1 |
=head1 NAME |
2 |
|
3 |
IO::AIO - Asynchronous Input/Output |
4 |
|
5 |
=head1 SYNOPSIS |
6 |
|
7 |
use IO::AIO; |
8 |
|
9 |
aio_open "/etc/passwd", O_RDONLY, 0, sub { |
10 |
my ($fh) = @_; |
11 |
... |
12 |
}; |
13 |
|
14 |
aio_unlink "/tmp/file", sub { }; |
15 |
|
16 |
aio_read $fh, 30000, 1024, $buffer, 0, sub { |
17 |
$_[0] > 0 or die "read error: $!"; |
18 |
}; |
19 |
|
20 |
# version 2+ has request and group objects |
21 |
use IO::AIO 2; |
22 |
|
23 |
aioreq_pri 4; # give next request a very high priority |
24 |
my $req = aio_unlink "/tmp/file", sub { }; |
25 |
$req->cancel; # cancel request if still in queue |
26 |
|
27 |
my $grp = aio_group sub { print "all stats done\n" }; |
28 |
add $grp aio_stat "..." for ...; |
29 |
|
30 |
# AnyEvent integration |
31 |
open my $fh, "<&=" . IO::AIO::poll_fileno or die "$!"; |
32 |
my $w = AnyEvent->io (fh => $fh, poll => 'r', cb => sub { IO::AIO::poll_cb }); |
33 |
|
34 |
# Event integration |
35 |
Event->io (fd => IO::AIO::poll_fileno, |
36 |
poll => 'r', |
37 |
cb => \&IO::AIO::poll_cb); |
38 |
|
39 |
# Glib/Gtk2 integration |
40 |
add_watch Glib::IO IO::AIO::poll_fileno, |
41 |
in => sub { IO::AIO::poll_cb; 1 }; |
42 |
|
43 |
# Tk integration |
44 |
Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "", |
45 |
readable => \&IO::AIO::poll_cb); |
46 |
|
47 |
# Danga::Socket integration |
48 |
Danga::Socket->AddOtherFds (IO::AIO::poll_fileno => |
49 |
\&IO::AIO::poll_cb); |
50 |
|
51 |
=head1 DESCRIPTION |
52 |
|
53 |
This module implements asynchronous I/O using whatever means your |
54 |
operating system supports. |
55 |
|
56 |
In this version, a number of threads are started that execute your |
57 |
requests and signal their completion. You don't need thread support |
58 |
in perl, and the threads created by this module will not be visible |
59 |
to perl. In the future, this module might make use of the native aio |
60 |
functions available on many operating systems. However, they are often |
61 |
not well-supported or restricted (Linux doesn't allow them on normal |
62 |
files currently, for example), and they would only support aio_read and |
63 |
aio_write, so the remaining functionality would have to be implemented |
64 |
using threads anyway. |
65 |
|
66 |
Although the module will work with in the presence of other (Perl-) |
67 |
threads, it is currently not reentrant in any way, so use appropriate |
68 |
locking yourself, always call C<poll_cb> from within the same thread, or |
69 |
never call C<poll_cb> (or other C<aio_> functions) recursively. |
70 |
|
71 |
=head1 REQUEST ANATOMY AND LIFETIME |
72 |
|
73 |
Every C<aio_*> function creates a request. which is a C data structure not |
74 |
directly visible to Perl. |
75 |
|
76 |
If called in non-void context, every request function returns a Perl |
77 |
object representing the request. In void context, nothing is returned, |
78 |
which saves a bit of memory. |
79 |
|
80 |
The perl object is a fairly standard ref-to-hash object. The hash contents |
81 |
are not used by IO::AIO so you are free to store anything you like in it. |
82 |
|
83 |
During their existance, aio requests travel through the following states, |
84 |
in order: |
85 |
|
86 |
=over 4 |
87 |
|
88 |
=item ready |
89 |
|
90 |
Immediately after a request is created it is put into the ready state, |
91 |
waiting for a thread to execute it. |
92 |
|
93 |
=item execute |
94 |
|
95 |
A thread has accepted the request for processing and is currently |
96 |
executing it (e.g. blocking in read). |
97 |
|
98 |
=item pending |
99 |
|
100 |
The request has been executed and is waiting for result processing. |
101 |
|
102 |
While request submission and execution is fully asynchronous, result |
103 |
processing is not and relies on the perl interpreter calling C<poll_cb> |
104 |
(or another function with the same effect). |
105 |
|
106 |
=item result |
107 |
|
108 |
The request results are processed synchronously by C<poll_cb>. |
109 |
|
110 |
The C<poll_cb> function will process all outstanding aio requests by |
111 |
calling their callbacks, freeing memory associated with them and managing |
112 |
any groups they are contained in. |
113 |
|
114 |
=item done |
115 |
|
116 |
Request has reached the end of its lifetime and holds no resources anymore |
117 |
(except possibly for the Perl object, but its connection to the actual |
118 |
aio request is severed and calling its methods will either do nothing or |
119 |
result in a runtime error). |
120 |
|
121 |
=cut |
122 |
|
123 |
package IO::AIO; |
124 |
|
125 |
no warnings; |
126 |
use strict 'vars'; |
127 |
|
128 |
use base 'Exporter'; |
129 |
|
130 |
BEGIN { |
131 |
our $VERSION = '2.0'; |
132 |
|
133 |
our @AIO_REQ = qw(aio_sendfile aio_read aio_write aio_open aio_close aio_stat |
134 |
aio_lstat aio_unlink aio_rmdir aio_readdir aio_scandir aio_symlink |
135 |
aio_fsync aio_fdatasync aio_readahead aio_rename aio_link aio_move |
