual number of bytes read.
		350
		351	If C<$offset> is undefined, then the current file descriptor offset will
		352	be used (and updated), otherwise the file descriptor offset will not be
		353	changed by these calls.
		354
		355	If C<$length> is undefined in C<aio_write>, use the remaining length of
		356	C<$data>.
		357
		358	If C<$dataoffset> is less than zero, it will be counted from the end of
		359	C<$data>.
		360
		361	The C<$data> scalar I<MUST NOT> be modified in any way while the request
		362	is outstanding. Modifying it can result in segfaults or World War III (if
		363	the necessary/optional hardware is installed).
		364
134	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, strating at	365	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at
135	offset C<0> within the scalar:	366	offset C<0> within the scalar:
136		367
137	aio_read $fh, 7, 15, $buffer, 0, sub {	368	aio_read $fh, 7, 15, $buffer, 0, sub {
138	$_[0] > 0 or die "read error: $!";	369	$_[0] > 0 or die "read error: $!";
139	print "read $_[0] bytes: <$buffer>\n";	370	print "read $_[0] bytes: <$buffer>\n";
140	};	371	};
141		372
		373
		374	=item aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
		375
		376	Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts
		377	reading at byte offset C<$in_offset>, and starts writing at the current
		378	file offset of C<$out_fh>. Because of that, it is not safe to issue more
		379	than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each
		380	other.
		381
		382	This call tries to make use of a native C<sendfile> syscall to provide
		383	zero-copy operation. For this to work, C<$out_fh> should refer to a
		384	socket, and C<$in_fh> should refer to mmap'able file.
		385
		386	If the native sendfile call fails or is not implemented, it will be
		387	emulated, so you can call C<aio_sendfile> on any type of filehandle
		388	regardless of the limitations of the operating system.
		389
		390	Please note, however, that C<aio_sendfile> can read more bytes from
		391	C<$in_fh> than are written, and there is no way to find out how many
		392	bytes have been read from C<aio_sendfile> alone, as C<aio_sendfile> only
		393	provides the number of bytes written to C<$out_fh>. Only if the result
		394	value equals C<$length> one can assume that C<$length> bytes have been
		395	read.
		396
		397
142	=item aio_readahead $fh,$offset,$length, $callback	398	=item aio_readahead $fh,$offset,$length, $callback->($retval)
143		399
144	Asynchronously reads the specified byte range into the page cache, using
145	the C<readahead> syscall. If that syscall doesn't exist the status will be
146	C<-1> and C<$!> is set to ENOSYS.
147
148	readahead() populates the page cache with data from a file so that	400	C<aio_readahead> populates the page cache with data from a file so that
149	subsequent reads from that file will not block on disk I/O. The C<$offset>	401	subsequent reads from that file will not block on disk I/O. The C<$offset>
150	argument specifies the starting point from which data is to be read and	402	argument specifies the starting point from which data is to be read and
151	C<$length> specifies the number of bytes to be read. I/O is performed in	403	C<$length> specifies the number of bytes to be read. I/O is performed in
152	whole pages, so that offset is effectively rounded down to a page boundary	404	whole pages, so that offset is effectively rounded down to a page boundary
153	and bytes are read up to the next page boundary greater than or equal to	405	and bytes are read up to the next page boundary greater than or equal to
154	(off-set+length). aio_readahead() does not read beyond the end of the	406	(off-set+length). C<aio_readahead> does not read beyond the end of the
155	file. The current file offset of the file is left unchanged.	407	file. The current file offset of the file is left unchanged.
156		408
		409	If that syscall doesn't exist (likely if your OS isn't Linux) it will be
		410	emulated by simply reading the data, which would have a similar effect.
		411
		412
157	=item aio_stat $fh_or_path, $callback	413	=item aio_stat $fh_or_path, $callback->($status)
158		414
159	=item aio_lstat $fh, $callback	415	=item aio_lstat $fh, $callback->($status)
160		416
161	Works like perl's C<stat> or C<lstat> in void context. The callback will	417	Works like perl's C<stat> or C<lstat> in void context. The callback will
162	be called after the stat and the results will be available using C<stat _>	418	be called after the stat and the results will be available using C<stat _>
163	or C<-s _> etc...	419	or C<-s _> etc...
164		420
…		…
174	aio_stat "/etc/passwd", sub {	430	aio_stat "/etc/passwd", sub {
175	$_[0] and die "stat failed: $!";	431	$_[0] and die "stat failed: $!";
176	print "size is ", -s _, "\n";	432	print "size is ", -s _, "\n";
177	};	433	};
178		434
		435
		436	=item aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
		437
		438	Works like perl's C<utime> function (including the special case of $atime
		439	and $mtime being undef). Fractional times are supported if the underlying
		440	syscalls support them.
		441
		442	When called with a pathname, uses utimes(2) if available, otherwise
		443	utime(2). If called on a file descriptor, uses futimes(2) if available,
		444	otherwise returns ENOSYS, so this is not portable.
		445
		446	Examples:
		447
		448	# set atime and mtime to current time (basically touch(1)):
		449	aio_utime "path", undef, undef;
		450	# set atime to current time and mtime to beginning of the epoch:
		451	aio_utime "path", time, undef; # undef==0
		452
		453
		454	=item aio_chown $fh_or_path, $uid, $gid, $callback->($status)
		455
		456	Works like perl's C<chown> function, except that C<undef> for either $uid
		457	or $gid is being interpreted as "do not change" (but -1 can also be used).
