[ViewVC] Diff of: cvs/IO-AIO/AIO.pm

Comparing IO-AIO/AIO.pm (file contents):
Revision 1.9 by root, Sun Jul 10 22:20:55 2005 UTC vs.
Revision 1.85 by root, Sat Oct 28 01:40:30 2006 UTC

…		…
15		15
16	aio_read $fh, 30000, 1024, $buffer, 0, sub {	16	aio_read $fh, 30000, 1024, $buffer, 0, sub {
17	$_[0] > 0 or die "read error: $!";	17	$_[0] > 0 or die "read error: $!";
18	};	18	};
19		19
20	# Event	20	# version 2+ has request and group objects
		21	use IO::AIO 2;
		22
		23	aioreq_pri 4; # give next request a very high priority
		24	my $req = aio_unlink "/tmp/file", sub { };
		25	$req->cancel; # cancel request if still in queue
		26
		27	my $grp = aio_group sub { print "all stats done\n" };
		28	add $grp aio_stat "..." for ...;
		29
		30	# AnyEvent integration
		31	open my $fh, "<&=" . IO::AIO::poll_fileno or die "$!";
		32	my $w = AnyEvent->io (fh => $fh, poll => 'r', cb => sub { IO::AIO::poll_cb });
		33
		34	# Event integration
21	Event->io (fd => IO::AIO::poll_fileno,	35	Event->io (fd => IO::AIO::poll_fileno,
22	poll => 'r',	36	poll => 'r',
23	cb => \&IO::AIO::poll_cb);	37	cb => \&IO::AIO::poll_cb);
24		38
25	# Glib/Gtk2	39	# Glib/Gtk2 integration
26	add_watch Glib::IO IO::AIO::poll_fileno,	40	add_watch Glib::IO IO::AIO::poll_fileno,
27	\&IO::AIO::poll_cb;	41	in => sub { IO::AIO::poll_cb; 1 };
28		42
29	# Tk	43	# Tk integration
30	Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",	44	Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",
31	readable => \&IO::AIO::poll_cb);	45	readable => \&IO::AIO::poll_cb);
32		46
		47	# Danga::Socket integration
		48	Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
		49	\&IO::AIO::poll_cb);
		50
33	=head1 DESCRIPTION	51	=head1 DESCRIPTION
34		52
35	This module implements asynchronous I/O using whatever means your	53	This module implements asynchronous I/O using whatever means your
36	operating system supports.	54	operating system supports.
37		55
		56	Asynchronous means that operations that can normally block your program
		57	(e.g. reading from disk) will be done asynchronously: the operation
		58	will still block, but you can do something else in the meantime. This
		59	is extremely useful for programs that need to stay interactive even
		60	when doing heavy I/O (GUI programs, high performance network servers
		61	etc.), but can also be used to easily do operations in parallel that are
		62	normally done sequentially, e.g. stat'ing many files, which is much faster
		63	on a RAID volume or over NFS when you do a number of stat operations
		64	concurrently.
		65
		66	While this works on all types of file descriptors (for example sockets),
		67	using these functions on file descriptors that support nonblocking
		68	operation (again, sockets, pipes etc.) is very inefficient. Use an event
		69	loop for that (such as the L<Event\|Event> module): IO::AIO will naturally
		70	fit into such an event loop itself.
		71
38	Currently, a number of threads are started that execute your read/writes	72	In this version, a number of threads are started that execute your
39	and signal their completion. You don't need thread support in your libc or	73	requests and signal their completion. You don't need thread support
40	perl, and the threads created by this module will not be visible to the	74	in perl, and the threads created by this module will not be visible
41	pthreads library. In the future, this module might make use of the native	75	to perl. In the future, this module might make use of the native aio
42	aio functions available on many operating systems. However, they are often	76	functions available on many operating systems. However, they are often
43	not well-supported (Linux doesn't allow them on normal files currently,	77	not well-supported or restricted (GNU/Linux doesn't allow them on normal
44	for example), and they would only support aio_read and aio_write, so the	78	files currently, for example), and they would only support aio_read and
45	remaining functionality would have to be implemented using threads anyway.	79	aio_write, so the remaining functionality would have to be implemented
		80	using threads anyway.
46		81
47	Although the module will work with in the presence of other threads, it is	82	Although the module will work with in the presence of other (Perl-)
48	currently not reentrant, so use appropriate locking yourself.	83	threads, it is currently not reentrant in any way, so use appropriate
		84	locking yourself, always call C<poll_cb> from within the same thread, or
		85	never call C<poll_cb> (or other C<aio_> functions) recursively.
		86
		87	=head1 REQUEST ANATOMY AND LIFETIME
		88
		89	Every C<aio_*> function creates a request. which is a C data structure not
		90	directly visible to Perl.
		91
		92	If called in non-void context, every request function returns a Perl
		93	object representing the request. In void context, nothing is returned,
		94	which saves a bit of memory.
		95
		96	The perl object is a fairly standard ref-to-hash object. The hash contents
		97	are not used by IO::AIO so you are free to store anything you like in it.
		98
		99	During their existance, aio requests travel through the following states,
		100	in order:
		101
		102	=over 4
		103
		104	=item ready
		105
		106	Immediately after a request is created it is put into the ready state,
		107	waiting for a thread to execute it.
		108
		109	=item execute
		110
		111	A thread has accepted the request for processing and is currently
		112	executing it (e.g. blocking in read).
		113
		114	=item pending
		115
		116	The request has been executed and is waiting for result processing.
		117
		118	While request submission and execution is fully asynchronous, result
		119	processing is not and relies on the perl interpreter calling C<poll_cb>
		120	(or another function with the same effect).
		121
		122	=item result
		123
		124	The request results are processed synchronously by C<poll_cb>.
		125
		126	The C<poll_cb> function will process all outstanding aio requests by
		127	calling their callbacks, freeing memory associated with them and managing
		128	any groups they are contained in.