136 |
aio_group aio_nop); |
137 |
our @EXPORT = (@AIO_REQ, qw(aioreq_pri aioreq_nice)); |
138 |
our @EXPORT_OK = qw(poll_fileno poll_cb poll_wait flush |
139 |
min_parallel max_parallel nreqs nready npending); |
140 |
|
141 |
@IO::AIO::GRP::ISA = 'IO::AIO::REQ'; |
142 |
|
143 |
require XSLoader; |
144 |
XSLoader::load ("IO::AIO", $VERSION); |
145 |
} |
146 |
|
147 |
=head1 FUNCTIONS |
148 |
|
149 |
=head2 AIO FUNCTIONS |
150 |
|
151 |
All the C<aio_*> calls are more or less thin wrappers around the syscall |
152 |
with the same name (sans C<aio_>). The arguments are similar or identical, |
153 |
and they all accept an additional (and optional) C<$callback> argument |
154 |
which must be a code reference. This code reference will get called with |
155 |
the syscall return code (e.g. most syscalls return C<-1> on error, unlike |
156 |
perl, which usually delivers "false") as it's sole argument when the given |
157 |
syscall has been executed asynchronously. |
158 |
|
159 |
All functions expecting a filehandle keep a copy of the filehandle |
160 |
internally until the request has finished. |
161 |
|
162 |
All requests return objects of type L<IO::AIO::REQ> that allow further |
163 |
manipulation of those requests while they are in-flight. |
164 |
|
165 |
The pathnames you pass to these routines I<must> be absolute and |
166 |
encoded in byte form. The reason for the former is that at the time the |
167 |
request is being executed, the current working directory could have |
168 |
changed. Alternatively, you can make sure that you never change the |
169 |
current working directory. |
170 |
|
171 |
To encode pathnames to byte form, either make sure you either: a) |
172 |
always pass in filenames you got from outside (command line, readdir |
173 |
etc.), b) are ASCII or ISO 8859-1, c) use the Encode module and encode |
174 |
your pathnames to the locale (or other) encoding in effect in the user |
175 |
environment, d) use Glib::filename_from_unicode on unicode filenames or e) |
176 |
use something else. |
177 |
|
178 |
=over 4 |
179 |
|
180 |
=item $prev_pri = aioreq_pri [$pri] |
181 |
|
182 |
Returns the priority value that would be used for the next request and, if |
183 |
C<$pri> is given, sets the priority for the next aio request. |
184 |
|
185 |
The default priority is C<0>, the minimum and maximum priorities are C<-4> |
186 |
and C<4>, respectively. Requests with higher priority will be serviced |
187 |
first. |
188 |
|
189 |
The priority will be reset to C<0> after each call to one of the C<aio_*> |
190 |
functions. |
191 |
|
192 |
Example: open a file with low priority, then read something from it with |
193 |
higher priority so the read request is serviced before other low priority |
194 |
open requests (potentially spamming the cache): |
195 |
|
196 |
aioreq_pri -3; |
197 |
aio_open ..., sub { |
198 |
return unless $_[0]; |
199 |
|
200 |
aioreq_pri -2; |
201 |
aio_read $_[0], ..., sub { |
202 |
... |
203 |
}; |
204 |
}; |
205 |
|
206 |
=item aioreq_nice $pri_adjust |
207 |
|
208 |
Similar to C<aioreq_pri>, but subtracts the given value from the current |
209 |
priority, so effects are cumulative. |
210 |
|
211 |
=item aio_open $pathname, $flags, $mode, $callback->($fh) |
212 |
|
213 |
Asynchronously open or create a file and call the callback with a newly |
214 |
created filehandle for the file. |
215 |
|
216 |
The pathname passed to C<aio_open> must be absolute. See API NOTES, above, |
217 |
for an explanation. |
218 |
|
219 |
The C<$flags> argument is a bitmask. See the C<Fcntl> module for a |
220 |
list. They are the same as used by C<sysopen>. |
221 |
|
222 |
Likewise, C<$mode> specifies the mode of the newly created file, if it |
223 |
didn't exist and C<O_CREAT> has been given, just like perl's C<sysopen>, |
224 |
except that it is mandatory (i.e. use C<0> if you don't create new files, |
225 |
and C<0666> or C<0777> if you do). |
226 |
|
227 |
Example: |
228 |
|
229 |
aio_open "/etc/passwd", O_RDONLY, 0, sub { |
230 |
if ($_[0]) { |
231 |
print "open successful, fh is $_[0]\n"; |
232 |
... |
233 |
} else { |
234 |
die "open failed: $!\n"; |
235 |
} |
236 |
}; |
237 |
|
238 |
=item aio_close $fh, $callback->($status) |
239 |
|
240 |
Asynchronously close a file and call the callback with the result |
241 |
code. I<WARNING:> although accepted, you should not pass in a perl |
242 |
filehandle here, as perl will likely close the file descriptor another |
243 |
time when the filehandle is destroyed. Normally, you can safely call perls |
244 |
C<close> or just let filehandles go out of scope. |
245 |
|
246 |
This is supposed to be a bug in the API, so that might change. It's |
247 |
therefore best to avoid this function. |
248 |
|
249 |
=item aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval) |
250 |
|
251 |
=item aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval) |