		458
		459	Examples:
		460
		461	# same as "chown root path" in the shell:
		462	aio_chown "path", 0, -1;
		463	# same as above:
		464	aio_chown "path", 0, undef;
		465
		466
		467	=item aio_truncate $fh_or_path, $offset, $callback->($status)
		468
		469	Works like truncate(2) or ftruncate(2).
		470
		471
		472	=item aio_chmod $fh_or_path, $mode, $callback->($status)
		473
		474	Works like perl's C<chmod> function.
		475
		476
179	=item aio_unlink $pathname, $callback	477	=item aio_unlink $pathname, $callback->($status)
180		478
181	Asynchronously unlink (delete) a file and call the callback with the	479	Asynchronously unlink (delete) a file and call the callback with the
182	result code.	480	result code.
183		481
		482
		483	=item aio_mknod $path, $mode, $dev, $callback->($status)
		484
		485	[EXPERIMENTAL]
		486
		487	Asynchronously create a device node (or fifo). See mknod(2).
		488
		489	The only (POSIX-) portable way of calling this function is:
		490
		491	aio_mknod $path, IO::AIO::S_IFIFO | $mode, 0, sub { ...
		492
		493
		494	=item aio_link $srcpath, $dstpath, $callback->($status)
		495
		496	Asynchronously create a new link to the existing object at C<$srcpath> at
		497	the path C<$dstpath> and call the callback with the result code.
		498
		499
		500	=item aio_symlink $srcpath, $dstpath, $callback->($status)
		501
		502	Asynchronously create a new symbolic link to the existing object at C<$srcpath> at
		503	the path C<$dstpath> and call the callback with the result code.
		504
		505
		506	=item aio_readlink $path, $callback->($link)
		507
		508	Asynchronously read the symlink specified by C<$path> and pass it to
		509	the callback. If an error occurs, nothing or undef gets passed to the
		510	callback.
		511
		512
		513	=item aio_rename $srcpath, $dstpath, $callback->($status)
		514
		515	Asynchronously rename the object at C<$srcpath> to C<$dstpath>, just as
		516	rename(2) and call the callback with the result code.
		517
		518
		519	=item aio_mkdir $pathname, $mode, $callback->($status)
		520
		521	Asynchronously mkdir (create) a directory and call the callback with
		522	the result code. C<$mode> will be modified by the umask at the time the
		523	request is executed, so do not change your umask.
		524
		525
		526	=item aio_rmdir $pathname, $callback->($status)
		527
		528	Asynchronously rmdir (delete) a directory and call the callback with the
		529	result code.
		530
		531
		532	=item aio_readdir $pathname, $callback->($entries)
		533
		534	Unlike the POSIX call of the same name, C<aio_readdir> reads an entire
		535	directory (i.e. opendir + readdir + closedir). The entries will not be
		536	sorted, and will B<NOT> include the C<.> and C<..> entries.
		537
		538	The callback a single argument which is either C<undef> or an array-ref
		539	with the filenames.
		540
		541
		542	=item aio_load $path, $data, $callback->($status)
		543
		544	This is a composite request that tries to fully load the given file into
		545	memory. Status is the same as with aio_read.
		546
		547	=cut
		548
		549	sub aio_load($$;$) {
		550	my ($path, undef, $cb) = @_;
		551	my $data = \$_[1];
		552
		553	my $pri = aioreq_pri;
		554	my $grp = aio_group $cb;
		555
		556	aioreq_pri $pri;
		557	add $grp aio_open $path, O_RDONLY, 0, sub {
		558	my $fh = shift
		559	or return $grp->result (-1);
		560
		561	aioreq_pri $pri;
		562	add $grp aio_read $fh, 0, (-s $fh), $$data, 0, sub {
		563	$grp->result ($_[0]);
		564	};
		565	};
		566
		567	$grp
		568	}
		569
		570	=item aio_copy $srcpath, $dstpath, $callback->($status)
		571
		572	Try to copy the I<file> (directories not supported as either source or
		573	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		574	the C<0> (error) or C<-1> ok.
		575
		576	This is a composite request that creates the destination file with
		577	mode 0200 and copies the contents of the source file into it using
		578	C<aio_sendfile>, followed by restoring atime, mtime, access mode and
		579	uid/gid, in that order.
		580
		581	If an error occurs, the partial destination file will be unlinked, if
		582	possible, except when setting atime, mtime, access mode and uid/gid, where
		583	errors are being ignored.
		584
		585	=cut
		586
		587	sub aio_copy($$;$) {
		588	my ($src, $dst, $cb) = @_;
		589
		590	my $pri = aioreq_pri;
		591	my $grp = aio_group $cb;
		592
		593	aioreq_pri $pri;
		594	add $grp aio_open $src, O_RDONLY, 0, sub {
		595	if (my $src_fh = $_[0]) {
		596	my @stat = stat $src_fh;
		597
		598	aioreq_pri $pri;
		599	add $grp aio_open $dst, O_CREAT | O_WRONLY | O_TRUNC, 0200, sub {
		600	if (my $dst_fh = $_[0]) {
		601	aioreq_pri $pri;
		602	add $grp aio_sendfile $dst_fh, $src_fh, 0, $stat[7], sub {
		603	if ($_[0] == $stat[7]) {
		604	$grp->result (0);
		605	close $src_fh;
		606
		607	# those should not normally block. should. should.
