ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/IO-AIO/AIO.pm

Comparing IO-AIO/AIO.pm (file contents):
Revision 1.2 by root, Sun Jul 10 18:16:49 2005 UTC vs.
Revision 1.220 by root, Sun Apr 1 17:46:02 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", IO::AIO::O_RDONLY, 0, sub {
		10	my $fh = shift
		11	or die "/etc/passwd: $!";
		12	...
		13	};
		14
		15	aio_unlink "/tmp/file", sub { };
		16
		17	aio_read $fh, 30000, 1024, $buffer, 0, sub {
		18	$_[0] > 0 or die "read error: $!";
		19	};
		20
		21	# version 2+ has request and group objects
		22	use IO::AIO 2;
		23
		24	aioreq_pri 4; # give next request a very high priority
		25	my $req = aio_unlink "/tmp/file", sub { };
		26	$req->cancel; # cancel request if still in queue
		27
		28	my $grp = aio_group sub { print "all stats done\n" };
		29	add $grp aio_stat "..." for ...;
		30
9	=head1 DESCRIPTION	31	=head1 DESCRIPTION
10		32
11	This module implements asynchronous I/O using whatever means your	33	This module implements asynchronous I/O using whatever means your
12	operating system supports.	34	operating system supports. It is implemented as an interface to C<libeio>
		35	(L<http://software.schmorp.de/pkg/libeio.html>).
13		36
		37	Asynchronous means that operations that can normally block your program
		38	(e.g. reading from disk) will be done asynchronously: the operation
		39	will still block, but you can do something else in the meantime. This
		40	is extremely useful for programs that need to stay interactive even
		41	when doing heavy I/O (GUI programs, high performance network servers
		42	etc.), but can also be used to easily do operations in parallel that are
		43	normally done sequentially, e.g. stat'ing many files, which is much faster
		44	on a RAID volume or over NFS when you do a number of stat operations
		45	concurrently.
		46
		47	While most of this works on all types of file descriptors (for
		48	example sockets), using these functions on file descriptors that
		49	support nonblocking operation (again, sockets, pipes etc.) is
		50	very inefficient. Use an event loop for that (such as the L<EV>
		51	module): IO::AIO will naturally fit into such an event loop itself.
		52
14	Currently, a number of threads are started that execute your read/writes	53	In this version, a number of threads are started that execute your
15	and signal their completion. You don't need thread support in your libc or	54	requests and signal their completion. You don't need thread support
16	perl, and the threads created by this module will not be visible to the	55	in perl, and the threads created by this module will not be visible
17	pthreads library. In the future, this module might make use of the native	56	to perl. In the future, this module might make use of the native aio
18	aio functions available on many operating systems. However, they are often	57	functions available on many operating systems. However, they are often
19	not well-supported (Linux doesn't allow them on normal files currently,	58	not well-supported or restricted (GNU/Linux doesn't allow them on normal
20	for example), and they would only support aio_read and aio_write, so the	59	files currently, for example), and they would only support aio_read and
21	remaining functionality would have to be implemented using threads anyway.	60	aio_write, so the remaining functionality would have to be implemented
		61	using threads anyway.
22		62
23	Although the module will work with in the presence of other threads, it is	63	Although the module will work in the presence of other (Perl-) threads,
24	currently not reentrant, so use appropriate locking yourself.	64	it is currently not reentrant in any way, so use appropriate locking
		65	yourself, always call C<poll_cb> from within the same thread, or never
		66	call C<poll_cb> (or other C<aio_> functions) recursively.
		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
		161
		162	=cut
		163
		164	package IO::AIO;
		165
		166	use Carp ();
		167
		168	use common::sense;
		169
		170	use base 'Exporter';
		171
		172	BEGIN {
		173	our $VERSION = '4.13';
		174
		175	our @AIO_REQ = qw(aio_sendfile aio_seek aio_read aio_write aio_open aio_close
		176	aio_stat aio_lstat aio_unlink aio_rmdir aio_readdir aio_readdirx
		177	aio_scandir aio_symlink aio_readlink aio_realpath aio_sync
		178	aio_fsync aio_syncfs aio_fdatasync aio_sync_file_range aio_fallocate
		179	aio_pathsync aio_readahead
		180	aio_rename aio_link aio_move aio_copy aio_group
		181	aio_nop aio_mknod aio_load aio_rmtree aio_mkdir aio_chown
		182	aio_chmod aio_utime aio_truncate
		183	aio_msync aio_mtouch aio_mlock aio_mlockall
		184	aio_statvfs
		185	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';
		198
		199	require XSLoader;
		200	XSLoader::load ("IO::AIO", $VERSION);
		201	}
		202
		203	=head1 FUNCTIONS
		204
		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, $length = undef
		280	IO::AIO::munlockall
25		281
26	=head2 API NOTES	282	=head2 API NOTES
27		283
28	All the C<aio_*> calls are more or less thin wrappers around the syscall	284	All the C<aio_*> calls are more or less thin wrappers around the syscall
29	with the same name (sans C<aio_>). The arguments are similar or identical,	285	with the same name (sans C<aio_>). The arguments are similar or identical,
30	and they all accept an additional C<$callback> argument which must be	286	and they all accept an additional (and optional) C<$callback> argument
31	a code reference. This code reference will get called with the syscall	287	which must be a code reference. This code reference will be called after
		288	the syscall has been executed in an asynchronous fashion. The results
		289	of the request will be passed as arguments to the callback (and, if an
		290	error occured, in C<$!>) - for most requests the syscall return code (e.g.
32	return code (e.g. most syscalls return C<-1> on error, unlike perl, which	291	most syscalls return C<-1> on error, unlike perl, which usually delivers
33	usually delivers "false") as it's sole argument when the given syscall has	292	"false").
34	been executed asynchronously.
35		293
36	All functions that expect a filehandle will also accept a file descriptor.	294	Some requests (such as C<aio_readdir>) pass the actual results and
		295	communicate failures by passing C<undef>.
37		296
		297	All functions expecting a filehandle keep a copy of the filehandle
		298	internally until the request has finished.
		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
38	The filenames you pass to these routines I<must> be absolute. The reason	303	The pathnames you pass to these routines I<should> be absolute. The
39	is that at the time the request is being executed, the current working	304	reason for this is that at the time the request is being executed, the
40	directory could have changed. Alternatively, you can make sure that you	305	current working directory could have changed. Alternatively, you can
41	never change the current working directory.	306	make sure that you never change the current working directory anywhere
		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.
		311
		312	To encode pathnames as octets, either make sure you either: a) always pass
		313	in filenames you got from outside (command line, readdir etc.) without
		314	tinkering, b) are in your native filesystem encoding, c) use the Encode
		315	module and encode your pathnames to the locale (or other) encoding in
		316	effect in the user environment, d) use Glib::filename_from_unicode on
		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
42		324
43	=over 4	325	=over 4
44		326
45	=cut	327	=item $prev_pri = aioreq_pri [$pri]
46		328
47	package IO::AIO;	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.
48		331
49	use base 'Exporter';	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.
50		335
51	use Fcntl ();	336	The priority will be reset to C<0> after each call to one of the C<aio_*>
		337	functions.
52		338
53	BEGIN {	339	Example: open a file with low priority, then read something from it with
54	$VERSION = 0.1;	340	higher priority so the read request is serviced before other low priority
		341	open requests (potentially spamming the cache):
55		342
56	@EXPORT = qw(aio_read aio_write aio_open aio_close aio_stat aio_lstat aio_unlink	343	aioreq_pri -3;
57	aio_fsync aio_fdatasync aio_readahead);	344	aio_open ..., sub {
58	@EXPORT_OK = qw(poll_fileno poll_cb min_parallel max_parallel nreqs);	345	return unless $_[0];
59		346
60	require XSLoader;	347	aioreq_pri -2;
61	XSLoader::load IO::AIO, $VERSION;	348	aio_read $_[0], ..., sub {
62	}	349	...
		350	};
		351	};
63		352
64	=item IO::AIO::min_parallel $nthreads
65		353
66	Set the minimum number of AIO threads to C<$nthreads>. The default is	354	=item aioreq_nice $pri_adjust
67	C<1>, which means a single asynchronous operation can be done at one time
68	(the number of outstanding operations, however, is unlimited).
69		355
70	It is recommended to keep the number of threads low, as some linux	356	Similar to C<aioreq_pri>, but subtracts the given value from the current
71	kernel versions will scale negatively with the number of threads (higher	357	priority, so the effect is cumulative.
72	parallelity => MUCH higher latency).
73		358
74	Under normal circumstances you don't need to call this function, as this
75	module automatically starts a single async thread.
76		359
77	=item IO::AIO::max_parallel $nthreads
78
79	Sets the maximum number of AIO threads to C<$nthreads>. If more than
80	the specified number of threads are currently running, kill them. This
81	function blocks until the limit is reached.
82
83	This module automatically runs C<max_parallel 0> at program end, to ensure
84	that all threads are killed and that there are no outstanding requests.
85
86	Under normal circumstances you don't need to call this function.
87
88	=item $fileno = IO::AIO::poll_fileno
89
90	Return the I<request result pipe filehandle>. This filehandle must be
91	polled for reading by some mechanism outside this module (e.g. Event
92	or select, see below). If the pipe becomes readable you have to call
93	C<poll_cb> to check the results.
94
95	See C<poll_cb> for an example.
96
97	=item IO::AIO::poll_cb
98
99	Process all outstanding events on the result pipe. You have to call this
100	regularly. Returns the number of events processed. Returns immediately
101	when no events are outstanding.
102
103	You can use Event to multiplex, e.g.:
104
105	Event->io (fd => IO::AIO::poll_fileno,
106	poll => 'r', async => 1,
107	cb => \&IO::AIO::poll_cb);
108
109	=item IO::AIO::poll_wait
110
111	Wait till the result filehandle becomes ready for reading (simply does a
112	select on the filehandle. This is useful if you want to synchronously wait
113	for some requests to finish).