252 |
|
253 |
Reads or writes C<length> bytes from the specified C<fh> and C<offset> |
254 |
into the scalar given by C<data> and offset C<dataoffset> and calls the |
255 |
callback without the actual number of bytes read (or -1 on error, just |
256 |
like the syscall). |
257 |
|
258 |
The C<$data> scalar I<MUST NOT> be modified in any way while the request |
259 |
is outstanding. Modifying it can result in segfaults or WW3 (if the |
260 |
necessary/optional hardware is installed). |
261 |
|
262 |
Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at |
263 |
offset C<0> within the scalar: |
264 |
|
265 |
aio_read $fh, 7, 15, $buffer, 0, sub { |
266 |
$_[0] > 0 or die "read error: $!"; |
267 |
print "read $_[0] bytes: <$buffer>\n"; |
268 |
}; |
269 |
|
270 |
=item aio_move $srcpath, $dstpath, $callback->($status) |
271 |
|
272 |
Try to move the I<file> (directories not supported as either source or |
273 |
destination) from C<$srcpath> to C<$dstpath> and call the callback with |
274 |
the C<0> (error) or C<-1> ok. |
275 |
|
276 |
This is a composite request that tries to rename(2) the file first. If |
277 |
rename files with C<EXDEV>, it creates the destination file with mode 0200 |
278 |
and copies the contents of the source file into it using C<aio_sendfile>, |
279 |
followed by restoring atime, mtime, access mode and uid/gid, in that |
280 |
order, and unlinking the C<$srcpath>. |
281 |
|
282 |
If an error occurs, the partial destination file will be unlinked, if |
283 |
possible, except when setting atime, mtime, access mode and uid/gid, where |
284 |
errors are being ignored. |
285 |
|
286 |
=cut |
287 |
|
288 |
sub aio_move($$$) { |
289 |
my ($src, $dst, $cb) = @_; |
290 |
|
291 |
my $pri = aioreq_pri; |
292 |
my $grp = aio_group $cb; |
293 |
|
294 |
aioreq_pri $pri; |
295 |
add $grp aio_rename $src, $dst, sub { |
296 |
if ($_[0] && $! == EXDEV) { |
297 |
aioreq_pri $pri; |
298 |
add $grp aio_open $src, O_RDONLY, 0, sub { |
299 |
if (my $src_fh = $_[0]) { |
300 |
my @stat = stat $src_fh; |
301 |
|
302 |
aioreq_pri $pri; |
303 |
add $grp aio_open $dst, O_WRONLY, 0200, sub { |
304 |
if (my $dst_fh = $_[0]) { |
305 |
aioreq_pri $pri; |
306 |
add $grp aio_sendfile $dst_fh, $src_fh, 0, $stat[7], sub { |
307 |
close $src_fh; |
308 |
|
309 |
if ($_[0] == $stat[7]) { |
310 |
utime $stat[8], $stat[9], $dst; |
311 |
chmod $stat[2] & 07777, $dst_fh; |
312 |
chown $stat[4], $stat[5], $dst_fh; |
313 |
close $dst_fh; |
314 |
|
315 |
aioreq_pri $pri; |
316 |
add $grp aio_unlink $src, sub { |
317 |
$grp->result ($_[0]); |
318 |
}; |
319 |
} else { |
320 |
my $errno = $!; |
321 |
aioreq_pri $pri; |
322 |
add $grp aio_unlink $dst, sub { |
323 |
$! = $errno; |
324 |
$grp->result (-1); |
325 |
}; |
326 |
} |
327 |
}; |
328 |
} else { |
329 |
$grp->result (-1); |
330 |
} |
331 |
}, |
332 |
|
333 |
} else { |
334 |
$grp->result (-1); |
335 |
} |
336 |
}; |
337 |
} else { |
338 |
$grp->result ($_[0]); |
339 |
} |
340 |
}; |
341 |
|
342 |
$grp |
343 |
} |
344 |
|
345 |
=item aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval) |
346 |
|
347 |
Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts |
348 |
reading at byte offset C<$in_offset>, and starts writing at the current |
349 |
file offset of C<$out_fh>. Because of that, it is not safe to issue more |
350 |
than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each |
351 |
other. |
352 |
|
353 |
This call tries to make use of a native C<sendfile> syscall to provide |
354 |
zero-copy operation. For this to work, C<$out_fh> should refer to a |
355 |
socket, and C<$in_fh> should refer to mmap'able file. |
356 |
|
357 |
If the native sendfile call fails or is not implemented, it will be |
358 |
emulated, so you can call C<aio_sendfile> on any type of filehandle |
359 |
regardless of the limitations of the operating system. |
360 |
|
361 |
Please note, however, that C<aio_sendfile> can read more bytes from |
362 |
C<$in_fh> than are written, and there is no way to find out how many |
363 |
bytes have been read from C<aio_sendfile> alone, as C<aio_sendfile> only |
364 |
provides the number of bytes written to C<$out_fh>. Only if the result |
365 |
value equals C<$length> one can assume that C<$length> bytes have been |
366 |
read. |
367 |
|
368 |
=item aio_readahead $fh,$offset,$length, $callback->($retval) |
369 |
|
370 |
C<aio_readahead> populates the page cache with data from a file so that |
371 |
subsequent reads from that file will not block on disk I/O. The C<$offset> |
372 |
argument specifies the starting point from which data is to be read and |
373 |
C<$length> specifies the number of bytes to be read. I/O is performed in |
374 |
whole pages, so that offset is effectively rounded down to a page boundary |
375 |
and bytes are read up to the next page boundary greater than or equal to |
376 |
(off-set+length). C<aio_readahead> does not read beyond the end of the |
