ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/IO-AIO/AIO.pm
(Generate patch)

Comparing IO-AIO/AIO.pm (file contents):
Revision 1.1 by root, Sun Jul 10 17:07:44 2005 UTC vs.
Revision 1.270 by root, Fri Jun 23 03:23:19 2017 UTC

1=head1 NAME 1=head1 NAME
2 2
3IO::AIO - Asynchronous Input/Output 3IO::AIO - Asynchronous/Advanced Input/Output
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
11This module implements asynchronous I/O using whatever means your 33This module implements asynchronous I/O using whatever means your
12operating system supports. Currently, it falls back to Linux::AIO if that 34operating system supports. It is implemented as an interface to C<libeio>
13module is available, or uses pthreads to emulato aio functionality. 35(L<http://software.schmorp.de/pkg/libeio.html>).
14 36
37Asynchronous means that operations that can normally block your program
38(e.g. reading from disk) will be done asynchronously: the operation
39will still block, but you can do something else in the meantime. This
40is extremely useful for programs that need to stay interactive even
41when doing heavy I/O (GUI programs, high performance network servers
42etc.), but can also be used to easily do operations in parallel that are
43normally done sequentially, e.g. stat'ing many files, which is much faster
44on a RAID volume or over NFS when you do a number of stat operations
45concurrently.
46
47While most of this works on all types of file descriptors (for
48example sockets), using these functions on file descriptors that
49support nonblocking operation (again, sockets, pipes etc.) is
50very inefficient. Use an event loop for that (such as the L<EV>
51module): IO::AIO will naturally fit into such an event loop itself.
52
15Currently, in this module a number of threads are started that execute 53In this version, a number of threads are started that execute your
16your read/writes and signal their completion. You don't need thread 54requests and signal their completion. You don't need thread support
17support in your libc or perl, and the threads created by this module will 55in perl, and the threads created by this module will not be visible
18not be visible to the pthreads library. 56to perl. In the future, this module might make use of the native aio
57functions available on many operating systems. However, they are often
58not well-supported or restricted (GNU/Linux doesn't allow them on normal
59files currently, for example), and they would only support aio_read and
60aio_write, so the remaining functionality would have to be implemented
61using threads anyway.
19 62
63In addition to asynchronous I/O, this module also exports some rather
64arcane interfaces, such as C<madvise> or linux's C<splice> system call,
65which is why the C<A> in C<AIO> can also mean I<advanced>.
66
20Although the module will work with in the presence of other threads, it is 67Although the module will work in the presence of other (Perl-) threads,
21not reentrant, so use appropriate locking yourself. 68it is currently not reentrant in any way, so use appropriate locking
69yourself, always call C<poll_cb> from within the same thread, or never
70call C<poll_cb> (or other C<aio_> functions) recursively.
71
72=head2 EXAMPLE
73
74This is a simple example that uses the EV module and loads
75F</etc/passwd> asynchronously:
76
77 use EV;
78 use IO::AIO;
79
80 # register the IO::AIO callback with EV
81 my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
82
83 # queue the request to open /etc/passwd
84 aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub {
85 my $fh = shift
86 or die "error while opening: $!";
87
88 # stat'ing filehandles is generally non-blocking
89 my $size = -s $fh;
90
91 # queue a request to read the file
92 my $contents;
93 aio_read $fh, 0, $size, $contents, 0, sub {
94 $_[0] == $size
95 or die "short read: $!";
96
97 close $fh;
98
99 # file contents now in $contents
100 print $contents;
101
102 # exit event loop and program
103 EV::break;
104 };
105 };
106
107 # possibly queue up other requests, or open GUI windows,
108 # check for sockets etc. etc.
109
110 # process events as long as there are some:
111 EV::run;
112
113=head1 REQUEST ANATOMY AND LIFETIME
114
115Every C<aio_*> function creates a request. which is a C data structure not
116directly visible to Perl.
117
118If called in non-void context, every request function returns a Perl
119object representing the request. In void context, nothing is returned,
120which saves a bit of memory.
121
122The perl object is a fairly standard ref-to-hash object. The hash contents
123are not used by IO::AIO so you are free to store anything you like in it.
124
125During their existance, aio requests travel through the following states,
126in order:
127
128=over 4
129
130=item ready
131
132Immediately after a request is created it is put into the ready state,
133waiting for a thread to execute it.
134
135=item execute
136
137A thread has accepted the request for processing and is currently
138executing it (e.g. blocking in read).
139
140=item pending
141
142The request has been executed and is waiting for result processing.
143
144While request submission and execution is fully asynchronous, result
145processing is not and relies on the perl interpreter calling C<poll_cb>
146(or another function with the same effect).
147
148=item result
149
150The request results are processed synchronously by C<poll_cb>.
151
152The C<poll_cb> function will process all outstanding aio requests by
153calling their callbacks, freeing memory associated with them and managing
154any groups they are contained in.
155
156=item done
157
158Request has reached the end of its lifetime and holds no resources anymore
159(except possibly for the Perl object, but its connection to the actual
160aio request is severed and calling its methods will either do nothing or
161result in a runtime error).
162
163=back
164
165=cut
166
167package IO::AIO;
168
169use Carp ();
170
171use common::sense;
172
173use base 'Exporter';
174
175BEGIN {
176 our $VERSION = 4.35;
177
178 our @AIO_REQ = qw(aio_sendfile aio_seek aio_read aio_write aio_open aio_close
179 aio_stat aio_lstat aio_unlink aio_rmdir aio_readdir aio_readdirx
180 aio_scandir aio_symlink aio_readlink aio_realpath aio_fcntl aio_ioctl
181 aio_sync aio_fsync aio_syncfs aio_fdatasync aio_sync_file_range
182 aio_pathsync aio_readahead aio_fiemap aio_allocate
183 aio_rename aio_rename2 aio_link aio_move aio_copy aio_group
184 aio_nop aio_mknod aio_load aio_rmtree aio_mkdir aio_chown
185 aio_chmod aio_utime aio_truncate
186 aio_msync aio_mtouch aio_mlock aio_mlockall
187 aio_statvfs
188 aio_wd);
189
190 our @EXPORT = (@AIO_REQ, qw(aioreq_pri aioreq_nice));
191 our @EXPORT_OK = qw(poll_fileno poll_cb poll_wait flush
192 min_parallel max_parallel max_idle idle_timeout
193 nreqs nready npending nthreads
194 max_poll_time max_poll_reqs
195 sendfile fadvise madvise
196 mmap munmap munlock munlockall);
197
198 push @AIO_REQ, qw(aio_busy); # not exported
199
200 @IO::AIO::GRP::ISA = 'IO::AIO::REQ';
201
202 require XSLoader;
203 XSLoader::load ("IO::AIO", $VERSION);
204}
205
206=head1 FUNCTIONS
207
208=head2 QUICK OVERVIEW
209
210This section simply lists the prototypes most of the functions for
211quick reference. See the following sections for function-by-function
212documentation.
213
214 aio_wd $pathname, $callback->($wd)
215 aio_open $pathname, $flags, $mode, $callback->($fh)
216 aio_close $fh, $callback->($status)
217 aio_seek $fh,$offset,$whence, $callback->($offs)
218 aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
219 aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
220 aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
221 aio_readahead $fh,$offset,$length, $callback->($retval)
222 aio_stat $fh_or_path, $callback->($status)
223 aio_lstat $fh, $callback->($status)
224 aio_statvfs $fh_or_path, $callback->($statvfs)
225 aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
226 aio_chown $fh_or_path, $uid, $gid, $callback->($status)
227 aio_chmod $fh_or_path, $mode, $callback->($status)
228 aio_truncate $fh_or_path, $offset, $callback->($status)
229 aio_allocate $fh, $mode, $offset, $len, $callback->($status)
230 aio_fiemap $fh, $start, $length, $flags, $count, $cb->(\@extents)
231 aio_unlink $pathname, $callback->($status)
232 aio_mknod $pathname, $mode, $dev, $callback->($status)
233 aio_link $srcpath, $dstpath, $callback->($status)
234 aio_symlink $srcpath, $dstpath, $callback->($status)
235 aio_readlink $pathname, $callback->($link)
236 aio_realpath $pathname, $callback->($path)
237 aio_rename $srcpath, $dstpath, $callback->($status)
238 aio_rename2 $srcpath, $dstpath, $flags, $callback->($status)
239 aio_mkdir $pathname, $mode, $callback->($status)
240 aio_rmdir $pathname, $callback->($status)
241 aio_readdir $pathname, $callback->($entries)
242 aio_readdirx $pathname, $flags, $callback->($entries, $flags)
243 IO::AIO::READDIR_DENTS IO::AIO::READDIR_DIRS_FIRST
244 IO::AIO::READDIR_STAT_ORDER IO::AIO::READDIR_FOUND_UNKNOWN
245 aio_scandir $pathname, $maxreq, $callback->($dirs, $nondirs)
246 aio_load $pathname, $data, $callback->($status)
247 aio_copy $srcpath, $dstpath, $callback->($status)
248 aio_move $srcpath, $dstpath, $callback->($status)
249 aio_rmtree $pathname, $callback->($status)
250 aio_fcntl $fh, $cmd, $arg, $callback->($status)
251 aio_ioctl $fh, $request, $buf, $callback->($status)
252 aio_sync $callback->($status)
253 aio_syncfs $fh, $callback->($status)
254 aio_fsync $fh, $callback->($status)
255 aio_fdatasync $fh, $callback->($status)
256 aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
257 aio_pathsync $pathname, $callback->($status)
258 aio_msync $scalar, $offset = 0, $length = undef, flags = MS_SYNC, $callback->($status)
259 aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
260 aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status)
261 aio_mlockall $flags, $callback->($status)
262 aio_group $callback->(...)
263 aio_nop $callback->()
264
265 $prev_pri = aioreq_pri [$pri]
266 aioreq_nice $pri_adjust
267
268 IO::AIO::poll_wait
269 IO::AIO::poll_cb
270 IO::AIO::poll
271 IO::AIO::flush
272 IO::AIO::max_poll_reqs $nreqs
273 IO::AIO::max_poll_time $seconds
274 IO::AIO::min_parallel $nthreads
275 IO::AIO::max_parallel $nthreads
276 IO::AIO::max_idle $nthreads
277 IO::AIO::idle_timeout $seconds
278 IO::AIO::max_outstanding $maxreqs
279 IO::AIO::nreqs
280 IO::AIO::nready
281 IO::AIO::npending
282
283 IO::AIO::sendfile $ofh, $ifh, $offset, $count
284 IO::AIO::fadvise $fh, $offset, $len, $advice
285 IO::AIO::mmap $scalar, $length, $prot, $flags[, $fh[, $offset]]
286 IO::AIO::munmap $scalar
287 IO::AIO::madvise $scalar, $offset, $length, $advice
288 IO::AIO::mprotect $scalar, $offset, $length, $protect
289 IO::AIO::munlock $scalar, $offset = 0, $length = undef
290 IO::AIO::munlockall
22 291
23=head2 API NOTES 292=head2 API NOTES
24 293
25All the C<aio_*> calls are more or less thin wrappers around the syscall 294All the C<aio_*> calls are more or less thin wrappers around the syscall
26with the same name (sans C<aio_>). The arguments are similar or identical, 295with the same name (sans C<aio_>). The arguments are similar or identical,
27and they all accept an additional C<$callback> argument which must be 296and they all accept an additional (and optional) C<$callback> argument
28a code reference. This code reference will get called with the syscall 297which must be a code reference. This code reference will be called after
298the syscall has been executed in an asynchronous fashion. The results
299of the request will be passed as arguments to the callback (and, if an
300error occured, in C<$!>) - for most requests the syscall return code (e.g.
29return code (e.g. most syscalls return C<-1> on error, unlike perl, which 301most syscalls return C<-1> on error, unlike perl, which usually delivers
30usually delivers "false") as it's sole argument when the given syscall has 302"false").
31been executed asynchronously.
32 303
33All functions that expect a filehandle will also accept a file descriptor. 304Some requests (such as C<aio_readdir>) pass the actual results and
305communicate failures by passing C<undef>.
34 306
307All functions expecting a filehandle keep a copy of the filehandle
308internally until the request has finished.
309
310All functions return request objects of type L<IO::AIO::REQ> that allow
311further manipulation of those requests while they are in-flight.