114
115	See C<nreqs> for an example.
116
117	=item IO::AIO::nreqs
118
119	Returns the number of requests currently outstanding.
120
121	Example: wait till there are no outstanding requests anymore:
122
123	IO::AIO::poll_wait, IO::AIO::poll_cb
124	while IO::AIO::nreqs;
125
126	=item aio_open $pathname, $flags, $mode, $callback	360	=item aio_open $pathname, $flags, $mode, $callback->($fh)
127		361
128	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
129	created filehandle for the file.	363	created filehandle for the file.
130		364
131	The pathname passed to C<aio_open> must be absolute. See API NOTES, above,	365	The pathname passed to C<aio_open> must be absolute. See API NOTES, above,
132	for an explanation.	366	for an explanation.
133		367
134	The C<$mode> argument is a bitmask. See the C<Fcntl> module for a	368	The C<$flags> argument is a bitmask. See the C<Fcntl> module for a
135	list. They are the same as used in C<sysopen>.	369	list. They are the same as used by C<sysopen>.
		370
		371	Likewise, C<$mode> specifies the mode of the newly created file, if it
		372	didn't exist and C<O_CREAT> has been given, just like perl's C<sysopen>,
		373	except that it is mandatory (i.e. use C<0> if you don't create new files,
		374	and C<0666> or C<0777> if you do). Note that the C<$mode> will be modified
		375	by the umask in effect then the request is being executed, so better never
		376	change the umask.
136		377
137	Example:	378	Example:
138		379
139	aio_open "/etc/passwd", O_RDONLY, 0, sub {	380	aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub {
140	if ($_[0]) {	381	if ($_[0]) {
141	print "open successful, fh is $_[0]\n";	382	print "open successful, fh is $_[0]\n";
142	...	383	...
143	} else {	384	} else {
144	die "open failed: $!\n";	385	die "open failed: $!\n";
145	}	386	}
146	};	387	};
147		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
148	=item aio_close $fh, $callback	399	=item aio_close $fh, $callback->($status)
149		400
150	Asynchronously close a file and call the callback with the result	401	Asynchronously close a file and call the callback with the result
151	code. I<WARNING:> although accepted, you should not pass in a perl	402	code.
152	filehandle here, as perl will likely close the file descriptor itself when
153	the filehandle is destroyed. Normally, you can safely call perls C<close>
154	or just let filehandles go out of scope.
155		403
		404	Unfortunately, you can't do this to perl. Perl I<insists> very strongly on
		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
156	=item aio_read $fh,$offset,$length, $data,$dataoffset,$callback	430	=item aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
157		431
158	=item aio_write $fh,$offset,$length, $data,$dataoffset,$callback	432	=item aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
159		433
160	Reads or writes C<length> bytes from the specified C<fh> and C<offset>	434	Reads or writes C<$length> bytes from or to the specified C<$fh> and
161	into the scalar given by C<data> and offset C<dataoffset> and calls the	435	C<$offset> into the scalar given by C<$data> and offset C<$dataoffset>
162	callback without the actual number of bytes read (or -1 on error, just	436	and calls the callback without the actual number of bytes read (or -1 on
163	like the syscall).	437	error, just like the syscall).
164		438
		439	C<aio_read> will, like C<sysread>, shrink or grow the C<$data> scalar to
		440	offset plus the actual number of bytes read.
		441
		442	If C<$offset> is undefined, then the current file descriptor offset will
		443	be used (and updated), otherwise the file descriptor offset will not be
		444	changed by these calls.
		445
		446	If C<$length> is undefined in C<aio_write>, use the remaining length of
		447	C<$data>.
		448
		449	If C<$dataoffset> is less than zero, it will be counted from the end of
		450	C<$data>.
		451
		452	The C<$data> scalar I<MUST NOT> be modified in any way while the request
		453	is outstanding. Modifying it can result in segfaults or World War III (if
		454	the necessary/optional hardware is installed).
		455
165	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, strating at	456	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at
166	offset C<0> within the scalar:	457	offset C<0> within the scalar:
167		458
168	aio_read $fh, 7, 15, $buffer, 0, sub {	459	aio_read $fh, 7, 15, $buffer, 0, sub {
169	$_[0] >= 0 or die "read error: $!";	460	$_[0] > 0 or die "read error: $!";
170	print "read <$buffer>\n";	461	print "read $_[0] bytes: <$buffer>\n";
171	};	462	};
172		463
		464
		465	=item aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
		466
		467	Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts
		468	reading at byte offset C<$in_offset>, and starts writing at the current
		469	file offset of C<$out_fh>. Because of that, it is not safe to issue more
		470	than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each
		471	other. The same C<$in_fh> works fine though, as this function does not
		472	move or use the file offset of C<$in_fh>.
		473
		474	Please note that C<aio_sendfile> can read more bytes from C<$in_fh> than
		475	are written, and there is no way to find out how many more bytes have been
		476	read from C<aio_sendfile> alone, as C<aio_sendfile> only provides the
		477	number of bytes written to C<$out_fh>. Only if the result value equals
		478	C<$length> one can assume that C<$length> bytes have been read.
		479
		480	Unlike with other C<aio_> functions, it makes a lot of sense to use
		481	C<aio_sendfile> on non-blocking sockets, as long as one end (typically
		482	the C<$in_fh>) is a file - the file I/O will then be asynchronous, while
		483	the socket I/O will be non-blocking. Note, however, that you can run
		484	into a trap where C<aio_sendfile> reads some data with readahead, then
		485	fails to write all data, and when the socket is ready the next time, the
		486	data in the cache is already lost, forcing C<aio_sendfile> to again hit
		487	the disk. Explicit C<aio_read> + C<aio_write> let's you better control
		488	resource usage.
		489
		490	This call tries to make use of a native C<sendfile>-like syscall to
		491	provide zero-copy operation. For this to work, C<$out_fh> should refer to
		492	a socket, and C<$in_fh> should refer to an mmap'able file.
		493
		494	If a native sendfile cannot be found or it fails with C<ENOSYS>,
		495	C<EINVAL>, C<ENOTSUP>, C<EOPNOTSUPP>, C<EAFNOSUPPORT>, C<EPROTOTYPE> or
		496	C<ENOTSOCK>, it will be emulated, so you can call C<aio_sendfile> on any
		497	type of filehandle regardless of the limitations of the operating system.
		498
		499	As native sendfile syscalls (as practically any non-POSIX interface hacked
		500	together in a hurry to improve benchmark numbers) tend to be rather buggy
		501	on many systems, this implementation tries to work around some known bugs
		502	in Linux and FreeBSD kernels (probably others, too), but that might fail,
		503	so you really really should check the return value of C<aio_sendfile> -
		504	fewre bytes than expected might have been transferred.
		505
		506
173	=item aio_readahead $fh,$offset,$length, $callback	507	=item aio_readahead $fh,$offset,$length, $callback->($retval)
174		508
175	Asynchronously reads the specified byte range into the page cache, using
176	the C<readahead> syscall. If that syscall doesn't exist the status will be
177	C<-1> and C<$!> is set to ENOSYS.
178
179	readahead() populates the page cache with data from a file so that	509	C<aio_readahead> populates the page cache with data from a file so that
180	subsequent reads from that file will not block on disk I/O. The C<$offset>	510	subsequent reads from that file will not block on disk I/O. The C<$offset>
181	argument specifies the starting point from which data is to be read and	511	argument specifies the starting point from which data is to be read and
182	C<$length> specifies the number of bytes to be read. I/O is performed in	512	C<$length> specifies the number of bytes to be read. I/O is performed in
183	whole pages, so that offset is effectively rounded down to a page boundary	513	whole pages, so that offset is effectively rounded down to a page boundary
184	and bytes are read up to the next page boundary greater than or equal to	514	and bytes are read up to the next page boundary greater than or equal to
185	(off-set+length). aio_readahead() does not read beyond the end of the	515	(off-set+length). C<aio_readahead> does not read beyond the end of the
186	file. The current file offset of the file is left unchanged.	516	file. The current file offset of the file is left unchanged.
187		517
		518	If that syscall doesn't exist (likely if your OS isn't Linux) it will be
		519	emulated by simply reading the data, which would have a similar effect.
		520
		521
188	=item aio_stat $fh_or_path, $callback	522	=item aio_stat $fh_or_path, $callback->($status)
189		523
190	=item aio_lstat $fh, $callback	524	=item aio_lstat $fh, $callback->($status)
191		525
192	Works like perl's C<stat> or C<lstat> in void context. The callback will	526	Works like perl's C<stat> or C<lstat> in void context. The callback will
193	be called after the stat and the results will be available using C<stat _>	527	be called after the stat and the results will be available using C<stat _>
194	or C<-s _> etc...	528	or C<-s _> etc...
195		529
…		…
197	for an explanation.	531	for an explanation.
198		532
199	Currently, the stats are always 64-bit-stats, i.e. instead of returning an	533	Currently, the stats are always 64-bit-stats, i.e. instead of returning an
200	error when stat'ing a large file, the results will be silently truncated	534	error when stat'ing a large file, the results will be silently truncated
201	unless perl itself is compiled with large file support.	535	unless perl itself is compiled with large file support.
		536
		537	To help interpret the mode and dev/rdev stat values, IO::AIO offers the
		538	following constants and functions (if not implemented, the constants will
		539	be C<0> and the functions will either C<croak> or fall back on traditional
		540	behaviour).
		541
		542	C<S_IFMT>, C<S_IFIFO>, C<S_IFCHR>, C<S_IFBLK>, C<S_IFLNK>, C<S_IFREG>,
		543	C<S_IFDIR>, C<S_IFWHT>, C<S_IFSOCK>, C<IO::AIO::major $dev_t>,
		544	C<IO::AIO::minor $dev_t>, C<IO::AIO::makedev $major, $minor>.
202		545
203	Example: Print the length of F</etc/passwd>:	546	Example: Print the length of F</etc/passwd>:
204		547
205	aio_stat "/etc/passwd", sub {	548	aio_stat "/etc/passwd", sub {
206	$_[0] and die "stat failed: $!";	549	$_[0] and die "stat failed: $!";
207	print "size is ", -s _, "\n";	550	print "size is ", -s _, "\n";
208	};	551	};
209		552
		553
		554	=item aio_statvfs $fh_or_path, $callback->($statvfs)
		555
		556	Works like the POSIX C<statvfs> or C<fstatvfs> syscalls, depending on
		557	whether a file handle or path was passed.