377 |
file. The current file offset of the file is left unchanged. |
378 |
|
379 |
If that syscall doesn't exist (likely if your OS isn't Linux) it will be |
380 |
emulated by simply reading the data, which would have a similar effect. |
381 |
|
382 |
=item aio_stat $fh_or_path, $callback->($status) |
383 |
|
384 |
=item aio_lstat $fh, $callback->($status) |
385 |
|
386 |
Works like perl's C<stat> or C<lstat> in void context. The callback will |
387 |
be called after the stat and the results will be available using C<stat _> |
388 |
or C<-s _> etc... |
389 |
|
390 |
The pathname passed to C<aio_stat> must be absolute. See API NOTES, above, |
391 |
for an explanation. |
392 |
|
393 |
Currently, the stats are always 64-bit-stats, i.e. instead of returning an |
394 |
error when stat'ing a large file, the results will be silently truncated |
395 |
unless perl itself is compiled with large file support. |
396 |
|
397 |
Example: Print the length of F</etc/passwd>: |
398 |
|
399 |
aio_stat "/etc/passwd", sub { |
400 |
$_[0] and die "stat failed: $!"; |
401 |
print "size is ", -s _, "\n"; |
402 |
}; |
403 |
|
404 |
=item aio_unlink $pathname, $callback->($status) |
405 |
|
406 |
Asynchronously unlink (delete) a file and call the callback with the |
407 |
result code. |
408 |
|
409 |
=item aio_link $srcpath, $dstpath, $callback->($status) |
410 |
|
411 |
Asynchronously create a new link to the existing object at C<$srcpath> at |
412 |
the path C<$dstpath> and call the callback with the result code. |
413 |
|
414 |
=item aio_symlink $srcpath, $dstpath, $callback->($status) |
415 |
|
416 |
Asynchronously create a new symbolic link to the existing object at C<$srcpath> at |
417 |
the path C<$dstpath> and call the callback with the result code. |
418 |
|
419 |
=item aio_rename $srcpath, $dstpath, $callback->($status) |
420 |
|
421 |
Asynchronously rename the object at C<$srcpath> to C<$dstpath>, just as |
422 |
rename(2) and call the callback with the result code. |
423 |
|
424 |
=item aio_rmdir $pathname, $callback->($status) |
425 |
|
426 |
Asynchronously rmdir (delete) a directory and call the callback with the |
427 |
result code. |
428 |
|
429 |
=item aio_readdir $pathname, $callback->($entries) |
430 |
|
431 |
Unlike the POSIX call of the same name, C<aio_readdir> reads an entire |
432 |
directory (i.e. opendir + readdir + closedir). The entries will not be |
433 |
sorted, and will B<NOT> include the C<.> and C<..> entries. |
434 |
|
435 |
The callback a single argument which is either C<undef> or an array-ref |
436 |
with the filenames. |
437 |
|
438 |
=item aio_scandir $path, $maxreq, $callback->($dirs, $nondirs) |
439 |
|
440 |
Scans a directory (similar to C<aio_readdir>) but additionally tries to |
441 |
efficiently separate the entries of directory C<$path> into two sets of |
442 |
names, directories you can recurse into (directories), and ones you cannot |
443 |
recurse into (everything else, including symlinks to directories). |
444 |
|
445 |
C<aio_scandir> is a composite request that creates of many sub requests_ |
446 |
C<$maxreq> specifies the maximum number of outstanding aio requests that |
447 |
this function generates. If it is C<< <= 0 >>, then a suitable default |
448 |
will be chosen (currently 6). |
449 |
|
450 |
On error, the callback is called without arguments, otherwise it receives |
451 |
two array-refs with path-relative entry names. |
452 |
|
453 |
Example: |
454 |
|
455 |
aio_scandir $dir, 0, sub { |
456 |
my ($dirs, $nondirs) = @_; |
457 |
print "real directories: @$dirs\n"; |
458 |
print "everything else: @$nondirs\n"; |
459 |
}; |
460 |
|
461 |
Implementation notes. |
462 |
|
463 |
The C<aio_readdir> cannot be avoided, but C<stat()>'ing every entry can. |
464 |
|
465 |
After reading the directory, the modification time, size etc. of the |
466 |
directory before and after the readdir is checked, and if they match (and |
467 |
isn't the current time), the link count will be used to decide how many |
468 |
entries are directories (if >= 2). Otherwise, no knowledge of the number |
469 |
of subdirectories will be assumed. |
470 |
|
471 |
Then entries will be sorted into likely directories (everything without |
472 |
a non-initial dot currently) and likely non-directories (everything |
473 |
else). Then every entry plus an appended C</.> will be C<stat>'ed, |
474 |
likely directories first. If that succeeds, it assumes that the entry |
475 |
is a directory or a symlink to directory (which will be checked |
476 |
seperately). This is often faster than stat'ing the entry itself because |
477 |
filesystems might detect the type of the entry without reading the inode |
478 |
data (e.g. ext2fs filetype feature). |
479 |
|
480 |
If the known number of directories (link count - 2) has been reached, the |
481 |
rest of the entries is assumed to be non-directories. |
482 |
|
483 |