312
35The filenames you pass to these routines I<must> be absolute. The reason 313The pathnames you pass to these routines I<should> be absolute. The
36is that at the time the request is being executed, the current working 314reason for this is that at the time the request is being executed, the
37directory could have changed. Alternatively, you can make sure that you 315current working directory could have changed. Alternatively, you can
38never change the current working directory. 316make sure that you never change the current working directory anywhere
317in the program and then use relative paths. You can also take advantage
318of IO::AIOs working directory abstraction, that lets you specify paths
319relative to some previously-opened "working directory object" - see the
320description of the C<IO::AIO::WD> class later in this document.
321
322To encode pathnames as octets, either make sure you either: a) always pass
323in filenames you got from outside (command line, readdir etc.) without
324tinkering, b) are in your native filesystem encoding, c) use the Encode
325module and encode your pathnames to the locale (or other) encoding in
326effect in the user environment, d) use Glib::filename_from_unicode on
327unicode filenames or e) use something else to ensure your scalar has the
328correct contents.
329
330This works, btw. independent of the internal UTF-8 bit, which IO::AIO
331handles correctly whether it is set or not.
332
333=head2 AIO REQUEST FUNCTIONS
39 334
40=over 4 335=over 4
41 336
42=cut 337=item $prev_pri = aioreq_pri [$pri]
43 338
44package IO::AIO; 339Returns the priority value that would be used for the next request and, if
340C<$pri> is given, sets the priority for the next aio request.
45 341
46use base 'Exporter'; 342The default priority is C<0>, the minimum and maximum priorities are C<-4>
343and C<4>, respectively. Requests with higher priority will be serviced
344first.
47 345
48BEGIN { 346The priority will be reset to C<0> after each call to one of the C<aio_*>
49 $VERSION = 0.1; 347functions.
50 348
51 @EXPORT = qw(aio_read aio_write aio_open aio_close aio_stat aio_lstat aio_unlink 349Example: open a file with low priority, then read something from it with
52 aio_fsync aio_fdatasync aio_readahead); 350higher priority so the read request is serviced before other low priority
53 @EXPORT_OK = qw(poll_fileno poll_cb min_parallel max_parallel nreqs); 351open requests (potentially spamming the cache):
54 352
55 require XSLoader; 353 aioreq_pri -3;
56 XSLoader::load IO::AIO, $VERSION; 354 aio_open ..., sub {
57} 355 return unless $_[0];
58 356
59=item IO::AIO::min_parallel $nthreads 357 aioreq_pri -2;
358 aio_read $_[0], ..., sub {
359 ...
360 };
361 };
60 362
61Set the minimum number of AIO threads to C<$nthreads>. The default is
62C<1>, which means a single asynchronous operation can be done at one time
63(the number of outstanding operations, however, is unlimited).
64 363
65It is recommended to keep the number of threads low, as some linux 364=item aioreq_nice $pri_adjust
66kernel versions will scale negatively with the number of threads (higher
67parallelity => MUCH higher latency).
68 365
69Under normal circumstances you don't need to call this function, as this 366Similar to C<aioreq_pri>, but subtracts the given value from the current
70module automatically starts a single async thread. 367priority, so the effect is cumulative.
71 368
72=item IO::AIO::max_parallel $nthreads
73 369
74Sets the maximum number of AIO threads to C<$nthreads>. If more than
75the specified number of threads are currently running, kill them. This
76function blocks until the limit is reached.
77
78This module automatically runs C<max_parallel 0> at program end, to ensure
79that all threads are killed and that there are no outstanding requests.
80
81Under normal circumstances you don't need to call this function.
82
83=item $fileno = IO::AIO::poll_fileno
84
85Return the I<request result pipe filehandle>. This filehandle must be
86polled for reading by some mechanism outside this module (e.g. Event
87or select, see below). If the pipe becomes readable you have to call
88C<poll_cb> to check the results.
89
90See C<poll_cb> for an example.
91
92=item IO::AIO::poll_cb
93
94Process all outstanding events on the result pipe. You have to call this
95regularly. Returns the number of events processed. Returns immediately
96when no events are outstanding.
97
98You can use Event to multiplex, e.g.:
99
100 Event->io (fd => IO::AIO::poll_fileno,
101 poll => 'r', async => 1,
102 cb => \&IO::AIO::poll_cb);
103
104=item IO::AIO::poll_wait
105
106Wait till the result filehandle becomes ready for reading (simply does a
107select on the filehandle. This is useful if you want to synchronously wait
108for some requests to finish).
109
110See C<nreqs> for an example.
111
112=item IO::AIO::nreqs
113
114Returns the number of requests currently outstanding.
115
116Example: wait till there are no outstanding requests anymore:
117
118 IO::AIO::poll_wait, IO::AIO::poll_cb
119 while IO::AIO::nreqs;
120
121=item aio_open $pathname, $flags, $mode, $callback 370=item aio_open $pathname, $flags, $mode, $callback->($fh)
122 371
123Asynchronously open or create a file and call the callback with the 372Asynchronously open or create a file and call the callback with a newly
124filedescriptor (NOT a perl filehandle, sorry for that, but watch out, this 373created filehandle for the file (or C<undef> in case of an error).
125might change in the future).
126 374
127The pathname passed to C<aio_open> must be absolute. See API NOTES, above, 375The pathname passed to C<aio_open> must be absolute. See API NOTES, above,
128for an explanation. 376for an explanation.
129 377
130The C<$mode> argument is a bitmask. See the C<Fcntl> module for a 378The C<$flags> argument is a bitmask. See the C<Fcntl> module for a
131list. They are the same as used in C<sysopen>. 379list. They are the same as used by C<sysopen>.
380
381Likewise, C<$mode> specifies the mode of the newly created file, if it
382didn't exist and C<O_CREAT> has been given, just like perl's C<sysopen>,
383except that it is mandatory (i.e. use C<0> if you don't create new files,
384and C<0666> or C<0777> if you do). Note that the C<$mode> will be modified
385by the umask in effect then the request is being executed, so better never
386change the umask.
132 387
133Example: 388Example:
134 389
135 aio_open "/etc/passwd", O_RDONLY, 0, sub { 390 aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub {
136 if ($_[0] >= 0) { 391 if ($_[0]) {
137 open my $fh, "<&$_[0]"; # create a copy for perl
138 aio_close $_[0], sub { }; # close the aio handle
139 print "open successful, fh is $fh\n"; 392 print "open successful, fh is $_[0]\n";
140 ... 393 ...
141 } else { 394 } else {
142 die "open failed: $!\n"; 395 die "open failed: $!\n";
143 } 396 }
144 }; 397 };
145 398
399In addition to all the common open modes/flags (C<O_RDONLY>, C<O_WRONLY>,
400C<O_RDWR>, C<O_CREAT>, C<O_TRUNC>, C<O_EXCL> and C<O_APPEND>), the
401following POSIX and non-POSIX constants are available (missing ones on
402your system are, as usual, C<0>):
403
404C<O_ASYNC>, C<O_DIRECT>, C<O_NOATIME>, C<O_CLOEXEC>, C<O_NOCTTY>, C<O_NOFOLLOW>,
405C<O_NONBLOCK>, C<O_EXEC>, C<O_SEARCH>, C<O_DIRECTORY>, C<O_DSYNC>,
406C<O_RSYNC>, C<O_SYNC>, C<O_PATH>, C<O_TMPFILE>, and C<O_TTY_INIT>.
407
408
146=item aio_close $fh, $callback 409=item aio_close $fh, $callback->($status)
147 410
148Asynchronously close a file and call the callback with the result code. 411Asynchronously close a file and call the callback with the result
412code.
149 413
414Unfortunately, you can't do this to perl. Perl I<insists> very strongly on
415closing the file descriptor associated with the filehandle itself.
416
417Therefore, C<aio_close> will not close the filehandle - instead it will
418use dup2 to overwrite the file descriptor with the write-end of a pipe
419(the pipe fd will be created on demand and will be cached).
420
421Or in other words: the file descriptor will be closed, but it will not be
422free for reuse until the perl filehandle is closed.
423
424=cut
425
426=item aio_seek $fh, $offset, $whence, $callback->($offs)
427
428Seeks the filehandle to the new C<$offset>, similarly to perl's
429C<sysseek>. The C<$whence> can use the traditional values (C<0> for
430C<IO::AIO::SEEK_SET>, C<1> for C<IO::AIO::SEEK_CUR> or C<2> for
431C<IO::AIO::SEEK_END>).
432
433The resulting absolute offset will be passed to the callback, or C<-1> in
434case of an error.
435
436In theory, the C<$whence> constants could be different than the
437corresponding values from L<Fcntl>, but perl guarantees they are the same,
438so don't panic.
439
440As a GNU/Linux (and maybe Solaris) extension, also the constants
441C<IO::AIO::SEEK_DATA> and C<IO::AIO::SEEK_HOLE> are available, if they
442could be found. No guarantees about suitability for use in C<aio_seek> or
443Perl's C<sysseek> can be made though, although I would naively assume they
444"just work".
445
150=item aio_read $fh,$offset,$length, $data,$dataoffset,$callback 446=item aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
151 447
152=item aio_write $fh,$offset,$length, $data,$dataoffset,$callback 448=item aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
153 449
154Reads or writes C<length> bytes from the specified C<fh> and C<offset> 450Reads or writes C<$length> bytes from or to the specified C<$fh> and
155into the scalar given by C<data> and offset C<dataoffset> and calls the 451C<$offset> into the scalar given by C<$data> and offset C<$dataoffset> and
156callback without the actual number of bytes read (or -1 on error, just 452calls the callback with the actual number of bytes transferred (or -1 on
157like the syscall). 453error, just like the syscall).
158 454
455C<aio_read> will, like C<sysread>, shrink or grow the C<$data> scalar to
456offset plus the actual number of bytes read.
457
458If C<$offset> is undefined, then the current file descriptor offset will
459be used (and updated), otherwise the file descriptor offset will not be
460changed by these calls.
461
462If C<$length> is undefined in C<aio_write>, use the remaining length of
463C<$data>.
464
465If C<$dataoffset> is less than zero, it will be counted from the end of
466C<$data>.
467
468The C<$data> scalar I<MUST NOT> be modified in any way while the request
469is outstanding. Modifying it can result in segfaults or World War III (if
470the necessary/optional hardware is installed).
471
159Example: Read 15 bytes at offset 7 into scalar C<$buffer>, strating at 472Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at
160offset C<0> within the scalar: 473offset C<0> within the scalar:
161 474
162 aio_read $fh, 7, 15, $buffer, 0, sub { 475 aio_read $fh, 7, 15, $buffer, 0, sub {
163 $_[0] >= 0 or die "read error: $!"; 476 $_[0] > 0 or die "read error: $!";
164 print "read <$buffer>\n"; 477 print "read $_[0] bytes: <$buffer>\n";
165 }; 478 };
166 479
480
481=item aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
482
483Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts
484reading at byte offset C<$in_offset>, and starts writing at the current
485file offset of C<$out_fh>. Because of that, it is not safe to issue more
486than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each
487other. The same C<$in_fh> works fine though, as this function does not
488move or use the file offset of C<$in_fh>.
489
490Please note that C<aio_sendfile> can read more bytes from C<$in_fh> than
491are written, and there is no way to find out how many more bytes have been
492read from C<aio_sendfile> alone, as C<aio_sendfile> only provides the
493number of bytes written to C<$out_fh>. Only if the result value equals
494C<$length> one can assume that C<$length> bytes have been read.
495
496Unlike with other C<aio_> functions, it makes a lot of sense to use
497C<aio_sendfile> on non-blocking sockets, as long as one end (typically
498the C<$in_fh>) is a file - the file I/O will then be asynchronous, while
499the socket I/O will be non-blocking. Note, however, that you can run
500into a trap where C<aio_sendfile> reads some data with readahead, then
501fails to write all data, and when the socket is ready the next time, the
502data in the cache is already lost, forcing C<aio_sendfile> to again hit
503the disk. Explicit C<aio_read> + C<aio_write> let's you better control
504resource usage.
505
506This call tries to make use of a native C<sendfile>-like syscall to
507provide zero-copy operation. For this to work, C<$out_fh> should refer to
508a socket, and C<$in_fh> should refer to an mmap'able file.