		558
		559	On success, the callback is passed a hash reference with the following
		560	members: C<bsize>, C<frsize>, C<blocks>, C<bfree>, C<bavail>, C<files>,
		561	C<ffree>, C<favail>, C<fsid>, C<flag> and C<namemax>. On failure, C<undef>
		562	is passed.
		563
		564	The following POSIX IO::AIO::ST_* constants are defined: C<ST_RDONLY> and
		565	C<ST_NOSUID>.
		566
		567	The following non-POSIX IO::AIO::ST_* flag masks are defined to
		568	their correct value when available, or to C<0> on systems that do
		569	not support them: C<ST_NODEV>, C<ST_NOEXEC>, C<ST_SYNCHRONOUS>,
		570	C<ST_MANDLOCK>, C<ST_WRITE>, C<ST_APPEND>, C<ST_IMMUTABLE>, C<ST_NOATIME>,
		571	C<ST_NODIRATIME> and C<ST_RELATIME>.
		572
		573	Example: stat C</wd> and dump out the data if successful.
		574
		575	aio_statvfs "/wd", sub {
		576	my $f = $_[0]
		577	or die "statvfs: $!";
		578
		579	use Data::Dumper;
		580	say Dumper $f;
		581	};
		582
		583	# result:
		584	{
		585	bsize => 1024,
		586	bfree => 4333064312,
		587	blocks => 10253828096,
		588	files => 2050765568,
		589	flag => 4096,
		590	favail => 2042092649,
		591	bavail => 4333064312,
		592	ffree => 2042092649,
		593	namemax => 255,
		594	frsize => 1024,
		595	fsid => 1810
		596	}
		597
		598
		599	=item aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
		600
		601	Works like perl's C<utime> function (including the special case of $atime
		602	and $mtime being undef). Fractional times are supported if the underlying
		603	syscalls support them.
		604
		605	When called with a pathname, uses utimes(2) if available, otherwise
		606	utime(2). If called on a file descriptor, uses futimes(2) if available,
		607	otherwise returns ENOSYS, so this is not portable.
		608
		609	Examples:
		610
		611	# set atime and mtime to current time (basically touch(1)):
		612	aio_utime "path", undef, undef;
		613	# set atime to current time and mtime to beginning of the epoch:
		614	aio_utime "path", time, undef; # undef==0
		615
		616
		617	=item aio_chown $fh_or_path, $uid, $gid, $callback->($status)
		618
		619	Works like perl's C<chown> function, except that C<undef> for either $uid
		620	or $gid is being interpreted as "do not change" (but -1 can also be used).
		621
		622	Examples:
		623
		624	# same as "chown root path" in the shell:
		625	aio_chown "path", 0, -1;
		626	# same as above:
		627	aio_chown "path", 0, undef;
		628
		629
		630	=item aio_truncate $fh_or_path, $offset, $callback->($status)
		631
		632	Works like truncate(2) or ftruncate(2).
		633
		634
		635	=item aio_chmod $fh_or_path, $mode, $callback->($status)
		636
		637	Works like perl's C<chmod> function.
		638
		639
210	=item aio_unlink $pathname, $callback	640	=item aio_unlink $pathname, $callback->($status)
211		641
212	Asynchronously unlink (delete) a file and call the callback with the	642	Asynchronously unlink (delete) a file and call the callback with the
213	result code.	643	result code.
214		644
		645
		646	=item aio_mknod $pathname, $mode, $dev, $callback->($status)
		647
		648	[EXPERIMENTAL]
		649
		650	Asynchronously create a device node (or fifo). See mknod(2).
		651
		652	The only (POSIX-) portable way of calling this function is:
		653
		654	aio_mknod $pathname, IO::AIO::S_IFIFO | $mode, 0, sub { ...
		655
		656	See C<aio_stat> for info about some potentially helpful extra constants
		657	and functions.
		658
		659	=item aio_link $srcpath, $dstpath, $callback->($status)
		660
		661	Asynchronously create a new link to the existing object at C<$srcpath> at
		662	the path C<$dstpath> and call the callback with the result code.
		663
		664
		665	=item aio_symlink $srcpath, $dstpath, $callback->($status)
		666
		667	Asynchronously create a new symbolic 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_readlink $pathname, $callback->($link)
		672
		673	Asynchronously read the symlink specified by C<$path> and pass it to
		674	the callback. If an error occurs, nothing or undef gets passed to the
		675	callback.
		676
		677
		678	=item aio_realpath $pathname, $callback->($path)
		679
		680	Asynchronously make the path absolute and resolve any symlinks in
		681	C<$path>. The resulting path only consists of directories (Same as
		682	L<Cwd::realpath>).
		683
		684	This request can be used to get the absolute path of the current working
		685	directory by passing it a path of F<.> (a single dot).
		686
		687
		688	=item aio_rename $srcpath, $dstpath, $callback->($status)
		689
		690	Asynchronously rename the object at C<$srcpath> to C<$dstpath>, just as
		691	rename(2) and call the callback with the result code.
		692
		693
		694	=item aio_mkdir $pathname, $mode, $callback->($status)
		695
		696	Asynchronously mkdir (create) a directory and call the callback with
		697	the result code. C<$mode> will be modified by the umask at the time the
		698	request is executed, so do not change your umask.
		699
		700
		701	=item aio_rmdir $pathname, $callback->($status)
		702
		703	Asynchronously rmdir (delete) a directory and call the callback with the
		704	result code.
		705
		706
		707	=item aio_readdir $pathname, $callback->($entries)
		708
		709	Unlike the POSIX call of the same name, C<aio_readdir> reads an entire
		710	directory (i.e. opendir + readdir + closedir). The entries will not be
		711	sorted, and will B<NOT> include the C<.> and C<..> entries.
		712
		713	The callback is passed a single argument which is either C<undef> or an
		714	array-ref with the filenames.
		715
		716
		717	=item aio_readdirx $pathname, $flags, $callback->($entries, $flags)
		718
		719	Quite similar to C<aio_readdir>, but the C<$flags> argument allows one to
		720	tune behaviour and output format. In case of an error, C<$entries> will be
		721	C<undef>.
		722
		723	The flags are a combination of the following constants, ORed together (the
		724	flags will also be passed to the callback, possibly modified):
		725
		726	=over 4
		727
		728	=item IO::AIO::READDIR_DENTS
		729
		730	When this flag is off, then the callback gets an arrayref consisting of
		731	names only (as with C<aio_readdir>), otherwise it gets an arrayref with
		732	C<[$name, $type, $inode]> arrayrefs, each describing a single directory
		733	entry in more detail.
		734
		735	C<$name> is the name of the entry.
		736
		737	C<$type> is one of the C<IO::AIO::DT_xxx> constants:
		738
		739	C<IO::AIO::DT_UNKNOWN>, C<IO::AIO::DT_FIFO>, C<IO::AIO::DT_CHR>, C<IO::AIO::DT_DIR>,
		740	C<IO::AIO::DT_BLK>, C<IO::AIO::DT_REG>, C<IO::AIO::DT_LNK>, C<IO::AIO::DT_SOCK>,
		741	C<IO::AIO::DT_WHT>.
		742
		743	C<IO::AIO::DT_UNKNOWN> means just that: readdir does not know. If you need to
		744	know, you have to run stat yourself. Also, for speed reasons, the C<$type>
		745	scalars are read-only: you can not modify them.
		746
		747	C<$inode> is the inode number (which might not be exact on systems with 64
		748	bit inode numbers and 32 bit perls). This field has unspecified content on
		749	systems that do not deliver the inode information.
		750
		751	=item IO::AIO::READDIR_DIRS_FIRST
		752
		753	When this flag is set, then the names will be returned in an order where
		754	likely directories come first, in optimal stat order. This is useful when
		755	you need to quickly find directories, or you want to find all directories
		756	while avoiding to stat() each entry.
		757
		758	If the system returns type information in readdir, then this is used
		759	to find directories directly. Otherwise, likely directories are names
		760	beginning with ".", or otherwise names with no dots, of which names with
		761	short names are tried first.
		762
		763	=item IO::AIO::READDIR_STAT_ORDER
		764
		765	When this flag is set, then the names will be returned in an order
		766	suitable for stat()'ing each one. That is, when you plan to stat()
		767	all files in the given directory, then the returned order will likely
		768	be fastest.
		769
		770	If both this flag and C<IO::AIO::READDIR_DIRS_FIRST> are specified, then
		771	the likely dirs come first, resulting in a less optimal stat order.
		772
		773	=item IO::AIO::READDIR_FOUND_UNKNOWN
		774
		775	This flag should not be set when calling C<aio_readdirx>. Instead, it
		776	is being set by C<aio_readdirx>, when any of the C<$type>'s found were
		777	C<IO::AIO::DT_UNKNOWN>. The absence of this flag therefore indicates that all
		778	C<$type>'s are known, which can be used to speed up some algorithms.
		779
		780	=back
		781
		782
		783	=item aio_load $pathname, $data, $callback->($status)
		784
		785	This is a composite request that tries to fully load the given file into
		786	memory. Status is the same as with aio_read.
		787
		788	=cut
		789
		790	sub aio_load($$;$) {
		791	my ($path, undef, $cb) = @_;
		792	my $data = \$_[1];
		793
		794	my $pri = aioreq_pri;
		795	my $grp = aio_group $cb;
		796
		797	aioreq_pri $pri;
		798	add $grp aio_open $path, O_RDONLY, 0, sub {
		799	my $fh = shift
		800	or return $grp->result (-1);
		801
		802	aioreq_pri $pri;
		803	add $grp aio_read $fh, 0, (-s $fh), $$data, 0, sub {
		804	$grp->result ($_[0]);
		805	};
		806	};
		807
		808	$grp
		809	}
		810
		811	=item aio_copy $srcpath, $dstpath, $callback->($status)
		812
		813	Try to copy the I<file> (directories not supported as either source or
		814	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		815	a status of C<0> (ok) or C<-1> (error, see C<$!>).