		608	utime $stat[8], $stat[9], $dst;
		609	chmod $stat[2] & 07777, $dst_fh;
		610	chown $stat[4], $stat[5], $dst_fh;
		611
		612	aioreq_pri $pri;
		613	add $grp aio_close $dst_fh;
		614	} else {
		615	$grp->result (-1);
		616	close $src_fh;
		617	close $dst_fh;
		618
		619	aioreq $pri;
		620	add $grp aio_unlink $dst;
		621	}
		622	};
		623	} else {
		624	$grp->result (-1);
		625	}
		626	},
		627
		628	} else {
		629	$grp->result (-1);
		630	}
		631	};
		632
		633	$grp
		634	}
		635
		636	=item aio_move $srcpath, $dstpath, $callback->($status)
		637
		638	Try to move the I<file> (directories not supported as either source or
		639	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		640	the C<0> (error) or C<-1> ok.
		641
		642	This is a composite request that tries to rename(2) the file first; if
		643	rename fails with C<EXDEV>, it copies the file with C<aio_copy> and, if
		644	that is successful, unlinks the C<$srcpath>.
		645
		646	=cut
		647
		648	sub aio_move($$;$) {
		649	my ($src, $dst, $cb) = @_;
		650
		651	my $pri = aioreq_pri;
		652	my $grp = aio_group $cb;
		653
		654	aioreq_pri $pri;
		655	add $grp aio_rename $src, $dst, sub {
		656	if ($_[0] && $! == EXDEV) {
		657	aioreq_pri $pri;
		658	add $grp aio_copy $src, $dst, sub {
		659	$grp->result ($_[0]);
		660
		661	if (!$_[0]) {
		662	aioreq_pri $pri;
		663	add $grp aio_unlink $src;
		664	}
		665	};
		666	} else {
		667	$grp->result ($_[0]);
		668	}
		669	};
		670
		671	$grp
		672	}
		673
		674	=item aio_scandir $path, $maxreq, $callback->($dirs, $nondirs)
		675
		676	Scans a directory (similar to C<aio_readdir>) but additionally tries to
		677	efficiently separate the entries of directory C<$path> into two sets of
		678	names, directories you can recurse into (directories), and ones you cannot
		679	recurse into (everything else, including symlinks to directories).
		680
		681	C<aio_scandir> is a composite request that creates of many sub requests_
		682	C<$maxreq> specifies the maximum number of outstanding aio requests that
		683	this function generates. If it is C<< <= 0 >>, then a suitable default
		684	will be chosen (currently 4).
		685
		686	On error, the callback is called without arguments, otherwise it receives
		687	two array-refs with path-relative entry names.
		688
		689	Example:
		690
		691	aio_scandir $dir, 0, sub {
		692	my ($dirs, $nondirs) = @_;
		693	print "real directories: @$dirs\n";
		694	print "everything else: @$nondirs\n";
		695	};
		696
		697	Implementation notes.
		698
		699	The C<aio_readdir> cannot be avoided, but C<stat()>'ing every entry can.
		700
		701	After reading the directory, the modification time, size etc. of the
		702	directory before and after the readdir is checked, and if they match (and
		703	isn't the current time), the link count will be used to decide how many
		704	entries are directories (if >= 2). Otherwise, no knowledge of the number
		705	of subdirectories will be assumed.
		706
		707	Then entries will be sorted into likely directories (everything without
		708	a non-initial dot currently) and likely non-directories (everything
		709	else). Then every entry plus an appended C</.> will be C<stat>'ed,
		710	likely directories first. If that succeeds, it assumes that the entry
		711	is a directory or a symlink to directory (which will be checked
		712	seperately). This is often faster than stat'ing the entry itself because
		713	filesystems might detect the type of the entry without reading the inode
		714	data (e.g. ext2fs filetype feature).
		715
		716	If the known number of directories (link count - 2) has been reached, the
		717	rest of the entries is assumed to be non-directories.
		718
		719	This only works with certainty on POSIX (= UNIX) filesystems, which
		720	fortunately are the vast majority of filesystems around.
		721
		722	It will also likely work on non-POSIX filesystems with reduced efficiency
		723	as those tend to return 0 or 1 as link counts, which disables the
		724	directory counting heuristic.