		129
		130	=item done
		131
		132	Request has reached the end of its lifetime and holds no resources anymore
		133	(except possibly for the Perl object, but its connection to the actual
		134	aio request is severed and calling its methods will either do nothing or
		135	result in a runtime error).
49		136
50	=cut	137	=cut
51		138
52	package IO::AIO;	139	package IO::AIO;
53		140
		141	no warnings;
		142	use strict 'vars';
		143
54	use base 'Exporter';	144	use base 'Exporter';
55		145
56	use Fcntl ();
57
58	BEGIN {	146	BEGIN {
59	$VERSION = 0.2;	147	our $VERSION = '2.0';
60		148
61	@EXPORT = qw(aio_read aio_write aio_open aio_close aio_stat aio_lstat aio_unlink	149	our @AIO_REQ = qw(aio_sendfile aio_read aio_write aio_open aio_close aio_stat
62	aio_fsync aio_fdatasync aio_readahead);	150	aio_lstat aio_unlink aio_rmdir aio_readdir aio_scandir aio_symlink
63	@EXPORT_OK = qw(poll_fileno poll_cb min_parallel max_parallel max_outstanding nreqs);	151	aio_fsync aio_fdatasync aio_readahead aio_rename aio_link aio_move
		152	aio_copy aio_group aio_nop aio_mknod);
		153	our @EXPORT = (@AIO_REQ, qw(aioreq_pri aioreq_nice));
		154	our @EXPORT_OK = qw(poll_fileno poll_cb poll_wait flush
		155	min_parallel max_parallel nreqs nready npending);
		156
		157	@IO::AIO::GRP::ISA = 'IO::AIO::REQ';
64		158
65	require XSLoader;	159	require XSLoader;
66	XSLoader::load IO::AIO, $VERSION;	160	XSLoader::load ("IO::AIO", $VERSION);
67	}	161	}
68		162
69	=head1 FUNCTIONS	163	=head1 FUNCTIONS
70		164
71	=head2 AIO FUNCTIONS	165	=head2 AIO FUNCTIONS
72		166
73	All the C<aio_*> calls are more or less thin wrappers around the syscall	167	All the C<aio_*> calls are more or less thin wrappers around the syscall
74	with the same name (sans C<aio_>). The arguments are similar or identical,	168	with the same name (sans C<aio_>). The arguments are similar or identical,
75	and they all accept an additional C<$callback> argument which must be	169	and they all accept an additional (and optional) C<$callback> argument
76	a code reference. This code reference will get called with the syscall	170	which must be a code reference. This code reference will get called with
77	return code (e.g. most syscalls return C<-1> on error, unlike perl, which	171	the syscall return code (e.g. most syscalls return C<-1> on error, unlike
78	usually delivers "false") as it's sole argument when the given syscall has	172	perl, which usually delivers "false") as it's sole argument when the given
79	been executed asynchronously.	173	syscall has been executed asynchronously.
80		174
81	All functions that expect a filehandle will also accept a file descriptor.	175	All functions expecting a filehandle keep a copy of the filehandle
		176	internally until the request has finished.
82		177
		178	All requests return objects of type L<IO::AIO::REQ> that allow further
		179	manipulation of those requests while they are in-flight.
		180
83	The filenames you pass to these routines I<must> be absolute. The reason	181	The pathnames you pass to these routines I<must> be absolute and
84	is that at the time the request is being executed, the current working	182	encoded in byte form. The reason for the former is that at the time the
85	directory could have changed. Alternatively, you can make sure that you	183	request is being executed, the current working directory could have
		184	changed. Alternatively, you can make sure that you never change the
86	never change the current working directory.	185	current working directory.
		186
		187	To encode pathnames to byte form, either make sure you either: a)
		188	always pass in filenames you got from outside (command line, readdir
		189	etc.), b) are ASCII or ISO 8859-1, c) use the Encode module and encode
		190	your pathnames to the locale (or other) encoding in effect in the user
		191	environment, d) use Glib::filename_from_unicode on unicode filenames or e)
		192	use something else.
87		193
88	=over 4	194	=over 4
89		195
		196	=item $prev_pri = aioreq_pri [$pri]
		197
		198	Returns the priority value that would be used for the next request and, if
		199	C<$pri> is given, sets the priority for the next aio request.
		200
		201	The default priority is C<0>, the minimum and maximum priorities are C<-4>
		202	and C<4>, respectively. Requests with higher priority will be serviced
		203	first.
		204
		205	The priority will be reset to C<0> after each call to one of the C<aio_*>
		206	functions.
		207
		208	Example: open a file with low priority, then read something from it with
		209	higher priority so the read request is serviced before other low priority
		210	open requests (potentially spamming the cache):
		211
		212	aioreq_pri -3;
		213	aio_open ..., sub {
		214	return unless $_[0];
		215
		216	aioreq_pri -2;
		217	aio_read $_[0], ..., sub {
		218	...
		219	};
		220	};
		221
		222	=item aioreq_nice $pri_adjust
		223
		224	Similar to C<aioreq_pri>, but subtracts the given value from the current
		225	priority, so effects are cumulative.
		226
90	=item aio_open $pathname, $flags, $mode, $callback	227	=item aio_open $pathname, $flags, $mode, $callback->($fh)
91		228
92	Asynchronously open or create a file and call the callback with a newly	229	Asynchronously open or create a file and call the callback with a newly
93	created filehandle for the file.	230	created filehandle for the file.
94		231
95	The pathname passed to C<aio_open> must be absolute. See API NOTES, above,	232	The pathname passed to C<aio_open> must be absolute. See API NOTES, above,
96	for an explanation.	233	for an explanation.
97		234
98	The C<$mode> argument is a bitmask. See the C<Fcntl> module for a	235	The C<$flags> argument is a bitmask. See the C<Fcntl> module for a
99	list. They are the same as used in C<sysopen>.	236	list. They are the same as used by C<sysopen>.
		237
		238	Likewise, C<$mode> specifies the mode of the newly created file, if it
		239	didn't exist and C<O_CREAT> has been given, just like perl's C<sysopen>,
		240	except that it is mandatory (i.e. use C<0> if you don't create new files,
		241	and C<0666> or C<0777> if you do).