		816
		817	This is a composite request that creates the destination file with
		818	mode 0200 and copies the contents of the source file into it using
		819	C<aio_sendfile>, followed by restoring atime, mtime, access mode and
		820	uid/gid, in that order.
		821
		822	If an error occurs, the partial destination file will be unlinked, if
		823	possible, except when setting atime, mtime, access mode and uid/gid, where
		824	errors are being ignored.
		825
		826	=cut
		827
		828	sub aio_copy($$;$) {
		829	my ($src, $dst, $cb) = @_;
		830
		831	my $pri = aioreq_pri;
		832	my $grp = aio_group $cb;
		833
		834	aioreq_pri $pri;
		835	add $grp aio_open $src, O_RDONLY, 0, sub {
		836	if (my $src_fh = $_[0]) {
		837	my @stat = stat $src_fh; # hmm, might block over nfs?
		838
		839	aioreq_pri $pri;
		840	add $grp aio_open $dst, O_CREAT | O_WRONLY | O_TRUNC, 0200, sub {
		841	if (my $dst_fh = $_[0]) {
		842	aioreq_pri $pri;
		843	add $grp aio_sendfile $dst_fh, $src_fh, 0, $stat[7], sub {
		844	if ($_[0] == $stat[7]) {
		845	$grp->result (0);
		846	close $src_fh;
		847
		848	my $ch = sub {
		849	aioreq_pri $pri;
		850	add $grp aio_chmod $dst_fh, $stat[2] & 07777, sub {
		851	aioreq_pri $pri;
		852	add $grp aio_chown $dst_fh, $stat[4], $stat[5], sub {
		853	aioreq_pri $pri;
		854	add $grp aio_close $dst_fh;
		855	}
		856	};
		857	};
		858
		859	aioreq_pri $pri;
		860	add $grp aio_utime $dst_fh, $stat[8], $stat[9], sub {
		861	if ($_[0] < 0 && $! == ENOSYS) {
		862	aioreq_pri $pri;
		863	add $grp aio_utime $dst, $stat[8], $stat[9], $ch;
		864	} else {
		865	$ch->();
		866	}
		867	};
		868	} else {
		869	$grp->result (-1);
		870	close $src_fh;
		871	close $dst_fh;
		872
		873	aioreq $pri;
		874	add $grp aio_unlink $dst;
		875	}
		876	};
		877	} else {
		878	$grp->result (-1);
		879	}
		880	},
		881
		882	} else {
		883	$grp->result (-1);
		884	}
		885	};
		886
		887	$grp
		888	}
		889
		890	=item aio_move $srcpath, $dstpath, $callback->($status)
		891
		892	Try to move the I<file> (directories not supported as either source or
		893	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		894	a status of C<0> (ok) or C<-1> (error, see C<$!>).
		895
		896	This is a composite request that tries to rename(2) the file first; if
		897	rename fails with C<EXDEV>, it copies the file with C<aio_copy> and, if
		898	that is successful, unlinks the C<$srcpath>.
		899
		900	=cut
		901
		902	sub aio_move($$;$) {
		903	my ($src, $dst, $cb) = @_;
		904
		905	my $pri = aioreq_pri;
		906	my $grp = aio_group $cb;
		907
		908	aioreq_pri $pri;
		909	add $grp aio_rename $src, $dst, sub {
		910	if ($_[0] && $! == EXDEV) {
		911	aioreq_pri $pri;
		912	add $grp aio_copy $src, $dst, sub {
		913	$grp->result ($_[0]);
		914
		915	unless ($_[0]) {
		916	aioreq_pri $pri;
		917	add $grp aio_unlink $src;
		918	}
		919	};
		920	} else {
		921	$grp->result ($_[0]);
		922	}
		923	};
		924
		925	$grp
		926	}
		927
		928	=item aio_scandir $pathname, $maxreq, $callback->($dirs, $nondirs)
		929
		930	Scans a directory (similar to C<aio_readdir>) but additionally tries to
		931	efficiently separate the entries of directory C<$path> into two sets of
		932	names, directories you can recurse into (directories), and ones you cannot
		933	recurse into (everything else, including symlinks to directories).
		934
		935	C<aio_scandir> is a composite request that creates of many sub requests_
		936	C<$maxreq> specifies the maximum number of outstanding aio requests that
		937	this function generates. If it is C<< <= 0 >>, then a suitable default
		938	will be chosen (currently 4).
		939
		940	On error, the callback is called without arguments, otherwise it receives
		941	two array-refs with path-relative entry names.
		942
		943	Example:
		944
		945	aio_scandir $dir, 0, sub {
		946	my ($dirs, $nondirs) = @_;
		947	print "real directories: @$dirs\n";
		948	print "everything else: @$nondirs\n";
		949	};
		950
		951	Implementation notes.
		952
		953	The C<aio_readdir> cannot be avoided, but C<stat()>'ing every entry can.
		954
		955	If readdir returns file type information, then this is used directly to
		956	find directories.
		957
		958	Otherwise, after reading the directory, the modification time, size etc.
		959	of the directory before and after the readdir is checked, and if they
		960	match (and isn't the current time), the link count will be used to decide
		961	how many entries are directories (if >= 2). Otherwise, no knowledge of the
		962	number of subdirectories will be assumed.
		963
		964	Then entries will be sorted into likely directories a non-initial dot
		965	currently) and likely non-directories (see C<aio_readdirx>). Then every
		966	entry plus an appended C</.> will be C<stat>'ed, likely directories first,
		967	in order of their inode numbers. If that succeeds, it assumes that the
		968	entry is a directory or a symlink to directory (which will be checked
		969	separately). This is often faster than stat'ing the entry itself because
		970	filesystems might detect the type of the entry without reading the inode
		971	data (e.g. ext2fs filetype feature), even on systems that cannot return
		972	the filetype information on readdir.
		973
		974	If the known number of directories (link count - 2) has been reached, the
		975	rest of the entries is assumed to be non-directories.
		976
		977	This only works with certainty on POSIX (= UNIX) filesystems, which
		978	fortunately are the vast majority of filesystems around.
		979
		980	It will also likely work on non-POSIX filesystems with reduced efficiency
		981	as those tend to return 0 or 1 as link counts, which disables the
		982	directory counting heuristic.
		983
		984	=cut
		985
		986	sub aio_scandir($$;$) {
		987	my ($path, $maxreq, $cb) = @_;
		988
		989	my $pri = aioreq_pri;
		990
		991	my $grp = aio_group $cb;
		992
		993	$maxreq = 4 if $maxreq <= 0;
		994
		995	# get a wd object
		996	aioreq_pri $pri;
		997	add $grp aio_wd $path, sub {
		998	$_[0]
		999	or return $grp->result ();
		1000
		1001	my $wd = [shift, "."];
		1002
		1003	# stat once
		1004	aioreq_pri $pri;
		1005	add $grp aio_stat $wd, sub {
		1006	return $grp->result () if $_[0];
		1007	my $now = time;
		1008	my $hash1 = join ":", (stat _)[0,1,3,7,9];
		1009
		1010	# read the directory entries
		1011	aioreq_pri $pri;
		1012	add $grp aio_readdirx $wd, READDIR_DIRS_FIRST, sub {
		1013	my $entries = shift
		1014	or return $grp->result ();
		1015
		1016	# stat the dir another time
		1017	aioreq_pri $pri;
		1018	add $grp aio_stat $wd, sub {
		1019	my $hash2 = join ":", (stat _)[0,1,3,7,9];
		1020
		1021	my $ndirs;
		1022
		1023	# take the slow route if anything looks fishy
		1024	if ($hash1 ne $hash2 or (stat _)[9] == $now) {
		1025	$ndirs = -1;
		1026	} else {
		1027	# if nlink == 2, we are finished
		1028	# for non-posix-fs's, we rely on nlink < 2
		1029	$ndirs = (stat _)[3] - 2
		1030	or return $grp->result ([], $entries);
		1031	}
		1032
		1033	my (@dirs, @nondirs);
		1034
		1035	my $statgrp = add $grp aio_group sub {
		1036	$grp->result (\@dirs, \@nondirs);
		1037	};
		1038
		1039	limit $statgrp $maxreq;
		1040	feed $statgrp sub {
		1041	return unless @$entries;
		1042	my $entry = shift @$entries;
		1043
		1044	aioreq_pri $pri;
		1045	$wd->[1] = "$entry/.";
		1046	add $statgrp aio_stat $wd, sub {
		1047	if ($_[0] < 0) {
		1048	push @nondirs, $entry;
		1049	} else {
		1050	# need to check for real directory
		1051	aioreq_pri $pri;
		1052	$wd->[1] = $entry;
		1053	add $statgrp aio_lstat $wd, sub {
		1054	if (-d _) {
		1055	push @dirs, $entry;
		1056
		1057	unless (--$ndirs) {
		1058	push @nondirs, @$entries;
		1059	feed $statgrp;
		1060	}
		1061	} else {
		1062	push @nondirs, $entry;
		1063	}
		1064	}
		1065	}
		1066	};
		1067	};
		1068	};
		1069	};
		1070	};
		1071	};
		1072
		1073	$grp
		1074	}
		1075
		1076	=item aio_rmtree $pathname, $callback->($status)
		1077
		1078	Delete a directory tree starting (and including) C<$path>, return the
		1079	status of the final C<rmdir> only. This is a composite request that
		1080	uses C<aio_scandir> to recurse into and rmdir directories, and unlink
		1081	everything else.