		725
		726	=cut
		727
		728	sub aio_scandir($$;$) {
		729	my ($path, $maxreq, $cb) = @_;
		730
		731	my $pri = aioreq_pri;
		732
		733	my $grp = aio_group $cb;
		734
		735	$maxreq = 4 if $maxreq <= 0;
		736
		737	# stat once
		738	aioreq_pri $pri;
		739	add $grp aio_stat $path, sub {
		740	return $grp->result () if $_[0];
		741	my $now = time;
		742	my $hash1 = join ":", (stat _)[0,1,3,7,9];
		743
		744	# read the directory entries
		745	aioreq_pri $pri;
		746	add $grp aio_readdir $path, sub {
		747	my $entries = shift
		748	or return $grp->result ();
		749
		750	# stat the dir another time
		751	aioreq_pri $pri;
		752	add $grp aio_stat $path, sub {
		753	my $hash2 = join ":", (stat _)[0,1,3,7,9];
		754
		755	my $ndirs;
		756
		757	# take the slow route if anything looks fishy
		758	if ($hash1 ne $hash2 or (stat _)[9] == $now) {
		759	$ndirs = -1;
		760	} else {
		761	# if nlink == 2, we are finished
		762	# on non-posix-fs's, we rely on nlink < 2
		763	$ndirs = (stat _)[3] - 2
		764	or return $grp->result ([], $entries);
		765	}
		766
		767	# sort into likely dirs and likely nondirs
		768	# dirs == files without ".", short entries first
		769	$entries = [map $_->[0],
		770	sort { $b->[1] cmp $a->[1] }
		771	map [$_, sprintf "%s%04d", (/.\./ ? "1" : "0"), length],
		772	@$entries];
		773
		774	my (@dirs, @nondirs);
		775
		776	my $statgrp = add $grp aio_group sub {
		777	$grp->result (\@dirs, \@nondirs);
		778	};
		779
		780	limit $statgrp $maxreq;
		781	feed $statgrp sub {
		782	return unless @$entries;
		783	my $entry = pop @$entries;
		784
		785	aioreq_pri $pri;
		786	add $statgrp aio_stat "$path/$entry/.", sub {
		787	if ($_[0] < 0) {
		788	push @nondirs, $entry;
		789	} else {
		790	# need to check for real directory
		791	aioreq_pri $pri;
		792	add $statgrp aio_lstat "$path/$entry", sub {
		793	if (-d _) {
		794	push @dirs, $entry;
		795
		796	unless (--$ndirs) {
		797	push @nondirs, @$entries;
		798	feed $statgrp;
		799	}
		800	} else {
		801	push @nondirs, $entry;
		802	}
		803	}
		804	}
		805	};
		806	};
		807	};
		808	};
		809	};
		810
		811	$grp
		812	}
		813
		814	=item aio_rmtree $path, $callback->($status)
		815
		816	Delete a directory tree starting (and including) C<$path>, return the
		817	status of the final C<rmdir> only. This is a composite request that
		818	uses C<aio_scandir> to recurse into and rmdir directories, and unlink
		819	everything else.
		820
		821	=cut
		822
		823	sub aio_rmtree;
		824	sub aio_rmtree($;$) {
		825	my ($path, $cb) = @_;
		826
		827	my $pri = aioreq_pri;
		828	my $grp = aio_group $cb;
		829
		830	aioreq_pri $pri;
		831	add $grp aio_scandir $path, 0, sub {
		832	my ($dirs, $nondirs) = @_;
		833
		834	my $dirgrp = aio_group sub {
		835	add $grp aio_rmdir $path, sub {
		836	$grp->result ($_[0]);
		837	};
		838	};
		839
		840	(aioreq_pri $pri), add $dirgrp aio_rmtree "$path/$_" for @$dirs;
		841	(aioreq_pri $pri), add $dirgrp aio_unlink "$path/$_" for @$nondirs;
		842
		843	add $grp $dirgrp;
		844	};
		845
		846	$grp
		847	}
		848
		849	=item aio_sync $callback->($status)
		850
		851	Asynchronously call sync and call the callback when finished.
		852
184	=item aio_fsync $fh, $callback	853	=item aio_fsync $fh, $callback->($status)
185		854
186	Asynchronously call fsync on the given filehandle and call the callback	855	Asynchronously call fsync on the given filehandle and call the callback
187	with the fsync result code.	856	with the fsync result code.
188		857
189	=item aio_fdatasync $fh, $callback	858	=item aio_fdatasync $fh, $callback->($status)
190		859
191	Asynchronously call fdatasync on the given filehandle and call the	860	Asynchronously call fdatasync on the given filehandle and call the
192	callback with the fdatasync result code.	861	callback with the fdatasync result code.
193		862
		863	If this call isn't available because your OS lacks it or it couldn't be
		864	detected, it will be emulated by calling C<fsync> instead.
		865
		866	=item aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
		867
		868	Sync the data portion of the file specified by C<$offset> and C<$length>
		869	to disk (but NOT the metadata), by calling the Linux-specific
		870	sync_file_range call. If sync_file_range is not available or it returns
		871	ENOSYS, then fdatasync or fsync is being substituted.
		872
		873	C<$flags> can be a combination of C<IO::AIO::SYNC_FILE_RANGE_WAIT_BEFORE>,
		874	C<IO::AIO::SYNC_FILE_RANGE_WRITE> and
		875	C<IO::AIO::SYNC_FILE_RANGE_WAIT_AFTER>: refer to the sync_file_range
		876	manpage for details.
		877
		878	=item aio_pathsync $path, $callback->($status)
		879
		880	This request tries to open, fsync and close the given path. This is a
		881	composite request intended to sync directories after directory operations
		882	(E.g. rename). This might not work on all operating systems or have any
		883	specific effect, but usually it makes sure that directory changes get
		884	written to disc. It works for anything that can be opened for read-only,
		885	not just directories.
		886
		887	Passes C<0> when everything went ok, and C<-1> on error.
		888
		889	=cut
		890
		891	sub aio_pathsync($;$) {
		892	my ($path, $cb) = @_;
		893
		894	my $pri = aioreq_pri;
		895	my $grp = aio_group $cb;
		896
		897	aioreq_pri $pri;
		898	add $grp aio_open $path, O_RDONLY, 0, sub {
		899	my ($fh) = @_;
		900	if ($fh) {
		901	aioreq_pri $pri;
		902	add $grp aio_fsync $fh, sub {
		903	$grp->result ($_[0]);
		904
		905	aioreq_pri $pri;
		906	add $grp aio_close $fh;
		907	};
		908	} else {
		909	$grp->result (-1);
		910	}
		911	};
		912
		913	$grp
		914	}
		915
		916	=item aio_group $callback->(...)