509
510If a native sendfile cannot be found or it fails with C<ENOSYS>,
511C<EINVAL>, C<ENOTSUP>, C<EOPNOTSUPP>, C<EAFNOSUPPORT>, C<EPROTOTYPE> or
512C<ENOTSOCK>, it will be emulated, so you can call C<aio_sendfile> on any
513type of filehandle regardless of the limitations of the operating system.
514
515As native sendfile syscalls (as practically any non-POSIX interface hacked
516together in a hurry to improve benchmark numbers) tend to be rather buggy
517on many systems, this implementation tries to work around some known bugs
518in Linux and FreeBSD kernels (probably others, too), but that might fail,
519so you really really should check the return value of C<aio_sendfile> -
520fewer bytes than expected might have been transferred.
521
522
167=item aio_readahead $fh,$offset,$length, $callback 523=item aio_readahead $fh,$offset,$length, $callback->($retval)
168 524
169Asynchronously reads the specified byte range into the page cache, using
170the C<readahead> syscall. If that syscall doesn't exist the status will be
171C<-1> and C<$!> is set to ENOSYS.
172
173readahead() populates the page cache with data from a file so that 525C<aio_readahead> populates the page cache with data from a file so that
174subsequent reads from that file will not block on disk I/O. The C<$offset> 526subsequent reads from that file will not block on disk I/O. The C<$offset>
175argument specifies the starting point from which data is to be read and 527argument specifies the starting point from which data is to be read and
176C<$length> specifies the number of bytes to be read. I/O is performed in 528C<$length> specifies the number of bytes to be read. I/O is performed in
177whole pages, so that offset is effectively rounded down to a page boundary 529whole pages, so that offset is effectively rounded down to a page boundary
178and bytes are read up to the next page boundary greater than or equal to 530and bytes are read up to the next page boundary greater than or equal to
179(off-set+length). aio_readahead() does not read beyond the end of the 531(off-set+length). C<aio_readahead> does not read beyond the end of the
180file. The current file offset of the file is left unchanged. 532file. The current file offset of the file is left unchanged.
181 533
534If that syscall doesn't exist (likely if your kernel isn't Linux) it will
535be emulated by simply reading the data, which would have a similar effect.
536
537
182=item aio_stat $fh_or_path, $callback 538=item aio_stat $fh_or_path, $callback->($status)
183 539
184=item aio_lstat $fh, $callback 540=item aio_lstat $fh, $callback->($status)
185 541
186Works like perl's C<stat> or C<lstat> in void context. The callback will 542Works like perl's C<stat> or C<lstat> in void context. The callback will
187be called after the stat and the results will be available using C<stat _> 543be called after the stat and the results will be available using C<stat _>
188or C<-s _> etc... 544or C<-s _> etc...
189 545
191for an explanation. 547for an explanation.
192 548
193Currently, the stats are always 64-bit-stats, i.e. instead of returning an 549Currently, the stats are always 64-bit-stats, i.e. instead of returning an
194error when stat'ing a large file, the results will be silently truncated 550error when stat'ing a large file, the results will be silently truncated
195unless perl itself is compiled with large file support. 551unless perl itself is compiled with large file support.
552
553To help interpret the mode and dev/rdev stat values, IO::AIO offers the
554following constants and functions (if not implemented, the constants will
555be C<0> and the functions will either C<croak> or fall back on traditional
556behaviour).
557
558C<S_IFMT>, C<S_IFIFO>, C<S_IFCHR>, C<S_IFBLK>, C<S_IFLNK>, C<S_IFREG>,
559C<S_IFDIR>, C<S_IFWHT>, C<S_IFSOCK>, C<IO::AIO::major $dev_t>,
560C<IO::AIO::minor $dev_t>, C<IO::AIO::makedev $major, $minor>.
196 561
197Example: Print the length of F</etc/passwd>: 562Example: Print the length of F</etc/passwd>:
198 563
199 aio_stat "/etc/passwd", sub { 564 aio_stat "/etc/passwd", sub {
200 $_[0] and die "stat failed: $!"; 565 $_[0] and die "stat failed: $!";
201 print "size is ", -s _, "\n"; 566 print "size is ", -s _, "\n";
202 }; 567 };
203 568
569
570=item aio_statvfs $fh_or_path, $callback->($statvfs)
571
572Works like the POSIX C<statvfs> or C<fstatvfs> syscalls, depending on
573whether a file handle or path was passed.
574
575On success, the callback is passed a hash reference with the following
576members: C<bsize>, C<frsize>, C<blocks>, C<bfree>, C<bavail>, C<files>,
577C<ffree>, C<favail>, C<fsid>, C<flag> and C<namemax>. On failure, C<undef>
578is passed.
579
580The following POSIX IO::AIO::ST_* constants are defined: C<ST_RDONLY> and
581C<ST_NOSUID>.
582
583The following non-POSIX IO::AIO::ST_* flag masks are defined to
584their correct value when available, or to C<0> on systems that do
585not support them: C<ST_NODEV>, C<ST_NOEXEC>, C<ST_SYNCHRONOUS>,
586C<ST_MANDLOCK>, C<ST_WRITE>, C<ST_APPEND>, C<ST_IMMUTABLE>, C<ST_NOATIME>,
587C<ST_NODIRATIME> and C<ST_RELATIME>.
588
589Example: stat C</wd> and dump out the data if successful.
590
591 aio_statvfs "/wd", sub {
592 my $f = $_[0]
593 or die "statvfs: $!";
594
595 use Data::Dumper;
596 say Dumper $f;
597 };
598
599 # result:
600 {
601 bsize => 1024,
602 bfree => 4333064312,
603 blocks => 10253828096,
604 files => 2050765568,
605 flag => 4096,
606 favail => 2042092649,
607 bavail => 4333064312,
608 ffree => 2042092649,
609 namemax => 255,
610 frsize => 1024,
611 fsid => 1810
612 }
613
614Here is a (likely partial - send me updates!) list of fsid values used by
615Linux - it is safe to hardcode these when C<$^O> is C<linux>:
616
617 0x0000adf5 adfs
618 0x0000adff affs
619 0x5346414f afs
620 0x09041934 anon-inode filesystem
621 0x00000187 autofs
622 0x42465331 befs
623 0x1badface bfs
624 0x42494e4d binfmt_misc
625 0x9123683e btrfs
626 0x0027e0eb cgroupfs
627 0xff534d42 cifs
628 0x73757245 coda
629 0x012ff7b7 coh
630 0x28cd3d45 cramfs
631 0x453dcd28 cramfs-wend (wrong endianness)
632 0x64626720 debugfs
633 0x00001373 devfs
634 0x00001cd1 devpts
635 0x0000f15f ecryptfs
636 0x00414a53 efs
637 0x0000137d ext
638 0x0000ef53 ext2/ext3/ext4
639 0x0000ef51 ext2
640 0xf2f52010 f2fs
641 0x00004006 fat
642 0x65735546 fuseblk
643 0x65735543 fusectl
644 0x0bad1dea futexfs
645 0x01161970 gfs2
646 0x47504653 gpfs
647 0x00004244 hfs
648 0xf995e849 hpfs
649 0x00c0ffee hostfs
650 0x958458f6 hugetlbfs
651 0x2bad1dea inotifyfs
652 0x00009660 isofs
653 0x000072b6 jffs2
654 0x3153464a jfs
655 0x6b414653 k-afs
656 0x0bd00bd0 lustre
657 0x0000137f minix
658 0x0000138f minix 30 char names
659 0x00002468 minix v2
660 0x00002478 minix v2 30 char names
661 0x00004d5a minix v3
662 0x19800202 mqueue
663 0x00004d44 msdos
664 0x0000564c novell
665 0x00006969 nfs
666 0x6e667364 nfsd
667 0x00003434 nilfs
668 0x5346544e ntfs
669 0x00009fa1 openprom
670 0x7461636F ocfs2
671 0x00009fa0 proc
672 0x6165676c pstorefs
673 0x0000002f qnx4
674 0x68191122 qnx6
675 0x858458f6 ramfs
676 0x52654973 reiserfs
677 0x00007275 romfs
678 0x67596969 rpc_pipefs
679 0x73636673 securityfs
680 0xf97cff8c selinux
681 0x0000517b smb
682 0x534f434b sockfs
683 0x73717368 squashfs
684 0x62656572 sysfs
685 0x012ff7b6 sysv2
686 0x012ff7b5 sysv4
687 0x01021994 tmpfs
688 0x15013346 udf
689 0x00011954 ufs
690 0x54190100 ufs byteswapped
691 0x00009fa2 usbdevfs
692 0x01021997 v9fs
693 0xa501fcf5 vxfs
694 0xabba1974 xenfs
695 0x012ff7b4 xenix
696 0x58465342 xfs
697 0x012fd16d xia
698
699=item aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
700
701Works like perl's C<utime> function (including the special case of $atime
702and $mtime being undef). Fractional times are supported if the underlying
703syscalls support them.
704
705When called with a pathname, uses utimes(2) if available, otherwise
706utime(2). If called on a file descriptor, uses futimes(2) if available,
707otherwise returns ENOSYS, so this is not portable.
708
709Examples:
710
711 # set atime and mtime to current time (basically touch(1)):
712 aio_utime "path", undef, undef;
713 # set atime to current time and mtime to beginning of the epoch:
714 aio_utime "path", time, undef; # undef==0
715
716
717=item aio_chown $fh_or_path, $uid, $gid, $callback->($status)
718
719Works like perl's C<chown> function, except that C<undef> for either $uid
720or $gid is being interpreted as "do not change" (but -1 can also be used).
721
722Examples:
723
724 # same as "chown root path" in the shell:
725 aio_chown "path", 0, -1;
726 # same as above:
727 aio_chown "path", 0, undef;
728
729
730=item aio_truncate $fh_or_path, $offset, $callback->($status)
731
732Works like truncate(2) or ftruncate(2).
733
734
735=item aio_allocate $fh, $mode, $offset, $len, $callback->($status)
736
737Allocates or frees disk space according to the C<$mode> argument. See the
738linux C<fallocate> documentation for details.
739
740C<$mode> is usually C<0> or C<IO::AIO::FALLOC_FL_KEEP_SIZE> to allocate
741space, or C<IO::AIO::FALLOC_FL_PUNCH_HOLE | IO::AIO::FALLOC_FL_KEEP_SIZE>,
742to deallocate a file range.
743
744IO::AIO also supports C<FALLOC_FL_COLLAPSE_RANGE>, to remove a range
745(without leaving a hole) and C<FALLOC_FL_ZERO_RANGE>, to zero a range (see
746your L<fallocate(2)> manpage).
747
748The file system block size used by C<fallocate> is presumably the
749C<f_bsize> returned by C<statvfs>.
750
751If C<fallocate> isn't available or cannot be emulated (currently no
752emulation will be attempted), passes C<-1> and sets C<$!> to C<ENOSYS>.
753
754
755=item aio_chmod $fh_or_path, $mode, $callback->($status)
756
757Works like perl's C<chmod> function.
758
759
204=item aio_unlink $pathname, $callback 760=item aio_unlink $pathname, $callback->($status)
205 761
206Asynchronously unlink (delete) a file and call the callback with the 762Asynchronously unlink (delete) a file and call the callback with the
207result code. 763result code.
208 764
765
766=item aio_mknod $pathname, $mode, $dev, $callback->($status)
767
768[EXPERIMENTAL]
769
770Asynchronously create a device node (or fifo). See mknod(2).
771
772The only (POSIX-) portable way of calling this function is:
773
774 aio_mknod $pathname, IO::AIO::S_IFIFO | $mode, 0, sub { ...
775
776See C<aio_stat> for info about some potentially helpful extra constants
777and functions.
778
779=item aio_link $srcpath, $dstpath, $callback->($status)
780
781Asynchronously create a new link to the existing object at C<$srcpath> at
782the path C<$dstpath> and call the callback with the result code.
783
784
785=item aio_symlink $srcpath, $dstpath, $callback->($status)
786
787Asynchronously create a new symbolic link to the existing object at C<$srcpath> at
788the path C<$dstpath> and call the callback with the result code.
789
790
791=item aio_readlink $pathname, $callback->($link)
792
793Asynchronously read the symlink specified by C<$path> and pass it to
794the callback. If an error occurs, nothing or undef gets passed to the
795callback.