		1082
		1083	=cut
		1084
		1085	sub aio_rmtree;
		1086	sub aio_rmtree($;$) {
		1087	my ($path, $cb) = @_;
		1088
		1089	my $pri = aioreq_pri;
		1090	my $grp = aio_group $cb;
		1091
		1092	aioreq_pri $pri;
		1093	add $grp aio_scandir $path, 0, sub {
		1094	my ($dirs, $nondirs) = @_;
		1095
		1096	my $dirgrp = aio_group sub {
		1097	add $grp aio_rmdir $path, sub {
		1098	$grp->result ($_[0]);
		1099	};
		1100	};
		1101
		1102	(aioreq_pri $pri), add $dirgrp aio_rmtree "$path/$_" for @$dirs;
		1103	(aioreq_pri $pri), add $dirgrp aio_unlink "$path/$_" for @$nondirs;
		1104
		1105	add $grp $dirgrp;
		1106	};
		1107
		1108	$grp
		1109	}
		1110
		1111	=item aio_sync $callback->($status)
		1112
		1113	Asynchronously call sync and call the callback when finished.
		1114
215	=item aio_fsync $fh, $callback	1115	=item aio_fsync $fh, $callback->($status)
216		1116
217	Asynchronously call fsync on the given filehandle and call the callback	1117	Asynchronously call fsync on the given filehandle and call the callback
218	with the fsync result code.	1118	with the fsync result code.
219		1119
220	=item aio_fdatasync $fh, $callback	1120	=item aio_fdatasync $fh, $callback->($status)
221		1121
222	Asynchronously call fdatasync on the given filehandle and call the	1122	Asynchronously call fdatasync on the given filehandle and call the
223	callback with the fdatasync result code.	1123	callback with the fdatasync result code.
224		1124
		1125	If this call isn't available because your OS lacks it or it couldn't be
		1126	detected, it will be emulated by calling C<fsync> instead.
		1127
		1128	=item aio_syncfs $fh, $callback->($status)
		1129
		1130	Asynchronously call the syncfs syscall to sync the filesystem associated
		1131	to the given filehandle and call the callback with the syncfs result
		1132	code. If syncfs is not available, calls sync(), but returns C<-1> and sets
		1133	errno to C<ENOSYS> nevertheless.
		1134
		1135	=item aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
		1136
		1137	Sync the data portion of the file specified by C<$offset> and C<$length>
		1138	to disk (but NOT the metadata), by calling the Linux-specific
		1139	sync_file_range call. If sync_file_range is not available or it returns
		1140	ENOSYS, then fdatasync or fsync is being substituted.
		1141
		1142	C<$flags> can be a combination of C<IO::AIO::SYNC_FILE_RANGE_WAIT_BEFORE>,
		1143	C<IO::AIO::SYNC_FILE_RANGE_WRITE> and
		1144	C<IO::AIO::SYNC_FILE_RANGE_WAIT_AFTER>: refer to the sync_file_range
		1145	manpage for details.
		1146
		1147	=item aio_pathsync $pathname, $callback->($status)
		1148
		1149	This request tries to open, fsync and close the given path. This is a
		1150	composite request intended to sync directories after directory operations
		1151	(E.g. rename). This might not work on all operating systems or have any
		1152	specific effect, but usually it makes sure that directory changes get
		1153	written to disc. It works for anything that can be opened for read-only,
		1154	not just directories.
		1155
		1156	Future versions of this function might fall back to other methods when
		1157	C<fsync> on the directory fails (such as calling C<sync>).
		1158
		1159	Passes C<0> when everything went ok, and C<-1> on error.
		1160
225	=cut	1161	=cut
226		1162
227	# support function to convert a fd into a perl filehandle	1163	sub aio_pathsync($;$) {
228	sub _fd2fh {	1164	my ($path, $cb) = @_;
229	return undef if $_[0] < 0;
230		1165
231	# try to be perl5.6-compatible	1166	my $pri = aioreq_pri;
232	local *AIO_FH;	1167	my $grp = aio_group $cb;
233	open AIO_FH, "+<&=$_[0]"
234	or return undef;
235		1168
236	*AIO_FH	1169	aioreq_pri $pri;
		1170	add $grp aio_open $path, O_RDONLY, 0, sub {
		1171	my ($fh) = @_;
		1172	if ($fh) {
		1173	aioreq_pri $pri;
		1174	add $grp aio_fsync $fh, sub {
		1175	$grp->result ($_[0]);
		1176
		1177	aioreq_pri $pri;
		1178	add $grp aio_close $fh;
		1179	};
		1180	} else {
		1181	$grp->result (-1);
		1182	}
		1183	};
		1184
		1185	$grp
237	}	1186	}
238		1187
		1188	=item aio_msync $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		1189
		1190	This is a rather advanced IO::AIO call, which only works on mmap(2)ed
		1191	scalars (see the C<IO::AIO::mmap> function, although it also works on data
		1192	scalars managed by the L<Sys::Mmap> or L<Mmap> modules, note that the
		1193	scalar must only be modified in-place while an aio operation is pending on
		1194	it).
		1195
		1196	It calls the C<msync> function of your OS, if available, with the memory
		1197	area starting at C<$offset> in the string and ending C<$length> bytes
		1198	later. If C<$length> is negative, counts from the end, and if C<$length>
		1199	is C<undef>, then it goes till the end of the string. The flags can be
		1200	a combination of C<IO::AIO::MS_ASYNC>, C<IO::AIO::MS_INVALIDATE> and
		1201	C<IO::AIO::MS_SYNC>.
		1202
		1203	=item aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
		1204
		1205	This is a rather advanced IO::AIO call, which works best on mmap(2)ed
		1206	scalars.
		1207
		1208	It touches (reads or writes) all memory pages in the specified
		1209	range inside the scalar. All caveats and parameters are the same
		1210	as for C<aio_msync>, above, except for flags, which must be either
		1211	C<0> (which reads all pages and ensures they are instantiated) or
		1212	C<IO::AIO::MT_MODIFY>, which modifies the memory page s(by reading and
		1213	writing an octet from it, which dirties the page).
		1214
		1215	=item aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status)
		1216
		1217	This is a rather advanced IO::AIO call, which works best on mmap(2)ed
		1218	scalars.
		1219
		1220	It reads in all the pages of the underlying storage into memory (if any)
		1221	and locks them, so they are not getting swapped/paged out or removed.
		1222
		1223	If C<$length> is undefined, then the scalar will be locked till the end.
		1224
		1225	On systems that do not implement C<mlock>, this function returns C<-1>
		1226	and sets errno to C<ENOSYS>.
		1227
		1228	Note that the corresponding C<munlock> is synchronous and is
		1229	documented under L<MISCELLANEOUS FUNCTIONS>.
		1230
		1231	Example: open a file, mmap and mlock it - both will be undone when
		1232	C<$data> gets destroyed.
		1233
		1234	open my $fh, "<", $path or die "$path: $!";
		1235	my $data;
		1236	IO::AIO::mmap $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh;
		1237	aio_mlock $data; # mlock in background
		1238
		1239	=item aio_mlockall $flags, $callback->($status)
		1240
		1241	Calls the C<mlockall> function with the given C<$flags> (a combination of
		1242	C<IO::AIO::MCL_CURRENT> and C<IO::AIO::MCL_FUTURE>).
		1243
		1244	On systems that do not implement C<mlockall>, this function returns C<-1>
		1245	and sets errno to C<ENOSYS>.
		1246
		1247	Note that the corresponding C<munlockall> is synchronous and is
		1248	documented under L<MISCELLANEOUS FUNCTIONS>.
		1249
		1250	Example: asynchronously lock all current and future pages into memory.
		1251
		1252	aio_mlockall IO::AIO::MCL_FUTURE;
		1253
		1254	=item aio_group $callback->(...)
		1255
		1256	This is a very special aio request: Instead of doing something, it is a
		1257	container for other aio requests, which is useful if you want to bundle
		1258	many requests into a single, composite, request with a definite callback
		1259	and the ability to cancel the whole request with its subrequests.
		1260
		1261	Returns an object of class L<IO::AIO::GRP>. See its documentation below
		1262	for more info.
		1263
		1264	Example:
		1265
		1266	my $grp = aio_group sub {
		1267	print "all stats done\n";
		1268	};
		1269
		1270	add $grp
		1271	(aio_stat ...),
		1272	(aio_stat ...),
		1273	...;
		1274
		1275	=item aio_nop $callback->()
		1276
		1277	This is a special request - it does nothing in itself and is only used for
		1278	side effects, such as when you want to add a dummy request to a group so
		1279	that finishing the requests in the group depends on executing the given
		1280	code.
		1281
		1282	While this request does nothing, it still goes through the execution
		1283	phase and still requires a worker thread. Thus, the callback will not
		1284	be executed immediately but only after other requests in the queue have
		1285	entered their execution phase. This can be used to measure request
		1286	latency.
		1287
		1288	=item IO::AIO::aio_busy $fractional_seconds, $callback->() *NOT EXPORTED*
		1289
		1290	Mainly used for debugging and benchmarking, this aio request puts one of
		1291	the request workers to sleep for the given time.
		1292
		1293	While it is theoretically handy to have simple I/O scheduling requests
		1294	like sleep and file handle readable/writable, the overhead this creates is
		1295	immense (it blocks a thread for a long time) so do not use this function
		1296	except to put your application under artificial I/O pressure.
		1297
		1298	=back
		1299
		1300
		1301	=head2 IO::AIO::WD - multiple working directories
		1302
		1303	Your process only has one current working directory, which is used by all
		1304	threads. This makes it hard to use relative paths (some other component
		1305	could call C<chdir> at any time, and it is hard to control when the path
		1306	will be used by IO::AIO).
		1307
		1308	One solution for this is to always use absolute paths. This usually works,
		1309	but can be quite slow (the kernel has to walk the whole path on every
		1310	access), and can also be a hassle to implement.
		1311
		1312	Newer POSIX systems have a number of functions (openat, fdopendir,
		1313	futimensat and so on) that make it possible to specify working directories
		1314	per operation.
		1315
		1316	For portability, and because the clowns who "designed", or shall I write,
		1317	perpetrated this new interface were obviously half-drunk, this abstraction
		1318	cannot be perfect, though.