This only works with certainty on POSIX (= UNIX) filesystems, which |
484 |
fortunately are the vast majority of filesystems around. |
485 |
|
486 |
It will also likely work on non-POSIX filesystems with reduced efficiency |
487 |
as those tend to return 0 or 1 as link counts, which disables the |
488 |
directory counting heuristic. |
489 |
|
490 |
=cut |
491 |
|
492 |
sub aio_scandir($$$) { |
493 |
my ($path, $maxreq, $cb) = @_; |
494 |
|
495 |
my $pri = aioreq_pri; |
496 |
|
497 |
my $grp = aio_group $cb; |
498 |
|
499 |
$maxreq = 6 if $maxreq <= 0; |
500 |
|
501 |
# stat once |
502 |
aioreq_pri $pri; |
503 |
add $grp aio_stat $path, sub { |
504 |
return $grp->result () if $_[0]; |
505 |
my $now = time; |
506 |
my $hash1 = join ":", (stat _)[0,1,3,7,9]; |
507 |
|
508 |
# read the directory entries |
509 |
aioreq_pri $pri; |
510 |
add $grp aio_readdir $path, sub { |
511 |
my $entries = shift |
512 |
or return $grp->result (); |
513 |
|
514 |
# stat the dir another time |
515 |
aioreq_pri $pri; |
516 |
add $grp aio_stat $path, sub { |
517 |
my $hash2 = join ":", (stat _)[0,1,3,7,9]; |
518 |
|
519 |
my $ndirs; |
520 |
|
521 |
# take the slow route if anything looks fishy |
522 |
if ($hash1 ne $hash2 or (stat _)[9] == $now) { |
523 |
$ndirs = -1; |
524 |
} else { |
525 |
# if nlink == 2, we are finished |
526 |
# on non-posix-fs's, we rely on nlink < 2 |
527 |
$ndirs = (stat _)[3] - 2 |
528 |
or return $grp->result ([], $entries); |
529 |
} |
530 |
|
531 |
# sort into likely dirs and likely nondirs |
532 |
# dirs == files without ".", short entries first |
533 |
$entries = [map $_->[0], |
534 |
sort { $b->[1] cmp $a->[1] } |
535 |
map [$_, sprintf "%s%04d", (/.\./ ? "1" : "0"), length], |
536 |
@$entries]; |
537 |
|
538 |
my (@dirs, @nondirs); |
539 |
|
540 |
my $statgrp = add $grp aio_group sub { |
541 |
$grp->result (\@dirs, \@nondirs); |
542 |
}; |
543 |
|
544 |
limit $statgrp $maxreq; |
545 |
feed $statgrp sub { |
546 |
return unless @$entries; |
547 |
my $entry = pop @$entries; |
548 |
|
549 |
aioreq_pri $pri; |
550 |
add $statgrp aio_stat "$path/$entry/.", sub { |
551 |
if ($_[0] < 0) { |
552 |
push @nondirs, $entry; |
553 |
} else { |
554 |
# need to check for real directory |
555 |
aioreq_pri $pri; |
556 |
add $statgrp aio_lstat "$path/$entry", sub { |
557 |
if (-d _) { |
558 |
push @dirs, $entry; |
559 |
|
560 |
unless (--$ndirs) { |
561 |
push @nondirs, @$entries; |
562 |
feed $statgrp; |
563 |
} |
564 |
} else { |
565 |
push @nondirs, $entry; |
566 |
} |
567 |
} |
568 |
} |
569 |
}; |
570 |
}; |
571 |
}; |
572 |
}; |
573 |
}; |
574 |
|
575 |
$grp |
576 |
} |
577 |
|
578 |
=item aio_fsync $fh, $callback->($status) |
579 |
|
580 |
Asynchronously call fsync on the given filehandle and call the callback |
581 |
with the fsync result code. |
582 |
|
583 |
=item aio_fdatasync $fh, $callback->($status) |
584 |
|
585 |
Asynchronously call fdatasync on the given filehandle and call the |
586 |
callback with the fdatasync result code. |
587 |
|
588 |
If this call isn't available because your OS lacks it or it couldn't be |
589 |
detected, it will be emulated by calling C<fsync> instead. |
590 |
|
591 |
=item aio_group $callback->(...) |
592 |
|
593 |
This is a very special aio request: Instead of doing something, it is a |
594 |
container for other aio requests, which is useful if you want to bundle |
595 |
many requests into a single, composite, request with a definite callback |
596 |
and the ability to cancel the whole request with its subrequests. |
597 |
|
598 |
Returns an object of class L<IO::AIO::GRP>. See its documentation below |
599 |
for more info. |
600 |
|
601 |
Example: |
602 |
|
603 |
my $grp = aio_group sub { |
604 |
print "all stats done\n"; |
605 |
}; |
606 |
|
607 |
add $grp |
608 |
(aio_stat ...), |
609 |
(aio_stat ...), |
610 |
...; |
611 |
|
612 |
=item aio_nop $callback->() |
613 |
|
614 |
This is a special request - it does nothing in itself and is only used for |
615 |
side effects, such as when you want to add a dummy request to a group so |
616 |
that finishing the requests in the group depends on executing the given |
617 |
code. |
618 |
|
619 |
While this request does nothing, it still goes through the execution |
620 |
phase and still requires a worker thread. Thus, the callback will not |
621 |
be executed immediately but only after other requests in the queue have |
622 |
entered their execution phase. This can be used to measure request |
623 |
latency. |
624 |
|
625 |
=item IO::AIO::aio_busy $fractional_seconds, $callback->() *NOT EXPORTED* |
626 |
|
627 |
Mainly used for debugging and benchmarking, this aio request puts one of |
628 |
the request workers to sleep for the given time. |
629 |
|
630 |
While it is theoretically handy to have simple I/O scheduling requests |
631 |
like sleep and file handle readable/writable, the overhead this creates is |
632 |