796
797
798=item aio_realpath $pathname, $callback->($path)
799
800Asynchronously make the path absolute and resolve any symlinks in
801C<$path>. The resulting path only consists of directories (same as
802L<Cwd::realpath>).
803
804This request can be used to get the absolute path of the current working
805directory by passing it a path of F<.> (a single dot).
806
807
808=item aio_rename $srcpath, $dstpath, $callback->($status)
809
810Asynchronously rename the object at C<$srcpath> to C<$dstpath>, just as
811rename(2) and call the callback with the result code.
812
813On systems that support the AIO::WD working directory abstraction
814natively, the case C<[$wd, "."]> as C<$srcpath> is specialcased - instead
815of failing, C<rename> is called on the absolute path of C<$wd>.
816
817
818=item aio_rename2 $srcpath, $dstpath, $flags, $callback->($status)
819
820Basically a version of C<aio_rename> with an additional C<$flags>
821argument. Calling this with C<$flags=0> is the same as calling
822C<aio_rename>.
823
824Non-zero flags are currently only supported on GNU/Linux systems that
825support renameat2. Other systems fail with C<ENOSYS> in this case.
826
827The following constants are available (missing ones are, as usual C<0>),
828see renameat2(2) for details:
829
830C<IO::AIO::RENAME_NOREPLACE>, C<IO::AIO::RENAME_EXCHANGE>
831and C<IO::AIO::RENAME_WHITEOUT>.
832
833
834=item aio_mkdir $pathname, $mode, $callback->($status)
835
836Asynchronously mkdir (create) a directory and call the callback with
837the result code. C<$mode> will be modified by the umask at the time the
838request is executed, so do not change your umask.
839
840
841=item aio_rmdir $pathname, $callback->($status)
842
843Asynchronously rmdir (delete) a directory and call the callback with the
844result code.
845
846On systems that support the AIO::WD working directory abstraction
847natively, the case C<[$wd, "."]> is specialcased - instead of failing,
848C<rmdir> is called on the absolute path of C<$wd>.
849
850
851=item aio_readdir $pathname, $callback->($entries)
852
853Unlike the POSIX call of the same name, C<aio_readdir> reads an entire
854directory (i.e. opendir + readdir + closedir). The entries will not be
855sorted, and will B<NOT> include the C<.> and C<..> entries.
856
857The callback is passed a single argument which is either C<undef> or an
858array-ref with the filenames.
859
860
861=item aio_readdirx $pathname, $flags, $callback->($entries, $flags)
862
863Quite similar to C<aio_readdir>, but the C<$flags> argument allows one to
864tune behaviour and output format. In case of an error, C<$entries> will be
865C<undef>.
866
867The flags are a combination of the following constants, ORed together (the
868flags will also be passed to the callback, possibly modified):
869
870=over 4
871
872=item IO::AIO::READDIR_DENTS
873
874When this flag is off, then the callback gets an arrayref consisting of
875names only (as with C<aio_readdir>), otherwise it gets an arrayref with
876C<[$name, $type, $inode]> arrayrefs, each describing a single directory
877entry in more detail.
878
879C<$name> is the name of the entry.
880
881C<$type> is one of the C<IO::AIO::DT_xxx> constants:
882
883C<IO::AIO::DT_UNKNOWN>, C<IO::AIO::DT_FIFO>, C<IO::AIO::DT_CHR>, C<IO::AIO::DT_DIR>,
884C<IO::AIO::DT_BLK>, C<IO::AIO::DT_REG>, C<IO::AIO::DT_LNK>, C<IO::AIO::DT_SOCK>,
885C<IO::AIO::DT_WHT>.
886
887C<IO::AIO::DT_UNKNOWN> means just that: readdir does not know. If you need to
888know, you have to run stat yourself. Also, for speed reasons, the C<$type>
889scalars are read-only: you can not modify them.
890
891C<$inode> is the inode number (which might not be exact on systems with 64
892bit inode numbers and 32 bit perls). This field has unspecified content on
893systems that do not deliver the inode information.
894
895=item IO::AIO::READDIR_DIRS_FIRST
896
897When this flag is set, then the names will be returned in an order where
898likely directories come first, in optimal stat order. This is useful when
899you need to quickly find directories, or you want to find all directories
900while avoiding to stat() each entry.
901
902If the system returns type information in readdir, then this is used
903to find directories directly. Otherwise, likely directories are names
904beginning with ".", or otherwise names with no dots, of which names with
905short names are tried first.
906
907=item IO::AIO::READDIR_STAT_ORDER
908
909When this flag is set, then the names will be returned in an order
910suitable for stat()'ing each one. That is, when you plan to stat()
911all files in the given directory, then the returned order will likely
912be fastest.
913
914If both this flag and C<IO::AIO::READDIR_DIRS_FIRST> are specified, then
915the likely dirs come first, resulting in a less optimal stat order.
916
917=item IO::AIO::READDIR_FOUND_UNKNOWN
918
919This flag should not be set when calling C<aio_readdirx>. Instead, it
920is being set by C<aio_readdirx>, when any of the C<$type>'s found were
921C<IO::AIO::DT_UNKNOWN>. The absence of this flag therefore indicates that all
922C<$type>'s are known, which can be used to speed up some algorithms.
923
924=back
925
926
927=item aio_load $pathname, $data, $callback->($status)
928
929This is a composite request that tries to fully load the given file into
930memory. Status is the same as with aio_read.
931
932=cut
933
934sub aio_load($$;$) {
935 my ($path, undef, $cb) = @_;
936 my $data = \$_[1];
937
938 my $pri = aioreq_pri;
939 my $grp = aio_group $cb;
940
941 aioreq_pri $pri;
942 add $grp aio_open $path, O_RDONLY, 0, sub {
943 my $fh = shift
944 or return $grp->result (-1);
945
946 aioreq_pri $pri;
947 add $grp aio_read $fh, 0, (-s $fh), $$data, 0, sub {
948 $grp->result ($_[0]);
949 };
950 };
951
952 $grp
953}
954
955=item aio_copy $srcpath, $dstpath, $callback->($status)
956
957Try to copy the I<file> (directories not supported as either source or
958destination) from C<$srcpath> to C<$dstpath> and call the callback with
959a status of C<0> (ok) or C<-1> (error, see C<$!>).
960
961This is a composite request that creates the destination file with
962mode 0200 and copies the contents of the source file into it using
963C<aio_sendfile>, followed by restoring atime, mtime, access mode and
964uid/gid, in that order.
965
966If an error occurs, the partial destination file will be unlinked, if
967possible, except when setting atime, mtime, access mode and uid/gid, where
968errors are being ignored.
969
970=cut
971
972sub aio_copy($$;$) {
973 my ($src, $dst, $cb) = @_;
974
975 my $pri = aioreq_pri;
976 my $grp = aio_group $cb;
977
978 aioreq_pri $pri;
979 add $grp aio_open $src, O_RDONLY, 0, sub {
980 if (my $src_fh = $_[0]) {
981 my @stat = stat $src_fh; # hmm, might block over nfs?
982
983 aioreq_pri $pri;
984 add $grp aio_open $dst, O_CREAT | O_WRONLY | O_TRUNC, 0200, sub {
985 if (my $dst_fh = $_[0]) {
986 aioreq_pri $pri;
987 add $grp aio_sendfile $dst_fh, $src_fh, 0, $stat[7], sub {
988 if ($_[0] == $stat[7]) {
989 $grp->result (0);
990 close $src_fh;
991
992 my $ch = sub {
993 aioreq_pri $pri;
994 add $grp aio_chmod $dst_fh, $stat[2] & 07777, sub {
995 aioreq_pri $pri;
996 add $grp aio_chown $dst_fh, $stat[4], $stat[5], sub {
997 aioreq_pri $pri;
998 add $grp aio_close $dst_fh;
999 }
1000 };
1001 };
1002
1003 aioreq_pri $pri;
1004 add $grp aio_utime $dst_fh, $stat[8], $stat[9], sub {
1005 if ($_[0] < 0 && $! == ENOSYS) {
1006 aioreq_pri $pri;
1007 add $grp aio_utime $dst, $stat[8], $stat[9], $ch;
1008 } else {
1009 $ch->();
1010 }
1011 };
1012 } else {
1013 $grp->result (-1);
1014 close $src_fh;
1015 close $dst_fh;
1016
1017 aioreq $pri;
1018 add $grp aio_unlink $dst;
1019 }
1020 };
1021 } else {
1022 $grp->result (-1);
1023 }
1024 },
1025
1026 } else {
1027 $grp->result (-1);
1028 }
1029 };
1030
1031 $grp
1032}
1033
1034=item aio_move $srcpath, $dstpath, $callback->($status)
1035
1036Try to move the I<file> (directories not supported as either source or
1037destination) from C<$srcpath> to C<$dstpath> and call the callback with
1038a status of C<0> (ok) or C<-1> (error, see C<$!>).
1039
1040This is a composite request that tries to rename(2) the file first; if
1041rename fails with C<EXDEV>, it copies the file with C<aio_copy> and, if
1042that is successful, unlinks the C<$srcpath>.
1043
1044=cut
1045
1046sub aio_move($$;$) {
1047 my ($src, $dst, $cb) = @_;
1048
1049 my $pri = aioreq_pri;
1050 my $grp = aio_group $cb;
1051
1052 aioreq_pri $pri;
1053 add $grp aio_rename $src, $dst, sub {
1054 if ($_[0] && $! == EXDEV) {
1055 aioreq_pri $pri;
1056 add $grp aio_copy $src, $dst, sub {
1057 $grp->result ($_[0]);
1058
1059 unless ($_[0]) {
1060 aioreq_pri $pri;
1061 add $grp aio_unlink $src;
1062 }
1063 };
1064 } else {
1065 $grp->result ($_[0]);
1066 }
1067 };
1068
1069 $grp
1070}
1071
1072=item aio_scandir $pathname, $maxreq, $callback->($dirs, $nondirs)
1073
1074Scans a directory (similar to C<aio_readdir>) but additionally tries to
1075efficiently separate the entries of directory C<$path> into two sets of
1076names, directories you can recurse into (directories), and ones you cannot
1077recurse into (everything else, including symlinks to directories).
1078
1079C<aio_scandir> is a composite request that creates of many sub requests_
1080C<$maxreq> specifies the maximum number of outstanding aio requests that
1081this function generates. If it is C<< <= 0 >>, then a suitable default
1082will be chosen (currently 4).
1083
1084On error, the callback is called without arguments, otherwise it receives
1085two array-refs with path-relative entry names.
1086
1087Example:
1088
1089 aio_scandir $dir, 0, sub {
1090 my ($dirs, $nondirs) = @_;
1091 print "real directories: @$dirs\n";
1092 print "everything else: @$nondirs\n";
1093 };
1094
1095Implementation notes.
1096
1097The C<aio_readdir> cannot be avoided, but C<stat()>'ing every entry can.
1098
1099If readdir returns file type information, then this is used directly to
1100find directories.
1101
1102Otherwise, after reading the directory, the modification time, size etc.
1103of the directory before and after the readdir is checked, and if they
1104match (and isn't the current time), the link count will be used to decide
1105how many entries are directories (if >= 2). Otherwise, no knowledge of the
1106number of subdirectories will be assumed.
1107
1108Then entries will be sorted into likely directories a non-initial dot
1109currently) and likely non-directories (see C<aio_readdirx>). Then every
1110entry plus an appended C</.> will be C<stat>'ed, likely directories first,
1111in order of their inode numbers. If that succeeds, it assumes that the
1112entry is a directory or a symlink to directory (which will be checked
1113separately). This is often faster than stat'ing the entry itself because
1114filesystems might detect the type of the entry without reading the inode
1115data (e.g. ext2fs filetype feature), even on systems that cannot return
1116the filetype information on readdir.
1117
1118If the known number of directories (link count - 2) has been reached, the
1119rest of the entries is assumed to be non-directories.
1120
1121This only works with certainty on POSIX (= UNIX) filesystems, which
1122fortunately are the vast majority of filesystems around.
1123
1124It will also likely work on non-POSIX filesystems with reduced efficiency
1125as those tend to return 0 or 1 as link counts, which disables the
1126directory counting heuristic.