		1319
		1320	IO::AIO allows you to convert directory paths into a so-called IO::AIO::WD
		1321	object. This object stores the canonicalised, absolute version of the
		1322	path, and on systems that allow it, also a directory file descriptor.
		1323
		1324	Everywhere where a pathname is accepted by IO::AIO (e.g. in C<aio_stat>
		1325	or C<aio_unlink>), one can specify an array reference with an IO::AIO::WD
		1326	object and a pathname instead (or the IO::AIO::WD object alone, which
		1327	gets interpreted as C<[$wd, "."]>). If the pathname is absolute, the
		1328	IO::AIO::WD object is ignored, otherwise the pathname is resolved relative
		1329	to that IO::AIO::WD object.
		1330
		1331	For example, to get a wd object for F</etc> and then stat F<passwd>
		1332	inside, you would write:
		1333
		1334	aio_wd "/etc", sub {
		1335	my $etcdir = shift;
		1336
		1337	# although $etcdir can be undef on error, there is generally no reason
		1338	# to check for errors here, as aio_stat will fail with ENOENT
		1339	# when $etcdir is undef.
		1340
		1341	aio_stat [$etcdir, "passwd"], sub {
		1342	# yay
		1343	};
		1344	};
		1345
		1346	That C<aio_wd> is a request and not a normal function shows that creating
		1347	an IO::AIO::WD object is itself a potentially blocking operation, which is
		1348	why it is done asynchronously.
		1349
		1350	To stat the directory obtained with C<aio_wd> above, one could write
		1351	either of the following three request calls:
		1352
		1353	aio_lstat "/etc" , sub { ... # pathname as normal string
		1354	aio_lstat [$wd, "."], sub { ... # "." relative to $wd (i.e. $wd itself)
		1355	aio_lstat $wd , sub { ... # shorthand for the previous
		1356
		1357	As with normal pathnames, IO::AIO keeps a copy of the working directory
		1358	object and the pathname string, so you could write the following without
		1359	causing any issues due to C<$path> getting reused:
		1360
		1361	my $path = [$wd, undef];
		1362
		1363	for my $name (qw(abc def ghi)) {
		1364	$path->[1] = $name;
		1365	aio_stat $path, sub {
		1366	# ...
		1367	};
		1368	}
		1369
		1370	There are some caveats: when directories get renamed (or deleted), the
		1371	pathname string doesn't change, so will point to the new directory (or
		1372	nowhere at all), while the directory fd, if available on the system,
		1373	will still point to the original directory. Most functions accepting a
		1374	pathname will use the directory fd on newer systems, and the string on
		1375	older systems. Some functions (such as realpath) will always rely on the
		1376	string form of the pathname.
		1377
		1378	So this fucntionality is mainly useful to get some protection against
		1379	C<chdir>, to easily get an absolute path out of a relative path for future
		1380	reference, and to speed up doing many operations in the same directory
		1381	(e.g. when stat'ing all files in a directory).
		1382
		1383	The following functions implement this working directory abstraction:
		1384
		1385	=over 4
		1386
		1387	=item aio_wd $pathname, $callback->($wd)
		1388
		1389	Asynchonously canonicalise the given pathname and convert it to an
		1390	IO::AIO::WD object representing it. If possible and supported on the
		1391	system, also open a directory fd to speed up pathname resolution relative
		1392	to this working directory.
		1393
		1394	If something goes wrong, then C<undef> is passwd to the callback instead
		1395	of a working directory object and C<$!> is set appropriately. Since
		1396	passing C<undef> as working directory component of a pathname fails the
		1397	request with C<ENOENT>, there is often no need for error checking in the
		1398	C<aio_wd> callback, as future requests using the value will fail in the
		1399	expected way.
		1400
		1401	If this call isn't available because your OS lacks it or it couldn't be
		1402	detected, it will be emulated by calling C<fsync> instead.
		1403
		1404	=item IO::AIO::CWD
		1405
		1406	This is a compiletime constant (object) that represents the process
		1407	current working directory.
		1408
		1409	Specifying this object as working directory object for a pathname is as
		1410	if the pathname would be specified directly, without a directory object,
		1411	e.g., these calls are functionally identical:
		1412
		1413	aio_stat "somefile", sub { ... };
		1414	aio_stat [IO::AIO::CWD, "somefile"], sub { ... };
		1415
		1416	=back
		1417
		1418
		1419	=head2 IO::AIO::REQ CLASS
		1420
		1421	All non-aggregate C<aio_*> functions return an object of this class when
		1422	called in non-void context.
		1423
		1424	=over 4
		1425
		1426	=item cancel $req
		1427
		1428	Cancels the request, if possible. Has the effect of skipping execution
		1429	when entering the B<execute> state and skipping calling the callback when
		1430	entering the the B<result> state, but will leave the request otherwise
		1431	untouched (with the exception of readdir). That means that requests that
		1432	currently execute will not be stopped and resources held by the request
		1433	will not be freed prematurely.
		1434
		1435	=item cb $req $callback->(...)
		1436
		1437	Replace (or simply set) the callback registered to the request.
		1438
		1439	=back
		1440
		1441	=head2 IO::AIO::GRP CLASS
		1442
		1443	This class is a subclass of L<IO::AIO::REQ>, so all its methods apply to
		1444	objects of this class, too.
		1445
		1446	A IO::AIO::GRP object is a special request that can contain multiple other
		1447	aio requests.
		1448
		1449	You create one by calling the C<aio_group> constructing function with a
		1450	callback that will be called when all contained requests have entered the
		1451	C<done> state:
		1452
		1453	my $grp = aio_group sub {
		1454	print "all requests are done\n";
		1455	};
		1456
		1457	You add requests by calling the C<add> method with one or more
		1458	C<IO::AIO::REQ> objects:
		1459
		1460	$grp->add (aio_unlink "...");
		1461
		1462	add $grp aio_stat "...", sub {
		1463	$_[0] or return $grp->result ("error");
		1464
		1465	# add another request dynamically, if first succeeded
		1466	add $grp aio_open "...", sub {
		1467	$grp->result ("ok");
		1468	};
		1469	};
		1470
		1471	This makes it very easy to create composite requests (see the source of
		1472	C<aio_move> for an application) that work and feel like simple requests.
		1473
		1474	=over 4
		1475
		1476	=item * The IO::AIO::GRP objects will be cleaned up during calls to
		1477	C<IO::AIO::poll_cb>, just like any other request.
		1478
		1479	=item * They can be canceled like any other request. Canceling will cancel not
		1480	only the request itself, but also all requests it contains.
		1481
		1482	=item * They can also can also be added to other IO::AIO::GRP objects.
		1483
		1484	=item * You must not add requests to a group from within the group callback (or
		1485	any later time).
		1486
		1487	=back
		1488
		1489	Their lifetime, simplified, looks like this: when they are empty, they
		1490	will finish very quickly. If they contain only requests that are in the
		1491	C<done> state, they will also finish. Otherwise they will continue to
		1492	exist.
		1493
		1494	That means after creating a group you have some time to add requests
		1495	(precisely before the callback has been invoked, which is only done within
		1496	the C<poll_cb>). And in the callbacks of those requests, you can add
		1497	further requests to the group. And only when all those requests have
		1498	finished will the the group itself finish.
		1499
		1500	=over 4
		1501
		1502	=item add $grp ...
		1503
		1504	=item $grp->add (...)
		1505
		1506	Add one or more requests to the group. Any type of L<IO::AIO::REQ> can
		1507	be added, including other groups, as long as you do not create circular
		1508	dependencies.
		1509
		1510	Returns all its arguments.
		1511
		1512	=item $grp->cancel_subs
		1513
		1514	Cancel all subrequests and clears any feeder, but not the group request
		1515	itself. Useful when you queued a lot of events but got a result early.
		1516
		1517	The group request will finish normally (you cannot add requests to the
		1518	group).
		1519
		1520	=item $grp->result (...)
		1521
		1522	Set the result value(s) that will be passed to the group callback when all
		1523	subrequests have finished and set the groups errno to the current value
		1524	of errno (just like calling C<errno> without an error number). By default,
		1525	no argument will be passed and errno is zero.
		1526
		1527	=item $grp->errno ([$errno])
		1528
		1529	Sets the group errno value to C<$errno>, or the current value of errno
		1530	when the argument is missing.
		1531
		1532	Every aio request has an associated errno value that is restored when
		1533	the callback is invoked. This method lets you change this value from its
		1534	default (0).
		1535
		1536	Calling C<result> will also set errno, so make sure you either set C<$!>
		1537	before the call to C<result>, or call c<errno> after it.
		1538
		1539	=item feed $grp $callback->($grp)
		1540
		1541	Sets a feeder/generator on this group: every group can have an attached
		1542	generator that generates requests if idle. The idea behind this is that,
		1543	although you could just queue as many requests as you want in a group,
		1544	this might starve other requests for a potentially long time. For example,
		1545	C<aio_scandir> might generate hundreds of thousands of C<aio_stat>
		1546	requests, delaying any later requests for a long time.
		1547
		1548	To avoid this, and allow incremental generation of requests, you can
		1549	instead a group and set a feeder on it that generates those requests. The
		1550	feed callback will be called whenever there are few enough (see C<limit>,
		1551	below) requests active in the group itself and is expected to queue more
		1552	requests.
		1553
		1554	The feed callback can queue as many requests as it likes (i.e. C<add> does
		1555	not impose any limits).
		1556
		1557	If the feed does not queue more requests when called, it will be
		1558	automatically removed from the group.
		1559
		1560	If the feed limit is C<0> when this method is called, it will be set to
		1561	C<2> automatically.
		1562
		1563	Example:
		1564
		1565	# stat all files in @files, but only ever use four aio requests concurrently:
		1566
		1567	my $grp = aio_group sub { print "finished\n" };
		1568	limit $grp 4;
		1569	feed $grp sub {
		1570	my $file = pop @files
		1571	or return;
		1572
		1573	add $grp aio_stat $file, sub { ... };
		1574	};
		1575
		1576	=item limit $grp $num
		1577
		1578	Sets the feeder limit for the group: The feeder will be called whenever
		1579	the group contains less than this many requests.