		917
		918	This is a very special aio request: Instead of doing something, it is a
		919	container for other aio requests, which is useful if you want to bundle
		920	many requests into a single, composite, request with a definite callback
		921	and the ability to cancel the whole request with its subrequests.
		922
		923	Returns an object of class L<IO::AIO::GRP>. See its documentation below
		924	for more info.
		925
		926	Example:
		927
		928	my $grp = aio_group sub {
		929	print "all stats done\n";
		930	};
		931
		932	add $grp
		933	(aio_stat ...),
		934	(aio_stat ...),
		935	...;
		936
		937	=item aio_nop $callback->()
		938
		939	This is a special request - it does nothing in itself and is only used for
		940	side effects, such as when you want to add a dummy request to a group so
		941	that finishing the requests in the group depends on executing the given
		942	code.
		943
		944	While this request does nothing, it still goes through the execution
		945	phase and still requires a worker thread. Thus, the callback will not
		946	be executed immediately but only after other requests in the queue have
		947	entered their execution phase. This can be used to measure request
		948	latency.
		949
		950	=item IO::AIO::aio_busy $fractional_seconds, $callback->() *NOT EXPORTED*
		951
		952	Mainly used for debugging and benchmarking, this aio request puts one of
		953	the request workers to sleep for the given time.
		954
		955	While it is theoretically handy to have simple I/O scheduling requests
		956	like sleep and file handle readable/writable, the overhead this creates is
		957	immense (it blocks a thread for a long time) so do not use this function
		958	except to put your application under artificial I/O pressure.
		959
194	=back	960	=back
195		961
		962	=head2 IO::AIO::REQ CLASS
		963
		964	All non-aggregate C<aio_*> functions return an object of this class when
		965	called in non-void context.
		966
		967	=over 4
		968
		969	=item cancel $req
		970
		971	Cancels the request, if possible. Has the effect of skipping execution
		972	when entering the B<execute> state and skipping calling the callback when
		973	entering the the B<result> state, but will leave the request otherwise
		974	untouched. That means that requests that currently execute will not be
		975	stopped and resources held by the request will not be freed prematurely.
		976
		977	=item cb $req $callback->(...)
		978
		979	Replace (or simply set) the callback registered to the request.
		980
		981	=back
		982
		983	=head2 IO::AIO::GRP CLASS
		984
		985	This class is a subclass of L<IO::AIO::REQ>, so all its methods apply to
		986	objects of this class, too.
		987
		988	A IO::AIO::GRP object is a special request that can contain multiple other
		989	aio requests.
		990
		991	You create one by calling the C<aio_group> constructing function with a
		992	callback that will be called when all contained requests have entered the
		993	C<done> state:
		994
		995	my $grp = aio_group sub {
		996	print "all requests are done\n";
		997	};
		998
		999	You add requests by calling the C<add> method with one or more
		1000	C<IO::AIO::REQ> objects:
		1001
		1002	$grp->add (aio_unlink "...");
		1003
		1004	add $grp aio_stat "...", sub {
		1005	$_[0] or return $grp->result ("error");
		1006
		1007	# add another request dynamically, if first succeeded
		1008	add $grp aio_open "...", sub {
		1009	$grp->result ("ok");
		1010	};
		1011	};
		1012
		1013	This makes it very easy to create composite requests (see the source of
		1014	C<aio_move> for an application) that work and feel like simple requests.
		1015
		1016	=over 4
		1017
		1018	=item * The IO::AIO::GRP objects will be cleaned up during calls to
		1019	C<IO::AIO::poll_cb>, just like any other request.
		1020
		1021	=item * They can be canceled like any other request. Canceling will cancel not
		1022	only the request itself, but also all requests it contains.
		1023
		1024	=item * They can also can also be added to other IO::AIO::GRP objects.
		1025
		1026	=item * You must not add requests to a group from within the group callback (or
		1027	any later time).
		1028
		1029	=back
		1030
		1031	Their lifetime, simplified, looks like this: when they are empty, they
		1032	will finish very quickly. If they contain only requests that are in the
		1033	C<done> state, they will also finish. Otherwise they will continue to
		1034	exist.
		1035
		1036	That means after creating a group you have some time to add requests
		1037	(precisely before the callback has been invoked, which is only done within
		1038	the C<poll_cb>). And in the callbacks of those requests, you can add
		1039	further requests to the group. And only when all those requests have
		1040	finished will the the group itself finish.
		1041
		1042	=over 4
		1043
		1044	=item add $grp ...
		1045
		1046	=item $grp->add (...)
		1047
		1048	Add one or more requests to the group. Any type of L<IO::AIO::REQ> can
		1049	be added, including other groups, as long as you do not create circular
		1050	dependencies.
		1051
		1052	Returns all its arguments.
		1053
		1054	=item $grp->cancel_subs
		1055
		1056	Cancel all subrequests and clears any feeder, but not the group request
		1057	itself. Useful when you queued a lot of events but got a result early.
		1058
		1059	=item $grp->result (...)
		1060
		1061	Set the result value(s) that will be passed to the group callback when all
		1062	subrequests have finished and set the groups errno to the current value
		1063	of errno (just like calling C<errno> without an error number). By default,
		1064	no argument will be passed and errno is zero.