1127
1128=cut
1129
1130sub aio_scandir($$;$) {
1131 my ($path, $maxreq, $cb) = @_;
1132
1133 my $pri = aioreq_pri;
1134
1135 my $grp = aio_group $cb;
1136
1137 $maxreq = 4 if $maxreq <= 0;
1138
1139 # get a wd object
1140 aioreq_pri $pri;
1141 add $grp aio_wd $path, sub {
1142 $_[0]
1143 or return $grp->result ();
1144
1145 my $wd = [shift, "."];
1146
1147 # stat once
1148 aioreq_pri $pri;
1149 add $grp aio_stat $wd, sub {
1150 return $grp->result () if $_[0];
1151 my $now = time;
1152 my $hash1 = join ":", (stat _)[0,1,3,7,9];
1153
1154 # read the directory entries
1155 aioreq_pri $pri;
1156 add $grp aio_readdirx $wd, READDIR_DIRS_FIRST, sub {
1157 my $entries = shift
1158 or return $grp->result ();
1159
1160 # stat the dir another time
1161 aioreq_pri $pri;
1162 add $grp aio_stat $wd, sub {
1163 my $hash2 = join ":", (stat _)[0,1,3,7,9];
1164
1165 my $ndirs;
1166
1167 # take the slow route if anything looks fishy
1168 if ($hash1 ne $hash2 or (stat _)[9] == $now) {
1169 $ndirs = -1;
1170 } else {
1171 # if nlink == 2, we are finished
1172 # for non-posix-fs's, we rely on nlink < 2
1173 $ndirs = (stat _)[3] - 2
1174 or return $grp->result ([], $entries);
1175 }
1176
1177 my (@dirs, @nondirs);
1178
1179 my $statgrp = add $grp aio_group sub {
1180 $grp->result (\@dirs, \@nondirs);
1181 };
1182
1183 limit $statgrp $maxreq;
1184 feed $statgrp sub {
1185 return unless @$entries;
1186 my $entry = shift @$entries;
1187
1188 aioreq_pri $pri;
1189 $wd->[1] = "$entry/.";
1190 add $statgrp aio_stat $wd, sub {
1191 if ($_[0] < 0) {
1192 push @nondirs, $entry;
1193 } else {
1194 # need to check for real directory
1195 aioreq_pri $pri;
1196 $wd->[1] = $entry;
1197 add $statgrp aio_lstat $wd, sub {
1198 if (-d _) {
1199 push @dirs, $entry;
1200
1201 unless (--$ndirs) {
1202 push @nondirs, @$entries;
1203 feed $statgrp;
1204 }
1205 } else {
1206 push @nondirs, $entry;
1207 }
1208 }
1209 }
1210 };
1211 };
1212 };
1213 };
1214 };
1215 };
1216
1217 $grp
1218}
1219
1220=item aio_rmtree $pathname, $callback->($status)
1221
1222Delete a directory tree starting (and including) C<$path>, return the
1223status of the final C<rmdir> only. This is a composite request that
1224uses C<aio_scandir> to recurse into and rmdir directories, and unlink
1225everything else.
1226
1227=cut
1228
1229sub aio_rmtree;
1230sub aio_rmtree($;$) {
1231 my ($path, $cb) = @_;
1232
1233 my $pri = aioreq_pri;
1234 my $grp = aio_group $cb;
1235
1236 aioreq_pri $pri;
1237 add $grp aio_scandir $path, 0, sub {
1238 my ($dirs, $nondirs) = @_;
1239
1240 my $dirgrp = aio_group sub {
1241 add $grp aio_rmdir $path, sub {
1242 $grp->result ($_[0]);
1243 };
1244 };
1245
1246 (aioreq_pri $pri), add $dirgrp aio_rmtree "$path/$_" for @$dirs;
1247 (aioreq_pri $pri), add $dirgrp aio_unlink "$path/$_" for @$nondirs;
1248
1249 add $grp $dirgrp;
1250 };
1251
1252 $grp
1253}
1254
1255=item aio_fcntl $fh, $cmd, $arg, $callback->($status)
1256
1257=item aio_ioctl $fh, $request, $buf, $callback->($status)
1258
1259These work just like the C<fcntl> and C<ioctl> built-in functions, except
1260they execute asynchronously and pass the return value to the callback.
1261
1262Both calls can be used for a lot of things, some of which make more sense
1263to run asynchronously in their own thread, while some others make less
1264sense. For example, calls that block waiting for external events, such
1265as locking, will also lock down an I/O thread while it is waiting, which
1266can deadlock the whole I/O system. At the same time, there might be no
1267alternative to using a thread to wait.
1268
1269So in general, you should only use these calls for things that do
1270(filesystem) I/O, not for things that wait for other events (network,
1271other processes), although if you are careful and know what you are doing,
1272you still can.
1273
1274The following constants are available (missing ones are, as usual C<0>):
1275
1276C<FIFREEZE>, C<FITHAW>, C<FITRIM>, C<FICLONE>, C<FICLONERANGE>, C<FIDEDUPERANGE>.
1277
1278C<FS_IOC_GETFLAGS>, C<FS_IOC_SETFLAGS>, C<FS_IOC_GETVERSION>, C<FS_IOC_SETVERSION>,
1279C<FS_IOC_FIEMAP>.
1280
1281C<FS_IOC_FSGETXATTR>, C<FS_IOC_FSSETXATTR>, C<FS_IOC_SET_ENCRYPTION_POLICY>,
1282C<FS_IOC_GET_ENCRYPTION_PWSALT>, C<FS_IOC_GET_ENCRYPTION_POLICY>, C<FS_KEY_DESCRIPTOR_SIZE>.
1283
1284C<FS_SECRM_FL>, C<FS_UNRM_FL>, C<FS_COMPR_FL>, C<FS_SYNC_FL>, C<FS_IMMUTABLE_FL>,
1285C<FS_APPEND_FL>, C<FS_NODUMP_FL>, C<FS_NOATIME_FL>, C<FS_DIRTY_FL>,
1286C<FS_COMPRBLK_FL>, C<FS_NOCOMP_FL>, C<FS_ENCRYPT_FL>, C<FS_BTREE_FL>,
1287C<FS_INDEX_FL>, C<FS_JOURNAL_DATA_FL>, C<FS_NOTAIL_FL>, C<FS_DIRSYNC_FL>, C<FS_TOPDIR_FL>,
1288C<FS_FL_USER_MODIFIABLE>.
1289
1290C<FS_XFLAG_REALTIME>, C<FS_XFLAG_PREALLOC>, C<FS_XFLAG_IMMUTABLE>, C<FS_XFLAG_APPEND>,
1291C<FS_XFLAG_SYNC>, C<FS_XFLAG_NOATIME>, C<FS_XFLAG_NODUMP>, C<FS_XFLAG_RTINHERIT>,
1292C<FS_XFLAG_PROJINHERIT>, C<FS_XFLAG_NOSYMLINKS>, C<FS_XFLAG_EXTSIZE>, C<FS_XFLAG_EXTSZINHERIT>,
1293C<FS_XFLAG_NODEFRAG>, C<FS_XFLAG_FILESTREAM>, C<FS_XFLAG_DAX>, C<FS_XFLAG_HASATTR>,
1294
1295=item aio_sync $callback->($status)
1296
1297Asynchronously call sync and call the callback when finished.
1298
209=item aio_fsync $fh, $callback 1299=item aio_fsync $fh, $callback->($status)
210 1300
211Asynchronously call fsync on the given filehandle and call the callback 1301Asynchronously call fsync on the given filehandle and call the callback
212with the fsync result code. 1302with the fsync result code.
213 1303
214=item aio_fdatasync $fh, $callback 1304=item aio_fdatasync $fh, $callback->($status)
215 1305
216Asynchronously call fdatasync on the given filehandle and call the 1306Asynchronously call fdatasync on the given filehandle and call the
217callback with the fdatasync result code. 1307callback with the fdatasync result code.
218 1308
1309If this call isn't available because your OS lacks it or it couldn't be
1310detected, it will be emulated by calling C<fsync> instead.
1311
1312=item aio_syncfs $fh, $callback->($status)
1313
1314Asynchronously call the syncfs syscall to sync the filesystem associated
1315to the given filehandle and call the callback with the syncfs result
1316code. If syncfs is not available, calls sync(), but returns C<-1> and sets
1317errno to C<ENOSYS> nevertheless.
1318
1319=item aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
1320
1321Sync the data portion of the file specified by C<$offset> and C<$length>
1322to disk (but NOT the metadata), by calling the Linux-specific
1323sync_file_range call. If sync_file_range is not available or it returns
1324ENOSYS, then fdatasync or fsync is being substituted.
1325
1326C<$flags> can be a combination of C<IO::AIO::SYNC_FILE_RANGE_WAIT_BEFORE>,
1327C<IO::AIO::SYNC_FILE_RANGE_WRITE> and
1328C<IO::AIO::SYNC_FILE_RANGE_WAIT_AFTER>: refer to the sync_file_range
1329manpage for details.
1330
1331=item aio_pathsync $pathname, $callback->($status)
1332
1333This request tries to open, fsync and close the given path. This is a
1334composite request intended to sync directories after directory operations
1335(E.g. rename). This might not work on all operating systems or have any
1336specific effect, but usually it makes sure that directory changes get
1337written to disc. It works for anything that can be opened for read-only,
1338not just directories.
1339
1340Future versions of this function might fall back to other methods when
1341C<fsync> on the directory fails (such as calling C<sync>).
1342
1343Passes C<0> when everything went ok, and C<-1> on error.
1344
219=cut 1345=cut
220 1346
221min_parallel 4; 1347sub aio_pathsync($;$) {
1348 my ($path, $cb) = @_;
222 1349
223END { 1350 my $pri = aioreq_pri;
224 max_parallel 0; 1351 my $grp = aio_group $cb;
1352
1353 aioreq_pri $pri;
1354 add $grp aio_open $path, O_RDONLY, 0, sub {
1355 my ($fh) = @_;
1356 if ($fh) {
1357 aioreq_pri $pri;
1358 add $grp aio_fsync $fh, sub {
1359 $grp->result ($_[0]);
1360
1361 aioreq_pri $pri;
1362 add $grp aio_close $fh;
1363 };
1364 } else {
1365 $grp->result (-1);
1366 }
1367 };
1368
1369 $grp
225} 1370}
226 1371
1372=item aio_msync $scalar, $offset = 0, $length = undef, flags = MS_SYNC, $callback->($status)
1373
1374This is a rather advanced IO::AIO call, which only works on mmap(2)ed
1375scalars (see the C<IO::AIO::mmap> function, although it also works on data
1376scalars managed by the L<Sys::Mmap> or L<Mmap> modules, note that the
1377scalar must only be modified in-place while an aio operation is pending on
1378it).
1379
1380It calls the C<msync> function of your OS, if available, with the memory
1381area starting at C<$offset> in the string and ending C<$length> bytes
1382later. If C<$length> is negative, counts from the end, and if C<$length>
1383is C<undef>, then it goes till the end of the string. The flags can be
1384either C<IO::AIO::MS_ASYNC> or C<IO::AIO::MS_SYNC>, plus an optional
1385C<IO::AIO::MS_INVALIDATE>.
1386
1387=item aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
1388
1389This is a rather advanced IO::AIO call, which works best on mmap(2)ed
1390scalars.
1391
1392It touches (reads or writes) all memory pages in the specified
1393range inside the scalar. All caveats and parameters are the same
1394as for C<aio_msync>, above, except for flags, which must be either
1395C<0> (which reads all pages and ensures they are instantiated) or
1396C<IO::AIO::MT_MODIFY>, which modifies the memory pages (by reading and
1397writing an octet from it, which dirties the page).
1398
1399=item aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status)
1400
1401This is a rather advanced IO::AIO call, which works best on mmap(2)ed
1402scalars.
1403
1404It reads in all the pages of the underlying storage into memory (if any)
1405and locks them, so they are not getting swapped/paged out or removed.
1406
1407If C<$length> is undefined, then the scalar will be locked till the end.
1408
1409On systems that do not implement C<mlock>, this function returns C<-1>
1410and sets errno to C<ENOSYS>.
1411
1412Note that the corresponding C<munlock> is synchronous and is
1413documented under L<MISCELLANEOUS FUNCTIONS>.
1414
1415Example: open a file, mmap and mlock it - both will be undone when
1416C<$data> gets destroyed.