immense (it blocks a thread for a long time) so do not use this function |
633 |
except to put your application under artificial I/O pressure. |
634 |
|
635 |
=back |
636 |
|
637 |
=head2 IO::AIO::REQ CLASS |
638 |
|
639 |
All non-aggregate C<aio_*> functions return an object of this class when |
640 |
called in non-void context. |
641 |
|
642 |
=over 4 |
643 |
|
644 |
=item cancel $req |
645 |
|
646 |
Cancels the request, if possible. Has the effect of skipping execution |
647 |
when entering the B<execute> state and skipping calling the callback when |
648 |
entering the the B<result> state, but will leave the request otherwise |
649 |
untouched. That means that requests that currently execute will not be |
650 |
stopped and resources held by the request will not be freed prematurely. |
651 |
|
652 |
=item cb $req $callback->(...) |
653 |
|
654 |
Replace (or simply set) the callback registered to the request. |
655 |
|
656 |
=back |
657 |
|
658 |
=head2 IO::AIO::GRP CLASS |
659 |
|
660 |
This class is a subclass of L<IO::AIO::REQ>, so all its methods apply to |
661 |
objects of this class, too. |
662 |
|
663 |
A IO::AIO::GRP object is a special request that can contain multiple other |
664 |
aio requests. |
665 |
|
666 |
You create one by calling the C<aio_group> constructing function with a |
667 |
callback that will be called when all contained requests have entered the |
668 |
C<done> state: |
669 |
|
670 |
my $grp = aio_group sub { |
671 |
print "all requests are done\n"; |
672 |
}; |
673 |
|
674 |
You add requests by calling the C<add> method with one or more |
675 |
C<IO::AIO::REQ> objects: |
676 |
|
677 |
$grp->add (aio_unlink "..."); |
678 |
|
679 |
add $grp aio_stat "...", sub { |
680 |
$_[0] or return $grp->result ("error"); |
681 |
|
682 |
# add another request dynamically, if first succeeded |
683 |
add $grp aio_open "...", sub { |
684 |
$grp->result ("ok"); |
685 |
}; |
686 |
}; |
687 |
|
688 |
This makes it very easy to create composite requests (see the source of |
689 |
C<aio_move> for an application) that work and feel like simple requests. |
690 |
|
691 |
=over 4 |
692 |
|
693 |
=item * The IO::AIO::GRP objects will be cleaned up during calls to |
694 |
C<IO::AIO::poll_cb>, just like any other request. |
695 |
|
696 |
=item * They can be canceled like any other request. Canceling will cancel not |
697 |
only the request itself, but also all requests it contains. |
698 |
|
699 |
=item * They can also can also be added to other IO::AIO::GRP objects. |
700 |
|
701 |
=item * You must not add requests to a group from within the group callback (or |
702 |
any later time). |
703 |
|
704 |
=back |
705 |
|
706 |
Their lifetime, simplified, looks like this: when they are empty, they |
707 |
will finish very quickly. If they contain only requests that are in the |
708 |
C<done> state, they will also finish. Otherwise they will continue to |
709 |
exist. |
710 |
|
711 |
That means after creating a group you have some time to add requests. And |
712 |
in the callbacks of those requests, you can add further requests to the |
713 |
group. And only when all those requests have finished will the the group |
714 |
itself finish. |
715 |
|
716 |
=over 4 |
717 |
|
718 |
=item add $grp ... |
719 |
|
720 |
=item $grp->add (...) |
721 |
|
722 |
Add one or more requests to the group. Any type of L<IO::AIO::REQ> can |
723 |
be added, including other groups, as long as you do not create circular |
724 |
dependencies. |
725 |
|
726 |
Returns all its arguments. |
727 |
|
728 |
=item $grp->cancel_subs |
729 |
|
730 |
Cancel all subrequests and clears any feeder, but not the group request |
731 |
itself. Useful when you queued a lot of events but got a result early. |
732 |
|
733 |
=item $grp->result (...) |
734 |
|
735 |
Set the result value(s) that will be passed to the group callback when all |
736 |
subrequests have finished and set thre groups errno to the current value |
737 |
of errno (just like calling C<errno> without an error number). By default, |
738 |
no argument will be passed and errno is zero. |
739 |
|
740 |
=item $grp->errno ([$errno]) |
741 |
|
742 |
Sets the group errno value to C<$errno>, or the current value of errno |
743 |
when the argument is missing. |
744 |
|
745 |
Every aio request has an associated errno value that is restored when |
746 |
the callback is invoked. This method lets you change this value from its |
747 |
default (0). |
748 |
|
749 |
Calling C<result> will also set errno, so make sure you either set C<$!> |
750 |
before the call to C<result>, or call c<errno> after it. |
751 |
|
752 |
=item feed $grp $callback->($grp) |
753 |
|
754 |
Sets a feeder/generator on this group: every group can have an attached |
755 |
generator that generates requests if idle. The idea behind this is that, |
756 |
although you could just queue as many requests as you want in a group, |
757 |
this might starve other requests for a potentially long time. For |