		1065
		1066	=item $grp->errno ([$errno])
		1067
		1068	Sets the group errno value to C<$errno>, or the current value of errno
		1069	when the argument is missing.
		1070
		1071	Every aio request has an associated errno value that is restored when
		1072	the callback is invoked. This method lets you change this value from its
		1073	default (0).
		1074
		1075	Calling C<result> will also set errno, so make sure you either set C<$!>
		1076	before the call to C<result>, or call c<errno> after it.
		1077
		1078	=item feed $grp $callback->($grp)
		1079
		1080	Sets a feeder/generator on this group: every group can have an attached
		1081	generator that generates requests if idle. The idea behind this is that,
		1082	although you could just queue as many requests as you want in a group,
		1083	this might starve other requests for a potentially long time. For example,
		1084	C<aio_scandir> might generate hundreds of thousands C<aio_stat> requests,
		1085	delaying any later requests for a long time.
		1086
		1087	To avoid this, and allow incremental generation of requests, you can
		1088	instead a group and set a feeder on it that generates those requests. The
		1089	feed callback will be called whenever there are few enough (see C<limit>,
		1090	below) requests active in the group itself and is expected to queue more
		1091	requests.
		1092
		1093	The feed callback can queue as many requests as it likes (i.e. C<add> does
		1094	not impose any limits).
		1095
		1096	If the feed does not queue more requests when called, it will be
		1097	automatically removed from the group.
		1098
		1099	If the feed limit is C<0> when this method is called, it will be set to
		1100	C<2> automatically.
		1101
		1102	Example:
		1103
		1104	# stat all files in @files, but only ever use four aio requests concurrently:
		1105
		1106	my $grp = aio_group sub { print "finished\n" };
		1107	limit $grp 4;
		1108	feed $grp sub {
		1109	my $file = pop @files
		1110	or return;
		1111
		1112	add $grp aio_stat $file, sub { ... };
		1113	};
		1114
		1115	=item limit $grp $num
		1116
		1117	Sets the feeder limit for the group: The feeder will be called whenever
		1118	the group contains less than this many requests.
		1119
		1120	Setting the limit to C<0> will pause the feeding process.
		1121
		1122	The default value for the limit is C<0>, but note that setting a feeder
		1123	automatically bumps it up to C<2>.
		1124
		1125	=back
		1126
196	=head2 SUPPORT FUNCTIONS	1127	=head2 SUPPORT FUNCTIONS
197		1128
		1129	=head3 EVENT PROCESSING AND EVENT LOOP INTEGRATION
		1130
198	=over 4	1131	=over 4
199		1132
200	=item $fileno = IO::AIO::poll_fileno	1133	=item $fileno = IO::AIO::poll_fileno
201		1134
202	Return the I<request result pipe filehandle>. This filehandle must be	1135	Return the I<request result pipe file descriptor>. This filehandle must be
203	polled for reading by some mechanism outside this module (e.g. Event	1136	polled for reading by some mechanism outside this module (e.g. Event or
204	or select, see below). If the pipe becomes readable you have to call	1137	select, see below or the SYNOPSIS). If the pipe becomes readable you have
205	C<poll_cb> to check the results.	1138	to call C<poll_cb> to check the results.
206		1139
207	See C<poll_cb> for an example.	1140	See C<poll_cb> for an example.
208		1141
209	=item IO::AIO::poll_cb	1142	=item IO::AIO::poll_cb
210		1143
211	Process all outstanding events on the result pipe. You have to call this	1144	Process some outstanding events on the result pipe. You have to call this
212	regularly. Returns the number of events processed. Returns immediately	1145	regularly. Returns C<0> if all events could be processed, or C<-1> if it
213	when no events are outstanding.	1146	returned earlier for whatever reason. Returns immediately when no events
		1147	are outstanding. The amount of events processed depends on the settings of
		1148	C<IO::AIO::max_poll_req> and C<IO::AIO::max_poll_time>.
214		1149
215	You can use Event to multiplex, e.g.:	1150	If not all requests were processed for whatever reason, the filehandle
		1151	will still be ready when C<poll_cb> returns, so normally you don't have to
		1152	do anything special to have it called later.
		1153
		1154	Example: Install an Event watcher that automatically calls
		1155	IO::AIO::poll_cb with high priority:
216		1156
217	Event->io (fd => IO::AIO::poll_fileno,	1157	Event->io (fd => IO::AIO::poll_fileno,
218	poll => 'r', async => 1,	1158	poll => 'r', async => 1,
219	cb => \&IO::AIO::poll_cb);	1159	cb => \&IO::AIO::poll_cb);
220		1160
		1161	=item IO::AIO::max_poll_reqs $nreqs
		1162
		1163	=item IO::AIO::max_poll_time $seconds
		1164
		1165	These set the maximum number of requests (default C<0>, meaning infinity)
		1166	that are being processed by C<IO::AIO::poll_cb> in one call, respectively
		1167	the maximum amount of time (default C<0>, meaning infinity) spent in
		1168	C<IO::AIO::poll_cb> to process requests (more correctly the mininum amount
		1169	of time C<poll_cb> is allowed to use).
		1170
		1171	Setting C<max_poll_time> to a non-zero value creates an overhead of one
		1172	syscall per request processed, which is not normally a problem unless your
		1173	callbacks are really really fast or your OS is really really slow (I am
		1174	not mentioning Solaris here). Using C<max_poll_reqs> incurs no overhead.