		1580
		1581	Setting the limit to C<0> will pause the feeding process.
		1582
		1583	The default value for the limit is C<0>, but note that setting a feeder
		1584	automatically bumps it up to C<2>.
		1585
		1586	=back
		1587
		1588	=head2 SUPPORT FUNCTIONS
		1589
		1590	=head3 EVENT PROCESSING AND EVENT LOOP INTEGRATION
		1591
		1592	=over 4
		1593
		1594	=item $fileno = IO::AIO::poll_fileno
		1595
		1596	Return the I<request result pipe file descriptor>. This filehandle must be
		1597	polled for reading by some mechanism outside this module (e.g. EV, Glib,
		1598	select and so on, see below or the SYNOPSIS). If the pipe becomes readable
		1599	you have to call C<poll_cb> to check the results.
		1600
		1601	See C<poll_cb> for an example.
		1602
		1603	=item IO::AIO::poll_cb
		1604
		1605	Process some outstanding events on the result pipe. You have to call
		1606	this regularly. Returns C<0> if all events could be processed (or there
		1607	were no events to process), or C<-1> if it returned earlier for whatever
		1608	reason. Returns immediately when no events are outstanding. The amount of
		1609	events processed depends on the settings of C<IO::AIO::max_poll_req> and
		1610	C<IO::AIO::max_poll_time>.
		1611
		1612	If not all requests were processed for whatever reason, the filehandle
		1613	will still be ready when C<poll_cb> returns, so normally you don't have to
		1614	do anything special to have it called later.
		1615
		1616	Apart from calling C<IO::AIO::poll_cb> when the event filehandle becomes
		1617	ready, it can be beneficial to call this function from loops which submit
		1618	a lot of requests, to make sure the results get processed when they become
		1619	available and not just when the loop is finished and the event loop takes
		1620	over again. This function returns very fast when there are no outstanding
		1621	requests.
		1622
		1623	Example: Install an Event watcher that automatically calls
		1624	IO::AIO::poll_cb with high priority (more examples can be found in the
		1625	SYNOPSIS section, at the top of this document):
		1626
		1627	Event->io (fd => IO::AIO::poll_fileno,
		1628	poll => 'r', async => 1,
		1629	cb => \&IO::AIO::poll_cb);
		1630
		1631	=item IO::AIO::poll_wait
		1632
		1633	If there are any outstanding requests and none of them in the result
		1634	phase, wait till the result filehandle becomes ready for reading (simply
		1635	does a C<select> on the filehandle. This is useful if you want to
		1636	synchronously wait for some requests to finish).
		1637
		1638	See C<nreqs> for an example.
		1639
		1640	=item IO::AIO::poll
		1641
		1642	Waits until some requests have been handled.
		1643
		1644	Returns the number of requests processed, but is otherwise strictly
		1645	equivalent to:
		1646
		1647	IO::AIO::poll_wait, IO::AIO::poll_cb
		1648
		1649	=item IO::AIO::flush
		1650
		1651	Wait till all outstanding AIO requests have been handled.
		1652
		1653	Strictly equivalent to:
		1654
		1655	IO::AIO::poll_wait, IO::AIO::poll_cb
		1656	while IO::AIO::nreqs;
		1657
		1658	=item IO::AIO::max_poll_reqs $nreqs
		1659
		1660	=item IO::AIO::max_poll_time $seconds
		1661
		1662	These set the maximum number of requests (default C<0>, meaning infinity)
		1663	that are being processed by C<IO::AIO::poll_cb> in one call, respectively
		1664	the maximum amount of time (default C<0>, meaning infinity) spent in
		1665	C<IO::AIO::poll_cb> to process requests (more correctly the mininum amount
		1666	of time C<poll_cb> is allowed to use).
		1667
		1668	Setting C<max_poll_time> to a non-zero value creates an overhead of one
		1669	syscall per request processed, which is not normally a problem unless your
		1670	callbacks are really really fast or your OS is really really slow (I am
		1671	not mentioning Solaris here). Using C<max_poll_reqs> incurs no overhead.
		1672
		1673	Setting these is useful if you want to ensure some level of
		1674	interactiveness when perl is not fast enough to process all requests in
		1675	time.
		1676
		1677	For interactive programs, values such as C<0.01> to C<0.1> should be fine.
		1678
		1679	Example: Install an Event watcher that automatically calls
		1680	IO::AIO::poll_cb with low priority, to ensure that other parts of the
		1681	program get the CPU sometimes even under high AIO load.
		1682
		1683	# try not to spend much more than 0.1s in poll_cb
		1684	IO::AIO::max_poll_time 0.1;
		1685
		1686	# use a low priority so other tasks have priority
		1687	Event->io (fd => IO::AIO::poll_fileno,
		1688	poll => 'r', nice => 1,
		1689	cb => &IO::AIO::poll_cb);
		1690
		1691	=back
		1692
		1693	=head3 CONTROLLING THE NUMBER OF THREADS
		1694
		1695	=over
		1696
		1697	=item IO::AIO::min_parallel $nthreads
		1698
		1699	Set the minimum number of AIO threads to C<$nthreads>. The current
		1700	default is C<8>, which means eight asynchronous operations can execute
		1701	concurrently at any one time (the number of outstanding requests,
		1702	however, is unlimited).
		1703
		1704	IO::AIO starts threads only on demand, when an AIO request is queued and
		1705	no free thread exists. Please note that queueing up a hundred requests can
		1706	create demand for a hundred threads, even if it turns out that everything
		1707	is in the cache and could have been processed faster by a single thread.
		1708
		1709	It is recommended to keep the number of threads relatively low, as some
		1710	Linux kernel versions will scale negatively with the number of threads
		1711	(higher parallelity => MUCH higher latency). With current Linux 2.6
		1712	versions, 4-32 threads should be fine.
		1713
		1714	Under most circumstances you don't need to call this function, as the
		1715	module selects a default that is suitable for low to moderate load.
		1716
		1717	=item IO::AIO::max_parallel $nthreads
		1718
		1719	Sets the maximum number of AIO threads to C<$nthreads>. If more than the
		1720	specified number of threads are currently running, this function kills
		1721	them. This function blocks until the limit is reached.
		1722
		1723	While C<$nthreads> are zero, aio requests get queued but not executed
		1724	until the number of threads has been increased again.
		1725
		1726	This module automatically runs C<max_parallel 0> at program end, to ensure
		1727	that all threads are killed and that there are no outstanding requests.
		1728
		1729	Under normal circumstances you don't need to call this function.
		1730
		1731	=item IO::AIO::max_idle $nthreads
		1732
		1733	Limit the number of threads (default: 4) that are allowed to idle
		1734	(i.e., threads that did not get a request to process within the idle
		1735	timeout (default: 10 seconds). That means if a thread becomes idle while
		1736	C<$nthreads> other threads are also idle, it will free its resources and
		1737	exit.
		1738
		1739	This is useful when you allow a large number of threads (e.g. 100 or 1000)
		1740	to allow for extremely high load situations, but want to free resources
		1741	under normal circumstances (1000 threads can easily consume 30MB of RAM).
		1742
		1743	The default is probably ok in most situations, especially if thread
		1744	creation is fast. If thread creation is very slow on your system you might
		1745	want to use larger values.
		1746
		1747	=item IO::AIO::idle_timeout $seconds
		1748
		1749	Sets the minimum idle timeout (default 10) after which worker threads are
		1750	allowed to exit. SEe C<IO::AIO::max_idle>.
		1751
		1752	=item IO::AIO::max_outstanding $maxreqs
		1753
		1754	Sets the maximum number of outstanding requests to C<$nreqs>. If
		1755	you do queue up more than this number of requests, the next call to
		1756	C<IO::AIO::poll_cb> (and other functions calling C<poll_cb>, such as
		1757	C<IO::AIO::flush> or C<IO::AIO::poll>) will block until the limit is no
		1758	longer exceeded.
		1759
		1760	In other words, this setting does not enforce a queue limit, but can be
		1761	used to make poll functions block if the limit is exceeded.
		1762
		1763	This is a very bad function to use in interactive programs because it
		1764	blocks, and a bad way to reduce concurrency because it is inexact: Better
		1765	use an C<aio_group> together with a feed callback.
		1766
		1767	It's main use is in scripts without an event loop - when you want to stat
		1768	a lot of files, you can write somehting like this:
		1769
		1770	IO::AIO::max_outstanding 32;
		1771
		1772	for my $path (...) {
		1773	aio_stat $path , ...;
		1774	IO::AIO::poll_cb;
		1775	}
		1776
		1777	IO::AIO::flush;
		1778
		1779	The call to C<poll_cb> inside the loop will normally return instantly, but
		1780	as soon as more thna C<32> reqeusts are in-flight, it will block until
		1781	some requests have been handled. This keeps the loop from pushing a large
		1782	number of C<aio_stat> requests onto the queue.
		1783
		1784	The default value for C<max_outstanding> is very large, so there is no
		1785	practical limit on the number of outstanding requests.
		1786
		1787	=back
		1788
		1789	=head3 STATISTICAL INFORMATION
		1790
		1791	=over
		1792
		1793	=item IO::AIO::nreqs
		1794
		1795	Returns the number of requests currently in the ready, execute or pending
		1796	states (i.e. for which their callback has not been invoked yet).
		1797
		1798	Example: wait till there are no outstanding requests anymore:
		1799
		1800	IO::AIO::poll_wait, IO::AIO::poll_cb
		1801	while IO::AIO::nreqs;
		1802
		1803	=item IO::AIO::nready
		1804
		1805	Returns the number of requests currently in the ready state (not yet
		1806	executed).
		1807
		1808	=item IO::AIO::npending
		1809
		1810	Returns the number of requests currently in the pending state (executed,
		1811	but not yet processed by poll_cb).
		1812
		1813	=back
		1814
		1815	=head3 MISCELLANEOUS FUNCTIONS
		1816
		1817	IO::AIO implements some functions that might be useful, but are not
		1818	asynchronous.