758 |
example, C<aio_scandir> might generate hundreds of thousands C<aio_stat> |
759 |
requests, delaying any later requests for a long time. |
760 |
|
761 |
To avoid this, and allow incremental generation of requests, you can |
762 |
instead a group and set a feeder on it that generates those requests. The |
763 |
feed callback will be called whenever there are few enough (see C<limit>, |
764 |
below) requests active in the group itself and is expected to queue more |
765 |
requests. |
766 |
|
767 |
The feed callback can queue as many requests as it likes (i.e. C<add> does |
768 |
not impose any limits). |
769 |
|
770 |
If the feed does not queue more requests when called, it will be |
771 |
automatically removed from the group. |
772 |
|
773 |
If the feed limit is C<0>, it will be set to C<2> automatically. |
774 |
|
775 |
Example: |
776 |
|
777 |
# stat all files in @files, but only ever use four aio requests concurrently: |
778 |
|
779 |
my $grp = aio_group sub { print "finished\n" }; |
780 |
limit $grp 4; |
781 |
feed $grp sub { |
782 |
my $file = pop @files |
783 |
or return; |
784 |
|
785 |
add $grp aio_stat $file, sub { ... }; |
786 |
}; |
787 |
|
788 |
=item limit $grp $num |
789 |
|
790 |
Sets the feeder limit for the group: The feeder will be called whenever |
791 |
the group contains less than this many requests. |
792 |
|
793 |
Setting the limit to C<0> will pause the feeding process. |
794 |
|
795 |
=back |
796 |
|
797 |
=head2 SUPPORT FUNCTIONS |
798 |
|
799 |
=over 4 |
800 |
|
801 |
=item $fileno = IO::AIO::poll_fileno |
802 |
|
803 |
Return the I<request result pipe file descriptor>. This filehandle must be |
804 |
polled for reading by some mechanism outside this module (e.g. Event or |
805 |
select, see below or the SYNOPSIS). If the pipe becomes readable you have |
806 |
to call C<poll_cb> to check the results. |
807 |
|
808 |
See C<poll_cb> for an example. |
809 |
|
810 |
=item IO::AIO::poll_cb |
811 |
|
812 |
Process all outstanding events on the result pipe. You have to call this |
813 |
regularly. Returns the number of events processed. Returns immediately |
814 |
when no events are outstanding. |
815 |
|
816 |
If not all requests were processed for whatever reason, the filehandle |
817 |
will still be ready when C<poll_cb> returns. |
818 |
|
819 |
Example: Install an Event watcher that automatically calls |
820 |
IO::AIO::poll_cb with high priority: |
821 |
|
822 |
Event->io (fd => IO::AIO::poll_fileno, |
823 |
poll => 'r', async => 1, |
824 |
cb => \&IO::AIO::poll_cb); |
825 |
|
826 |
=item IO::AIO::poll_some $max_requests |
827 |
|
828 |
Similar to C<poll_cb>, but only processes up to C<$max_requests> requests |
829 |
at a time. |
830 |
|
831 |
Useful if you want to ensure some level of interactiveness when perl is |
832 |
not fast enough to process all requests in time. |
833 |
|
834 |
Example: Install an Event watcher that automatically calls |
835 |
IO::AIO::poll_some with low priority, to ensure that other parts of the |
836 |
program get the CPU sometimes even under high AIO load. |
837 |
|
838 |
Event->io (fd => IO::AIO::poll_fileno, |
839 |
poll => 'r', nice => 1, |
840 |
cb => sub { IO::AIO::poll_some 256 }); |
841 |
|
842 |
=item IO::AIO::poll_wait |
843 |
|
844 |
Wait till the result filehandle becomes ready for reading (simply does a |
845 |
C<select> on the filehandle. This is useful if you want to synchronously wait |
846 |
for some requests to finish). |
847 |
|
848 |
See C<nreqs> for an example. |
849 |
|
850 |
=item IO::AIO::nreqs |
851 |
|
852 |
Returns the number of requests currently in the ready, execute or pending |
853 |
states (i.e. for which their callback has not been invoked yet). |
854 |
|
855 |
Example: wait till there are no outstanding requests anymore: |
856 |
|
857 |
IO::AIO::poll_wait, IO::AIO::poll_cb |
858 |
while IO::AIO::nreqs; |
859 |
|
860 |
=item IO::AIO::nready |
861 |
|
862 |
Returns the number of requests currently in the ready state (not yet |
863 |
executed). |
864 |
|
865 |
=item IO::AIO::npending |
866 |
|
867 |
Returns the number of requests currently in the pending state (executed, |
868 |
but not yet processed by poll_cb). |
869 |
|
870 |
=item IO::AIO::flush |
871 |
|
872 |
Wait till all outstanding AIO requests have been handled. |
873 |
|
874 |
Strictly equivalent to: |
875 |
|
876 |
IO::AIO::poll_wait, IO::AIO::poll_cb |
877 |
while IO::AIO::nreqs; |
878 |
|
879 |
=item IO::AIO::poll |
880 |
|
881 |
Waits until some requests have been handled. |
882 |
|
883 |
Strictly equivalent to: |
884 |
|
885 |
IO::AIO::poll_wait, IO::AIO::poll_cb |
886 |
if IO::AIO::nreqs; |
887 |
|
888 |
=item IO::AIO::min_parallel $nthreads |
889 |
|
890 |
Set the minimum number of AIO threads to C<$nthreads>. The current |
891 |
default is C<8>, which means eight asynchronous operations can execute |
892 |