		1175
		1176	Setting these is useful if you want to ensure some level of
		1177	interactiveness when perl is not fast enough to process all requests in
		1178	time.
		1179
		1180	For interactive programs, values such as C<0.01> to C<0.1> should be fine.
		1181
		1182	Example: Install an Event watcher that automatically calls
		1183	IO::AIO::poll_cb with low priority, to ensure that other parts of the
		1184	program get the CPU sometimes even under high AIO load.
		1185
		1186	# try not to spend much more than 0.1s in poll_cb
		1187	IO::AIO::max_poll_time 0.1;
		1188
		1189	# use a low priority so other tasks have priority
		1190	Event->io (fd => IO::AIO::poll_fileno,
		1191	poll => 'r', nice => 1,
		1192	cb => &IO::AIO::poll_cb);
		1193
221	=item IO::AIO::poll_wait	1194	=item IO::AIO::poll_wait
222		1195
		1196	If there are any outstanding requests and none of them in the result
223	Wait till the result filehandle becomes ready for reading (simply does a	1197	phase, wait till the result filehandle becomes ready for reading (simply
224	select on the filehandle. This is useful if you want to synchronously wait	1198	does a C<select> on the filehandle. This is useful if you want to
225	for some requests to finish).	1199	synchronously wait for some requests to finish).
226		1200
227	See C<nreqs> for an example.	1201	See C<nreqs> for an example.
228		1202
		1203	=item IO::AIO::poll
		1204
		1205	Waits until some requests have been handled.
		1206
		1207	Returns the number of requests processed, but is otherwise strictly
		1208	equivalent to:
		1209
		1210	IO::AIO::poll_wait, IO::AIO::poll_cb
		1211
229	=item IO::AIO::nreqs	1212	=item IO::AIO::flush
230		1213
231	Returns the number of requests currently outstanding.	1214	Wait till all outstanding AIO requests have been handled.
232		1215
233	Example: wait till there are no outstanding requests anymore:	1216	Strictly equivalent to:
234		1217
235	IO::AIO::poll_wait, IO::AIO::poll_cb	1218	IO::AIO::poll_wait, IO::AIO::poll_cb
236	while IO::AIO::nreqs;	1219	while IO::AIO::nreqs;
237		1220
238	=item IO::AIO::flush	1221	=back
239		1222
240	Wait till all outstanding AIO requests have been handled.	1223	=head3 CONTROLLING THE NUMBER OF THREADS
		1224
		1225	=over
241		1226
242	=item IO::AIO::min_parallel $nthreads	1227	=item IO::AIO::min_parallel $nthreads
243		1228
244	Set the minimum number of AIO threads to C<$nthreads>. The default is	1229	Set the minimum number of AIO threads to C<$nthreads>. The current
245	C<1>, which means a single asynchronous operation can be done at one time	1230	default is C<8>, which means eight asynchronous operations can execute
246	(the number of outstanding operations, however, is unlimited).	1231	concurrently at any one time (the number of outstanding requests,
		1232	however, is unlimited).
247		1233
		1234	IO::AIO starts threads only on demand, when an AIO request is queued and
		1235	no free thread exists. Please note that queueing up a hundred requests can
		1236	create demand for a hundred threads, even if it turns out that everything
		1237	is in the cache and could have been processed faster by a single thread.
		1238
248	It is recommended to keep the number of threads low, as some Linux	1239	It is recommended to keep the number of threads relatively low, as some
249	kernel versions will scale negatively with the number of threads (higher	1240	Linux kernel versions will scale negatively with the number of threads
250	parallelity => MUCH higher latency). With current Linux 2.6 versions, 4-32	1241	(higher parallelity => MUCH higher latency). With current Linux 2.6
251	threads should be fine.	1242	versions, 4-32 threads should be fine.
252		1243
253	Under normal circumstances you don't need to call this function, as this	1244	Under most circumstances you don't need to call this function, as the
254	module automatically starts some threads (the exact number might change,	1245	module selects a default that is suitable for low to moderate load.
255	and is currently 4).
256		1246
257	=item IO::AIO::max_parallel $nthreads	1247	=item IO::AIO::max_parallel $nthreads
258		1248
259	Sets the maximum number of AIO threads to C<$nthreads>. If more than	1249	Sets the maximum number of AIO threads to C<$nthreads>. If more than the
260	the specified number of threads are currently running, kill them. This	1250	specified number of threads are currently running, this function kills
261	function blocks until the limit is reached.	1251	them. This function blocks until the limit is reached.
		1252
		1253	While C<$nthreads> are zero, aio requests get queued but not executed
		1254	until the number of threads has been increased again.
262		1255
263	This module automatically runs C<max_parallel 0> at program end, to ensure	1256	This module automatically runs C<max_parallel 0> at program end, to ensure
264	that all threads are killed and that there are no outstanding requests.	1257	that all threads are killed and that there are no outstanding requests.
265		1258
266	Under normal circumstances you don't need to call this function.	1259	Under normal circumstances you don't need to call this function.
267		1260
		1261	=item IO::AIO::max_idle $nthreads
		1262
		1263	Limit the number of threads (default: 4) that are allowed to idle (i.e.,
		1264	threads that did not get a request to process within 10 seconds). That
		1265	means if a thread becomes idle while C<$nthreads> other threads are also
		1266	idle, it will free its resources and exit.