		1819
		1820	=over 4
		1821
		1822	=item IO::AIO::sendfile $ofh, $ifh, $offset, $count
		1823
		1824	Calls the C<eio_sendfile_sync> function, which is like C<aio_sendfile>,
		1825	but is blocking (this makes most sense if you know the input data is
		1826	likely cached already and the output filehandle is set to non-blocking
		1827	operations).
		1828
		1829	Returns the number of bytes copied, or C<-1> on error.
		1830
		1831	=item IO::AIO::fadvise $fh, $offset, $len, $advice
		1832
		1833	Simply calls the C<posix_fadvise> function (see its
		1834	manpage for details). The following advice constants are
		1835	available: C<IO::AIO::FADV_NORMAL>, C<IO::AIO::FADV_SEQUENTIAL>,
		1836	C<IO::AIO::FADV_RANDOM>, C<IO::AIO::FADV_NOREUSE>,
		1837	C<IO::AIO::FADV_WILLNEED>, C<IO::AIO::FADV_DONTNEED>.
		1838
		1839	On systems that do not implement C<posix_fadvise>, this function returns
		1840	ENOSYS, otherwise the return value of C<posix_fadvise>.
		1841
		1842	=item IO::AIO::madvise $scalar, $offset, $len, $advice
		1843
		1844	Simply calls the C<posix_madvise> function (see its
		1845	manpage for details). The following advice constants are
		1846	available: C<IO::AIO::MADV_NORMAL>, C<IO::AIO::MADV_SEQUENTIAL>,
		1847	C<IO::AIO::MADV_RANDOM>, C<IO::AIO::MADV_WILLNEED>, C<IO::AIO::MADV_DONTNEED>.
		1848
		1849	On systems that do not implement C<posix_madvise>, this function returns
		1850	ENOSYS, otherwise the return value of C<posix_madvise>.
		1851
		1852	=item IO::AIO::mprotect $scalar, $offset, $len, $protect
		1853
		1854	Simply calls the C<mprotect> function on the preferably AIO::mmap'ed
		1855	$scalar (see its manpage for details). The following protect
		1856	constants are available: C<IO::AIO::PROT_NONE>, C<IO::AIO::PROT_READ>,
		1857	C<IO::AIO::PROT_WRITE>, C<IO::AIO::PROT_EXEC>.
		1858
		1859	On systems that do not implement C<mprotect>, this function returns
		1860	ENOSYS, otherwise the return value of C<mprotect>.
		1861
		1862	=item IO::AIO::mmap $scalar, $length, $prot, $flags, $fh[, $offset]
		1863
		1864	Memory-maps a file (or anonymous memory range) and attaches it to the
		1865	given C<$scalar>, which will act like a string scalar.
		1866
		1867	The only operations allowed on the scalar are C<substr>/C<vec> that don't
		1868	change the string length, and most read-only operations such as copying it
		1869	or searching it with regexes and so on.
		1870
		1871	Anything else is unsafe and will, at best, result in memory leaks.
		1872
		1873	The memory map associated with the C<$scalar> is automatically removed
		1874	when the C<$scalar> is destroyed, or when the C<IO::AIO::mmap> or
		1875	C<IO::AIO::munmap> functions are called.
		1876
		1877	This calls the C<mmap>(2) function internally. See your system's manual
		1878	page for details on the C<$length>, C<$prot> and C<$flags> parameters.
		1879
		1880	The C<$length> must be larger than zero and smaller than the actual
		1881	filesize.
		1882
		1883	C<$prot> is a combination of C<IO::AIO::PROT_NONE>, C<IO::AIO::PROT_EXEC>,
		1884	C<IO::AIO::PROT_READ> and/or C<IO::AIO::PROT_WRITE>,
		1885
		1886	C<$flags> can be a combination of C<IO::AIO::MAP_SHARED> or
		1887	C<IO::AIO::MAP_PRIVATE>, or a number of system-specific flags (when
		1888	not available, the are defined as 0): C<IO::AIO::MAP_ANONYMOUS>
		1889	(which is set to C<MAP_ANON> if your system only provides this
		1890	constant), C<IO::AIO::MAP_HUGETLB>, C<IO::AIO::MAP_LOCKED>,
		1891	C<IO::AIO::MAP_NORESERVE>, C<IO::AIO::MAP_POPULATE> or
		1892	C<IO::AIO::MAP_NONBLOCK>
		1893
		1894	If C<$fh> is C<undef>, then a file descriptor of C<-1> is passed.
		1895
		1896	C<$offset> is the offset from the start of the file - it generally must be
		1897	a multiple of C<IO::AIO::PAGESIZE> and defaults to C<0>.
		1898
		1899	Example:
		1900
		1901	use Digest::MD5;
		1902	use IO::AIO;
		1903
		1904	open my $fh, "<verybigfile"
		1905	or die "$!";
		1906
		1907	IO::AIO::mmap my $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh
		1908	or die "verybigfile: $!";
		1909
		1910	my $fast_md5 = md5 $data;
		1911
		1912	=item IO::AIO::munmap $scalar
		1913
		1914	Removes a previous mmap and undefines the C<$scalar>.
		1915
		1916	=item IO::AIO::munlock $scalar, $offset = 0, $length = undef
		1917
		1918	Calls the C<munlock> function, undoing the effects of a previous
		1919	C<aio_mlock> call (see its description for details).
		1920
		1921	=item IO::AIO::munlockall
		1922
		1923	Calls the C<munlockall> function.
		1924
		1925	On systems that do not implement C<munlockall>, this function returns
		1926	ENOSYS, otherwise the return value of C<munlockall>.
		1927
		1928	=back
		1929
		1930	=cut
		1931
239	min_parallel 4;	1932	min_parallel 8;
240		1933
241	END {	1934	END { flush }
242	max_parallel 0;
243	}
244		1935
245	1;	1936	1;
246		1937
		1938	=head1 EVENT LOOP INTEGRATION
		1939
		1940	It is recommended to use L<AnyEvent::AIO> to integrate IO::AIO
		1941	automatically into many event loops:
		1942
		1943	# AnyEvent integration (EV, Event, Glib, Tk, POE, urxvt, pureperl...)
		1944	use AnyEvent::AIO;
		1945
		1946	You can also integrate IO::AIO manually into many event loops, here are
		1947	some examples of how to do this:
		1948
		1949	# EV integration
		1950	my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
		1951
		1952	# Event integration
		1953	Event->io (fd => IO::AIO::poll_fileno,
		1954	poll => 'r',
		1955	cb => \&IO::AIO::poll_cb);
		1956
		1957	# Glib/Gtk2 integration
		1958	add_watch Glib::IO IO::AIO::poll_fileno,
		1959	in => sub { IO::AIO::poll_cb; 1 };
		1960
		1961	# Tk integration
		1962	Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",
		1963	readable => \&IO::AIO::poll_cb);
		1964
		1965	# Danga::Socket integration
		1966	Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
		1967	\&IO::AIO::poll_cb);
		1968
		1969	=head2 FORK BEHAVIOUR
		1970
		1971	Usage of pthreads in a program changes the semantics of fork
		1972	considerably. Specifically, only async-safe functions can be called after
		1973	fork. Perl doesn't know about this, so in general, you cannot call fork
		1974	with defined behaviour in perl if pthreads are involved. IO::AIO uses
		1975	pthreads, so this applies, but many other extensions and (for inexplicable
		1976	reasons) perl itself often is linked against pthreads, so this limitation
		1977	applies to quite a lot of perls.
		1978
		1979	This module no longer tries to fight your OS, or POSIX. That means IO::AIO
		1980	only works in the process that loaded it. Forking is fully supported, but
		1981	using IO::AIO in the child is not.
		1982
		1983	You might get around by not I<using> IO::AIO before (or after)
		1984	forking. You could also try to call the L<IO::AIO::reinit> function in the
		1985	child:
		1986
		1987	=over 4
		1988
		1989	=item IO::AIO::reinit
		1990
		1991	Abandons all current requests and I/O threads and simply reinitialises all
		1992	data structures. This is not an operation supported by any standards, but
		1993	happens to work on GNU/Linux and some newer BSD systems.
		1994
		1995	The only reasonable use for this function is to call it after forking, if
		1996	C<IO::AIO> was used in the parent. Calling it while IO::AIO is active in
		1997	the process will result in undefined behaviour. Calling it at any time
		1998	will also result in any undefined (by POSIX) behaviour.
		1999
247	=back	2000	=back
248		2001
		2002	=head2 MEMORY USAGE
		2003
		2004	Per-request usage:
		2005
		2006	Each aio request uses - depending on your architecture - around 100-200
		2007	bytes of memory. In addition, stat requests need a stat buffer (possibly
		2008	a few hundred bytes), readdir requires a result buffer and so on. Perl
		2009	scalars and other data passed into aio requests will also be locked and
		2010	will consume memory till the request has entered the done state.
		2011
		2012	This is not awfully much, so queuing lots of requests is not usually a
		2013	problem.
		2014
		2015	Per-thread usage:
		2016
		2017	In the execution phase, some aio requests require more memory for
		2018	temporary buffers, and each thread requires a stack and other data
		2019	structures (usually around 16k-128k, depending on the OS).
		2020
249	=head1 BUGS	2021	=head1 KNOWN BUGS
250		2022
251	- could be optimized to use more semaphores instead of filehandles.	2023	Known bugs will be fixed in the next release.
252		2024
253	=head1 SEE ALSO	2025	=head1 SEE ALSO
254		2026
255	L<Coro>, L<Linux::AIO>.	2027	L<AnyEvent::AIO> for easy integration into event loops, L<Coro::AIO> for a
		2028	more natural syntax.
256		2029
257	=head1 AUTHOR	2030	=head1 AUTHOR
258		2031
259	Marc Lehmann <schmorp@schmorp.de>	2032	Marc Lehmann <schmorp@schmorp.de>
260	http://home.schmorp.de/	2033	http://home.schmorp.de/

Diff Legend

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