1417
1418 open my $fh, "<", $path or die "$path: $!";
1419 my $data;
1420 IO::AIO::mmap $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh;
1421 aio_mlock $data; # mlock in background
1422
1423=item aio_mlockall $flags, $callback->($status)
1424
1425Calls the C<mlockall> function with the given C<$flags> (a combination of
1426C<IO::AIO::MCL_CURRENT> and C<IO::AIO::MCL_FUTURE>).
1427
1428On systems that do not implement C<mlockall>, this function returns C<-1>
1429and sets errno to C<ENOSYS>.
1430
1431Note that the corresponding C<munlockall> is synchronous and is
1432documented under L<MISCELLANEOUS FUNCTIONS>.
1433
1434Example: asynchronously lock all current and future pages into memory.
1435
1436 aio_mlockall IO::AIO::MCL_FUTURE;
1437
1438=item aio_fiemap $fh, $start, $length, $flags, $count, $cb->(\@extents)
1439
1440Queries the extents of the given file (by calling the Linux C<FIEMAP>
1441ioctl, see L<http://cvs.schmorp.de/IO-AIO/doc/fiemap.txt> for details). If
1442the ioctl is not available on your OS, then this request will fail with
1443C<ENOSYS>.
1444
1445C<$start> is the starting offset to query extents for, C<$length> is the
1446size of the range to query - if it is C<undef>, then the whole file will
1447be queried.
1448
1449C<$flags> is a combination of flags (C<IO::AIO::FIEMAP_FLAG_SYNC> or
1450C<IO::AIO::FIEMAP_FLAG_XATTR> - C<IO::AIO::FIEMAP_FLAGS_COMPAT> is also
1451exported), and is normally C<0> or C<IO::AIO::FIEMAP_FLAG_SYNC> to query
1452the data portion.
1453
1454C<$count> is the maximum number of extent records to return. If it is
1455C<undef>, then IO::AIO queries all extents of the range. As a very special
1456case, if it is C<0>, then the callback receives the number of extents
1457instead of the extents themselves (which is unreliable, see below).
1458
1459If an error occurs, the callback receives no arguments. The special
1460C<errno> value C<IO::AIO::EBADR> is available to test for flag errors.
1461
1462Otherwise, the callback receives an array reference with extent
1463structures. Each extent structure is an array reference itself, with the
1464following members:
1465
1466 [$logical, $physical, $length, $flags]
1467
1468Flags is any combination of the following flag values (typically either C<0>
1469or C<IO::AIO::FIEMAP_EXTENT_LAST> (1)):
1470
1471C<IO::AIO::FIEMAP_EXTENT_LAST>, C<IO::AIO::FIEMAP_EXTENT_UNKNOWN>,
1472C<IO::AIO::FIEMAP_EXTENT_DELALLOC>, C<IO::AIO::FIEMAP_EXTENT_ENCODED>,
1473C<IO::AIO::FIEMAP_EXTENT_DATA_ENCRYPTED>, C<IO::AIO::FIEMAP_EXTENT_NOT_ALIGNED>,
1474C<IO::AIO::FIEMAP_EXTENT_DATA_INLINE>, C<IO::AIO::FIEMAP_EXTENT_DATA_TAIL>,
1475C<IO::AIO::FIEMAP_EXTENT_UNWRITTEN>, C<IO::AIO::FIEMAP_EXTENT_MERGED> or
1476C<IO::AIO::FIEMAP_EXTENT_SHARED>.
1477
1478At the time of this writing (Linux 3.2), this requets is unreliable unless
1479C<$count> is C<undef>, as the kernel has all sorts of bugs preventing
1480it to return all extents of a range for files with large number of
1481extents. The code works around all these issues if C<$count> is undef.
1482
1483=item aio_group $callback->(...)
1484
1485This is a very special aio request: Instead of doing something, it is a
1486container for other aio requests, which is useful if you want to bundle
1487many requests into a single, composite, request with a definite callback
1488and the ability to cancel the whole request with its subrequests.
1489
1490Returns an object of class L<IO::AIO::GRP>. See its documentation below
1491for more info.
1492
1493Example:
1494
1495 my $grp = aio_group sub {
1496 print "all stats done\n";
1497 };
1498
1499 add $grp
1500 (aio_stat ...),
1501 (aio_stat ...),
1502 ...;
1503
1504=item aio_nop $callback->()
1505
1506This is a special request - it does nothing in itself and is only used for
1507side effects, such as when you want to add a dummy request to a group so
1508that finishing the requests in the group depends on executing the given
1509code.
1510
1511While this request does nothing, it still goes through the execution
1512phase and still requires a worker thread. Thus, the callback will not
1513be executed immediately but only after other requests in the queue have
1514entered their execution phase. This can be used to measure request
1515latency.
1516
1517=item IO::AIO::aio_busy $fractional_seconds, $callback->() *NOT EXPORTED*
1518
1519Mainly used for debugging and benchmarking, this aio request puts one of
1520the request workers to sleep for the given time.
1521
1522While it is theoretically handy to have simple I/O scheduling requests
1523like sleep and file handle readable/writable, the overhead this creates is
1524immense (it blocks a thread for a long time) so do not use this function
1525except to put your application under artificial I/O pressure.
1526
1527=back
1528
1529
1530=head2 IO::AIO::WD - multiple working directories
1531
1532Your process only has one current working directory, which is used by all
1533threads. This makes it hard to use relative paths (some other component
1534could call C<chdir> at any time, and it is hard to control when the path
1535will be used by IO::AIO).
1536
1537One solution for this is to always use absolute paths. This usually works,
1538but can be quite slow (the kernel has to walk the whole path on every
1539access), and can also be a hassle to implement.
1540
1541Newer POSIX systems have a number of functions (openat, fdopendir,
1542futimensat and so on) that make it possible to specify working directories
1543per operation.
1544
1545For portability, and because the clowns who "designed", or shall I write,
1546perpetrated this new interface were obviously half-drunk, this abstraction
1547cannot be perfect, though.
1548
1549IO::AIO allows you to convert directory paths into a so-called IO::AIO::WD
1550object. This object stores the canonicalised, absolute version of the
1551path, and on systems that allow it, also a directory file descriptor.
1552
1553Everywhere where a pathname is accepted by IO::AIO (e.g. in C<aio_stat>
1554or C<aio_unlink>), one can specify an array reference with an IO::AIO::WD
1555object and a pathname instead (or the IO::AIO::WD object alone, which
1556gets interpreted as C<[$wd, "."]>). If the pathname is absolute, the
1557IO::AIO::WD object is ignored, otherwise the pathname is resolved relative
1558to that IO::AIO::WD object.
1559
1560For example, to get a wd object for F</etc> and then stat F<passwd>
1561inside, you would write:
1562
1563 aio_wd "/etc", sub {
1564 my $etcdir = shift;
1565
1566 # although $etcdir can be undef on error, there is generally no reason
1567 # to check for errors here, as aio_stat will fail with ENOENT
1568 # when $etcdir is undef.
1569
1570 aio_stat [$etcdir, "passwd"], sub {
1571 # yay
1572 };
1573 };
1574
1575The fact that C<aio_wd> is a request and not a normal function shows that
1576creating an IO::AIO::WD object is itself a potentially blocking operation,
1577which is why it is done asynchronously.
1578
1579To stat the directory obtained with C<aio_wd> above, one could write
1580either of the following three request calls:
1581
1582 aio_lstat "/etc" , sub { ... # pathname as normal string
1583 aio_lstat [$wd, "."], sub { ... # "." relative to $wd (i.e. $wd itself)
1584 aio_lstat $wd , sub { ... # shorthand for the previous
1585
1586As with normal pathnames, IO::AIO keeps a copy of the working directory
1587object and the pathname string, so you could write the following without
1588causing any issues due to C<$path> getting reused:
1589
1590 my $path = [$wd, undef];
1591
1592 for my $name (qw(abc def ghi)) {
1593 $path->[1] = $name;
1594 aio_stat $path, sub {
1595 # ...
1596 };
1597 }
1598
1599There are some caveats: when directories get renamed (or deleted), the
1600pathname string doesn't change, so will point to the new directory (or
1601nowhere at all), while the directory fd, if available on the system,
1602will still point to the original directory. Most functions accepting a
1603pathname will use the directory fd on newer systems, and the string on
1604older systems. Some functions (such as realpath) will always rely on the
1605string form of the pathname.
1606
1607So this functionality is mainly useful to get some protection against
1608C<chdir>, to easily get an absolute path out of a relative path for future
1609reference, and to speed up doing many operations in the same directory
1610(e.g. when stat'ing all files in a directory).
1611
1612The following functions implement this working directory abstraction:
1613
1614=over 4
1615
1616=item aio_wd $pathname, $callback->($wd)
1617
1618Asynchonously canonicalise the given pathname and convert it to an
1619IO::AIO::WD object representing it. If possible and supported on the
1620system, also open a directory fd to speed up pathname resolution relative
1621to this working directory.
1622
1623If something goes wrong, then C<undef> is passwd to the callback instead
1624of a working directory object and C<$!> is set appropriately. Since
1625passing C<undef> as working directory component of a pathname fails the
1626request with C<ENOENT>, there is often no need for error checking in the
1627C<aio_wd> callback, as future requests using the value will fail in the
1628expected way.
1629
1630=item IO::AIO::CWD
1631
1632This is a compiletime constant (object) that represents the process
1633current working directory.
1634
1635Specifying this object as working directory object for a pathname is as if
1636the pathname would be specified directly, without a directory object. For
1637example, these calls are functionally identical:
1638
1639 aio_stat "somefile", sub { ... };
1640 aio_stat [IO::AIO::CWD, "somefile"], sub { ... };
1641
1642=back
1643
1644To recover the path associated with an IO::AIO::WD object, you can use
1645C<aio_realpath>:
1646
1647 aio_realpath $wd, sub {
1648 warn "path is $_[0]\n";
1649 };
1650
1651Currently, C<aio_statvfs> always, and C<aio_rename> and C<aio_rmdir>
1652sometimes, fall back to using an absolue path.
1653
1654=head2 IO::AIO::REQ CLASS
1655
1656All non-aggregate C<aio_*> functions return an object of this class when
1657called in non-void context.
1658
1659=over 4
1660
1661=item cancel $req
1662
1663Cancels the request, if possible. Has the effect of skipping execution
1664when entering the B<execute> state and skipping calling the callback when
1665entering the the B<result> state, but will leave the request otherwise
1666untouched (with the exception of readdir). That means that requests that
1667currently execute will not be stopped and resources held by the request
1668will not be freed prematurely.
1669
1670=item cb $req $callback->(...)
1671
1672Replace (or simply set) the callback registered to the request.
1673
1674=back
1675
1676=head2 IO::AIO::GRP CLASS
1677
1678This class is a subclass of L<IO::AIO::REQ>, so all its methods apply to
1679objects of this class, too.
1680
1681A IO::AIO::GRP object is a special request that can contain multiple other
1682aio requests.
1683
1684You create one by calling the C<aio_group> constructing function with a
1685callback that will be called when all contained requests have entered the
1686C<done> state:
1687
1688 my $grp = aio_group sub {
1689 print "all requests are done\n";
1690 };
1691
1692You add requests by calling the C<add> method with one or more
1693C<IO::AIO::REQ> objects:
1694
1695 $grp->add (aio_unlink "...");
1696
1697 add $grp aio_stat "...", sub {
1698 $_[0] or return $grp->result ("error");
1699
1700 # add another request dynamically, if first succeeded
1701 add $grp aio_open "...", sub {
1702 $grp->result ("ok");
1703 };
1704 };
1705
1706This makes it very easy to create composite requests (see the source of
1707C<aio_move> for an application) that work and feel like simple requests.
1708
1709=over 4
1710
1711=item * The IO::AIO::GRP objects will be cleaned up during calls to
1712C<IO::AIO::poll_cb>, just like any other request.
1713
1714=item * They can be canceled like any other request. Canceling will cancel not
1715only the request itself, but also all requests it contains.
1716
1717=item * They can also can also be added to other IO::AIO::GRP objects.
1718
1719=item * You must not add requests to a group from within the group callback (or
1720any later time).
1721
1722=back
1723
1724Their lifetime, simplified, looks like this: when they are empty, they
1725will finish very quickly. If they contain only requests that are in the
1726C<done> state, they will also finish. Otherwise they will continue to
1727exist.