concurrently at any one time (the number of outstanding requests, |
893 |
however, is unlimited). |
894 |
|
895 |
IO::AIO starts threads only on demand, when an AIO request is queued and |
896 |
no free thread exists. |
897 |
|
898 |
It is recommended to keep the number of threads relatively low, as some |
899 |
Linux kernel versions will scale negatively with the number of threads |
900 |
(higher parallelity => MUCH higher latency). With current Linux 2.6 |
901 |
versions, 4-32 threads should be fine. |
902 |
|
903 |
Under most circumstances you don't need to call this function, as the |
904 |
module selects a default that is suitable for low to moderate load. |
905 |
|
906 |
=item IO::AIO::max_parallel $nthreads |
907 |
|
908 |
Sets the maximum number of AIO threads to C<$nthreads>. If more than the |
909 |
specified number of threads are currently running, this function kills |
910 |
them. This function blocks until the limit is reached. |
911 |
|
912 |
While C<$nthreads> are zero, aio requests get queued but not executed |
913 |
until the number of threads has been increased again. |
914 |
|
915 |
This module automatically runs C<max_parallel 0> at program end, to ensure |
916 |
that all threads are killed and that there are no outstanding requests. |
917 |
|
918 |
Under normal circumstances you don't need to call this function. |
919 |
|
920 |
=item $oldmaxreqs = IO::AIO::max_outstanding $maxreqs |
921 |
|
922 |
This is a very bad function to use in interactive programs because it |
923 |
blocks, and a bad way to reduce concurrency because it is inexact: Better |
924 |
use an C<aio_group> together with a feed callback. |
925 |
|
926 |
Sets the maximum number of outstanding requests to C<$nreqs>. If you |
927 |
to queue up more than this number of requests, the next call to the |
928 |
C<poll_cb> (and C<poll_some> and other functions calling C<poll_cb>) |
929 |
function will block until the limit is no longer exceeded. |
930 |
|
931 |
The default value is very large, so there is no practical limit on the |
932 |
number of outstanding requests. |
933 |
|
934 |
You can still queue as many requests as you want. Therefore, |
935 |
C<max_oustsanding> is mainly useful in simple scripts (with low values) or |
936 |
as a stop gap to shield against fatal memory overflow (with large values). |
937 |
|
938 |
=back |
939 |
|
940 |
=cut |
941 |
|
942 |
# support function to convert a fd into a perl filehandle |
943 |
sub _fd2fh { |
944 |
return undef if $_[0] < 0; |
945 |
|
946 |
# try to generate nice filehandles |
947 |
my $sym = "IO::AIO::fd#$_[0]"; |
948 |
local *$sym; |
949 |
|
950 |
open *$sym, "+<&=$_[0]" # usually works under any unix |
951 |
or open *$sym, "<&=$_[0]" # cygwin needs this |
952 |
or open *$sym, ">&=$_[0]" # or this |
953 |
or return undef; |
954 |
|
955 |
*$sym |
956 |
} |
957 |
|
958 |
min_parallel 8; |
959 |
|
960 |
END { |
961 |
max_parallel 0; |
962 |
} |
963 |
|
964 |
1; |
965 |
|
966 |
=head2 FORK BEHAVIOUR |
967 |
|
968 |
This module should do "the right thing" when the process using it forks: |
969 |
|
970 |
Before the fork, IO::AIO enters a quiescent state where no requests |
971 |
can be added in other threads and no results will be processed. After |
972 |
the fork the parent simply leaves the quiescent state and continues |
973 |
request/result processing, while the child frees the request/result queue |
974 |
(so that the requests started before the fork will only be handled in the |
975 |
parent). Threads will be started on demand until the limit set in the |
976 |
parent process has been reached again. |
977 |
|
978 |
In short: the parent will, after a short pause, continue as if fork had |
979 |
not been called, while the child will act as if IO::AIO has not been used |
980 |
yet. |
981 |
|
982 |
=head2 MEMORY USAGE |
983 |
|
984 |
Per-request usage: |
985 |
|
986 |
Each aio request uses - depending on your architecture - around 100-200 |
987 |
bytes of memory. In addition, stat requests need a stat buffer (possibly |
988 |
a few hundred bytes), readdir requires a result buffer and so on. Perl |
989 |
scalars and other data passed into aio requests will also be locked and |
990 |
will consume memory till the request has entered the done state. |
991 |
|
992 |
This is now awfully much, so queuing lots of requests is not usually a |
993 |
problem. |
994 |
|
995 |
Per-thread usage: |
996 |
|
997 |
In the execution phase, some aio requests require more memory for |
998 |
temporary buffers, and each thread requires a stack and other data |
999 |
structures (usually around 16k-128k, depending on the OS). |
1000 |
|
1001 |
=head1 KNOWN BUGS |
1002 |
|
1003 |
Known bugs will be fixed in the next release. |
1004 |
|
1005 |
=head1 SEE ALSO |
1006 |
|
1007 |
L<Coro::AIO>. |
1008 |
|
1009 |
=head1 AUTHOR |
1010 |
|
1011 |
Marc Lehmann <schmorp@schmorp.de> |
1012 |
http://home.schmorp.de/ |
1013 |
|
1014 |
=cut |
1015 |
|