100		242
101	Example:	243	Example:
102		244
103	aio_open "/etc/passwd", O_RDONLY, 0, sub {	245	aio_open "/etc/passwd", O_RDONLY, 0, sub {
104	if ($_[0]) {	246	if ($_[0]) {
…		…
107	} else {	249	} else {
108	die "open failed: $!\n";	250	die "open failed: $!\n";
109	}	251	}
110	};	252	};
111		253
112	=item aio_close $fh, $callback	254	=item aio_close $fh, $callback->($status)
113		255
114	Asynchronously close a file and call the callback with the result	256	Asynchronously close a file and call the callback with the result
115	code. I<WARNING:> although accepted, you should not pass in a perl	257	code. I<WARNING:> although accepted, you should not pass in a perl
116	filehandle here, as perl will likely close the file descriptor itself when	258	filehandle here, as perl will likely close the file descriptor another
117	the filehandle is destroyed. Normally, you can safely call perls C<close>	259	time when the filehandle is destroyed. Normally, you can safely call perls
118	or just let filehandles go out of scope.	260	C<close> or just let filehandles go out of scope.
119		261
		262	This is supposed to be a bug in the API, so that might change. It's
		263	therefore best to avoid this function.
		264
120	=item aio_read $fh,$offset,$length, $data,$dataoffset,$callback	265	=item aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
121		266
122	=item aio_write $fh,$offset,$length, $data,$dataoffset,$callback	267	=item aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
123		268
124	Reads or writes C<length> bytes from the specified C<fh> and C<offset>	269	Reads or writes C<length> bytes from the specified C<fh> and C<offset>
125	into the scalar given by C<data> and offset C<dataoffset> and calls the	270	into the scalar given by C<data> and offset C<dataoffset> and calls the
126	callback without the actual number of bytes read (or -1 on error, just	271	callback without the actual number of bytes read (or -1 on error, just
127	like the syscall).	272	like the syscall).
128		273
		274	The C<$data> scalar I<MUST NOT> be modified in any way while the request
		275	is outstanding. Modifying it can result in segfaults or WW3 (if the
		276	necessary/optional hardware is installed).
		277
129	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, strating at	278	Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at
130	offset C<0> within the scalar:	279	offset C<0> within the scalar:
131		280
132	aio_read $fh, 7, 15, $buffer, 0, sub {	281	aio_read $fh, 7, 15, $buffer, 0, sub {
133	$_[0] > 0 or die "read error: $!";	282	$_[0] > 0 or die "read error: $!";
134	print "read $_[0] bytes: <$buffer>\n";	283	print "read $_[0] bytes: <$buffer>\n";
135	};	284	};
136		285
		286	=item aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
		287
		288	Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts
		289	reading at byte offset C<$in_offset>, and starts writing at the current
		290	file offset of C<$out_fh>. Because of that, it is not safe to issue more
		291	than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each
		292	other.
		293
		294	This call tries to make use of a native C<sendfile> syscall to provide
		295	zero-copy operation. For this to work, C<$out_fh> should refer to a
		296	socket, and C<$in_fh> should refer to mmap'able file.
		297
		298	If the native sendfile call fails or is not implemented, it will be
		299	emulated, so you can call C<aio_sendfile> on any type of filehandle
		300	regardless of the limitations of the operating system.
		301
		302	Please note, however, that C<aio_sendfile> can read more bytes from
		303	C<$in_fh> than are written, and there is no way to find out how many
		304	bytes have been read from C<aio_sendfile> alone, as C<aio_sendfile> only
		305	provides the number of bytes written to C<$out_fh>. Only if the result
		306	value equals C<$length> one can assume that C<$length> bytes have been
		307	read.
		308
137	=item aio_readahead $fh,$offset,$length, $callback	309	=item aio_readahead $fh,$offset,$length, $callback->($retval)
138		310
139	Asynchronously reads the specified byte range into the page cache, using
140	the C<readahead> syscall. If that syscall doesn't exist the status will be
141	C<-1> and C<$!> is set to ENOSYS.
142
143	readahead() populates the page cache with data from a file so that	311	C<aio_readahead> populates the page cache with data from a file so that
144	subsequent reads from that file will not block on disk I/O. The C<$offset>	312	subsequent reads from that file will not block on disk I/O. The C<$offset>
145	argument specifies the starting point from which data is to be read and	313	argument specifies the starting point from which data is to be read and
146	C<$length> specifies the number of bytes to be read. I/O is performed in	314	C<$length> specifies the number of bytes to be read. I/O is performed in
147	whole pages, so that offset is effectively rounded down to a page boundary	315	whole pages, so that offset is effectively rounded down to a page boundary
148	and bytes are read up to the next page boundary greater than or equal to	316	and bytes are read up to the next page boundary greater than or equal to
149	(off-set+length). aio_readahead() does not read beyond the end of the	317	(off-set+length). C<aio_readahead> does not read beyond the end of the
150	file. The current file offset of the file is left unchanged.	318	file. The current file offset of the file is left unchanged.
151		319
		320	If that syscall doesn't exist (likely if your OS isn't Linux) it will be
		321	emulated by simply reading the data, which would have a similar effect.
		322
152	=item aio_stat $fh_or_path, $callback	323	=item aio_stat $fh_or_path, $callback->($status)
153		324
154	=item aio_lstat $fh, $callback	325	=item aio_lstat $fh, $callback->($status)
155		326
156	Works like perl's C<stat> or C<lstat> in void context. The callback will	327	Works like perl's C<stat> or C<lstat> in void context. The callback will
157	be called after the stat and the results will be available using C<stat _>	328	be called after the stat and the results will be available using C<stat _>
158	or C<-s _> etc...	329	or C<-s _> etc...