1728
1729That means after creating a group you have some time to add requests
1730(precisely before the callback has been invoked, which is only done within
1731the C<poll_cb>). And in the callbacks of those requests, you can add
1732further requests to the group. And only when all those requests have
1733finished will the the group itself finish.
1734
1735=over 4
1736
1737=item add $grp ...
1738
1739=item $grp->add (...)
1740
1741Add one or more requests to the group. Any type of L<IO::AIO::REQ> can
1742be added, including other groups, as long as you do not create circular
1743dependencies.
1744
1745Returns all its arguments.
1746
1747=item $grp->cancel_subs
1748
1749Cancel all subrequests and clears any feeder, but not the group request
1750itself. Useful when you queued a lot of events but got a result early.
1751
1752The group request will finish normally (you cannot add requests to the
1753group).
1754
1755=item $grp->result (...)
1756
1757Set the result value(s) that will be passed to the group callback when all
1758subrequests have finished and set the groups errno to the current value
1759of errno (just like calling C<errno> without an error number). By default,
1760no argument will be passed and errno is zero.
1761
1762=item $grp->errno ([$errno])
1763
1764Sets the group errno value to C<$errno>, or the current value of errno
1765when the argument is missing.
1766
1767Every aio request has an associated errno value that is restored when
1768the callback is invoked. This method lets you change this value from its
1769default (0).
1770
1771Calling C<result> will also set errno, so make sure you either set C<$!>
1772before the call to C<result>, or call c<errno> after it.
1773
1774=item feed $grp $callback->($grp)
1775
1776Sets a feeder/generator on this group: every group can have an attached
1777generator that generates requests if idle. The idea behind this is that,
1778although you could just queue as many requests as you want in a group,
1779this might starve other requests for a potentially long time. For example,
1780C<aio_scandir> might generate hundreds of thousands of C<aio_stat>
1781requests, delaying any later requests for a long time.
1782
1783To avoid this, and allow incremental generation of requests, you can
1784instead a group and set a feeder on it that generates those requests. The
1785feed callback will be called whenever there are few enough (see C<limit>,
1786below) requests active in the group itself and is expected to queue more
1787requests.
1788
1789The feed callback can queue as many requests as it likes (i.e. C<add> does
1790not impose any limits).
1791
1792If the feed does not queue more requests when called, it will be
1793automatically removed from the group.
1794
1795If the feed limit is C<0> when this method is called, it will be set to
1796C<2> automatically.
1797
1798Example:
1799
1800 # stat all files in @files, but only ever use four aio requests concurrently:
1801
1802 my $grp = aio_group sub { print "finished\n" };
1803 limit $grp 4;
1804 feed $grp sub {
1805 my $file = pop @files
1806 or return;
1807
1808 add $grp aio_stat $file, sub { ... };
1809 };
1810
1811=item limit $grp $num
1812
1813Sets the feeder limit for the group: The feeder will be called whenever
1814the group contains less than this many requests.
1815
1816Setting the limit to C<0> will pause the feeding process.
1817
1818The default value for the limit is C<0>, but note that setting a feeder
1819automatically bumps it up to C<2>.
1820
1821=back
1822
1823=head2 SUPPORT FUNCTIONS
1824
1825=head3 EVENT PROCESSING AND EVENT LOOP INTEGRATION
1826
1827=over 4
1828
1829=item $fileno = IO::AIO::poll_fileno
1830
1831Return the I<request result pipe file descriptor>. This filehandle must be
1832polled for reading by some mechanism outside this module (e.g. EV, Glib,
1833select and so on, see below or the SYNOPSIS). If the pipe becomes readable
1834you have to call C<poll_cb> to check the results.
1835
1836See C<poll_cb> for an example.
1837
1838=item IO::AIO::poll_cb
1839
1840Process some requests that have reached the result phase (i.e. they have
1841been executed but the results are not yet reported). You have to call
1842this "regularly" to finish outstanding requests.
1843
1844Returns C<0> if all events could be processed (or there were no
1845events to process), or C<-1> if it returned earlier for whatever
1846reason. Returns immediately when no events are outstanding. The amount
1847of events processed depends on the settings of C<IO::AIO::max_poll_req>,
1848C<IO::AIO::max_poll_time> and C<IO::AIO::max_outstanding>.
1849
1850If not all requests were processed for whatever reason, the poll file
1851descriptor will still be ready when C<poll_cb> returns, so normally you
1852don't have to do anything special to have it called later.
1853
1854Apart from calling C<IO::AIO::poll_cb> when the event filehandle becomes
1855ready, it can be beneficial to call this function from loops which submit
1856a lot of requests, to make sure the results get processed when they become
1857available and not just when the loop is finished and the event loop takes
1858over again. This function returns very fast when there are no outstanding
1859requests.
1860
1861Example: Install an Event watcher that automatically calls
1862IO::AIO::poll_cb with high priority (more examples can be found in the
1863SYNOPSIS section, at the top of this document):
1864
1865 Event->io (fd => IO::AIO::poll_fileno,
1866 poll => 'r', async => 1,
1867 cb => \&IO::AIO::poll_cb);
1868
1869=item IO::AIO::poll_wait
1870
1871Wait until either at least one request is in the result phase or no
1872requests are outstanding anymore.
1873
1874This is useful if you want to synchronously wait for some requests to
1875become ready, without actually handling them.
1876
1877See C<nreqs> for an example.
1878
1879=item IO::AIO::poll
1880
1881Waits until some requests have been handled.
1882
1883Returns the number of requests processed, but is otherwise strictly
1884equivalent to:
1885
1886 IO::AIO::poll_wait, IO::AIO::poll_cb
1887
1888=item IO::AIO::flush
1889
1890Wait till all outstanding AIO requests have been handled.
1891
1892Strictly equivalent to:
1893
1894 IO::AIO::poll_wait, IO::AIO::poll_cb
1895 while IO::AIO::nreqs;
1896
1897=item IO::AIO::max_poll_reqs $nreqs
1898
1899=item IO::AIO::max_poll_time $seconds
1900
1901These set the maximum number of requests (default C<0>, meaning infinity)
1902that are being processed by C<IO::AIO::poll_cb> in one call, respectively
1903the maximum amount of time (default C<0>, meaning infinity) spent in
1904C<IO::AIO::poll_cb> to process requests (more correctly the mininum amount
1905of time C<poll_cb> is allowed to use).
1906
1907Setting C<max_poll_time> to a non-zero value creates an overhead of one
1908syscall per request processed, which is not normally a problem unless your
1909callbacks are really really fast or your OS is really really slow (I am
1910not mentioning Solaris here). Using C<max_poll_reqs> incurs no overhead.
1911
1912Setting these is useful if you want to ensure some level of
1913interactiveness when perl is not fast enough to process all requests in
1914time.
1915
1916For interactive programs, values such as C<0.01> to C<0.1> should be fine.
1917
1918Example: Install an Event watcher that automatically calls
1919IO::AIO::poll_cb with low priority, to ensure that other parts of the
1920program get the CPU sometimes even under high AIO load.
1921
1922 # try not to spend much more than 0.1s in poll_cb
1923 IO::AIO::max_poll_time 0.1;
1924
1925 # use a low priority so other tasks have priority
1926 Event->io (fd => IO::AIO::poll_fileno,
1927 poll => 'r', nice => 1,
1928 cb => &IO::AIO::poll_cb);
1929
1930=back
1931
1932=head3 CONTROLLING THE NUMBER OF THREADS
1933
1934=over
1935
1936=item IO::AIO::min_parallel $nthreads
1937
1938Set the minimum number of AIO threads to C<$nthreads>. The current
1939default is C<8>, which means eight asynchronous operations can execute
1940concurrently at any one time (the number of outstanding requests,
1941however, is unlimited).
1942
1943IO::AIO starts threads only on demand, when an AIO request is queued and
1944no free thread exists. Please note that queueing up a hundred requests can
1945create demand for a hundred threads, even if it turns out that everything
1946is in the cache and could have been processed faster by a single thread.
1947
1948It is recommended to keep the number of threads relatively low, as some
1949Linux kernel versions will scale negatively with the number of threads
1950(higher parallelity => MUCH higher latency). With current Linux 2.6
1951versions, 4-32 threads should be fine.
1952
1953Under most circumstances you don't need to call this function, as the
1954module selects a default that is suitable for low to moderate load.
1955
1956=item IO::AIO::max_parallel $nthreads
1957
1958Sets the maximum number of AIO threads to C<$nthreads>. If more than the
1959specified number of threads are currently running, this function kills
1960them. This function blocks until the limit is reached.
1961
1962While C<$nthreads> are zero, aio requests get queued but not executed
1963until the number of threads has been increased again.
1964
1965This module automatically runs C<max_parallel 0> at program end, to ensure
1966that all threads are killed and that there are no outstanding requests.
1967
1968Under normal circumstances you don't need to call this function.
1969
1970=item IO::AIO::max_idle $nthreads
1971
1972Limit the number of threads (default: 4) that are allowed to idle
1973(i.e., threads that did not get a request to process within the idle
1974timeout (default: 10 seconds). That means if a thread becomes idle while
1975C<$nthreads> other threads are also idle, it will free its resources and
1976exit.
1977
1978This is useful when you allow a large number of threads (e.g. 100 or 1000)
1979to allow for extremely high load situations, but want to free resources
1980under normal circumstances (1000 threads can easily consume 30MB of RAM).
1981
1982The default is probably ok in most situations, especially if thread
1983creation is fast. If thread creation is very slow on your system you might
1984want to use larger values.
1985
1986=item IO::AIO::idle_timeout $seconds
1987
1988Sets the minimum idle timeout (default 10) after which worker threads are
1989allowed to exit. SEe C<IO::AIO::max_idle>.
1990
1991=item IO::AIO::max_outstanding $maxreqs
1992
1993Sets the maximum number of outstanding requests to C<$nreqs>. If
1994you do queue up more than this number of requests, the next call to
1995C<IO::AIO::poll_cb> (and other functions calling C<poll_cb>, such as
1996C<IO::AIO::flush> or C<IO::AIO::poll>) will block until the limit is no
1997longer exceeded.
1998
1999In other words, this setting does not enforce a queue limit, but can be
2000used to make poll functions block if the limit is exceeded.
2001
2002This is a very bad function to use in interactive programs because it
2003blocks, and a bad way to reduce concurrency because it is inexact: Better
2004use an C<aio_group> together with a feed callback.
2005
2006Its main use is in scripts without an event loop - when you want to stat
2007a lot of files, you can write somehting like this:
2008
2009 IO::AIO::max_outstanding 32;
2010
2011 for my $path (...) {
2012 aio_stat $path , ...;
2013 IO::AIO::poll_cb;
2014 }
2015
2016 IO::AIO::flush;
2017
2018The call to C<poll_cb> inside the loop will normally return instantly, but
2019as soon as more thna C<32> reqeusts are in-flight, it will block until
2020some requests have been handled. This keeps the loop from pushing a large
2021number of C<aio_stat> requests onto the queue.
2022
2023The default value for C<max_outstanding> is very large, so there is no
2024practical limit on the number of outstanding requests.
2025
2026=back
2027
2028=head3 STATISTICAL INFORMATION
2029
2030=over
2031
2032=item IO::AIO::nreqs
2033
2034Returns the number of requests currently in the ready, execute or pending
2035states (i.e. for which their callback has not been invoked yet).
2036
2037Example: wait till there are no outstanding requests anymore:
2038
2039 IO::AIO::poll_wait, IO::AIO::poll_cb
2040 while IO::AIO::nreqs;
2041
2042=item IO::AIO::nready
2043
2044Returns the number of requests currently in the ready state (not yet
2045executed).
2046
2047=item IO::AIO::npending
2048
2049Returns the number of requests currently in the pending state (executed,
2050but not yet processed by poll_cb).
2051
2052=back
2053
2054=head3 MISCELLANEOUS FUNCTIONS
2055
2056IO::AIO implements some functions that are useful when you want to use
2057some "Advanced I/O" function not available to in Perl, without going the
2058"Asynchronous I/O" route. Many of these have an asynchronous C<aio_*>
2059counterpart.