		1267
		1268	This is useful when you allow a large number of threads (e.g. 100 or 1000)
		1269	to allow for extremely high load situations, but want to free resources
		1270	under normal circumstances (1000 threads can easily consume 30MB of RAM).
		1271
		1272	The default is probably ok in most situations, especially if thread
		1273	creation is fast. If thread creation is very slow on your system you might
		1274	want to use larger values.
		1275
268	=item $oldnreqs = IO::AIO::max_outstanding $nreqs	1276	=item IO::AIO::max_outstanding $maxreqs
		1277
		1278	This is a very bad function to use in interactive programs because it
		1279	blocks, and a bad way to reduce concurrency because it is inexact: Better
		1280	use an C<aio_group> together with a feed callback.
269		1281
270	Sets the maximum number of outstanding requests to C<$nreqs>. If you	1282	Sets the maximum number of outstanding requests to C<$nreqs>. If you
271	try to queue up more than this number of requests, the caller will block until	1283	do queue up more than this number of requests, the next call to the
272	some requests have been handled.	1284	C<poll_cb> (and C<poll_some> and other functions calling C<poll_cb>)
		1285	function will block until the limit is no longer exceeded.
273		1286
274	The default is very large, so normally there is no practical limit. If you	1287	The default value is very large, so there is no practical limit on the
275	queue up many requests in a loop it it often improves speed if you set	1288	number of outstanding requests.
276	this to a relatively low number, such as C<100>.
277		1289
278	Under normal circumstances you don't need to call this function.	1290	You can still queue as many requests as you want. Therefore,
		1291	C<max_outstanding> is mainly useful in simple scripts (with low values) or
		1292	as a stop gap to shield against fatal memory overflow (with large values).
279		1293
280	=back	1294	=back
281		1295
		1296	=head3 STATISTICAL INFORMATION
		1297
		1298	=over
		1299
		1300	=item IO::AIO::nreqs
		1301
		1302	Returns the number of requests currently in the ready, execute or pending
		1303	states (i.e. for which their callback has not been invoked yet).
		1304
		1305	Example: wait till there are no outstanding requests anymore:
		1306
		1307	IO::AIO::poll_wait, IO::AIO::poll_cb
		1308	while IO::AIO::nreqs;
		1309
		1310	=item IO::AIO::nready
		1311
		1312	Returns the number of requests currently in the ready state (not yet
		1313	executed).
		1314
		1315	=item IO::AIO::npending
		1316
		1317	Returns the number of requests currently in the pending state (executed,
		1318	but not yet processed by poll_cb).
		1319
		1320	=back
		1321
282	=cut	1322	=cut
283		1323
284	# support function to convert a fd into a perl filehandle
285	sub _fd2fh {
286	return undef if $_[0] < 0;
287
288	# try to be perl5.6-compatible
289	local *AIO_FH;
290	open AIO_FH, "+<&=$_[0]"
291	or return undef;
292
293	*AIO_FH
294	}
295
296	min_parallel 4;	1324	min_parallel 8;
297		1325
298	END {	1326	END { flush }
299	max_parallel 0;
300	}
301		1327
302	1;	1328	1;
303		1329
		1330	=head2 FORK BEHAVIOUR
		1331
		1332	This module should do "the right thing" when the process using it forks:
		1333
		1334	Before the fork, IO::AIO enters a quiescent state where no requests
		1335	can be added in other threads and no results will be processed. After
		1336	the fork the parent simply leaves the quiescent state and continues
		1337	request/result processing, while the child frees the request/result queue
		1338	(so that the requests started before the fork will only be handled in the
		1339	parent). Threads will be started on demand until the limit set in the
		1340	parent process has been reached again.
		1341
		1342	In short: the parent will, after a short pause, continue as if fork had
		1343	not been called, while the child will act as if IO::AIO has not been used
		1344	yet.
		1345
		1346	=head2 MEMORY USAGE
		1347
		1348	Per-request usage:
		1349
		1350	Each aio request uses - depending on your architecture - around 100-200
		1351	bytes of memory. In addition, stat requests need a stat buffer (possibly
		1352	a few hundred bytes), readdir requires a result buffer and so on. Perl
		1353	scalars and other data passed into aio requests will also be locked and
		1354	will consume memory till the request has entered the done state.
		1355
		1356	This is not awfully much, so queuing lots of requests is not usually a
		1357	problem.
		1358
		1359	Per-thread usage:
		1360
		1361	In the execution phase, some aio requests require more memory for
		1362	temporary buffers, and each thread requires a stack and other data
		1363	structures (usually around 16k-128k, depending on the OS).
		1364
		1365	=head1 KNOWN BUGS
		1366
		1367	Known bugs will be fixed in the next release.
		1368
304	=head1 SEE ALSO	1369	=head1 SEE ALSO
305		1370
306	L<Coro>, L<Linux::AIO>.	1371	L<AnyEvent::AIO> for easy integration into event loops, L<Coro::AIO> for a
		1372	more natural syntax.
307		1373
308	=head1 AUTHOR	1374	=head1 AUTHOR
309		1375
310	Marc Lehmann <schmorp@schmorp.de>	1376	Marc Lehmann <schmorp@schmorp.de>
311	http://home.schmorp.de/	1377	http://home.schmorp.de/

Diff Legend

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