159		330
…		…
169	aio_stat "/etc/passwd", sub {	340	aio_stat "/etc/passwd", sub {
170	$_[0] and die "stat failed: $!";	341	$_[0] and die "stat failed: $!";
171	print "size is ", -s _, "\n";	342	print "size is ", -s _, "\n";
172	};	343	};
173		344
174	=item aio_unlink $pathname, $callback	345	=item aio_unlink $pathname, $callback->($status)
175		346
176	Asynchronously unlink (delete) a file and call the callback with the	347	Asynchronously unlink (delete) a file and call the callback with the
177	result code.	348	result code.
178		349
		350	=item aio_mknod $path, $mode, $dev, $callback->($status)
		351
		352	Asynchronously create a device node (or fifo). See mknod(2).
		353
		354	The only portable (POSIX) way of calling this function is:
		355
		356	aio_mknod $path, IO::AIO::S_IFIFO \| $mode, 0, sub { ...
		357
		358	=item aio_link $srcpath, $dstpath, $callback->($status)
		359
		360	Asynchronously create a new link to the existing object at C<$srcpath> at
		361	the path C<$dstpath> and call the callback with the result code.
		362
		363	=item aio_symlink $srcpath, $dstpath, $callback->($status)
		364
		365	Asynchronously create a new symbolic link to the existing object at C<$srcpath> at
		366	the path C<$dstpath> and call the callback with the result code.
		367
		368	=item aio_rename $srcpath, $dstpath, $callback->($status)
		369
		370	Asynchronously rename the object at C<$srcpath> to C<$dstpath>, just as
		371	rename(2) and call the callback with the result code.
		372
		373	=item aio_rmdir $pathname, $callback->($status)
		374
		375	Asynchronously rmdir (delete) a directory and call the callback with the
		376	result code.
		377
		378	=item aio_readdir $pathname, $callback->($entries)
		379
		380	Unlike the POSIX call of the same name, C<aio_readdir> reads an entire
		381	directory (i.e. opendir + readdir + closedir). The entries will not be
		382	sorted, and will B<NOT> include the C<.> and C<..> entries.
		383
		384	The callback a single argument which is either C<undef> or an array-ref
		385	with the filenames.
		386
		387	=item aio_copy $srcpath, $dstpath, $callback->($status)
		388
		389	Try to copy the I<file> (directories not supported as either source or
		390	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		391	the C<0> (error) or C<-1> ok.
		392
		393	This is a composite request that it creates the destination file with
		394	mode 0200 and copies the contents of the source file into it using
		395	C<aio_sendfile>, followed by restoring atime, mtime, access mode and
		396	uid/gid, in that order.
		397
		398	If an error occurs, the partial destination file will be unlinked, if
		399	possible, except when setting atime, mtime, access mode and uid/gid, where
		400	errors are being ignored.
		401
		402	=cut
		403
		404	sub aio_copy($$;$) {
		405	my ($src, $dst, $cb) = @_;
		406
		407	my $pri = aioreq_pri;
		408	my $grp = aio_group $cb;
		409
		410	aioreq_pri $pri;
		411	add $grp aio_open $src, O_RDONLY, 0, sub {
		412	if (my $src_fh = $_[0]) {
		413	my @stat = stat $src_fh;
		414
		415	aioreq_pri $pri;
		416	add $grp aio_open $dst, O_CREAT \| O_WRONLY \| O_TRUNC, 0200, sub {
		417	if (my $dst_fh = $_[0]) {
		418	aioreq_pri $pri;
		419	add $grp aio_sendfile $dst_fh, $src_fh, 0, $stat[7], sub {
		420	if ($_[0] == $stat[7]) {
		421	$grp->result (0);
		422	close $src_fh;
		423
		424	# those should not normally block. should. should.
		425	utime $stat[8], $stat[9], $dst;
		426	chmod $stat[2] & 07777, $dst_fh;
		427	chown $stat[4], $stat[5], $dst_fh;
		428	close $dst_fh;
		429	} else {
		430	$grp->result (-1);
		431	close $src_fh;
		432	close $dst_fh;
		433
		434	aioreq $pri;
		435	add $grp aio_unlink $dst;
		436	}
		437	};
		438	} else {
		439	$grp->result (-1);
		440	}
		441	},
		442
		443	} else {
		444	$grp->result (-1);
		445	}
		446	};
		447
		448	$grp
		449	}
		450
		451	=item aio_move $srcpath, $dstpath, $callback->($status)
		452
		453	Try to move the I<file> (directories not supported as either source or
		454	destination) from C<$srcpath> to C<$dstpath> and call the callback with
		455	the C<0> (error) or C<-1> ok.
		456
		457	This is a composite request that tries to rename(2) the file first. If
		458	rename files with C<EXDEV>, it copies the file with C<aio_copy> and, if
		459	that is successful, unlinking the C<$srcpath>.
		460
		461	=cut
		462
		463	sub aio_move($$;$) {
		464	my ($src, $dst, $cb) = @_;
		465
		466	my $pri = aioreq_pri;
		467	my $grp = aio_group $cb;
		468
		469	aioreq_pri $pri;
		470	add $grp aio_rename $src, $dst, sub {
		471	if ($_[0] && $! == EXDEV) {
		472	aioreq_pri $pri;
		473	add $grp aio_copy $src, $dst, sub {
		474	$grp->result ($_[0]);
		475
		476	if (!$_[0]) {
		477	aioreq_pri $pri;
		478	add $grp aio_unlink $src;
		479	}
		480	};
		481	} else {
		482	$grp->result ($_[0]);
		483	}
		484	};
		485
		486	$grp
		487	}
		488
		489	=item aio_scandir $path, $maxreq, $callback->($dirs, $nondirs)
		490
		491	Scans a directory (similar to C<aio_readdir>) but additionally tries to
		492	efficiently separate the entries of directory C<$path> into two sets of
		493	names, directories you can recurse into (directories), and ones you cannot
		494	recurse into (everything else, including symlinks to directories).
		495
		496	C<aio_scandir> is a composite request that creates of many sub requests_
		497	C<$maxreq> specifies the maximum number of outstanding aio requests that
		498	this function generates. If it is C<< <= 0 >>, then a suitable default
		499	will be chosen (currently 4).
		500
		501	On error, the callback is called without arguments, otherwise it receives
		502	two array-refs with path-relative entry names.
		503
		504	Example:
		505
		506	aio_scandir $dir, 0, sub {
		507	my ($dirs, $nondirs) = @_;
		508	print "real directories: @$dirs\n";
		509	print "everything else: @$nondirs\n";
		510	};
		511
		512	Implementation notes.
		513
		514	The C<aio_readdir> cannot be avoided, but C<stat()>'ing every entry can.
		515
		516	After reading the directory, the modification time, size etc. of the
		517	directory before and after the readdir is checked, and if they match (and
		518	isn't the current time), the link count will be used to decide how many
		519	entries are directories (if >= 2). Otherwise, no knowledge of the number
		520	of subdirectories will be assumed.
		521
		522	Then entries will be sorted into likely directories (everything without
		523	a non-initial dot currently) and likely non-directories (everything
		524	else). Then every entry plus an appended C</.> will be C<stat>'ed,
		525	likely directories first. If that succeeds, it assumes that the entry
		526	is a directory or a symlink to directory (which will be checked
		527	seperately). This is often faster than stat'ing the entry itself because
		528	filesystems might detect the type of the entry without reading the inode
		529	data (e.g. ext2fs filetype feature).
		530
		531	If the known number of directories (link count - 2) has been reached, the
		532	rest of the entries is assumed to be non-directories.
		533
		534	This only works with certainty on POSIX (= UNIX) filesystems, which
		535	fortunately are the vast majority of filesystems around.
		536
		537	It will also likely work on non-POSIX filesystems with reduced efficiency
		538	as those tend to return 0 or 1 as link counts, which disables the
		539	directory counting heuristic.
		540
		541	=cut
		542
		543	sub aio_scandir($$$) {
		544	my ($path, $maxreq, $cb) = @_;
		545
		546	my $pri = aioreq_pri;
		547
		548	my $grp = aio_group $cb;
		549
		550	$maxreq = 4 if $maxreq <= 0;
		551
		552	# stat once
		553	aioreq_pri $pri;
		554	add $grp aio_stat $path, sub {
		555	return $grp->result () if $_[0];
		556	my $now = time;
		557	my $hash1 = join ":", (stat _)[0,1,3,7,9];
		558
		559	# read the directory entries
		560	aioreq_pri $pri;
		561	add $grp aio_readdir $path, sub {
		562	my $entries = shift
		563	or return $grp->result ();
		564
		565	# stat the dir another time
		566	aioreq_pri $pri;
		567	add $grp aio_stat $path, sub {
		568	my $hash2 = join ":", (stat _)[0,1,3,7,9];
		569
		570	my $ndirs;
		571
		572	# take the slow route if anything looks fishy
		573	if ($hash1 ne $hash2 or (stat _)[9] == $now) {
		574	$ndirs = -1;
		575	} else {
		576	# if nlink == 2, we are finished
		577	# on non-posix-fs's, we rely on nlink < 2
		578	$ndirs = (stat _)[3] - 2
		579	or return $grp->result ([], $entries);
		580	}
		581
		582	# sort into likely dirs and likely nondirs
		583	# dirs == files without ".", short entries first
		584	$entries = [map $_->[0],
		585	sort { $b->[1] cmp $a->[1] }
		586	map [$_, sprintf "%s%04d", (/.\./ ? "1" : "0"), length],
		587	@$entries];
		588
		589	my (@dirs, @nondirs);
		590
		591	my $statgrp = add $grp aio_group sub {
		592	$grp->result (\@dirs, \@nondirs);
		593	};
		594
		595	limit $statgrp $maxreq;
		596	feed $statgrp sub {
		597	return unless @$entries;
		598	my $entry = pop @$entries;
		599
		600	aioreq_pri $pri;
		601	add $statgrp aio_stat "$path/$entry/.", sub {
		602	if ($_[0] < 0) {
		603	push @nondirs, $entry;
		604	} else {
		605	# need to check for real directory
		606	aioreq_pri $pri;
		607	add $statgrp aio_lstat "$path/$entry", sub {
		608	if (-d _) {
		609	push @dirs, $entry;
		610
		611	unless (--$ndirs) {
		612	push @nondirs, @$entries;
		613	feed $statgrp;
		614	}
		615	} else {
		616	push @nondirs, $entry;
		617	}
		618	}
		619	}
		620	};
		621	};
		622	};
		623	};
		624	};
		625
		626	$grp
		627	}
		628
179	=item aio_fsync $fh, $callback	629	=item aio_fsync $fh, $callback->($status)
180		630
181	Asynchronously call fsync on the given filehandle and call the callback	631	Asynchronously call fsync on the given filehandle and call the callback
182	with the fsync result code.	632	with the fsync result code.
183		633
184	=item aio_fdatasync $fh, $callback	634	=item aio_fdatasync $fh, $callback->($status)
185		635
186	Asynchronously call fdatasync on the given filehandle and call the	636	Asynchronously call fdatasync on the given filehandle and call the
187	callback with the fdatasync result code.	637	callback with the fdatasync result code.
188		638
		639	If this call isn't available because your OS lacks it or it couldn't be
		640	detected, it will be emulated by calling C<fsync> instead.
		641
		642	=item aio_group $callback->(...)
		643
		644	This is a very special aio request: Instead of doing something, it is a
		645	container for other aio requests, which is useful if you want to bundle
		646	many requests into a single, composite, request with a definite callback
		647	and the ability to cancel the whole request with its subrequests.
		648
		649	Returns an object of class L<IO::AIO::GRP>. See its documentation below
		650	for more info.
		651
		652	Example:
		653
		654	my $grp = aio_group sub {
		655	print "all stats done\n";
		656	};
		657
		658	add $grp
		659	(aio_stat ...),
		660	(aio_stat ...),
		661	...;
		662
		663	=item aio_nop $callback->()
		664
		665	This is a special request - it does nothing in itself and is only used for
		666	side effects, such as when you want to add a dummy request to a group so
		667	that finishing the requests in the group depends on executing the given
		668	code.
		669
		670	While this request does nothing, it still goes through the execution
		671	phase and still requires a worker thread. Thus, the callback will not
		672	be executed immediately but only after other requests in the queue have
		673	entered their execution phase. This can be used to measure request
		674	latency.
		675
		676	=item IO::AIO::aio_busy $fractional_seconds, $callback->() NOT EXPORTED
		677
		678	Mainly used for debugging and benchmarking, this aio request puts one of
		679	the request workers to sleep for the given time.
		680
		681	While it is theoretically handy to have simple I/O scheduling requests
		682	like sleep and file handle readable/writable, the overhead this creates is
		683	immense (it blocks a thread for a long time) so do not use this function
		684	except to put your application under artificial I/O pressure.
		685
189	=back	686	=back
190		687
		688	=head2 IO::AIO::REQ CLASS
		689
		690	All non-aggregate C<aio_*> functions return an object of this class when
		691	called in non-void context.
		692
		693	=over 4
		694
		695	=item cancel $req
		696
		697	Cancels the request, if possible. Has the effect of skipping execution
		698	when entering the B<execute> state and skipping calling the callback when
		699	entering the the B<result> state, but will leave the request otherwise
		700	untouched. That means that requests that currently execute will not be
		701	stopped and resources held by the request will not be freed prematurely.
		702
		703	=item cb $req $callback->(...)
		704
		705	Replace (or simply set) the callback registered to the request.
		706
		707	=back
		708
		709	=head2 IO::AIO::GRP CLASS
		710
		711	This class is a subclass of L<IO::AIO::REQ>, so all its methods apply to
		712	objects of this class, too.
		713
		714	A IO::AIO::GRP object is a special request that can contain multiple other
		715	aio requests.
		716
		717	You create one by calling the C<aio_group> constructing function with a
		718	callback that will be called when all contained requests have entered the
		719	C<done> state:
		720
		721	my $grp = aio_group sub {
		722	print "all requests are done\n";
		723	};
		724
		725	You add requests by calling the C<add> method with one or more
		726	C<IO::AIO::REQ> objects:
		727
		728	$grp->add (aio_unlink "...");
		729
		730	add $grp aio_stat "...", sub {
		731	$_[0] or return $grp->result ("error");
		732
		733	# add another request dynamically, if first succeeded
		734	add $grp aio_open "...", sub {
		735	$grp->result ("ok");
		736	};
		737	};
		738
		739	This makes it very easy to create composite requests (see the source of
		740	C<aio_move> for an application) that work and feel like simple requests.
		741
		742	=over 4
		743
		744	=item * The IO::AIO::GRP objects will be cleaned up during calls to
		745	C<IO::AIO::poll_cb>, just like any other request.
		746
		747	=item * They can be canceled like any other request. Canceling will cancel not
		748	only the request itself, but also all requests it contains.
		749
		750	=item * They can also can also be added to other IO::AIO::GRP objects.
		751
		752	=item * You must not add requests to a group from within the group callback (or
		753	any later time).
		754
		755	=back
		756
		757	Their lifetime, simplified, looks like this: when they are empty, they
		758	will finish very quickly. If they contain only requests that are in the
		759	C<done> state, they will also finish. Otherwise they will continue to
		760	exist.
		761
		762	That means after creating a group you have some time to add requests. And
		763	in the callbacks of those requests, you can add further requests to the
		764	group. And only when all those requests have finished will the the group
		765	itself finish.
		766
		767	=over 4
		768
		769	=item add $grp ...
		770
		771	=item $grp->add (...)
		772
		773	Add one or more requests to the group. Any type of L<IO::AIO::REQ> can
		774	be added, including other groups, as long as you do not create circular
		775	dependencies.
		776
		777	Returns all its arguments.
		778
		779	=item $grp->cancel_subs
		780
		781	Cancel all subrequests and clears any feeder, but not the group request
		782	itself. Useful when you queued a lot of events but got a result early.
		783
		784	=item $grp->result (...)
		785
		786	Set the result value(s) that will be passed to the group callback when all
		787	subrequests have finished and set thre groups errno to the current value
		788	of errno (just like calling C<errno> without an error number). By default,
		789	no argument will be passed and errno is zero.
		790
		791	=item $grp->errno ([$errno])
		792
		793	Sets the group errno value to C<$errno>, or the current value of errno
		794	when the argument is missing.
		795
		796	Every aio request has an associated errno value that is restored when
		797	the callback is invoked. This method lets you change this value from its
		798	default (0).
		799
		800	Calling C<result> will also set errno, so make sure you either set C<$!>
		801	before the call to C<result>, or call c<errno> after it.
		802
		803	=item feed $grp $callback->($grp)
		804
		805	Sets a feeder/generator on this group: every group can have an attached
		806	generator that generates requests if idle. The idea behind this is that,
		807	although you could just queue as many requests as you want in a group,
		808	this might starve other requests for a potentially long time. For
		809	example, C<aio_scandir> might generate hundreds of thousands C<aio_stat>
		810	requests, delaying any later requests for a long time.
		811
		812	To avoid this, and allow incremental generation of requests, you can
		813	instead a group and set a feeder on it that generates those requests. The
		814	feed callback will be called whenever there are few enough (see C<limit>,
		815	below) requests active in the group itself and is expected to queue more
		816	requests.
		817
		818	The feed callback can queue as many requests as it likes (i.e. C<add> does
		819	not impose any limits).
		820
		821	If the feed does not queue more requests when called, it will be
		822	automatically removed from the group.
		823
		824	If the feed limit is C<0>, it will be set to C<2> automatically.
		825
		826	Example:
		827
		828	# stat all files in @files, but only ever use four aio requests concurrently:
		829
		830	my $grp = aio_group sub { print "finished\n" };
		831	limit $grp 4;
		832	feed $grp sub {
		833	my $file = pop @files
		834	or return;
		835
		836	add $grp aio_stat $file, sub { ... };
		837	};
		838
		839	=item limit $grp $num
		840
		841	Sets the feeder limit for the group: The feeder will be called whenever
		842	the group contains less than this many requests.
		843
		844	Setting the limit to C<0> will pause the feeding process.
		845
		846	=back
		847
191	=head2 SUPPORT FUNCTIONS	848	=head2 SUPPORT FUNCTIONS
192		849
193	=over 4	850	=over 4
194		851
195	=item $fileno = IO::AIO::poll_fileno	852	=item $fileno = IO::AIO::poll_fileno
196		853
197	Return the I<request result pipe filehandle>. This filehandle must be	854	Return the I<request result pipe file descriptor>. This filehandle must be
198	polled for reading by some mechanism outside this module (e.g. Event	855	polled for reading by some mechanism outside this module (e.g. Event or
199	or select, see below). If the pipe becomes readable you have to call	856	select, see below or the SYNOPSIS). If the pipe becomes readable you have
200	C<poll_cb> to check the results.	857	to call C<poll_cb> to check the results.
201		858
202	See C<poll_cb> for an example.	859	See C<poll_cb> for an example.
203		860
204	=item IO::AIO::poll_cb	861	=item IO::AIO::poll_cb
205		862
206	Process all outstanding events on the result pipe. You have to call this	863	Process all outstanding events on the result pipe. You have to call this
207	regularly. Returns the number of events processed. Returns immediately	864	regularly. Returns the number of events processed. Returns immediately
208	when no events are outstanding.	865	when no events are outstanding.
209		866
210	You can use Event to multiplex, e.g.:	867	If not all requests were processed for whatever reason, the filehandle
		868	will still be ready when C<poll_cb> returns.
		869
		870	Example: Install an Event watcher that automatically calls
		871	IO::AIO::poll_cb with high priority:
211		872
212	Event->io (fd => IO::AIO::poll_fileno,	873	Event->io (fd => IO::AIO::poll_fileno,
213	poll => 'r', async => 1,	874	poll => 'r', async => 1,
214	cb => \&IO::AIO::poll_cb);	875	cb => \&IO::AIO::poll_cb);
215		876
		877	=item IO::AIO::poll_some $max_requests
		878
		879	Similar to C<poll_cb>, but only processes up to C<$max_requests> requests
		880	at a time.
		881
		882	Useful if you want to ensure some level of interactiveness when perl is
		883	not fast enough to process all requests in time.
		884
		885	Example: Install an Event watcher that automatically calls
		886	IO::AIO::poll_some with low priority, to ensure that other parts of the
		887	program get the CPU sometimes even under high AIO load.
		888
		889	Event->io (fd => IO::AIO::poll_fileno,
		890	poll => 'r', nice => 1,
		891	cb => sub { IO::AIO::poll_some 256 });
		892
216	=item IO::AIO::poll_wait	893	=item IO::AIO::poll_wait
217		894
218	Wait till the result filehandle becomes ready for reading (simply does a	895	Wait till the result filehandle becomes ready for reading (simply does a
219	select on the filehandle. This is useful if you want to synchronously wait	896	C<select> on the filehandle. This is useful if you want to synchronously wait
220	for some requests to finish).	897	for some requests to finish).
221		898
222	See C<nreqs> for an example.	899	See C<nreqs> for an example.
223		900
224	=item IO::AIO::nreqs	901	=item IO::AIO::nreqs
225		902
226	Returns the number of requests currently outstanding.	903	Returns the number of requests currently in the ready, execute or pending
		904	states (i.e. for which their callback has not been invoked yet).
227		905
228	Example: wait till there are no outstanding requests anymore:	906	Example: wait till there are no outstanding requests anymore:
229		907
230	IO::AIO::poll_wait, IO::AIO::poll_cb	908	IO::AIO::poll_wait, IO::AIO::poll_cb
231	while IO::AIO::nreqs;	909	while IO::AIO::nreqs;
232		910
		911	=item IO::AIO::nready
		912
		913	Returns the number of requests currently in the ready state (not yet
		914	executed).
		915
		916	=item IO::AIO::npending
		917
		918	Returns the number of requests currently in the pending state (executed,
		919	but not yet processed by poll_cb).
		920
		921	=item IO::AIO::flush
		922
		923	Wait till all outstanding AIO requests have been handled.
		924
		925	Strictly equivalent to:
		926
		927	IO::AIO::poll_wait, IO::AIO::poll_cb
		928	while IO::AIO::nreqs;
		929
		930	=item IO::AIO::poll
		931
		932	Waits until some requests have been handled.
		933
		934	Strictly equivalent to:
		935
		936	IO::AIO::poll_wait, IO::AIO::poll_cb
		937	if IO::AIO::nreqs;
		938
233	=item IO::AIO::min_parallel $nthreads	939	=item IO::AIO::min_parallel $nthreads
234		940
235	Set the minimum number of AIO threads to C<$nthreads>. The default is	941	Set the minimum number of AIO threads to C<$nthreads>. The current
236	C<1>, which means a single asynchronous operation can be done at one time	942	default is C<8>, which means eight asynchronous operations can execute
237	(the number of outstanding operations, however, is unlimited).	943	concurrently at any one time (the number of outstanding requests,
		944	however, is unlimited).
238		945
		946	IO::AIO starts threads only on demand, when an AIO request is queued and
		947	no free thread exists.
		948
239	It is recommended to keep the number of threads low, as some Linux	949	It is recommended to keep the number of threads relatively low, as some
240	kernel versions will scale negatively with the number of threads (higher	950	Linux kernel versions will scale negatively with the number of threads
241	parallelity => MUCH higher latency). With current Linux 2.6 versions, 4-32	951	(higher parallelity => MUCH higher latency). With current Linux 2.6
242	threads should be fine.	952	versions, 4-32 threads should be fine.
243		953
244	Under normal circumstances you don't need to call this function, as this	954	Under most circumstances you don't need to call this function, as the
245	module automatically starts some threads (the exact number might change,	955	module selects a default that is suitable for low to moderate load.
246	and is currently 4).
247		956
248	=item IO::AIO::max_parallel $nthreads	957	=item IO::AIO::max_parallel $nthreads
249		958
250	Sets the maximum number of AIO threads to C<$nthreads>. If more than	959	Sets the maximum number of AIO threads to C<$nthreads>. If more than the
251	the specified number of threads are currently running, kill them. This	960	specified number of threads are currently running, this function kills
252	function blocks until the limit is reached.	961	them. This function blocks until the limit is reached.
		962
		963	While C<$nthreads> are zero, aio requests get queued but not executed
		964	until the number of threads has been increased again.
253		965
254	This module automatically runs C<max_parallel 0> at program end, to ensure	966	This module automatically runs C<max_parallel 0> at program end, to ensure
255	that all threads are killed and that there are no outstanding requests.	967	that all threads are killed and that there are no outstanding requests.
256		968
257	Under normal circumstances you don't need to call this function.	969	Under normal circumstances you don't need to call this function.
258		970
259	=item $oldnreqs = IO::AIO::max_outstanding $nreqs	971	=item $oldmaxreqs = IO::AIO::max_outstanding $maxreqs
		972
		973	This is a very bad function to use in interactive programs because it
		974	blocks, and a bad way to reduce concurrency because it is inexact: Better
		975	use an C<aio_group> together with a feed callback.
260		976
261	Sets the maximum number of outstanding requests to C<$nreqs>. If you	977	Sets the maximum number of outstanding requests to C<$nreqs>. If you
262	try to queue up more than this number of requests, the caller will block until	978	to queue up more than this number of requests, the next call to the
263	some requests have been handled.	979	C<poll_cb> (and C<poll_some> and other functions calling C<poll_cb>)
		980	function will block until the limit is no longer exceeded.
264		981
265	The default is very large, so normally there is no practical limit. If you	982	The default value is very large, so there is no practical limit on the
266	queue up many requests in a loop it it often improves speed if you set	983	number of outstanding requests.
267	this to a relatively low number, such as C<100>.
268		984
269	Under normal circumstances you don't need to call this function.	985	You can still queue as many requests as you want. Therefore,
		986	C<max_oustsanding> is mainly useful in simple scripts (with low values) or
		987	as a stop gap to shield against fatal memory overflow (with large values).
270		988
271	=back	989	=back
272		990
273	=cut	991	=cut
274		992
275	# support function to convert a fd into a perl filehandle	993	# support function to convert a fd into a perl filehandle
276	sub _fd2fh {	994	sub _fd2fh {
277	return undef if $_[0] < 0;	995	return undef if $_[0] < 0;
278		996
279	# try to be perl5.6-compatible	997	# try to generate nice filehandles
280	local *AIO_FH;	998	my $sym = "IO::AIO::fd#$_[0]";
281	open AIO_FH, "+<&=$_[0]"	999	local *$sym;
		1000
		1001	open *$sym, "+<&=$_[0]" # usually works under any unix
		1002	or open *$sym, "<&=$_[0]" # cygwin needs this
		1003	or open *$sym, ">&=$_[0]" # or this
282	or return undef;	1004	or return undef;
283		1005
284	*AIO_FH	1006	*$sym
285	}	1007	}
286		1008
287	min_parallel 4;	1009	min_parallel 8;
288		1010
289	END {	1011	END {
290	max_parallel 0;	1012	min_parallel 1;
291	}	1013	flush;
		1014	};
292		1015
293	1;	1016	1;
294		1017
		1018	=head2 FORK BEHAVIOUR
		1019
		1020	This module should do "the right thing" when the process using it forks:
		1021
		1022	Before the fork, IO::AIO enters a quiescent state where no requests
		1023	can be added in other threads and no results will be processed. After
		1024	the fork the parent simply leaves the quiescent state and continues
		1025	request/result processing, while the child frees the request/result queue
		1026	(so that the requests started before the fork will only be handled in the
		1027	parent). Threads will be started on demand until the limit set in the
		1028	parent process has been reached again.
		1029
		1030	In short: the parent will, after a short pause, continue as if fork had
		1031	not been called, while the child will act as if IO::AIO has not been used
		1032	yet.
		1033
		1034	=head2 MEMORY USAGE
		1035
		1036	Per-request usage:
		1037
		1038	Each aio request uses - depending on your architecture - around 100-200
		1039	bytes of memory. In addition, stat requests need a stat buffer (possibly
		1040	a few hundred bytes), readdir requires a result buffer and so on. Perl
		1041	scalars and other data passed into aio requests will also be locked and
		1042	will consume memory till the request has entered the done state.
		1043
		1044	This is now awfully much, so queuing lots of requests is not usually a
		1045	problem.
		1046
		1047	Per-thread usage:
		1048
		1049	In the execution phase, some aio requests require more memory for
		1050	temporary buffers, and each thread requires a stack and other data
		1051	structures (usually around 16k-128k, depending on the OS).
		1052
		1053	=head1 KNOWN BUGS
		1054
		1055	Known bugs will be fixed in the next release.
		1056
295	=head1 SEE ALSO	1057	=head1 SEE ALSO
296		1058
297	L<Coro>, L<Linux::AIO>.	1059	L<Coro::AIO>.
298		1060
299	=head1 AUTHOR	1061	=head1 AUTHOR
300		1062
301	Marc Lehmann <schmorp@schmorp.de>	1063	Marc Lehmann <schmorp@schmorp.de>
302	http://home.schmorp.de/	1064	http://home.schmorp.de/

Diff Legend

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

Comparing IO-AIO/AIO.pm (file contents): Revision 1.9 by root, Sun Jul 10 22:20:55 2005 UTC vs. Revision 1.85 by root, Sat Oct 28 01:40:30 2006 UTC

Diff Legend

Comparing IO-AIO/AIO.pm (file contents):
Revision 1.9 by root, Sun Jul 10 22:20:55 2005 UTC vs.
Revision 1.85 by root, Sat Oct 28 01:40:30 2006 UTC