2060
2061=over 4
2062
2063=item IO::AIO::sendfile $ofh, $ifh, $offset, $count
2064
2065Calls the C<eio_sendfile_sync> function, which is like C<aio_sendfile>,
2066but is blocking (this makes most sense if you know the input data is
2067likely cached already and the output filehandle is set to non-blocking
2068operations).
2069
2070Returns the number of bytes copied, or C<-1> on error.
2071
2072=item IO::AIO::fadvise $fh, $offset, $len, $advice
2073
2074Simply calls the C<posix_fadvise> function (see its
2075manpage for details). The following advice constants are
2076available: C<IO::AIO::FADV_NORMAL>, C<IO::AIO::FADV_SEQUENTIAL>,
2077C<IO::AIO::FADV_RANDOM>, C<IO::AIO::FADV_NOREUSE>,
2078C<IO::AIO::FADV_WILLNEED>, C<IO::AIO::FADV_DONTNEED>.
2079
2080On systems that do not implement C<posix_fadvise>, this function returns
2081ENOSYS, otherwise the return value of C<posix_fadvise>.
2082
2083=item IO::AIO::madvise $scalar, $offset, $len, $advice
2084
2085Simply calls the C<posix_madvise> function (see its
2086manpage for details). The following advice constants are
2087available: C<IO::AIO::MADV_NORMAL>, C<IO::AIO::MADV_SEQUENTIAL>,
2088C<IO::AIO::MADV_RANDOM>, C<IO::AIO::MADV_WILLNEED>, C<IO::AIO::MADV_DONTNEED>,
2089C<IO::AIO::MADV_FREE>.
2090
2091If C<$offset> is negative, counts from the end. If C<$length> is negative,
2092the remaining length of the C<$scalar> is used. If possible, C<$length>
2093will be reduced to fit into the C<$scalar>.
2094
2095On systems that do not implement C<posix_madvise>, this function returns
2096ENOSYS, otherwise the return value of C<posix_madvise>.
2097
2098=item IO::AIO::mprotect $scalar, $offset, $len, $protect
2099
2100Simply calls the C<mprotect> function on the preferably AIO::mmap'ed
2101$scalar (see its manpage for details). The following protect
2102constants are available: C<IO::AIO::PROT_NONE>, C<IO::AIO::PROT_READ>,
2103C<IO::AIO::PROT_WRITE>, C<IO::AIO::PROT_EXEC>.
2104
2105If C<$offset> is negative, counts from the end. If C<$length> is negative,
2106the remaining length of the C<$scalar> is used. If possible, C<$length>
2107will be reduced to fit into the C<$scalar>.
2108
2109On systems that do not implement C<mprotect>, this function returns
2110ENOSYS, otherwise the return value of C<mprotect>.
2111
2112=item IO::AIO::mmap $scalar, $length, $prot, $flags, $fh[, $offset]
2113
2114Memory-maps a file (or anonymous memory range) and attaches it to the
2115given C<$scalar>, which will act like a string scalar. Returns true on
2116success, and false otherwise.
2117
2118The scalar must exist, but its contents do not matter - this means you
2119cannot use a nonexistant array or hash element. When in doubt, C<undef>
2120the scalar first.
2121
2122The only operations allowed on the mmapped scalar are C<substr>/C<vec>,
2123which don't change the string length, and most read-only operations such
2124as copying it or searching it with regexes and so on.
2125
2126Anything else is unsafe and will, at best, result in memory leaks.
2127
2128The memory map associated with the C<$scalar> is automatically removed
2129when the C<$scalar> is undef'd or destroyed, or when the C<IO::AIO::mmap>
2130or C<IO::AIO::munmap> functions are called on it.
2131
2132This calls the C<mmap>(2) function internally. See your system's manual
2133page for details on the C<$length>, C<$prot> and C<$flags> parameters.
2134
2135The C<$length> must be larger than zero and smaller than the actual
2136filesize.
2137
2138C<$prot> is a combination of C<IO::AIO::PROT_NONE>, C<IO::AIO::PROT_EXEC>,
2139C<IO::AIO::PROT_READ> and/or C<IO::AIO::PROT_WRITE>,
2140
2141C<$flags> can be a combination of
2142C<IO::AIO::MAP_SHARED> or
2143C<IO::AIO::MAP_PRIVATE>,
2144or a number of system-specific flags (when not available, the are C<0>):
2145C<IO::AIO::MAP_ANONYMOUS> (which is set to C<MAP_ANON> if your system only provides this constant),
2146C<IO::AIO::MAP_LOCKED>,
2147C<IO::AIO::MAP_NORESERVE>,
2148C<IO::AIO::MAP_POPULATE>,
2149C<IO::AIO::MAP_NONBLOCK>,
2150C<IO::AIO::MAP_FIXED>,
2151C<IO::AIO::MAP_GROWSDOWN>,
2152C<IO::AIO::MAP_32BIT>,
2153C<IO::AIO::MAP_HUGETLB> or
2154C<IO::AIO::MAP_STACK>.
2155
2156If C<$fh> is C<undef>, then a file descriptor of C<-1> is passed.
2157
2158C<$offset> is the offset from the start of the file - it generally must be
2159a multiple of C<IO::AIO::PAGESIZE> and defaults to C<0>.
2160
2161Example:
2162
2163 use Digest::MD5;
2164 use IO::AIO;
2165
2166 open my $fh, "<verybigfile"
2167 or die "$!";
2168
2169 IO::AIO::mmap my $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh
2170 or die "verybigfile: $!";
2171
2172 my $fast_md5 = md5 $data;
2173
2174=item IO::AIO::munmap $scalar
2175
2176Removes a previous mmap and undefines the C<$scalar>.
2177
2178=item IO::AIO::munlock $scalar, $offset = 0, $length = undef
2179
2180Calls the C<munlock> function, undoing the effects of a previous
2181C<aio_mlock> call (see its description for details).
2182
2183=item IO::AIO::munlockall
2184
2185Calls the C<munlockall> function.
2186
2187On systems that do not implement C<munlockall>, this function returns
2188ENOSYS, otherwise the return value of C<munlockall>.
2189
2190=item IO::AIO::splice $r_fh, $r_off, $w_fh, $w_off, $length, $flags
2191
2192Calls the GNU/Linux C<splice(2)> syscall, if available. If C<$r_off> or
2193C<$w_off> are C<undef>, then C<NULL> is passed for these, otherwise they
2194should be the file offset.
2195
2196C<$r_fh> and C<$w_fh> should not refer to the same file, as splice might
2197silently corrupt the data in this case.
2198
2199The following symbol flag values are available: C<IO::AIO::SPLICE_F_MOVE>,
2200C<IO::AIO::SPLICE_F_NONBLOCK>, C<IO::AIO::SPLICE_F_MORE> and
2201C<IO::AIO::SPLICE_F_GIFT>.
2202
2203See the C<splice(2)> manpage for details.
2204
2205=item IO::AIO::tee $r_fh, $w_fh, $length, $flags
2206
2207Calls the GNU/Linux C<tee(2)> syscall, see its manpage and the
2208description for C<IO::AIO::splice> above for details.
2209
2210=item $actual_size = IO::AIO::pipesize $r_fh[, $new_size]
2211
2212Attempts to query or change the pipe buffer size. Obviously works only
2213on pipes, and currently works only on GNU/Linux systems, and fails with
2214C<-1>/C<ENOSYS> everywhere else. If anybody knows how to influence pipe buffer
2215size on other systems, drop me a note.
2216
2217=item ($rfh, $wfh) = IO::AIO::pipe2 [$flags]
2218
2219This is a direct interface to the Linux L<pipe2(2)> system call. If
2220C<$flags> is missing or C<0>, then this should be the same as a call to
2221perl's built-in C<pipe> function and create a new pipe, and works on
2222systems that lack the pipe2 syscall. On win32, this case invokes C<_pipe
2223(..., 4096, O_BINARY)>.
2224
2225If C<$flags> is non-zero, it tries to invoke the pipe2 system call with
2226the given flags (Linux 2.6.27, glibc 2.9).
2227
2228On success, the read and write file handles are returned.
2229
2230On error, nothing will be returned. If the pipe2 syscall is missing and
2231C<$flags> is non-zero, fails with C<ENOSYS>.
2232
2233Please refer to L<pipe2(2)> for more info on the C<$flags>, but at the
2234time of this writing, C<IO::AIO::O_CLOEXEC>, C<IO::AIO::O_NONBLOCK> and
2235C<IO::AIO::O_DIRECT> (Linux 3.4, for packet-based pipes) were supported.
2236
2237=back
2238
2239=cut
2240
2241min_parallel 8;
2242
2243END { flush }
2244
2271; 22451;
228 2246
2247=head1 EVENT LOOP INTEGRATION
2248
2249It is recommended to use L<AnyEvent::AIO> to integrate IO::AIO
2250automatically into many event loops:
2251
2252 # AnyEvent integration (EV, Event, Glib, Tk, POE, urxvt, pureperl...)
2253 use AnyEvent::AIO;
2254
2255You can also integrate IO::AIO manually into many event loops, here are
2256some examples of how to do this:
2257
2258 # EV integration
2259 my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
2260
2261 # Event integration
2262 Event->io (fd => IO::AIO::poll_fileno,
2263 poll => 'r',
2264 cb => \&IO::AIO::poll_cb);
2265
2266 # Glib/Gtk2 integration
2267 add_watch Glib::IO IO::AIO::poll_fileno,
2268 in => sub { IO::AIO::poll_cb; 1 };
2269
2270 # Tk integration
2271 Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",
2272 readable => \&IO::AIO::poll_cb);
2273
2274 # Danga::Socket integration
2275 Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
2276 \&IO::AIO::poll_cb);
2277
2278=head2 FORK BEHAVIOUR
2279
2280Usage of pthreads in a program changes the semantics of fork
2281considerably. Specifically, only async-safe functions can be called after
2282fork. Perl doesn't know about this, so in general, you cannot call fork
2283with defined behaviour in perl if pthreads are involved. IO::AIO uses
2284pthreads, so this applies, but many other extensions and (for inexplicable
2285reasons) perl itself often is linked against pthreads, so this limitation
2286applies to quite a lot of perls.
2287
2288This module no longer tries to fight your OS, or POSIX. That means IO::AIO
2289only works in the process that loaded it. Forking is fully supported, but
2290using IO::AIO in the child is not.
2291
2292You might get around by not I<using> IO::AIO before (or after)
2293forking. You could also try to call the L<IO::AIO::reinit> function in the
2294child:
2295
2296=over 4
2297
2298=item IO::AIO::reinit
2299
2300Abandons all current requests and I/O threads and simply reinitialises all
2301data structures. This is not an operation supported by any standards, but
2302happens to work on GNU/Linux and some newer BSD systems.
2303
2304The only reasonable use for this function is to call it after forking, if
2305C<IO::AIO> was used in the parent. Calling it while IO::AIO is active in
2306the process will result in undefined behaviour. Calling it at any time
2307will also result in any undefined (by POSIX) behaviour.
2308
229=back 2309=back
230 2310
2311=head2 MEMORY USAGE
2312
2313Per-request usage:
2314
2315Each aio request uses - depending on your architecture - around 100-200
2316bytes of memory. In addition, stat requests need a stat buffer (possibly
2317a few hundred bytes), readdir requires a result buffer and so on. Perl
2318scalars and other data passed into aio requests will also be locked and
2319will consume memory till the request has entered the done state.
2320
2321This is not awfully much, so queuing lots of requests is not usually a
2322problem.
2323
2324Per-thread usage:
2325
2326In the execution phase, some aio requests require more memory for
2327temporary buffers, and each thread requires a stack and other data
2328structures (usually around 16k-128k, depending on the OS).
2329
231=head1 BUGS 2330=head1 KNOWN BUGS
232 2331
233 - aio_open gives a fd, but all other functions expect a perl filehandle. 2332Known bugs will be fixed in the next release.
234 2333
235=head1 SEE ALSO 2334=head1 SEE ALSO
236 2335
237L<Coro>, L<Linux::AIO>. 2336L<AnyEvent::AIO> for easy integration into event loops, L<Coro::AIO> for a
2337more natural syntax.
238 2338
239=head1 AUTHOR 2339=head1 AUTHOR
240 2340
241 Marc Lehmann <schmorp@schmorp.de> 2341 Marc Lehmann <schmorp@schmorp.de>
242 http://home.schmorp.de/ 2342 http://home.schmorp.de/

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines